Intra-Operative Bioprinting of Hard, Soft, and Hard/Soft Composite Tissues for Craniomaxillofacial Reconstruction

Abstract

Reconstruction of complex craniomaxillofacial (CMF) defects is challenging due to the highly organized layering of multiple tissue types. Such compartmentalization necessitates the precise and effective use of cells and other biologics to recapitulate the native tissue anatomy. In this study, intra-operative bioprinting (IOB) of different CMF tissues, including bone, skin, and composite (hard/soft) tissues, is demonstrated directly on rats in a surgical setting. A novel extrudable osteogenic hard tissue ink is introduced, which induced substantial bone regeneration, with ≈80% bone coverage area of calvarial defects in 6 weeks. Using droplet-based bioprinting, the soft tissue ink accelerated the reconstruction of full-thickness skin defects and facilitated up to 60% wound closure in 6 days. Most importantly, the use of a hybrid IOB approach is unveiled to reconstitute hard/soft composite tissues in a stratified arrangement with controlled spatial bioink deposition conforming the shape of a new composite defect model, which resulted in ≈80% skin wound closure in 10 days and 50% bone coverage area at Week 6. The presented approach will be absolutely unique in the clinical realm of CMF defects and will have a significant impact on translating bioprinting technologies into the clinic in the future.

Graphical Abstract

In this work, intra-operative bioprinting of different craniomaxillofacial tissues, including skin (soft), bone (hard), and composite (hard/soft) tissues, is demonstrated on rats in a surgical setting. The newly developed extrudable bioink supports osteogenic differentiation of progenitors and induces bone tissue formation in vivo. Furthermore, with droplet-based bioprinting, the soft-tissue bioink facilitates faster wound healing.

1. Introduction

Craniomaxillofacial (CMF) defects present challenging reconstructive dilemma for surgeons. For example, stroke can lead to surgery and affected 4.3 million adults in the United States (US) between 1999 and 2008 with 14.5 per 10 000 patients requiring decompressive craniotomy for elevated intracranial pressure.^[1] CMF injuries like these often require multiple surgeries for wound care treatment and brain protection. These surgeries serve to provide functional restoration of the blood-brain barrier, to re-establish the protective cranial skeleton and to create vascularized closure of the scalp while addressing aesthetic concerns.^[2] Several approaches have been utilized for the regeneration of bone (i.e., autografts,^[3] allografts,^[4] and ceramic and polymeric scaffolds^[5]) and skin (i.e., autografts, decellularized graft-like commercial products and polymeric scaffolds)^[6] to address these defects. However, when these approaches are used in clinical settings, multiple problems limit their effective use, including i) infections and pathogenic immune responses pertaining to allografts,^[7] ii) donor site morbidity and limited availability of autograft tissue, especially bone,^[3] and iii) poor control at the hard/soft tissue interface in the case of scaffolds fabricated via conventional approaches that limit their effective use.^[8] The most challenging clinical dilemma remains the repair of composite CMF defects. These defects necessitate the precise placement of multiple tissues in a stratified manner to recapitulate the native tissue anatomy and function. Surgeons continue to struggle with reconstruction of these defects using traditional approaches.^[8]

Bioprinting directly into an injured site during a surgical intervention is termed “intra-operative bioprinting (IOB),” also known as in-situ or in-vivo bioprinting. IOB is a highly promising technology in which the defect information can be rapidly acquired through 3D scanning and then repaired via bioprinting on a live subject within the same surgical intervention.^[9,10] In patients needing tissue resection, debridement, or anatomic reduction of a fracture, the ability to scan and bioprint immediately following surgical preparation of the defect site has great potential to improve the precision and efficiency of these procedures. Furthermore, IOB aims to reduce the need for extensive and expensive procedures such as culturing of scaffolds in vitro or refining the shape of prefabricated scaffolds in the operating room to accurately conform to the defect. These procedures can deteriorate structural integrity of constructs with extensive handling, deactivate biologics and may lead to breaks in sterility. IOB can reduce the manual interventions and eliminate swelling and change in morphology during the construct fabrication, by immediately delivering constructs to the defect site for accurate personalized reconstruction.^[11] In the case of reconstructing hard/soft composite tissues, a stratified organization of multiple tissue compartments with fine control on their interfaces is difficult to engineer.^[12] Thus, IOB could be an effective approach enabling complex tissue heterogeneity in an anatomically accurate and cosmetically appropriate manner.^[13]

IOB method has been previously demonstrated to repair calvarial bone defects using laser-based bioprinting (LBB),^[14,15] where nano-hydroxyapatite (nHA) combined with collagen was used as an ink and deposited into the cranial defects of mice, induced ≈10% and ≈15% bone formation in 1 and 2 months post printing, respectively.^[15] In general, the LBB method is not a rapid process, where the deposition of biomaterials into a defect side is exceptionally slow. Due to the complexity of the laser set-up and its impractical nature during surgical settings, its potential for clinical translation can be limited. Additionally, repair of skin,^[16] cartilage,^[17] and muscle^[18] defects using a handheld bioprinting system was demonstrated by directly delivering bioinks into the injury site. Despite manual filling can be utilized for certain applications such as dental filling, advanced IOB technologies will be essential for precise reconstruction of CMF defects. Thus, handheld bioprinter systems may not be ideal due to their inherent limitations such as non-uniform deposition, inaccurate stacking of multi-layer constructs, and most importantly, the lack of precision in reconstruction of anatomically-correct shapes.^[19] The complex shape of the skull requires precision, which would be extremely suitable for a bioprinting approach. The various contour zones of the cranium, where convexities blend into concavities present unique reconstructive challenges especially when symmetry with the unaffected contralateral side is demanded. More importantly, stratified arrangement of highly thin soft tissue layers (where layer thicknesses are in the order of a few hundred microns) in the CMF area necessitate the use of advanced technologies, such as IOB, for successful restoration of the tissue. In addition, such precise stratified arrangement also supports successful crosslinking of the bioink in situ as accomplished in the presented study. IOB has been used previously for dorsal skin reconstruction in mice and porcine models;^[9,16,20] however, the use of extrusion heads in those studies limit the bioprinting resolution, which is highly critical, particularly for reconstruction of hard/soft tissue interfaces.

