Enzyme-induced mineralization of calcium carbonate in 3D printable granular hydrogels

TL;DR

Researchers developed a 3D printable material that mimics bone by combining organic polymers with calcium carbonate, offering biocompatible and strong scaffolds for tissue engineering.

Contribution

A new 3D printable, biocompatible composite is introduced using enzyme-induced mineralization without living bacteria.

Findings

The mineralized scaffolds achieved up to 92 wt% calcium carbonate with 56 vol% porosity.

The material exhibited compressive strength up to 4 MPa and modulus of 56 MPa, comparable to human trabecular bone.

The composites showed low cytotoxicity, making them suitable for biomedical applications.

Abstract

Many biological materials, such as bone, are organic-inorganic composites, made from a polymeric matrix that supports biomineralization under mild conditions. These materials are usually composed of a small set of abundant components like polysaccharides, proteins, and minerals, and exhibit a remarkable combination of density normalized stiffness, toughness, and functionality. Producing bio-inspired synthetic porous composites with a similar combination of properties through energy-efficient processes still presents an unmet challenge. Some aspects of this challenge can be addressed using living bacteria that induce biomineralization. However, living bacteria limit biomedical applications, especially in vivo, require careful handling, and are costly. To address these limitations, we introduce enzyme-containing granular precursors exclusively made from naturally sourced polymers. These precursors can be cast or direct ink written into cm-sized structures before they are mineralized under benign conditions to reach CaCO3 contents up to 92 wt% with a porosity of 56 vol%. The resulting mineralized scaffolds exhibit a compressive strength up to 4 MPa (specific: 5.2 MPa·cm3·g− 1) and a compressive modulus of 56 MPa (specific: 72 MPa·cm3·g− 1). Although measured in the dry state, these values fall within the range reported for human trabecular bone with similar porosities (50–90% (Morgan et al., Annu Rev Biomed Eng 20:119–43, 91; Tanoto et al., Extreme Mech Lett 73:102265, 92). The resulting biomineral-organic composites show low cytotoxicity. These findings highlight the potential of this approach to 3D print biocompatible CaCO3-based composites under mild conditions. We envisage this formulation to open up new possibilities for tissue engineering.

The online version contains supplementary material available at 10.1007/s42114-026-01662-5.

Figures

Schematic illustration of the fabrication of an enzyme-containing granular ink. i) Urease is encapsulated into a bulk hydrogel (ii) that is cryo-milled to produce fragments. (iii) The fragments are suspended in an alginate solution and jammed. iv) The enzyme-containing granular ink is 3D printed (v) before it is soaked in a solution containing urea and calcium chloride (vi) to mineralize the scaffold. Mineralization is triggered by the enzyme-induced hydrolysis of urea. vii) Photograph of a 3D printed mineralized *Nautilus* half-shell. Scale bar is 2 cm

Characterization of microgel fragments. a Compression curves of bulk gelatin (pink) and κ-carrageenan (green) before fragmentation. The extracted compression moduli are shown in the inset. b Stress relaxation of gelatin and κ-carrageenan bulk hydrogels. c Optical microscopy image of (i) gelatin- (i) and (ii) κ-carrageenan-based fragments with the corresponding size distributions. d Frequency sweeps of jammed gelatin and κ-carrageenan fragments. e Shear recovery measurements showing the self-healing ability of the ink if subjected to alternate low (0.3%, white areas) and high (80%, shaded areas

CaCO_3_ characterization. a Optical microscopy images of extruded filaments composed of gelatin- (pink) and κ-carrageenan-based (green) fragments showing the growth of the minerals over time. All the scale bars are 500 μm. b Polarized optical microscopy images of CaCO_3_ minerals grown in the gelatin- (top) and κ-carrageenan-based granular scaffold (bottom) after 4 days of incubation. The birefringence is more evident in the gelatin-based scaffold where the CaCO_3_ crystals are bigger, as shown in the close-up. All scale bars are 200 μm. c XRD spectra of CaCO_3_ grown in (i) gelatin- and (ii)

Structural characterization of CaCO_3_-organic scaffolds. a FTIR spectra of (i) CaCO_3_-gelatin scaffold as prepared (gray), after 1 day (light pink), 2 (dark pink), 3 (purple), and 4 days (dark purple) of mineralization. Pure gelatin is shown as a control (black). FTIR spectra of (ii) CaCO_3_-κ-carrageenan scaffolds as prepared (gray), after 1 day (mint green), 2 (green), 3 (forest green), and 4 days (dark green) of mineralization. Pure κ-carrageenan is shown as a control (black). The purple shade indicates the enzyme C = O peak, the gray shades indicate the calcite peaks, and the red shade i

Mechanical properties of CaCO_3_-organic composites. a Compression curves of CaCO_3_-gelatin composites as prepared (gray curve), after 1 day (light pink), 2 (dark pink), 3 (purple), and 4 days (dark purple) of mineralization. b Compression curves of CaCO_3_-κ-carrageenan scaffolds as prepared (gray curve), after 1 day (mint green), 2 (green), 3 (forest green), and 4 days (dark green) of mineralization. c Specific compressive modulus and (d) specific compressive strength of CaCO_3_-gelatin (squares in pink palette) and CaCO_3_-κ-carrageenan composites (circles in green palette) as a function o

Cell viability and morphology of CaCO_3_-gelatin and CaCO_3_-κ-carrageenan scaffolds over time. a Live/Dead staining (green: live cells (calcein), red: dead cells (propidium iodine)) and SEM micrographs of cells (pointed with white arrows) cultured onto CaCO_3_-gelatin and CaCO_3_-κ-carrageenan-based scaffolds on Day 1 and Day 7, showing differences in cell distribution, viability, and morphology. b Quantification of cell viability (%) at day 1 and day 7, demonstrating higher viability in gelatin-based hydrogels (pink) compared to κ-carrageenan-based hydrogels (green) at both time points. c Vi

Full text

Introduction

Many biological materials are made from a limited number of naturally sourced, abundant components such as polysaccharides, proteins, and minerals, through energy-efficient processes that are conducted under ambient conditions. Despite these severe limitations in material selection and processing conditions, many biological materials exhibit a fascinating combination of density normalized stiffness, toughness, and functionality [1]. A prime example is nacre, which contains 95 wt% aragonite, a calcium carbonate (CaCO_3_) polymorph that is surrounded by a proteinaceous organic matrix [2]. The brick-and-mortar architecture of nacre, composed of stiff aragonite platelets and a soft protein-based organic matrix, provides high stiffness and exceptional toughness—enhancing the latter by up to 1000-fold compared to pure aragonite crystals [2–5]. Similarly, spines of sea urchins exhibit well-defined structures spanning multiple length scales that render them damage tolerant. These spines are composed of the thermodynamically most stable calcium carbonate polymorph, calcite, amorphous calcium carbonate (ACC), and organic molecules [6, 7]. Nature frequently uses such strong and tough mineral-based materials as defense shields [8], structural support [9], or for buoyancy regulation [10].

Calcium carbonate is one of the most abundant minerals: it constitutes up to 4 wt% of Earth’s crust [11]. This mineral primarily forms through the sedimentation of skeletal constituents that originate from marine organisms, such as shellfish [12], which contain up to 95 wt% CaCO_3_ [13]. To reduce the density of minerals, which is essential for an energy-efficient locomotion of creatures, natural biominerals often contain pores. A paradigmatic example of porous biominerals is bones that incorporate pores of different sizes and shapes to minimize their density while maximizing their strength and torsion resistance [4, 14].

