Volume proportions of A/C component

The degree of substitution (DS) of GelMA can be easily modified by changing the ratios of gelatin and methacrylic acid. Higher DS and precursor solution concentration result in higher mechanical properties of the photocrosslinked structures28,29. According to our experience, 5% (w/v) low-DS GelMA (EFL-GM-30) has low mechanical properties and high biocompatibility. Thus, it was selected to electrospray GelMA microgels (A30/5, 500 μm), which has been verified to own very uniform diameters (Supplementary Note 4 and Supplementary Fig. 6). 20% (w/v) high-DS GelMA (EFL-GM-300) can stand compressive stress, such as bone tissue, whereas 20% (w/v) low-DS GelMA (EFL-GM-30) can stand stretching stress, such as tendon tissue. Thus, they were selected as C component (C300/20, C30/20). For comparison, 5% (w/v) low-DS GelMA (EFL-GM-30) was also selected as C component (C30/5).

We assumed that C component exactly infiltrated the vacancy among A component so that the microgels could generate enough internal friction to avoid collapse before crosslinking. Accumulated A30/5 (with GFP) in the cell sieve were repeatedly soaked in C30/5 (with RFP) and lifted to filter the redundant C30/5 (Fig. 2a). C30/5 spontaneously entered the vacancy among A30/5 with capillary force. The crosslinked A–C bioink showed that A component closely attached to each other (Fig. 2b, c). The volume proportion of A component was analyzed red/green areas as 73.215 ± 2.312% (Supplementary Note 2), which was similar to the atomic space utilization in hexagonal closest packed (HCP) in crystal chemistry (74.05%) (Fig. 2d) in that A component displayed standard spheroid shape under floatage and surface tension in liquid C component and the system tended to own the lowest energy under gravity.

Emergency bag for portability

An emergency bag was designed to meet the storage and portability requirements (Fig. 2e). A component loading cells was immersed in cryopreservation medium and stored in a freezing tube as previously described by Ouyang et al. 30. C component was also stored in another freezing tube. The volume proportions of A/C component in one kit were 74.05%/25.95%, respectively. Both freezing tubes were stored in a container filled with liquid nitrogen.

In an accident, the freezing tubes were removed from the container and thawed with a tiny heating pad at 37 °C powered by a portable USB battery (Fig. 2f). For very urgent accidents without heating devices, they could be directly heated using the body temperature. Certainly, the cold injury on the patient’s skin should be noticed. Then, the cryopreservation medium was removed by a syringe and a napkin. A component was transferred to C component with a thin spoon and stirred uniformly, followed by transfer A–C bioink into a new syringe with cone printing nozzle for in-situ bioprinting with professional in-situ bioprinters or just hands, followed by photocrosslinking with 405-nm blue flashlight, which has been verified to be safe and widely used in current clinical scenes, such as tooth photocuring, blue light treatment of neonatal jaundice, etc.

In practice, the total volume of bioink is determined by the volumes and quantities of the applied freezing tubes. At A–C bioink production end, the prepared A–C bioink can be loaded with different types of freezing tubes according to the production requirements. Moreover, at the rescuing end, in terms of the quantities of freezing tubes, rescuers can definitely take as many freezing tubes loading A–C bioink as possible according to the patient injury situation. Therefore, with the further development and mass production of A–C bioink, this bioink system would be feasible for the in-situ bioprinting of tissue defect as large as possible.

Rheological robustness and mechanism

A–C bioink should own high rheological robustness in different temperature conditions to adapt to different accident situations of in-situ treatment. A30/5–C30/20, A30/5–C300/20, A30/5–C30/5 and C bioink C30/20, C300/20, C30/5 were prepared. 4, 24 and 37 °C were selected, under which GelMA precursor solution would be in excessive gelation, optimized sol–gel and excessive solization state for extruding bioprinting, respectively31,32.