Ideal bioinks for IOB are anticipated to maintain their shape and mechanical integrity at physiological temperature when delivered to the defect site.^[21] To date, different types of bioinks have been used for IOB.^[13,19] For instance, a ultraviolet (UV)-crosslinkable bioink has been directly extruded into osteochondral defects in rabbits allowing photo-polymerization in situ.^[22] Despite the quick crosslinking ability of bioink, photo-polymerization is not preferable as UV may damage the exposed surrounding tissue site.^[21] The other demonstrated bioinks relied on the ionic- or enzymatic-crosslinking.^[9,16,20] The ionic- or enzymatic-crosslinking may not be desirable for IOB utilizing an extrusion head, because accumulated blood in the defect site may interfere with the bioink at the nozzle tip causing nozzle clogging and non-uniform crosslinking problems.^[23,24] Collagen, a thermally-crosslinkable hydrogel, was also utilized for IOB due to its favorable properties in supporting tissue regeneration;^[16,25,26] however, its deposition in situ is not trivial to control.

2. Results and Discussion

In this study, we explored IOB for the reconstruction of bone, skin, and composite (hard/soft) tissues, in the CMF zone of 12 weeks-old inbred rats (Figure 1). First, we introduced a new bioink, henceforth named “hard tissue ink (HT-ink),” which is a paste-like shear-thinning bioink enabling extrusion-based bioprinting (EBB) of bone both in vitro and in vivo. HT-ink was constituted of high concentration collagen, chitosan (CS), nHA particles (nHAp), and β-Glycerophosphate disodium salt hydrate (β-GP), where all components were identified in the attenuated total reflection Fourier-transform infrared spectroscopy (ATR-FTIR) indicating no new bond formation (Figure 2a). β-GP was particularly used as an osteogenesis induction reagent for differentiation of mesenchymal stem cells.^[27] Even though any external crosslinker was not required for the HT-ink pre-, during-, or post-bioprinting, we achieved temperature-induced physical crosslinking and a rapid solution-to-gelation phase transition as it is known that CS and β-GP facilitates such when the temperature goes over 34 °C.^[28] The HT-ink was prepared and extruded at room temperature directly onto the heated surface (mimicking body temperature) for crosslinking purposes in vitro. Therefore, abovementioned ingredients improved the bioprintability and shape fidelity of collagen. In addition, collagen fibrillogenesis and its physical crosslinking becomes effective when the collagen is neutralized and at 37 °C by forming thicker and stronger fibrous structure.^[29,30] To determine whether HT-ink was appropriate for EBB, we evaluated its rheological properties. As IOB of the HT-ink was performed without cell encapsulation, the rheology study was performed with the acellular HT-ink. The viscosity (η) decreased with increased shear rate $\left(\dot{\gamma }\right)$ demonstrating the shear thinning behavior of the HT-ink, which is a desirable feature for EBB^[31] (Figure S1a, Supporting Information). The HT-ink showed a dominant storage modulus (G′) and exhibited a clear plateau of G′ and loss modulus (G″), which is commonly referred to as the linear viscoelastic (LVE) region. The LVE was observed at the strain range from 0.01% to 1.78%. Yield stress, or the minimum stress required to initiate the flow during extrusion, was determined to be ≈25 Pa (Figure 2b). Prior to the yield stress, HT-ink exhibited elastic behavior; however, with increasing shear rates, the non-covalent interactions in the gel were broken. G″ exceeded G′ after a crossover point at ≈121% strain, where the HT-ink showed viscous flow behavior thereafter. The frequency sweep test revealed that G′ was higher than G″ at the frequency range of 0.1–20 Hz (Figure S1b, Supporting Information). In addition, both G′ and G″ were strongly dependent on frequency at higher frequencies and the complex viscosity decreased linearly with increasing frequency from 0.01 to 20 Hz (Figure S1c, Supporting Information). After determining the rheological behavior of the HT-ink, we optimized the extrusion parameters and tested the response of the HT-ink under different extrusion pressure (ranging from 80 to 140 kPa) with a bioprinting speed of 400 mm min⁻¹ and a nozzle tip diameter of 410 μm (as listed in Table S1, Supporting Information) using an in-house developed Multi-arm BioPrinter (MABP).^[32](Figure S2, Supporting Information). In order to test the printability of the HT-ink, bioprinting was performed in a grid pattern as individual filaments were distinguishable compared to those in bulk constructs. The HT-ink yielded highly reproducible results with well-defined filaments in moderate fidelity without requiring an external crosslinker (Figures 2c,d; Video S1, Supporting Information). The printability of two-layer constructs resulted in a circularity degree of 0.798 ± 0.06, which was close to that for a perfect squared shape (0.785)^[33] (Figure S3, Supporting Information). In order to determine the reproducibility, four-layer bioprinted acellular constructs were used to measure their calcium (Ca) and phosphate (P) content. The inductively-coupled plasma mass spectrometer (ICP-MS) results showed that there was a total mass of 20.47 ± 1.72 (wt%) and 11.87 ± 1.04 (wt%) for Ca and P, respectively (n = 3, Figure 2e). The data displayed that the material distribution among individual constructs was similar. Immediately after fabrication, mechanical properties of acellular cylindrical constructs, bioprinted in a spiral pattern with a 100% infilled density (similar to the defect shape), were evaluated through unconfined uniaxial compression testing. Although most of the constructs did not resist load until 10% strain, beyond that the stress-strain relationship demonstrated a linear region followed by nonlinearly increasing resistance after 40% strain. The constructs demonstrated a Young’s modulus of 8.2 ± 1.4 kPa (n = 3, Figure 2f).