Inspired by the excellent density-normalized mechanical properties of such biological minerals, synthetic mineral-based composites have been introduced. Nacre-inspired examples that display a high fracture toughness include cement [15] or glass [16] platelets that have been infiltrated with synthetic polymers [15, 16]. However, in many cases, these synthetic composites are produced through energy-consuming processes that include high-temperature treatments or exposure to vacuum. This shortcoming can be addressed using foams or emulsions as templates [17] or through freeze-casting [18]. However, many of these methods still involve thermal treatments to rigidify the structures, and those that do not require thermal treatments involve exposure to vacuum or freezing steps that are also energy-consuming, and far from the biogenic mechanism of living organisms. To address these limitations and inspired by nature, the mineralization of organic scaffolds has been triggered by living organisms like bacteria. For example, ureolytic bacteria such as Sporosarcina Pasteurii [19] have been used for microbially induced calcium carbonate precipitation (MICP) [20] for geo-environmental [21, 22] and construction applications to form bulk clay [23] and CaCO_3_-based ceramics [24]. The mineral formation can be spatially confined by embedding bacteria in soft gels that can be cast into simple shapes [25]. If formulated as bacteria-loaded microgels, these materials can even be direct ink written into more complex 3D structures before being mineralized [26]. However, formulations containing living bacteria such as Sporosarcina Pasteurii (S. Pasteurii) are delicate to handle and costly, limiting biomedical applications and hampering in vivo implementations [27, 28]. To address these limitations, bacteria can be replaced by enzymes such as urease, which are more cost effective. Urease has thus far mainly been used for sand cementation [29]. Recently, urease has been used to stiffen synthetic hydrogels by precipitating FeCO_3_ [30], amorphous [31] and crystalline CaCO_3_ [32, 33]. However, the processing of these structures was limited to casting.

Here, we introduce a biocompatible granular ink that can be direct ink written into cm-sized intricate structures, which are subsequently mineralized under ambient conditions to result in composites containing up to 92 wt% CaCO_3_. These composites possess open pores whose volume accounts for up to 56% of the entire structure, thereby resulting in low densities. The high mineral content combined with the high porosity results in a specific compressive strength of dry samples of 5.2 MPa·g^− 1^·cm^3^ and a specific modulus up to 72 MPa·g^− 1^·cm^3^. These values, although measured in the dry state, are within the range of human trabecular bone that exhibits a similar porosity [34–36]. This combination of assets is achieved by formulating urease-loaded hydrogel microfragments exclusively from naturally-sourced polymers. These enzyme-containing fragments are jammed to form an ink that can be direct ink written in air and at room temperature before the scaffold is mineralized. The structure, mechanical properties, and morphology of the forming CaCO_3_ can be tuned with the composition of the microfragments. The resulting mineralized scaffolds exhibit low cytotoxicity towards osteoblasts.

Results and discussion

We use hydrogels as mineralization scaffolds. To ensure good biocompatibility of our formulation, we employ naturally sourced polymers, namely gelatin and κ-carrageenan. Both polymers exhibit a temperature-induced gelation. Gelatin contains motifs that favor cell attachment [37–39], while κ-carrageenan’s sulfate groups have a high affinity to cations, including Ca^2+^ [40]. We therefore expect the interfaces between κ-carrageenan and CaCO_3_ to be stronger than those between gelatin and CaCO_3_. Note that κ-carrageenan is significantly softer than gelatin, such that it is better suited for low-load applications including cell/drug delivery carriers [41] or scaffolds in protected sites. By contrast, the stiffer gelatin-based scaffolds have the potential to be used in load-bearing tissue engineering applications. To enable mineralization of these scaffolds, we functionalize them with urease.

Precursor solutions for gelatin and κ-carrageenan cannot be direct ink written close to room temperature if no additives [42–45] or thermal gradients in the printer [46] are present. To enable 3D printing of these hydrogels at room temperature, we formulate them as microfragments. We produce microfragments from bulk hydrogels through cryo-milling to minimize risks for residues that might hamper cell viability, as shown in Fig. 1i [47, 48]. The microfragments are polydisperse with irregular shapes, as sketched in Fig. 1ii. Fragments are firmly connected through a second network formed by soaking them in phosphate-buffered saline (PBS) containing 4 wt% alginate (Fig. 1iii). Alginate offers an additional benefit: it has a high affinity to Ca^2+^, such that it also serves as a Ca^2+^ source during mineralization [26]. Because alginate is present in both formulations, its egg-box Ca^2+^ sequestration likely biases local supersaturation and early polymorph selection in gelatin and κ-carrageenan systems to a similar extent. The enzyme-loaded microfragments, which are surrounded by alginate, are jammed to obtain an ink that can be 3D printed in air at room temperature. The 3D printed structure is incubated in a solution containing 1.75 M urea and 1 M CaCl_2_. Urease contained within the microfragments catalyzes the hydrolysis of urea into ammonia (NH_3_) and CO_2_. Because the microfragments are immersed in aqueous solutions, these products partially transform into ammonium (NH^4+^) and carbonate ions (CO_3_^2−^) [22], which, in the presence of Ca^2+^, initiate the mineralization of the hydrogel, as graphically shown in Fig. 1iv-vi. The mineralization is completed within 4 days, as demonstrated by the cm-sized model of a Nautilus half-shell in Fig. 1vii and detailed in Fig. S1a-b.

We employ gelatin as a hydrogel because it contains motifs that facilitate cell adhesion, migration, differentiation, and proliferation [37–39]. Gelatin is a denatured collagen derivative that is liquid at temperatures above 36 °C. We produce gelatin from an aqueous solution containing 25 wt% gelatin and 10 mg/g urease at 40 °C. If cooled to room temperature, gelatin structurally changes from coiled to a helix, resulting in its solidification [49].

Fig. 1. Schematic illustration of the fabrication of an enzyme-containing granular ink. i) Urease is encapsulated into a bulk hydrogel (ii) that is cryo-milled to produce fragments. (iii) The fragments are suspended in an alginate solution and jammed. iv) The enzyme-containing granular ink is 3D printed (v) before it is soaked in a solution containing urea and calcium chloride (vi) to mineralize the scaffold. Mineralization is triggered by the enzyme-induced hydrolysis of urea. vii) Photograph of a 3D printed mineralized Nautilus half-shell. Scale bar is 2 cm

The stiffness of the polymer network depends on its mesh size, which can affect the in situ growth of inorganic inclusions [50–52]. Because of the relevance of the stiffness for the mineral formation, we quantify the compressive modulus of the bulk hydrogels the fragments are made from [53]. The bulk gel comprising 25 wt% gelatin exhibits a compression modulus of 0.13 MPa, as shown by the pink curve in Fig. 2a. To assess if the gel relaxes stress, we perform stress relaxation tests by subjecting the gel to 1% strain and quantifying the time required to decrease the stress by 50%, τ_1/2_. The relaxation time τ_1/2_ of gelatin is 298 s, as shown in pink in Fig. 2b. This relaxation time is similar to that of various living tissues [54, 55].