Flow sweep results indicated both A–C bioinks and C bioinks had shear-thinning feature under the three selected temperatures (Fig. 3c), meeting the basic requirement of extruding bioprinting. This was because GelMA microgels were separated from each other and the friction among them disappeared at a high shear rate. Furthermore, the orientation of discrete GelMA molecules in C component tended to be coincident and molecule twining was reduced (Fig. 3a). According to published studies9,33, the system mainly composed of microgels had Bingham fluid properties. The flow sweep data of A–C bioinks were further fitted with the Bingham fluid model as follows:

$$\sigma =\sigma _\rm y+\eta _\rm B\dot\gamma$$

(1)

in which $$\sigma$$, $$\sigma _\rm y$$, $$\eta _\rm B$$ and $$\dot\gamma$$ is stress, yield stress, Bingham viscosity, shear rate, respectively. All A–C bioinks showed Bingham fluid feature and linear relationship between stress and shear rate above certain stress (Fig. 3d, e). The yield stress increased at 4 °C due to the excessive gelation of C component during which GelMA molecules would form collagen-like spirochete, spirochete aggregate and aggregate network in succession (Fig. 3b). The solid/fluid feature below/above yield stress was due to the internal friction among A component the malposition of microgels beyond the static friction, respectively.

Bioink should own fluidity to form filaments from nozzles and display solid feature to maintain the shape. The oscillation frequency testing results showed that all A–C bioinks under different temperatures owned a solid/fluid feature under low/high frequency, respectively, benefiting by Bingham fluid property (Fig. 3f). Remarkably, traditional mono-component GelMA bioink would transfer to solization and cannot maintain shape at body temperature (37 °C). However, even under 37 °C, A–C bioink still showed solid state under low frequency, verifying its feasibility for depositing on wounds. The thixotropy results by adding periodically varied oscillation amplitudes (200% and 1%) indicating the state transfer of A–C bioink was rapid and obvious, confirming its rapid self-healing speed. (Fig. 3g)

Thermo-sensitive bioinks need time to achieve stable state under certain temperature. Besides, for a certain temperature, the sol–gel state at certain time would be affected by the previous state and totally different during the temperature-increasing/decreasing process. Flow temperature ramp testing of A–C and C bioink was carried out to examine the viscosity variation in repeatedly increasing/decreasing temperatures (4–39 °C, 5 °C/min) for three times (I–III) (Fig. 3h). The results showed that even the viscosity curves of the temperature-increasing/decreasing process of the same bioink were not superimposed, verifying poor state reversibility and stability of thermo-sensitive bioink. However, compared to C bioink, the zones surrounded by the viscosity curves in the continuous three tests of A–C bioink were nearly superimposed, indicating A–C bioink could effectively avoid the effect of the previous state. Furthermore, for such a wide temperature changing range, the viscosity of C bioink stretched 4–5 magnitudes, whereas that of A–C bioink maintained inside 1 magnitude because of the dominant role of microgels. Remarkably, promising bioprinting temperatures of 20–24 °C were also the dramatical viscosity-changing range. It could be imagined that once a tiny temperature floating in this range happened during the in-situ bioprinting, the viscosity of traditional bioink would sharply vary and bring unpredictable risk while A–C bioink can perfectly maintain the viscosity stability to a great extent and guarantee the treatment validity.

Printability in simulated in-situ bioprinting

A simulated in-situ bioprinting scene was established to evaluate the printability of A–C bioink from the extruding and deposition states. For extruding state, three environment temperatures (4, 24, 37 °C) and A30/5–C300/20 and C300/20 were selected and extruded at a constant flow rate. (Fig. 4a). At 37 °C, C bioink was in an excessive solization state and form droplets. At 4 °C, C bioink was in an excessive gelation state and had become hydrogel bulk in the syringe, intermittently forming fragment. C bioink showed good printability only at 24 °C and formed uniform filaments. However, A–C bioink, thanks to its great rheological robustness, could form uniform filaments at the three temperatures. For deposition state, the environment temperature was set as 24 °C to achieve the best extruding state. To simulate the patients’ wounds, the receiving platform was set as 37 °C (body temperature) and some food coloring solution (blood) was daubed (Fig. 4b). The deposited A–C bioink could perfectly maintain the 3D structure while C bioink gradually melted and mixed with the “blood”. Thus, A–C bioink would show great adaptation to the complex conditions around wounds. Moreover, A–C bioink was successfully printed with a commercial 3D bioprinter to establish a 3D cube with 12 layers and 7.5 mm height in the roughly controlled environment temperature (30 °C) and on the “wound” receiving basement. (Fig. 4c, Supplementary Notes 5, 9 and Supplementary Video 3).