Figure 1.

Intra-operative bioprinting (IOB) for craniomaxillofacial reconstruction. a) General schematic of IOB process using different 3D bioprinting modalities after 3D scanning to reconstruct various anatomically-correct tissue types, including bone, skin, and composite in a surgical setting. b) IOB of bone performed on rats with two critical-sized (5 mm) calvarial defects for a total of three groups including i) empty, ii) HT-ink-only, and iii) HT-ink+rhBMP2 via EBB technology. c) IOB of skin performed on rats with two 6-mm full-thickness skin defects on crania for a total of four groups including i) negative control, ii) ST-ink, iii) rDFs-laden ST-ink, and iv) rDFs-laden ST-ink (for dermis) and ST-ink+KGF (for epidermis) via DBB technology. d) IOB of composite tissues, including bone and skin, performed on rats with two composite defects using the HT-ink+rhBMP2 for bone, ST-ink-only for the barrier, rDFs-laden ST-ink for the dermis and ST-ink+KGF for the epidermis using both EBB and DBB modalities.

Figure 2.

In-vitro bioprinting of the HT-ink. a) Assessment of the biochemical composition via ATR-FTIR spectroscopy for the HT-ink and its composition. b) Rheological analysis of the HT-ink, where G′ and G″ were measured using amplitude sweep test from 0.01% to 1000% strain. c) Dual-layer bioprinted acellular constructs and d) their morphological appearance on SEM. e) Calcium and phosphate measurement within 3D bioprinted acellular constructs using ICP-MS (n = 3). f) The stress-strain relationship and Young’s modulus of 3D bioprinted acellular bone construct under compression (n = 3). g) LIVE/DEAD staining and h) % cell viability of 3D bioprinted rBMSCs-laden constructs on Days 1 and 7 (n = 3). i) Assessment of in-vitro osteogenic differentiation of rBMSCs within 3D bioprinted rBMSCs-laden constructs using RT-PCR demonstrating fold-change in gene expression for RUNX2, ALP, OPN, and OCN at Days 7, 14, and 21 for the following groups: the control group (2D seeded rBMSCs on TCPs cultured with osteogenic medium (O.M.)) and 3D bioprinted rBMSCs-laden bone constructs cultured in growth medium (G.M.) or O.M. j) Immunofluorescence staining of earlier osteogenic differentiation marker, RUNX2 (green) and late-osteogenic marker BSP (cyan) as well as F-ACTIN showing actin cytoskeleton in cells (red) and cell nuclei (blue) on Days 7, 14, and 21 in G.M. and O.M for 3D bioprinted rBMSCs-laden constructs. Error bars indicate mean ± s.d., p* < 0.05, p** < 0.01, and p*** < 0.001.

We then tested the cytocompatibility of the HT-ink by encapsulating primary rat bone marrow stem cells (rBMSCs) at a concentration of 5 million cells mL⁻¹. Primary rBMSC-laden HT-ink was bioprinted into grid structures and rBMSCs spread within the printed filaments (Figure 2g). Cell viability after bioprinting was greater than 90% and then increased to more than 95% in a week (n = 3, Figure 2h). rBMSCs significantly proliferated between Days 4 and 7, which were 6- to 7.5-fold greater than those observed on Day 1 (n = 6, Figure S4, Supporting Information). In addition, we tested the osteogenic potential of the HT-ink, where rBMSCs-laden constructs were cultured in osteogenic medium (O.M.) or growth medium (G.M.). Scanning electron microscopy (SEM) images were taken to visualize the changes in the extra-cellular matrix (ECM) morphology of contructs cultured in 3D in G.M. or O.M. over time, where bone nodules were observed in ECM in both G.M. and O.M. (Figure S5, Supporting Information). Runt-related transcription factor 2 (RUNX2) gene expression was upregulated in 3D cultures (both in G.M. and O.M.) during a 3-week incubation period compared to rBMSCs cultured on tissue culture plates (TCPs) in 2D in O.M. (control) (Figure 2i). We also observed that rBMSCs in 3D bioprinted constructs treated with G.M. exhibited more proliferative behavior as cells tended to migrate out of the constructs and degraded the constructs substantially compared to the constructs treated with O.M. (Figure S6, Supporting Information). At Week 1, early-stage marker alkaline phosphatase (ALP) was upregulated by 17- and 18-fold in 3D G.M. and O.M. compared to the control, respectively. The gene expression of osteopontin (OPN), which is a middle stage osteogenic differentiation marker, showed upregulation in 3D bioprinted bone constructs that were treated with either G.M. or O.M., compared to that of the control group. 3D Constructs treated with O.M. on Days 14 and 21 showed significant increase in osteocalcin (OCN) expression compared to other groups. 3D Constructs in O.M. demonstrated the highest osteogenic gene expression profile. Furthermore, 3D constructs cultured in G.M. or O.M. for 1 week expressed RUNX2 sub-nuclei protein in their ECM (Figure 2j; Figure S6, Supporting Information). On the other hand, bone sialoprotein (BSP), which is expressed during late-stage osteogenic differentation, was used to visualize the deposited mineralized bone matrix of rBMSCs when cultured in G.M. or O.M. The BSP signals were becoming increasingly detectable for both in G.M. and O.M. cultures after 14 days (Figure 2j; Figure S6, Supporting Information). These findings suggest that the HT-ink, even without the presence of O.M., exhibited favorable osteogenic properties for encapsulated rBMSCs including ALP expression before mineralization, OPN upregulation during the proliferative stage and prior to the expression of OCN, and then finally OCN upregulation when the cells differentiated into osteoblasts inducing apatite mineralization in vitro. The mineralized bone matrix deposited during the differentiation of rBMSCs were not quantified due to its interference with the nHAp in the HT-ink. Differentiation pathway of the rBMSCs into the osteogenic cell lineage may vary depending on the micro-environment, age of the cells, differences in the cell population,^[34] which might be the reasons for the OPN upregulation in 3D culture starting at Week 1. Overall, 3D constructs cultured in G.M. supported osteoblastogenesis without requiring any exogenous osteogenic induction.