To enable direct ink writing of hydrogels, we cryo-mill them into fragments with an average equivalent diameter of 73 µm, as shown in Fig. 2c. These fragments are swollen in PBS containing 4 wt% alginate. To allow direct ink writing of the microfragments, they are jammed through centrifugation, as detailed in the Experimental Section. The resulting ink is shear thinning and displays a fast stress recovery, as shown in Fig. 2d-e. Moreover, the strain at the yield point, where the storage modulus G’ is equal to the loss modulus G’’, is 26%, as shown in Fig. 2f. This low yield point enables printing at low pressures that are readily accessible with commercially available bioprinters, such that these microfragment-based inks are well-suited for direct ink writing [47, 56–58], in good agreement with inks composed of jammed synthetic microfragments [47, 59–65].

To assess whether the enzyme activity and consequently its ability to induce mineralization are affected by the fragmentation and encapsulation, we quantify the activity through conductivity measurements [66]. The conductance of gelatin-based fragments increases almost linearly with time, reaching a plateau after 30 min, as shown in Fig. S2a. These results suggest high enzyme activity even after fragmentation. Indeed, the enzymatic activity is preserved during cryo-milling, as shown by the time-dependent urea-hydrolysis rate curve in Fig. S2b. Note that the activity of enzymes within gelatin microfragments is lower than that of free enzymes, indicating that encapsulation and cryo-milling decreases enzyme activity, as shown in Fig. S2c.

Fig. 2. Characterization of microgel fragments. a Compression curves of bulk gelatin (pink) and κ-carrageenan (green) before fragmentation. The extracted compression moduli are shown in the inset. b Stress relaxation of gelatin and κ-carrageenan bulk hydrogels. c Optical microscopy image of (i) gelatin- (i) and (ii) κ-carrageenan-based fragments with the corresponding size distributions. d Frequency sweeps of jammed gelatin and κ-carrageenan fragments. e Shear recovery measurements showing the self-healing ability of the ink if subjected to alternate low (0.3%, white areas) and high (80%, shaded areas) strains. f Amplitude sweeps of the jammed gelatin and κ-carrageenan fragments. Data represent mean ± SD (n = 3)

To assess if the granular gelatin-based structure can be mineralized, we incubate it in a solution containing CaCl_2_ and urea and visualize the mineral formation with optical microscopy. Within 18 h, we observe some minerals within the extruded filament. The density of minerals increases with incubation time, as shown in Fig. 3a. These minerals are crystalline, as indicated by their birefringence revealed with polarized optical microscopy in Fig. 3b.

Calcium carbonate exists in different polymorphs that exhibit distinct mechanical properties [18, 67, 68]. To evaluate the structure of the formed CaCO_3_, we perform X-ray diffraction (XRD). Within 24 h of incubation, we observe diffraction peaks at 29.6°, 39.5°, 43°, and 47.6° and 48.6°, characteristic of calcite [24], as shown in Fig. 3a. This result is supported by scanning electron microscopy (SEM) analysis, which reveals cubic crystals, as shown in Fig. 3d. Fourier Transform Infrared Spectroscopy (FTIR) further confirms the calcite structure with peaks at 712 cm^− 1^, 874 cm^− 1^ and 1396 cm^− 1^, as shown in Fig. 4a. Note that the peak at 1396 cm^− 1^ is slightly shifted from the usual asymmetric stretch vibration at 1425 cm^− 1^ possibly due to hydration [69]. After 4 days of mineralization, small diffraction peaks at 24.9°, 27.2°, 32.8°, and 36° characteristic of vaterite [24], appear in the XRD spectra, as shown in Fig. 3c. These findings are in contradiction to those reported for S. Pasteurii bacteria containing gelatin-based microparticles, which formed almost the same quantity of vaterite and calcite after 1 day of mineralization, albeit after 4 days of mineralization, the majority of the mineral was also calcite [26]. A possible reason for this discrepancy during the early stages of mineral formation is that cell-free enzymes display a high activity even immediately after the microgel-based ink has been formulated. By contrast, living bacteria initially often exhibit a low activity and require a metabolic recovery period of 48 h, during which their activity steadily increases. During this transition period, vaterite can form. After the metabolic recovery of the bacteria, they produce the thermodynamically most stable polymorph, calcite [26, 66]. To assess if there are significant fractions of amorphous calcium carbonate (ACC) present in our samples, we perform FTIR analysis on them. We do not observe any clear peaks at 860 cm^–1^ and 700 cm^− 1^, which would be characteristic for ACC [67], suggesting that our samples do not entail significant amounts of ACC.

Fig. 3. CaCO_3_ characterization. a Optical microscopy images of extruded filaments composed of gelatin- (pink) and κ-carrageenan-based (green) fragments showing the growth of the minerals over time. All the scale bars are 500 μm. b Polarized optical microscopy images of CaCO_3_ minerals grown in the gelatin- (top) and κ-carrageenan-based granular scaffold (bottom) after 4 days of incubation. The birefringence is more evident in the gelatin-based scaffold where the CaCO_3_ crystals are bigger, as shown in the close-up. All scale bars are 200 μm. c XRD spectra of CaCO_3_ grown in (i) gelatin- and (ii) κ-carrageenan-based granular gels measured after 1 day (D1), 2 (D2), 3 (D3), and 4 (D4) days of mineralization. The gray shades indicate the peaks characteristic for calcite (29.6 °, 39.5°, 43°, 47.6°, and 48.6°), the blue shades those characteristic for vaterite (24.9°, 27.2°, 32.8°, and 36°). d SEM image of CaCO_3_ crystals grown in the (i) gelatin- and (ii) κ-carrageenan-based granular scaffold showing the cubic calcite, labeled with white arrows, and residues of dry polymer, highlighted by green arrows

To quantify the amount of formed minerals, we perform thermogravimetry analysis (TGA). Dried samples that had been mineralized for one day contain 79 wt% minerals, as shown in Fig. 4b. The mineral content increases with increasing incubation time until it plateaus after 4 days of mineralization. Remarkably, if mineralized for at least 4 days, dried composites contain up to 92 wt% minerals, as shown in Fig. 4b. This mineral content is similar or higher than that achieved in previously published mineralized hydrogels, indicating the potential of our approach to fabricate highly mineralized composites.

Functional groups present within the hydrogel can influence the CaCO_3_ formation and growth [67, 70]. To assess the influence of the hydrogel composition on the mineral formation, we formulate microgels from κ-carrageenan. κ-carrageenan is a polysaccharide that is derived from red algae and has shown minimal cytotoxicity [71]. It contains one sulfate per repeat unit, which can form complexes with different cations, including calcium [40]. If dispersed in hot water, κ-carrageenan attains a coil configuration that transitions into a helix upon cooling, resulting in its gelation [72, 73]. We formulate κ-carrageenan gels that have a compression modulus similar to that of gelatin, such that we expect them to exhibit a similar mesh size. To achieve this goal, we formulate the gels from a precursor solution containing 4 wt% κ-carrageenan; this gel exhibits a compression modulus of 0.1 MPa, as shown in the green curve in Fig. 2a. Yet, its relaxation time is with τ_1/2_ = 116 s shorter than that of gelatin, as shown in Fig. 2b. Cryo-milled κ-carrageenan fragments are slightly larger than the gelatin-based ones with an equivalence diameter of 92 μm, as shown in Fig. 2c. As a result of this difference, jammed κ-carrageenan fragments display a strain at the flow point of 10%, as shown in Fig. 2f.