Binding effect on the defect–hydrogel interface

In in-vitro bioprinting, 3D structures are crosslinked on the deposition platform, resulting in the lack of strong binding force with tissues after transplanting. During therapy, the shifting of the transplanted structures can be invalid or dangerous for patients. By contrast, in-situ deposited A–C bioink would form a strong binding force on the defect/hydrogel interface (Fig. 4d). This is because C component can display fluidity after contacting body temperature and easily infiltrate the defect vacancy, increasing the attaching sites on the interface and the internal friction (Fig. 4e). Furthermore, based on the special chemical properties of GelMA, A–C composite structures can build auxiliary interface force on the wound, which was probably due to photo-generated aldehyde groups bonding with the amino groups on the tissue surface34 (Fig. 4f). To distinctly observe the strong binding force, A30/5–C30/20 bioink was poured on fresh pig tendon (Fig. 4g), while A30/5–C300/20 bioink was poured on fresh pig rib (Fig. 4i). All-direction forces were added to A–C structures (Supplementary Video 4). A–C structures strongly attached to the tissue surface. To explore the binding force between A–C bioink and tissue surface, A30/5–C30/20 and A30/5–C300/20 were poured (about 2 mm height) and photocrosslinked between two pieces of fresh pig tendon (cross-section was about 3 cm × 1 cm) and two pieces of fresh pig ribs (cross-section was about 1.6 cm), respectively. The binding stress was tested by the method of stretching the tissue–hydrogel–tissue structure by clipping two pieces of tissue in the opposite direction at 1 mm/min. The binding stress between A30/5–C30/20 and pig tendon could reach above 4000 Pa (Fig. 4h) and the one between A30/5–C300/20 and pig rib could reach nearly 6000 Pa (Fig. 4j). Two steps occurred on A30/5–C300/20 and rib curve because A30/5–C300/20 was more suitable for compressing and would form crack during the stretching test. All the results indicated that A–C bioink would meet the requirement of strong binding force.

Mechanical properties of A–C composite structure

Compared to the traditional method for establishing composite structures, namely, printing strengthen scaffolds followed by casting soft hydrogel, the method based on A–C bioink has obvious superiority. Firstly, the structure design can be more flexible and strengthen scaffold network would spontaneously form around A component. Furthermore, because A/C component are both GelMA possessing carbon–carbon double bonds, during photocrosslinking, the unreacted carbon–carbon double bonds on the surface of GelMA microgels in electrospraying would break up and connect with the broken ones in C component, forming strong covalent bonds on A/C interface (Fig. 5a), which is absent in traditional methods.

The compressive properties of A30/5–C300/20, C300/20, A30/5–C30/5, C30/5 and tensile properties of A30/5–C30/20, C30/20, A30/5–C30/5, C30/5 were tested. The compressive modulus was 204.00, 2608.00, 16.26, and 3.73 kPa, respectively (Fig. 5b) and the Young’s modulus was 1.81, 11.98, 0.80, and 1.26 kPa, respectively (Fig. 5d), indicating C component obviously strengthened the mechanical properties of the printed structure, which was also proved by the visual observation (Fig. 5c–e).

The stress distribution inside A–C composite structure was analyzed with finite-element simulation. The structure was simplified as an HCP model (Fig. 5f). The displacement boundary condition in the tensile/compressive model was set as 50%/10%, respectively (Fig. 5g, h). Compared to C structure, high Von Mises stress distributed among the microgels in A–C composite structure, just like a strong scaffold “printed” inside, which made it possible to provide extracellular matrix (ECM) with similar mechanical environment in different A–C bioink types. By analogy with the unit cell research method in crystal chemistry, we innovatively set up the concept of “A–C unit cell” with the same incision method as HCP, displaying regular high-stress network packed the low-stress microgels inside both in compressive/tensile models.