After thorough understanding of the HT-ink physical properties and its bioprintability, cytocompability, and osteogenic potential, we demonstrated IOB using EBB under aseptic surgical conditions. IOB was initiated using MABP by surgically opening two, 5 mm-diameter calvarial defects (critical size) on a rat skull after the 3D scanning process, which enabled us to bioprint anatomically-correct bone constructs (see Method Section in the Supporting Information for details). Two critical-sized calvarial defect model is a well-established model for bone repair in the literature as repair of a size of one tenth of skull does not heal spontaneously during the lifetime.^[35] As we introduced a new composite defect model and aimed to minimize number of animals used in our study, we preferred to use a small animal model and have two defects per animal, respectively. Of course, a larger single defect of 8-mm on the rat could be used as an alternative.^[36] First, a laser scanner was used to acquire the geometry of two defects on the parietal bone (Video S2, Supporting Information). In our study, 3D scanning was performed first followed by transferring the bioprinting path plan to the bioprinter unit. For future work, scanning and bioprinting can be synergized and integrated seamlessly. The animal head position was maintained fixed during surgeries and the bioprinting process took less than 5 min only. For larger animals, where instability might be a concern (such as due to breathing), a six-axis bioprinter unit^[21] with an advanced motion sensor system^[37] can be utilized. The resolution for the 3D scanner was up to 100 μm with an accuracy of 30–50 μm, which was sufficient to successfully reconstitute the geometric features of 5-mm calvarial defects. We also tested the robustness of the scanning approach for more complex geometries, such as larger infinity-shape calvarial defects (Figure S8, Supporting Information), and the results demonstrated that we successfully scanned such defects and generated a path plan for bioink deposition.

IOB was then performed by bioprinting cylindrical bulk constructs using a spiral deposition pattern with an 100% infilled density to ensure complete filling of each defect, which took <1 min per construct (Figure 3a). In this study, we employed three groups including i) empty defect (control, n = 4), ii) HT-ink-only (n = 6), and iii) HT-ink-loaded with recombinant human bone morphogenetic protein-2 (rhBMP-2) (n = 10) at a concentration of 1 μg rhBMP-2 per defect^[38] (Figure 1b). rhBMP-2 is a commercially available and clinically tested protein and known to enhance bone repair and mineralization.^[39] After bioprinting, the periosteum was closed and skin was sutured. Six weeks post bioprinting, rats were euthanized for the evaluation of bone regeneration.

Figure 3.

IOB of bone for CMF regeneration. a)The HT-ink was extruded directly into critical-size calvarial defects on a rat model in a surgical setting after performing 3D scanning and path planning processes. b) Visualization of newly formed mineralized bone tissue on Week 6 via μCT. c) Characterization of the regenerated bone including bone volume divided by total volume (BV/TV) (%), normalized bone mineral density (BMD) (%), and bone coverage area (%) of empty defect (n = 4), HT-ink (n = 6), and HT-ink+rhBMP-2 (n = 10) groups. d) Histomorphometric characterization of sectioned skulls after decalcification using H&E (Left) and Masson’s Trichrome staining (Right) (HB: host bone; RB: regenerated bone; ST: soft tissue; Triangle: mature bone; Star: immature bone). e) Immunohistochemical (IHC) staining of sectioned skulls for RUNX2 and determination of its fluorescence intensity (n = 9) (orange arrows indicate RUNX2 expressed regions). Error bars indicate mean ± s.d., p* < 0.05, p** < 0.01, and p*** < 0.001.