To assess the influence of the hydrogel composition on the mineralization, we mineralize jammed, 3D printed urease-containing κ-carrageenan fragments and visualize the formation of the minerals with optical microscopy. Note that the activity of enzymes within κ-carrageenan fragments is lower than that of enzymes contained in gelatin, as shown in Fig. S2c. We assign the lower activity of enzymes within κ-carrageenan microfragments to the stronger electrostatic attraction between κ-carrageenan and urease, which affects enzyme conformation [74].

By analogy to the gelatin fragments, minerals start to form within κ-carrageenan fragments during the first 24 h and continue to form for at least 48 h, as shown in green in Fig. 3a. A close inspection of the optical micrographs reveals that minerals form within the κ-carrageenan filaments as well as at their surfaces. This rather homogeneous mineral distribution is in stark contrast to that observed for gelatin-based counterparts, where minerals are initially preferably located at the filament surface, as shown in Fig. 3a. Remarkably, polarized optical microscopy reveals a significantly smaller birefringence of the κ-carrageenan-based samples, as shown in Fig. 3b. These results suggest that the formed minerals are smaller or less crystalline.

Fig. 4. Structural characterization of CaCO_3_-organic scaffolds. a FTIR spectra of (i) CaCO_3_-gelatin scaffold as prepared (gray), after 1 day (light pink), 2 (dark pink), 3 (purple), and 4 days (dark purple) of mineralization. Pure gelatin is shown as a control (black). FTIR spectra of (ii) CaCO_3_-κ-carrageenan scaffolds as prepared (gray), after 1 day (mint green), 2 (green), 3 (forest green), and 4 days (dark green) of mineralization. Pure κ-carrageenan is shown as a control (black). The purple shade indicates the enzyme C = O peak, the gray shades indicate the calcite peaks, and the red shade indicates the sulfate group and the ether linkage of κ-carrageenan. b CaCO_3_ content of the biomineralized dried CaCO_3_-gelatin (squares in pink palette) and CaCO_3_-κ-carrageenan (circles in green palette) scaffolds as a function of incubation time. Data represent mean ± SD (n = 3). c) µCT scan of mineralized (i, ii) CaCO_3_-gelatin scaffold and (iii, iv) CaCO_3_-κ-carrageenan composites after (i, iv) 1 and (ii, v) 4 days of mineralization with (iii, vi) the two respective close-ups. The red filter is assigned to the organic gelatin and κ-carrageenan

To characterize the structure of the formed minerals, we perform XRD analysis on κ-carrageenan-based samples. The vast majority of the minerals formed after 4 days is calcite, as indicated by the XRD peaks at 29.6 °, 39.5°, 43°, 47.6°, and 48.6°, shown in Fig. 3c. This result is in good agreement with minerals formed in gelatin fragments. The peak positions of minerals formed in gelatin and κ-carrageenan are similar, but intensities diverge: κ-carrageenan’s sulfate-driven Ca²⁺ complexation concentrates ions and favors calcite formation due to locally higher supersaturations [16]. The more uniform Ca²⁺ distribution in gelatin results in the simultaneous formation of vaterite and calcite. The mean size of crystals grown in gelatin increases from 29 nm at day 1 to 51 nm at day 4. Similarly, that of minerals formed in κ-carrageenan increases from 26 nm to 51 nm, as shown in Fig. S3, suggesting similar crystal growth kinetics, as detailed in Table S1. Vaterite is not detectable at Day 1. By contrast, after four days of mineralization, gelatin-based composites contain vaterite with an average crystal size of ~ 25 nm, whereas κ-carrageenan-based samples contain vaterite crystals with an average size of ~ 43 nm. XRD analysis of κ-carrageenan-based samples reveals more intense diffraction peaks compared to peaks obtained from gelatin-based counterparts, as shown in Fig. 3c. This comparison suggests that the minerals formed within κ-carrageenan are crystalline, such that the reduced birefringence of κ-carrageenan-based samples likely does not originate from a lower crystallinity of the minerals.

To assess if the reduced birefringence is related to a lower mineral concentration, we perform TGA on these composites. The mineral formation kinetics are similar to that observed for gelatin-based counterparts, as shown in Fig. 4b, in good agreement with our XRD results. The degree of mineralization of κ-carrageenan-based samples is approximately 10% lower than that of gelatin-based counterparts. Yet, this slightly lower mineral content in κ-carrageenan-based samples cannot be the sole reason for the much lower birefringence of these samples.

To assess if the lower birefringence is related to the crystal size, we perform SEM analysis on mineralized samples. Indeed, the morphology of the formed crystals is distinct: crystals formed in gelatin-based microfragments exhibit an average edge length of 12 μm, whereas those formed in κ-carrageenan have an average edge length of 3 μm, as shown in Fig. 3d and in Fig. S4. We attribute this difference to the presence of sulfate groups in the κ-carrageenan fragments that strongly complex Ca^2+^, thereby lowering the nucleation barrier and increasing the nucleation-site density. The strong Ca^2+^- κ-carrageenan interaction slows down the crystal growth, eventually leading to a lower mineral content [75, 76]. By contrast, gelatin, which lacks sulfates but bears carboxylate, amine, amide, and hydroxyl groups that exhibit weaker interactions with Ca^2+^, forms fewer nuclei if incubated under identical conditions. These nuclei grow more rapidly than in κ-carrageenan, resulting in higher mineral contents [77, 78].

The stiffness and toughness of mineral-based composites are much higher if minerals form a percolating network [31, 67]. To test if the minerals grown within our granular system form a percolating network, we visualize the samples using micro computed tomography (µCT). These tomography images reveal a significant mineral content already after the first day of mineralization, as shown in Fig. 4c. The volume fraction of CaCO_3_ increases more slowly thereafter, from 29 vol% after 1 day of mineralization to 38 vol% after 4 days of mineralization, as shown in Fig. S5, well in agreement with our TGA, XRD, and optical microscopy data. A 3D rendering of the tomographic images suggests that samples that have been mineralized for 4 days contain a percolating mineral network, as shown in Fig. S6a. These mineralized samples only contain 6 vol% gelatin, as determined from µCT analysis. Urease-driven alkalinization leads to partial gelatin dissolution, as indicated by the lower polymer content of gelatin-based composites compared to that of κ-carrageenan-based counterparts, shown in Fig. S7. Note that gelatin does not dissolve if incubated in an aqueous solution containing CaCl_2_ and urea but lacking urease, even if incubated for up to 10 days, as shown in Fig. S7. These results suggest that the enzymatic activity increases the pH, resulting in a gelatin dissolution.