Osteogenesis of BMSCs in A component

A30/5 encapsulating bone marrow stromal cells (BMSCs) were electrosprayed and cultured to be further used as functional cell therapy units in A–C bioink. The cell viability on 1st, 4th, and 7th day showed to be above 90% with the LIVE/DEAD kit (Fig. 6a), indicating the basic biocompatibility of A30/5 and F-actin morphology displayed that BMSCs could gradually spread inside the 3D microenvironment (Fig. 6b). Furthermore, some A30/5 encapsulating BMSCs were cultured in osteogenic induction medium after three-day basic culturing. On the 7th day of induction, A30/5 was stained with alkaline phosphatase (ALP) and showed that the induced BMSCs had entered the early osteogenic stage (Fig. 6c). On the 21st day of the induction, A30/5 was stained with Alizarin Red S (ARS) and showed that the induced BMSCs had entered the late osteogenic stage and produced calcium nodules inside the A30/5 (Fig. 6d), which verified its osteogenic differentiation ability and potential therapy effect. Moreover, A30/5 with BMSCs would experience shear force from printing nozzle during in-situ printing, which could cause cell apoptosis. From the results of apoptosis testing with flow cytometry (Fig. 6e), apoptosis caused by extrusion were not obvious, demonstrating the soft environment formed by A30/5 could protest the encapsulated BMSCs from shear force during extrusion and guarantee the biological function.

Bone regeneration in cranial defects

To examine the therapeutic action of A–C bioink, A30/5–C300/20 bioink with or without BMSCs was deposited directly inside the rat cranial defect (column: diameter 5 mm and height 1.5 mm) and photocrosslinked (Fig. 7a). At 2nd week, micro-computed tomography revealed new bone formation in BMSC-loaded A–C group (Fig. 7c). However, no obvious new bone was formed in blank group, and BMSC-unloaded group showed a very limited amount of bone regeneration. It was probably because hydrogels acted as a scaffold for relevant cells on the original defect location and provided more growing space. Furthermore, BMSC-loaded A–C group exhibited better bone regeneration efficacy with a higher BV/TV (Fig. 7b). At 4th week, the bone almost completely bridged the injured site in BMSC-loaded A–C group, and BMSC-unloaded group also exhibited more new bone tissue growth. The BV/TV values increased with time in all groups and both BMSC-loaded and BMSC-unloaded groups were significantly higher than that of blank group. Consistent with radiographic examination, histological analysis with hematoxylin and eosin (Fig. 7d) and Masson trichrome staining (Fig. 7e) at 2nd weeks revealed new bone formation with the largest surface area in the BMSC-loaded group sample, and no marked bone formation in blank and BMSC-unloaded groups. At 4th week, bone formation increased in all groups. Histological observations at higher magnifications confirmed that the neo-formed bone with typical structure in BMSC-loaded group showed many more regions of new mature bone formation than the other groups, indicating BMSC-loaded A–C bioink could promote endogenous bone formation in a critical-size rat cranial defect model.

In-situ cranial repair of different defect morphologies

The actual clinical cases of organ defect are caused by all kinds of accidents, such as fires, traffic accidents, and military injuries. Thus, the 3D morphologies and sizes of the organ defects are very different. To examine the in situ bioprinting capability of A–C bioink in a clinical setting, four rat “patients” with cranial defects with approximately rectangular, square, trapezoid, and triangular shapes (1.5 mm height) were created with a dental trephine (Fig. 8a–c). The robotic arm system was selected as the in-situ bioprinting tool (Fig. 8b). Four rat “patients” were placed on the operating table. The 3D models of the defects were rebuilt with computer-aided design software, and the printing routine program was generated with slicing software and loaded into the controlling system of the robotic arm. A30/5–C300/20 bioink encapsulated BMSCs were in-situ deposited into the cranial defect of four “patients” and photocrosslinked with 405-nm blue light (Supplementary Video 5). After 6 weeks of implantation, micro-computed tomography revealed that new bone was formed from the edge toward the center of the defects in all “patients” (Fig. 8d, e), which verified the high feasibility of A–C bioink in in-situ bioprinting therapy.