Micro-computed tomography (μCT) results showed that HT-ink-only and HT-ink+rhBMP-2 groups demonstrated higher bone regeneration by facilitating more mineralized bone tissue formation compared to empty defects, where defects in HT-ink+rhBMP-2 group exhibited ≈80% closure (Figure 3b,c). Empty defects demonstrated limited bone regeneration, restricted to periphery edges of the defects. HT-ink-only and HT-ink+rhBMP-2 groups yielded ≈1.9-fold (p = 0.001) and 1.7-fold (p = 0.003) greater new bone volume to total bone volume (BV/TV) % compared to that for the empty defect group, respectively (Figure 3c (Left)). Normalized bone mineral density (BMD), that is, density of the regenerated bone normalized to the density of native bone, exhibited ≈1.4- and ≈1.5-fold increase in HT-ink-only (p = 0.003) and HT-ink+rhBMP-2 (p = 0.001) groups compared to the empty defects, respectively (Figure 3c (Middle)). In addition, bone coverage area (%) for the HT-ink-only and HT-ink+rhBMP-2 groups showed ≈1.9-fold increase compared to the empty defect group (p = 0.000) (Figure 3c (Right)). The HT-ink improved bone regeneration in vivo; however, we did not observe a significant improvement in bone regeneration when we incorporated rhBMP-2 into HT-ink compared to the HT-ink-only group. This could be due to the fact that the release of rhBMP-2 could be slow in vivo^[40] as our in-vitro release study revealed that rhBMP-2 release from bioprinted constructs at 37 °C exhibited sustained release with no burst effect, where only ≈17% of rhBMP-2 was released in 4 weeks (n = 5, Figure S7a, Supporting Information). The slow release could be due to the high binding affinity of rhBMP-2^[41,42] to nHAp or its physical entrapment^[41] within the network of the HT-ink. Furthermore, our in-vitro degradation study demonstrated a slow degradation profile (Figures S7b,c, Supporting Information), which could contribute to the slow release of rhBMP-2. However, the rhBMP-2 addition into the HT-ink induced more mineralization compared to the empty defect group, which is also visualized using late osteogenic marker of BSP as an indication of mineralization (Figure S9, Supporting Information). These findings suggest that the HT-ink could be used as a carrier for recombinant proteins (and possibly gene-based growth factors) because the ink can facilitate their sustained release overcoming their short half-life. Bone regeneration was further evaluated using Haemotoxylin and Eosin (H&E) and Masson’s Trichrome staining (MTS) (Figure 3d). In H&E images, native bone (NB) was stained in a darker color compared to the regenerated bone (RB). On the other hand, soft tissue (ST) was displayed in pink. HT-ink-only and HT-ink+rhBMP-2 groups demonstrated bridging of the calvarial defect and thicker RB whereas considerable soft tissue formation was observed in the empty defect group. In addition, MTS demonstrated mainly soft tissue formation in the empty defect group and immature bone formation in both HT-ink-only and HT-ink+rhBMP-2 groups. Moreover, immunohistochemistry staining revealed that the bioprinted groups demonstrated significantly higher expression of RUNX2 compared to the empty defect group (n = 9, Figure 3e). Bone regeneration greatly relies on effective vascularization to transport waste, nutrients, and other essential molecules, necessary for accelerated bone tissue repair. In our study, cluster of differentiation 31 (CD31) staining was used to demonstrate blood vessels. At Week 6, the weakest CD31 signal was observed in the empty group, whereas HT-ink groups demonstrated stronger CD31 signal (Figure S10, Supporting Information). The HT-ink+rhBMP-2 group demonstrated the strongest CD31 signal demonstrating better vascularization, where the incorporation of rhBMP-2 would have been accelerated vascular network growth to repair of bone defect.^[43,44]

After successful demonstration of IOB of bone, we explored the use of IOB for full-thickness skin reconstruction. The skin consists of multiple thin stratified layers necessitating a higher bioprinting resolution. Previous studies for soft tissue bioprinting were based on EBB and relied on pressure- or motor-driven mechanisms to generate droplets to deliver them to the defect side(s).^[9,20] Here, we utilized a micro-solenoid valve system to eject droplets of soft tissue ink (ST-ink) with much better control on the droplet size and reproducibility for the soft tissue layers as compartments of skin are highly thin compared to the cranium.^[45] On the other hand, bioprinting of bone tissue constructs was performed with the EBB system as the utilized HT-ink had a higher viscosity and the bioprinting resolution was not considered an issue as the cranium was substantially thick compared to the compartments of skin.