Mineralized κ-carrageenan-based composites contain a higher polymer fraction: Samples mineralized for 4 days contain approximately 24 vol% κ-carrageenan, as shown qualitatively in Fig. 4c v and quantitatively in Fig. S5. We attribute the higher polymer content to the enhanced stability of κ-carrageenan under basic conditions compared to that of gelatin. κ-carrageenan is stable at mildly alkaline pH and degrades under acidic conditions [79]. Moreover, we expect κ-carrageenan to exhibit stronger interactions with the forming minerals than gelatin, hampering the removal of κ-carrageenan. To probe our expectation, we perform FTIR analysis. During the initial stages of mineralization, we observe an asymmetric stretch vibration of the sulfate group (O = S=O) at 1225 cm^-1^, as shown in Fig. 4a ii [80]. This peak disappears over time, suggesting a rather strong interaction with the ions present in the mineralization solution. To assess if the disappearance of the sulfate stretch vibration peak is a result of a partial dissolution of κ-carrageenan, we analyze the ether linkage (-C-O-C-) of the κ-carrageenan glycosidic backbone at 1035 cm^-1^. This peak remains unchanged during the mineralization, as shown in Fig. 4a ii, indicating that a significant amount of κ-carrageenan remains in the sample, in good agreement with our µCT results. The higher polymer fraction in κ-carrageenan-based samples might be a contributing reason for their slightly lower volume fraction of minerals, which reaches 32 vol% after 4 days of mineralization, as shown in Fig. S5. As a result of the higher polymer content, the porosity of κ-carrageenan-based composites is below that of gelatin-based counterparts, reaching up to 44 vol%, as shown in Fig. S5.

Minerals formed in granular κ-carrageenan-based composites also form a percolating network, as suggested by the 3D rendering of µCT images in Fig. S6b. Yet, the mineral network formed in κ-carrageenan-based gels exhibits a core-shell structure, as shown in Fig. 4c iii,** vi**. This core-shell structure is assigned to the sulfate-mediated Ca²⁺ complexation that triggers interfacial nucleation, which leads to the formation of a dense mineral shell that impedes mineralization of the core. By contrast, gelatin exhibits a weaker interaction with Ca^2+^ such that Ca^2+^ can diffuse more freely into the entire gelatin-based structure, leading to a more homogeneous mineralization [81].

A percolating mineral network typically strongly increases the stiffness and toughness of mineral-based composites [31, 67]. To assess the extent to which minerals formed within our granular scaffolds increase their stiffness and toughness, we perform compression measurements on them. Dry mineral-free gelatin-based granular scaffolds exhibit a compression modulus of 7 MPa. The compression modulus increases with mineralization time. For example, after one day of mineralization, the compression modulus reaches 26 MPa, and it further increases to 56 MPa after the scaffold has been mineralized for 4 days, as shown in Fig. S8a. However, the increase in stiffness is paired with a reduced strain at break such that the strength of the mineral-based composites is nearly constant at 4 MPa, independent of the mineralization time, as shown in Fig. S8b. At higher strains, mineral-containing composites exhibit a plateau in the stress-strain curve that is followed by a strong increase in the stress upon further compression, as shown in Fig. S9a-b. We assign the strong stress increase to a densification of the samples. The strain where densification starts decreases with increasing mineral fraction, indicating an embrittlement, as shown in Fig. 5a.

Mineralized samples are often used in wet conditions. To evaluate the effect of hydration on the mechanical properties, we soak mineralized samples in PBS before compression testing. As expected, the modulus of mineralized samples made of gelatin scaffolds decreases from 56 MPa measured in the dry state to 4 MPa in the hydrated state, as shown in Fig. S10. Similarly, the strength decreases from 4 MPa in the dry state to 0.2 MPa in the hydrated state.

To assess the influence of the mineral size on the mechanical properties of CaCO_3_-based composites, we perform compression tests on CaCO_3_-κ-carrageenan composites (Fig. 5b). These samples are significantly softer: Dried samples that have been mineralized for 4 days exhibit a compression modulus of 5.25 MPa and a strength of 0.35 MPa, as shown in Fig. 5c-d and Fig. S11. These values are approximately 11 times lower than those measured for mineralized gelatin-based counterparts. However, these samples soften less in wet state: their compression modulus decreases from 5.25 MPa in the dry state to around 10 MPa in the hydrated state, as shown in Fig. S10. Also the maximum strength at Day 4 of mineralization decreases less strongly than in gelatin-based counterparts, from 0.3 MPa for the dry state to 0.2 MPa for the hydrated state.

Our µCT results reveal a high porosity of our composites. To assess if the measured differences in stiffness and strength between gelatin and κ-carrageenan-based composites are solely related to a change in their degree of porosity, we quantify their density. The density of gelatin-based composites decreases from 1.3 g/cm^3^ to 0.68 g/cm^3^ upon mineralization for two days. We attribute this decrease to the partial dissolution of gelatin. The density subsequently increases to 0.78 if incubated for 4 days, as shown in Fig. S12. This density increase is assigned to an increased degree of mineralization. The density of κ-carrageenan-based composites also reaches a minimum of 0.79 g/cm^3^ during the second day of mineralization before it slightly increases to reach a maximum of 0.82 g/cm^3^ if mineralized for 4 days, as shown in Fig. S12. These values are lower than the theoretical density of calcite [82], mainly due to pores present in the composites, albeit polymer residues also contribute to a density reduction. We normalize the compression moduli of our samples by their density, resulting in specific compression moduli. The specific compression modulus of gelatin-derived composites increases from 33 to 72 MPa∙g^-1^∙cm^3^ if mineralized between 1 and 4 days, as shown in Fig. 5c. Similarly, the specific strength increases from 3.5 to 5.15 MPa∙g^-1^∙cm^3^ if mineralized between 1 and 4 days, as shown in Fig. 5d. Indeed, our composite exhibits a specific strength that exceeds those previously reported CaCO_3_-based composites possessing similar porosity by at least 26%, as summarized in the Ashby plot in Fig. 5e [26, 83–86]. We attribute the high specific strength of our composite to the high mineral content and homogeneous distribution of CaCO_3_ crystals throughout the sample. This assignment is supported by the lower specific strength of κ-carrageenan-based counterparts that contain less minerals; their specific strength peaks at 2.8 MPa∙g^-1^∙cm^3^ after 1 day of mineralization, as shown in Fig. 5d. Similarly, the compressive modulus of κ-carrageenan-based composites is lower than that of gelatin-based counterparts and peaks at 21.5 MPa∙g^-1^∙cm^3^ after 1 day of incubation.

Fig. 5. Mechanical properties of CaCO_3_-organic composites. a Compression curves of CaCO_3_-gelatin composites as prepared (gray curve), after 1 day (light pink), 2 (dark pink), 3 (purple), and 4 days (dark purple) of mineralization. b Compression curves of CaCO_3_-κ-carrageenan scaffolds as prepared (gray curve), after 1 day (mint green), 2 (green), 3 (forest green), and 4 days (dark green) of mineralization. c Specific compressive modulus and (d) specific compressive strength of CaCO_3_-gelatin (squares in pink palette) and CaCO_3_-κ-carrageenan composites (circles in green palette) as a function of the mineralization time. The values are obtained from the curves in (a) and (b). Data represent mean ± SD (n = 3). The gray shade qualitatively indicates the typical range for human trabecular bones [87–89]. e Ashby plot of the compressive strength of CaCO_3_-gelatin (light pink), CaCO_3_-κ-carrageenan (mint green) scaffolds obtained in this study with enzyme encapsulation, CaCO_3_-gelatin composites obtained with bacteria (brown) [26], CaCO_3_ composites obtained with emulsions (orange) [83], CaCO_3_ obtained with PU foam templates (red) [84], CaCO_3_-chitosan scaffolds obtained with in situ mineralization (light blue) [85], and CaCO_3_-based composites with needle-like crystals (blue) [86]