In this study, collagen was combined with fibrinogen to develop the ST-ink. First, viscosity measurements were performed to determine the rheological properties of the ST-ink and its components (Figure 4a). The neutralized collagen and ST-ink showed shear thinning behavior, whereas fibrinogen and thrombin, as well as the DPBS solvent (control), showed Newtonian behavior. Next, skin constructs were bioprinted using a custom-made droplet-based bioprinting (DBB) system (Figure S11, Supporting Information). The concentration of each ink was optimized to minimize clogging of micro-valves during jetting (Table S3, Supporting Information). Further, jetting of the ST-ink and its components with the use of micro-valves was demonstrated (Figure 4b), which subsequently broke up into streams of multiple droplets. Although the viscosity of neutralized collagen was ≈10 times greater than the viscosity of fibrinogen, the two other solutions and the ST-ink formed jets of comparable volume and ejection velocity because of the shear thinning nature of collagen.^[46] The behavior of droplets upon impacting the solid substrate has been described previously by Reynolds, Weber, Ohnesorge, and Bond numbers.^[47–50] The Bond number for droplets was <<1 and hence the impact of gravitational forces was negligible. Thus, the spreading of droplets on solid substrates was primarily driven by inertial or capillary forces. Although each jet broke up into a stream of droplets before landing on the substrate, the droplets did not splash upon the impact and coalesced into a single droplet on the substrate (Figure S12, Supporting Information). Spreading of droplets was primarily driven by the impact pressure and resisted by inertial oscillations (Figure 4c). The ST-ink solution required >10 min for complete polymerization and hence the interactions of droplets with the substrate changed from solid to liquid after deposition of the first layer. The liquid substrate interactions have been described by Leng et al.^[51] in terms of Froude and Weber numbers (Figure 4d). The droplets expelled with velocities closer to their terminal velocity and the deposition of alternating layers of the ST-ink and thrombin solution did not cause significant splashing. Similarly, the bulb and tail segments of the jets of the ST-ink and rat primary dermal fibroblasts (rDFs)-laden ST-ink were expelled at different velocities suggesting the breakup of jets into streams of multiple droplets (n = 3, Figure 4e (Top and Middle)). Further, the droplet volume of the ST-ink and rDFs-laden ST-ink expelled per each actuation voltage pulse decreased with increasing rDFs concentration suggesting that the incorporation of rDFs increased the viscosity of ST-ink (n = 5, Figure 4e (Bottom)). However, no significant splashing of droplets on both solid and liquid substrates was observed during the deposition of rDFs-laden ST-ink. To assess the mechanical properties of 3D bioprinted skin constructs (Figure 4f), compression testing was performed on manually fabricated and bioprinted fibrin and ST-ink constructs (Figure 4g). The bioprinted constructs were fabricated by depositing alternating layers of fibrinogen or ST-ink with thrombin solution. Incorporation of collagen within the ST-ink increased the Young’s modulus although no significant difference was observed among all groups. The Young’s modulus was calculated to be 1.70 ± 0.27 (kPa) and 1.81 ± 0.43 (kPa), and 1.95 ± 0.56 (kPa) and 2.39 ± 0.49 (kPa) for manually fabricated 3D fibrin and ST-ink constructs, and 3D bioprinted fibrin and ST-ink constructs, respectively. Inclusion of collagen in the ST-ink increased the strength, stability, and shape fidelity of bioprinted skin constructs due to the physical crosslinking of collagen at 37 °C.^[52] Additionally, bioprinted constructs, fabricated with fibrin or ST-ink, showed higher Young’s modulus compared to the manually prepared constructs due to the uniformity of bioprinted ink enabling homogenous and stabilized crosslinking within constructs. Constructs fabricated with 5 million rDFs per mL showed higher viability a day after bioprinting (n = 3, Figure 4h) compared to those fabricated with 1 million rDFs per mL. Thus, the ST-ink loaded with 5 million rDFs per ml was used for IOB purposes. The morphological images of the rDFs-laden constructs at Day 1 showed that rDFs attached to the constructs and had a spherical morphology; however, they spread from Day 3 onwards (Figure 4i). The circularity analysis demonstrated a reduction in the circularity of rDFs as their morphology changed over time due to spreading (n = 3, Figure 4j) Figure S10, Supporting Information). In addition, the proliferation rate of rDFs in 2D (control) and 3D environments showed similar growth trend suggesting no cytotoxicity in 3D constructs (n = 5, Figure 4k).

Figure 4.

In-vitro bioprinting of the ST-ink. a) Viscosity measurements for neutralized collagen, fibrinogen, ST-ink, thrombin, and DPBS solvent (control) as a function of shear rate at 25 °C. b) Jetting of ink solutions with a micro-valve device breaking up into streams of multiple droplets upon exiting the nozzle orifice. c) The interactions of droplets with the solid substrate, represented by Weber number of droplets as a function of their Ohnesorge number, drawn following the schematic diagram of Schiaffino and Sonin.^[47] d) The interactions of droplets with the liquid substrate, represented by Weber number of droplets as a function of their Froude number, drawn following the schematic diagram of Leng.^[51] e) The bulb velocity (n = 3), tail velocity (n = 3), and the droplet volume per pulse (n = 5) of the ST-ink with and without encapsulation of rDFs at concentrations of 1 and 5 million cells mL⁻¹ (n = 3, error bars; mean ± s.e.m.). f) A representative image of 3D bioprinted skin construct comprising of ≈20 layers of ST-ink and ≈20 layers of crosslinker (thrombin) solution at 1:1 ratio. g) Stress-strain curve of manually patterned or bioprinted acellular fibrin and ST-ink constructs (n = 3). h) Viability of rDFs within 3D bioprinted construct at 24 h post bioprinting for cell densities of 1 and 5 million rDFs per mL (n = 3). i) Fluorescent images of 3D bioprinted rDFs-laden constructs showing cell morphologies at Days 1, 3, and 5. j) Circularity measurements of rDFs within 3D bioprinted construct at Days 1, 3, and 5 (n = 3). k) Proliferation rate of rDFs cultured in 2D on TCPs and 3D bioprinted constructs from Day 1 through Day 5 (n = 5). Error bars indicate mean ± s.d., p* < 0.05, p** < 0.01, and p*** < 0.001.