To assess if we can tune mineral content by controlling the enzyme concentration within fragments, we vary this parameter from 5 to 20 mg·g^− 1^ while keeping the urea to Ca^2+^ ratio constant. The mineral fraction in κ-carrageenan scaffolds increases from 76 wt% for samples with an enzyme concentration of 5 mg·g^− 1^ to 92 wt% for samples containing an enzyme concentration of 20 mg·g^− 1^, as shown in Fig. S13a. As a result of the increasing mineral content with increasing enzyme concentration, the compression modulus and strength also increase, as shown in Fig. S13b–d. By contrast, gelatin-based scaffolds exhibit the highest mineral content if made from fragments containing 10 mg g^− 1^ enzymes, as shown in Fig. S14a. Accordingly, the compressive modulus and strength are highest if gelatin fragments contain 10 mg g^− 1^ enzymes, as shown in Fig. S13b-d. We assign this result to a homogeneous mineral distribution for all tested κ-carrageenan and gelatin scaffolds containing not more than 10 mg g^− 1^ enzymes. By contrast, gelatin samples containing 20 mg g^− 1^ enzymes possess a dense mineral shell and a polymer-rich core, as shown in µCT results, exemplified in Fig. S14.

The granular structure of the enzyme-containing polymeric scaffold enables its direct ink writing into intricate cm-sized structures. To demonstrate the potential of this ink, we print a brittle star using urease-loaded gelatin microfragments. We incubate the printed brittle star in the mineralization solution to rigidify the polymeric scaffolds. After 4 days of mineralization, we obtain a rigid brittle star whose shape is similar to that of the printed structure. The volume of the 3D printed structure increases by only 10% after 4 days of mineralization and drying, indicating that the mineralization does not significantly alter the shape of the polymeric scaffold, as shown in Fig. 6a. We leverage the injectability of the enzyme-containing ink to demonstrate its potential to heal defects. We intentionally break one of the limbs of the brittle star and re-connect it to its body using small amounts of urease-loaded gelatin microfragments. We add a red food dye to the ink for visualization. The structure is mineralized for another 4 days to obtain a brittle star with a repaired limb, as shown in Fig. 6b. Note that minerals only form within the newly added red ink and do not grow onto the previously formed minerals, as shown in Fig. 6c iv. This results hints at the good shape fidelity of our formulation. We also obtain good shape fidelity of mineralized structures if urease-loaded κ-carrageenan-based fragments are direct ink written into a cm-sized tower shell and subsequently mineralized, as shown in Fig. 6c i-ii or into a Stellaria Solaris shell, as shown in Fig. 6c iii,** iv**. In this case the 3D printed structures show a decrease in volume of approximately 14% after 4 days of mineralization. Importantly, we can print complex structures with cavities and small features at room temperature in air without the need for any sacrificial structures or baths, in contrast to emulsion-based [83] and granular bacteria-loaded [26] 3D printable inks.

Our results indicate that we can confine the mineralization to the granular scaffolds. However, even the fully mineralized scaffolds contain significant amounts of water. We quantify the volumetric change of mineralized structures upon drying, which potentially compromises their shape fidelity and structural stability. This is achieved by measuring the difference in the volume of the structure once removed from the incubating solution and after it has been dried. The volume of unmineralized gelatin-based granular scaffolds changes by 71% upon drying. The volumetric change decreases to 38% if the structure is mineralized for 1 day, to 8.5% if mineralized for 2 days and to 14% if mineralized for 4 days, as shown in Fig. 6d. These results indicate that within approximately 2 days of incubation, minerals form a percolating network that is sufficiently strong to sustain the weight of the structure such that it does not collapse upon drying. We obtain similar results for κ-carrageenan-based samples, as shown by the green symbols in Fig. 6d. We assign the greater volumetric change of fully mineralized κ-carrageenan-based scaffolds to their higher polymer content compared to gelatin-derived composites.

To demonstrate the load-bearing capability of 3D printed mineralized composites, we print square lattices, mineralize them, and perform compression tests on them. Gelatin-based composites exhibit a compressive modulus of 133 MPa, whereas κ-carrageenan-based lattices have a compression modulus of 71 MPa, as shown in Fig. S15. The compressive strength of gelatin-based samples is 7.8 MPa, whereas that of κ-carrageenan-based counterparts is 4.7 MPa, as shown in Fig. S15 and Tab. S2. Note that the compression modulus and strength of printed samples are slightly above the values measured for cast samples. We attribute this improvement in mechanical properties to a reduced defect density in printed samples.

Fig. 63D printing of the enzyme-containing granular inks. Photographs of (a) gelatin-based fragments containing urease printed in air into (i) a brittle star that is incubated (ii) in an aqueous solution containing CaCl_2_ and urea to trigger mineralization. (iii, iv) Photographs taken after four days of mineralization in the (iii) wet and (iv) dry state. b Mimicking the behavior of the brittle star fish, (i) the broken limb (ii-iii) can be re-attached by connecting the broken parts with new ink, (iii-iv) stained with red food dye for better visualization, and re-mineralizing. c κ-carrageenan-based fragments containing urease printed in air into (i) a tower shell, where (ii) the cavities are maintained after complete mineralization and drying. The same κ-carrageenan-based ink is used to 3D print a Stellaria solaris shell (iii, iv). All scale bars are 1 cm. d Volumetric shrinkage of the CaCO_3_-gelatin (squares in pink palette) and CaCO_3_-κ-carrageenan (circles in green palette) scaffolds as a function of the mineralization time. Data represent mean ± SD (n = 3)

We consider our samples to be potentially suitable for tissue engineering applications. To assess their biocompatibility, we culture MG-63 (osteoblasts-like cells) onto fully mineralized urease-containing microfragment-based scaffolds and characterize their viability and morphology over a 7-day culture period. Live/Dead staining reveals that CaCO_3_-gelatin and CaCO_3_-κ-carrageenan scaffolds support high cell viability at Days 1 and 7, as shown in Fig. 7a. Cell viability is consistently above 80% for both scaffold types, with gelatin scaffolds showing slightly higher viability, as shown in Fig. 7b. Low-magnification images and quantitative cell-density analysis confirm the increase in coverage from Day 1 to Day 7 across all samples, as shown in Fig. S16. To further assess cellular morphology, we perform SEM on the samples [90]. At Day 1, cells in gelatin and κ-carrageenan-based scaffolds exhibit a predominantly rounded shape, as indicated by the arrows in Fig. 7a. As expected, by Day 7, cells on gelatin scaffolds appear more spread and integrated into the calcite scaffold, resulting in larger cell areas and higher aspect ratios (ARs). In contrast, cells in κ-carrageenan scaffolds retain their spherical morphology with minimal changes, as shown in Fig. 7d and c. These results reveal good biocompatibility and higher cellular adhesion for the gelatin-based formulation, due to its cell-adhesive motifs. Note that the current study focuses on demonstrating the feasibility of fabricating stiff and porous CaCO_3_-based scaffolds. Studying the biocompatibility and degradation of these scaffolds in more depth constitutes an interesting avenue for further studies. Nevertheless, these results indicate the potential of our approach to design porous homogeneously mineralized scaffolds compatible with cells.