In this study, we showed IOB of the full-thickness skin using DBB and evaluated its efficacy for wound repair. Animals of the same strain and age used in the IOB of bone study were used for skin bioprinting purposes. Two 6-mm-wide circular defects on cranium of animals were created (Figure 5a, see Method Section in Supporting Information for details). A total of four groups including, i) negative control (empty defect), ii) ST-ink only, iii) rDFs-laden ST-ink, and iv) rDFs-laden ST-ink in the dermis followed by deposition of keratinocyte growth factor (KGF) (3.3 ng defect⁻¹)^[53] loaded in ST-ink in the epidermis (n = 6) were used (Figure 1c). DBB provided localized delivery of skin constructs and enabled stable crosslinking for both collagen and fibrinogen solutions due to its ability to precisely control the droplet size, deposition rate, and amount of bioink.^[45] Depending on the group type, the total skin IOB process took from 5 to 30 min. After surgery, the defects were photographed every other day until Day 28 (Figure 5b). Morphometric evaluation of wound healing was performed by calculating the percentage of wound closure with respect to the original size of wound (Figure 5c). Compared to the negative control at Day 6 (21%), ST-ink-containing groups (ST-ink, rDFs-laden ST-ink, rDFs-laden ST-ink+KGF) showed faster wound closure: 47%, 52%, and 60%, respectively. Similar to the wound closure results, ST-ink, rDFs-laden ST-ink, and rDFs-laden ST-ink+KGF groups exhibited faster re-epithelialization, 17%, 26%, and 26% at Day 6, compared to the negative control (1%), respectively (n = 6, Figure 5c). The closure of the dermis compartment was accelerated by the presence of bioprinted rDFs, which compensated for the slow migration of fibroblasts from the periphery of defects.^[54,55] Furthermore, we incorporated KGF deposited at the top layer, which was measured to be 200–300 μm in height. KGF plays a critical role in activating cell migration and stimulating the wound closure,^[56] while enhancing the barrier functionality of the repaired wound.^[56–58] Overall, the ST-ink containing groups facilitated an immediate treatment resulting in full scalp closure with anatomically-correct skin morphology in two weeks as opposed to the negative control. During the proliferation phase of wound healing process at the first week,^[59,60] the ST-ink supported rDFs attachment and deposition of ECM, which further stimulated collagenous matrix deposition^[61] (Figure S13, Supporting Information). To evaluate the quality of the repaired wound, we performed H&E staining, demonstrating the overall structure of the repaired wound (Figure 5d). At Week 4, ST-ink containing groups showed well-defined and organized connective tissue with a well-formed dermis layer, while the negative control showed less re-organization in dermis and limited stratification in epidermis layer (Figure S14a, Supporting Information). Next, MTS was performed to demonstrate collagen deposition at Week 4 (Figure S14b, Supporting Information). According to the collagen index determined from the MTS images, the group treated with rDFs-laden ST-ink+KGF represented the most densely packed collagen fibers as opposed to other groups, wherein the addition of rDFs further supported the deposition of collagen fibers compared to the groups lacking rDFs (n = 6, Figure 5e). We also identified blood vessels in all groups as highlighted with arrows in Figure 5d. Further vimentin and loricrin staining was performed to evaluate the presence of fibroblasts in the dermis and keratinocytes in the epidermis at Week 4, respectively^[62,63] (Figure 5d; Figure S14c, Supporting Information). The group without rDFs, including the negative control and ST-ink only group showed lower intensity of vimentin and loricrin. On the other hand, rDFs-laden ST-ink+KGF treated group showed the highest expression of vimentin (n = 4, Figure 5f) and the epidermal barrier protein loricrin (n = 4, Figure 5g). Interestingly, rDFs were densely populated around the epidermis area in the rDFs-laden ST-ink + KGF group. Based on the CD31 staining, we observed vascularization in all groups (Figure S15, Supporting Information) confirming the findings in Figure 5d). Furthermore, the mechanical integrity of the remodeled skin was evaluated using an insertion force test as we described before^[64] (Figure S16, Supporting Information). The rDFs-laden ST-ink+KGF groups showed the highest insertion force indicating the most intact organization of the regenerated skin in that group, which might be due to its faster healing and wound closure characteristics. (n = 7, Figure 5h). This suggests that the repaired skin in the negative control and ST-ink group could be weaker and less resistant to fracture than the other groups. By IOB of the ST-ink into the defects, 1) wounds were fulfilled with necessary components for healing to the fullest extent, 2) bi-layer full-thickness skin was reconstituted, and 3) the immediate coverage of the wound using the ST-ink prevented the desiccation during the healing process.

Figure 5.

IOB of full-thickness skin using DBB. a) IOB of skin in action showing skin defects before and after jetting of the ST-ink. b) Representative wound images showing wound healing until Week 4 for the following groups: i) negative control, ii) ST-ink only, iii) rDFs-laden ST-ink, and iv) rDFs-laden ST-ink+KGF (rDFs-laden ST-ink layer followed by ST-ink-loaded KGF on top). c) Percentage (%) of wound closure (top) and re-epithelization (bottom) (n = 6, error bars; mean ± s.e.m.). d) Histological images of the regenerated skin stained with H&E (Epi: epidermis layer; Der: dermis layer), collagen deposition in the regenerated skin determined using MTS (orange arrows indicate blood vessels), and IHC staining for Vimentin and Loricrin at Week 4. e) Collagen index determined from MTS images (n = 6). f,g) Expression of Vimentin (left) (n = 4) and Loricrin (right) (n = 4) fluorescence intensity. h) The needle insertion test performed on the harvested regenerated skin ex vivo at Week 4 (n = 7, error bars; mean ± s.e.m.). Error bars indicate mean ± s.d., p* < 0.05, p** < 0.01, and p*** < 0.001.

The reconstruction of skin defects on calvaria has been a challenge due to limited scalp mobility, the paucity of native hair bearing scalp available for reconstruction and the high physical and emotional toll associated with multiple surgical operations.^[65–67] Current surgical practice relies on traditional techniques such as the use of tissue expanders, skin grafts, and both local and free flaps.^[66,68,69] However, tissue expanders require staged procedures to be performed over many months to achieve the desired expansion resulting in thinner skin, which is more prone to breakdown. Local flaps, while ideal, are not suitable for defects over a critical size over 5 cm in humans. Skin grafts can typically only be used when intact periosteum is present at the base of the wound. While free-flaps do not require an intact periosteum, they require specialized micro-surgical intervention which can be accompanied by reconstructive failure in as much as 10% of cases and iatrogenic donor site morbidity.^[70,71] Therefore, IOB of skin has the potential to overcome the above shortcomings by precise reconstruction of skin compartments.