Fig. 7. Cell viability and morphology of CaCO_3_-gelatin and CaCO_3_-κ-carrageenan scaffolds over time. a Live/Dead staining (green: live cells (calcein), red: dead cells (propidium iodine)) and SEM micrographs of cells (pointed with white arrows) cultured onto CaCO_3_-gelatin and CaCO_3_-κ-carrageenan-based scaffolds on Day 1 and Day 7, showing differences in cell distribution, viability, and morphology. b Quantification of cell viability (%) at day 1 and day 7, demonstrating higher viability in gelatin-based hydrogels (pink) compared to κ-carrageenan-based hydrogels (green) at both time points. c Violin plots of cell area (µm²) indicating increased spreading of cells in CaCO_3_-gelatin (pink) scaffolds over time compared to CaCO_3_-κ-carrageenan (green). d Violin plots of cell aspect ratio (AR, arbitrary units), highlighting differences in cell elongation between the two scaffold types. Data are presented as mean ± SD and analyzed with unpaired t- test, significant differences are marked with **p < 0.01, ***p < 0.001, ****p < 0.0001

Conclusion

We introduce urease functionalized gelatin and κ-carrageenan microfragments that, if jammed, can be direct ink written into cm-sized 3D structures in air and at room temperature. These structures, reach CaCO_3_ contents up to 92 wt% and a porosity up to 56 vol% if mineralized for 4 days. The high mineral content and porosity impart the resulting composites a specific compressive modulus up to 72 MPa∙g^− 1^∙cm^3^ and a compressive strength up to 5.15 MPa∙g^− 1^∙cm^3^. These values exceed those of previously reported 3D printed CaCO_3_-based porous scaffolds. Our ink is composed of naturally sourced reagents, namely gelatin and algae-derived κ-carrageenan, and contains enzymes that are much more cost-effective and less resource-intensive to cultivate and maintain than ureolytic bacteria. We envision that these assets, combined with their biocompatibility, render our ink a promising platform for tissue engineering or to heal or repair broken minerals.

Materials and methods

Materials

κ-carrageenan (ThermoFischer, 431590250), gelatin from porcine skin (gelatin Type-A from porcine skin (gel strength 300, Sigma-Aldrich, G2500), urea (Sigma-Aldrich, 51456), urease from Canavalia ensiformis (Jack bean) (Sigma-Aldrich, U1500-20KU), alginic acid sodium salt from brown algae (Sigma-Aldrich, A1112), phosphate buffered saline (PBS) (Gibco), calcium chloride (Carl Roth, CN93.1), Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco), fibroblast growth factor 2 (FGF-2, PeproTech), fetal bovine serum (FBS) (Gibco), MG-63 cells (osteosarcoma cell line, ATCC #CRL-1427; ATCC, Wesel, Germany), Propidium Iodide (PI) (Fluka).

Microfragments preparation

Microfragments were prepared by mechanical grinding via cryomilling gelatin and κ-carrageenan bulk hydrogels that were prepared as follows. A solution was obtained by dissolving 25 wt% gelatin at 40 °C or 4 wt% κ-carrageenan in water at 70 °C. The concentrations have been selected to match fragment rheology and compressive behavior for direct ink writing, empirically tuning concentrations to avoid filament collapse/over-spreading. 10 mg of urease was weighed per g of gelatin solution and dissolved in 100 µL of water just before use. The enzyme was poured into a beaker containing the gelatin or κ-carrageenan solution (T = 40 °C) and stirred. The solution was poured into a glass mold and cooled to room temperature to solidify the gelatin or the κ-carrageenan. The obtained bulk gel with the encapsulated enzyme was stored at 4 °C. The bulk gels were fragmented with an oscillatory cryomill (Cryomill, Retsch) using 1 stainless steel ball (d = 25 mm). The cryomilling protocol is as follows: 3 min of precooling at 5 Hz; 1 milling cycle of 1 min at 30 Hz. The prepared fragments were freeze-dried and stored at 4 °C.

Ink preparation

Dry enzyme-loaded microfragments were resuspended in a 4 wt% alginate solution in PBS with a 1:4 volume ratio of dry fragments: alginate solution for 1 h at 4 °C. To remove the supernatant, the suspension was centrifuged for 10 min at 3700 rpm and 7 °C (Mega Star 1.6R, VWR). The obtained ink was stored at 4 °C and used for 3D printing and casting.

Ink biomineralization

Once 3D printed or cast into the final shape, the ink was gelled by exposing it dropwise to a large excess of 1 M CaCl_2_ solution until submerged. After 30 min the ink was transferred into the incubation aqueous solution containing 1 M CaCl_2_ and 1.5 M urea mixed in a 1:1 volume ratio. The solution was exchanged every 24 h for four days. After the fourth day, samples were removed from the incubation solution, soaked in water for 25 min, and dried in a vacuum chamber at room temperature overnight.

Rheology measurements

Rheological measurements were performed on a DHR-3 TA Instrument rheometer with a 8 mm diameter parallel plate steel geometry. All measurements were performed at 25 °C, with a 1000 μm gap and repeated on 3 distinct samples. Amplitude sweeps were performed at 10 rad∙s^− 1^ for a strain range of 0.01 to 200%. Frequency sweeps were performed within the linear viscoelastic regime at a strain value of 0.5% in a frequency range of 0.1–50 Hz. Shear recovery measurements were performed at 10 rad/s alternating 200 s at 0.3% with 200 s at 80% strain.

Stress relaxation measurements on bulk hydrogels were performed at the constant strain of 1%, a value picked in the linear viscoelastic regime, for 16 min. The stress is normalized at 0.1 s. All the represented data were obtained from three independent samples and are reported as mean ± SD.

XRD measurements

The regular theta-theta scans for all the samples were performed on a Panalytical Empyrean X-Ray polycrystalline diffractometer in Bragg-Brentano geometry, equipped with long-focused seal Cu X-Ray tube (λKα = 1.5418 Å), and PIXcel 1D X-Ray detector. The patterns were collected in continuous mode between 5 and 60 degrees (2θ), with the step-size of 0.02626 degree. The baseline removal and peak search were performed with the Peak Analyzer tool in OriginPro2021.

Crystallite size D was estimated per peak with the Debye–Scherrer equation:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:D=\frac{\mathrm{Kλ}}{{\mathrm{{\beta}}}_{\mathrm{hkl}}\mathrm{cosϑ}}$$\end{document}

with shape factor K = 0.94, λ = 1.5418 Å (Cu Kα, average), β_hkl_​ the instrument-corrected FWHM in radians of 2θ, and θ = 0.5 2θ. Values are reported in nm. For each day and phase, we report the mean ± standard deviation over the selected reflections. Instrumental correction and subsequent size calculation were executed in Match! 4 (Crystal Impact). All the crystallite sizes are reported in Tab S1.

Optical imaging

Optical microscopy images were obtained with a Nikon Eclipse TS100 microscope. The diameters of the microfragments were calculated as the average of the two diagonals measured with Image J. Polarized optical microscopy images were obtained with a polarizing microscope (ECLIPSE Ti-DH, Nikon).

Micro computed tomography

X-Ray µCT was performed on an Ultratom micro tomography system (RX-SOLUTIONS). The dry samples were scanned at a voxel size of 1.5 to 2.7 μm, with a voltage of 55 kV and a current of 100 mA. Amira-Avizo v2023.2. software was used for reconstruction, segmentation, and visualization. Image J is used to calculate the mineral, pore and gel fraction over the entire volume of each sample. All the results in Fig. S5 are reported as mean ± SD of all the obtained slices for each sample. All the measurements were performed on samples obtained by casting in a 4 mm x 2 mm mould.

FTIR characterization

FTIR spectra were measured on a Spectrum 3 spectrometer (PerkinElmer) in the attenuated total reflectance (ATR) mode, and corrected for the background and CO_2_ signals. Traces were acquired between 4000 and 650 cm^− 1^ at a resolution of 4 cm^− 1^. All the measurements were repeated on 3 distinct samples grinded in a fine powder.

Volume shrinkage and density measurements

The volume shrinkage percentage (Vs) was calculated as follow:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{V}_{s}=\:\frac{{V}_{i}-{V}_{f}}{{V}_{i}}\times\:100$$\end{document}

Where Vi is the initial volume calculated by measuring diameter, thickness and mass of 3 distinct cylinders as prepared in a 6 mm x 2 mm mold, after removing the excess water from the surface, and Vf is the final volume of the same set of 3 samples measured in the same way after overnight drying in a vacuum chamber.

Thermogravimetric analysis

Biomineralized samples were finely grinded in a mortar before testing with TGA (TGA 4000, PerkinElmer). The measurements were performed from 30 °C to 850 °C at a heating rate of 10 °C min^− 1^ with an air flow rate of 20 mL min^− 1^, and holding at 850 °C. The CaCO_3_ weight% (wt%CaCO_3_) was obtained with the following formula:

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{wt\%CaCO}_{3}=\:\frac{\varDelta\:wt\%}{{mCO}_{2}}\:\mathrm{x}\:{mCaCO}_{3}$$\end{document}

Where Δwt% is the weight% loss calculated between 600 °C and 850 °C, mCO2 is the molecular weight of CO_2_ (44.102 g∙mol^− 1^), and mCaCO3 is the molecular weight of CaCO_3_ (100.1 g∙mol^− 1^). All the samples were rinsed with water for 15 min and dried in a vacuum chamber overnight before testing. The represented data were obtained from at least three independent samples and are reported as mean ± SD. The full curves used for the calculations are reported in Fig. S17.

Compression measurements

Compression tests were performed on cylindrical samples obtained by casting the ink in 8 mm diameter and 6 mm tall molds. All the samples were rinsed with water for 15 min and dried in a vacuum chamber overnight before testing with a commercial machine (ZwickiLine, 5 kN load cell, Zwick Roell), compressed at a constant velocity of 5 mm/min until 80% strain was reached. All compression tests were performed in the dry state to maximize reproducibility; fully hydrated mechanical testing was not included due to poor reproducibility, which we assign to sample-to-sample variations in their degree of swelling. The compression modulus was calculated as the slope of the initial linear region, from 5% to 12% strain. The compression strength was obtained from the maximum load in the curve before 15% of deformation. All the reported results were obtained from at least three independent samples and are reported as mean ± SD.

3D printing

Jammed gelatin-alginate or κ-carrageenan-alginate microfragments loaded with urease enzyme were loaded in a 3 mL Luer lock syringe. The syringe was centrifuged at 3500 rpm for 1 min to remove trapped air bubbles at 7 °C. 3D printing was carried out with a commercial 3D bioprinter (BIO X, Cellink) at 12 mm/s and 60 kPa. The printing bed was maintained at 20 °C, unless stated differently. The granular ink was extruded through a blunt needle (20G, DT = 603 μm) using a pressure-driven piston, unless stated differently.

Cell expansion and seeding

MG-63 cells (osteosarcoma cell line, ATCC #CRL-1427; ATCC, Wesel, Germany) were expanded in a proliferation medium consisting of Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% (v/v) FBS, 100 µg/mL gentamicin (Gibco), and 5 ng/mL fibroblast growth factor 2 (FGF-2). Cells were maintained at 37 °C in a humidified atmosphere with 5% CO₂ and passaged upon reaching 80% confluency. Cells were trypsinized, resuspended in DMEM containing 10% (v/v) FBS and 10 µg/mL gentamicin, and collected by centrifugation (5 min at 500 × g). After counting, 100,000 cells were seeded on each composite construct (casted in a 4 mm x 2 mm mould) and cultured for 7 days. Constructs were sterilized by immersion in 70% ethanol for 10 min, followed by UV sterilization for 20 min.

Live/dead imaging

Samples were incubated in DMEM supplemented with 1:1000 Calcein AM (Invitrogen) and 1:500 Propidium Iodide (PI) for 40 min without FBS, followed by three washes with phosphate-buffered saline (PBS). Imaging was performed using a Leica SP8 microscope (Leica) equipped with a 10× dry objective. Z-stacks were acquired from the sample surface at 4 μm intervals, extending 100 μm into the sample. The images presented in this study were generated using maximum intensity z-projection.

Scanning electron microscopy (SEM)

SEM imaging on dry mineralized scaffolds at the end of the incubation period and on dry gelatin/alginate and κ-carrageenan/alginate controls containing the same concentration of enzyme were performed on a Zeiss Gemini 300 at an operating voltage of 3 kV with 10 nm Au/Pd coating using a secondary electron detector.

Samples seeded with cells, after the designated culture period, were washed with PBS and fixed with 4% paraformaldehyde (PFA) for 1 h. The samples were dehydrated through a graded ethanol series (20% to 95%), air-dried under a fume hood, and subsequently coated with a 10 nm carbon layer using a CCU-010 Carbon Coater (Safematic). SEM imaging was performed using a Zeiss Merlin microscope at an operating voltage of 2 kV.

Urease activity

The activity was calculated following previously reported protocols [66] with a conductivity meter (CDM210 MeterLab). The hydrolysis of urea catalyzed by urease produces ionic species that increase the electrical conductivity of the reaction medium. Urease activities were determined by measuring the relative change in conductivity of enzyme-containing samples exposed to 1.11 M urea for defined time intervals. The rate of conductivity increase (mS·min^− 1^) was converted to urea hydrolysis rate (mM·h⁻¹) using Eq. (1)

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Urea\;hydrolyzed\;\left(mM\right)-Conductivity\;\left(\mu S\;cm^{-1}\right)\times11.11$$\end{document}

The factor 11.11 represents the experimentally established proportionality constant between conductivity change and urea hydrolysis under standard conditions (1.11 M urea, 21–22 °C). We calculated the urea hydrolysis rate using Eq. (2):

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{lc}Urea\;hydrolysis\;rate\;\left(mMh^{-1}\right)=\\\frac{\triangle Conductivity\left(\mu S\;cm^{-1}\right)\;}{\triangle time\;\left(min\right)}\times\frac{10^{-3}mS}{1\;\mu S}\times\frac{60\;min}{1h}\times11.11\end{array}$$\end{document}

Finally, the specific activity (U·g⁻¹ or U·mg⁻¹) was calculated from the urea hydrolysis rate and normalized to the total mass of enzyme-containing fragments or the equivalent free enzyme (m) according to Eq. (3):

\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{array}{lc}Urease\;specific\;activity\;\left(U\;g^{-1}\right)=\\\frac{Urea\;hydrolosis\;rate\;\left(mMh^{-1}\right)\times V\left(mL\right)}{m\left(g\right)\times60}\end{array}$$\end{document}

One unit (U) corresponds to the hydrolysis of 1 µmol of urea per minute in a 10 mL assay volume.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary Material 3