We then fabricated composite constructs in vitro using the approach demonstrated in Figure S17a, Supporting Information, where the bone compartment was bioprinted followed by the skin compartment on top using rBMSC-laden HT-ink (at a cell density of 5 × 10⁶ rBMSCs mL⁻¹) and rDFs-laden ST-ink (at a density of 1.33 × 10⁵ rDFs mL⁻¹) using EBB and DBB, respectively. Bioprinted composite constructs, where bone and skin compartments were in direct contact (henceforth named the “direct contact” group) were maintained two weeks in culture. As a control group, bone and skin compartments were individually bioprinted next to each other with a distance apart and cultured in the same media. We analyzed cell proliferation and construct morphologically via SEM. The cell proliferation rate significantly increased from Day 0 to 14 for both groups (p = 0.000) (Figure S17b, Supporting Information). The direct contact group demonstrated a higher cell proliferation rate on Days 4 and 7 with respect to the control group but the cell proliferation rate was similar for both groups on Day 14. The surface morphology of the composite constructs was visualized using SEM, where the bone compartment demonstrated apparent peaks and valleys whereas the skin comportment exhibited reduced unevenness on the surface morphology (Figure S17c, Supporting Information). The clear boundary between both compartments was apparent through Day 14 as extensive fibrous network was deposited in the skin compartment, which can be attributed to the deposition of ECM from rDFs in the ST-ink as fibrin acts as a linkage protein between collagen type-I fibers and cells.^[72]

Upon demonstration of bone- and skin-only reconstruction and in-vitro evaluation of composite constructs, we explored the potential of IOB in the reconstruction of composite (hard/soft) tissues for the first time in the literature (Figure 1d). The conditions yielding the highest level of newly mineralized bone tissue and the fastest wound healing in bone- and skin-only defects were used for the composite tissue bioprinting, respectively. The impact of IOB was evaluated by surgically generating a novel composite defect model by first generating two 6-mm full-thickness skin defects on the cranium with a biopsy skin punch and then concentrically drilling two 5-mm critical-size calvarial defects directly below the skin into the parietal bone on rat skulls (Figure 6a; see Method Section in Supporting Information for details). The rationale for having the skin defect larger than that of bone is twofold: i) to delineate the skin thickness from that of bone during laser scanning and ii) to concentrically align the skin defect with respect to the bone defect as identified during our surgical study. Next, the following bioprinting process was performed: i) first, the cranium layer was bioprinted with the HT-ink loaded with 1 μg rhBMP-2/defect using EBB. ii) Then, a barrier layer made of ST-ink was bioprinted using DBB, iii) followed by bioprinting the dermis layer using the ST-ink loaded with rDFs (5 × 10⁶ cells mL⁻¹). iv) Finally, the ST-ink (loaded with 3.3 ng KGF/defect) was bioprinted to create the epidermis layer after the dermis layer was fully crosslinked (Figure 1d; Video S3, Supporting Information). In our experiments, no empty defect was used as a control group as such was unacceptably risky for rats due to the direct exposure of the brain with the outside environment. The rationale for the barrier layer was to minimize the chance of sedimentation of rDFs in the cranium layer due to gravity. In normal anatomy, different tissue interfaces are typically accompanied by an interface or barrier. For example, at the scalp from outward in, there is skin, dense connective tissue, aponeurosis, loose areolar connective tissue, periosteum, and bone. Our simplified barrier in this work was a representative interface between soft and hard tissue. Durable and safe reconstruction of a cranial defect is not achieved with only one tissue. Both bone and soft tissue are necessary to allow patients to safely reengage into society.

Figure 6.

IOB of composite tissues. a) Bioprinting of bone and skin constructs in a stratified manner on the cranium of 12-week old rats. b) After bioprinting of HT-ink+rhBMP-2 using EBB, a barrier layer made of ST-ink was deposited via DBB followed by bioprinting of the dermis with rDFs-laden ST-ink and then the epidermis with ST-ink+KGF via DBB. c) The Tegaderm was maintained for the first 12 days in order to protect the defect. After removing Tegaderm on Day 12, wound images were acquired to characterize percentages of wound closure and re-epithelization up to four weeks (n = 6, error bars; mean ± s.e.m.). d) Bone regeneration determined using μCT at Week 3 (n = 4) and Week 6 (n = 6) with BV/TV (%), normalized BMD (%) and bone coverage area (%). e) Histological and f) immunohistochemical characterization of the harvested composite tissues. Cytokeratin-10 (KT-10) and Vimentin were used to indicate epidermis and dermis layers, respectively, and RUNX2 was used to indicate pre-mature osteoblasts in the bone region at Weeks 3 and 6 (HB: host bone; RB: regenerated bone; NS: native skin; Triangle: mature bone; Epi; epidermis; Der; dermis). g) Collagen index of skin measured from MTS images at Weeks 3 and 6 (n = 6). h) Epidermis thickness measured from IHC stained images at Week 3 and 6 (n = 10). i) RUNX2 fluorescence intensity determined as the mean gray value from the stained bone samples (n = 9). Error bars indicate mean ± s.d., p* < 0.05, p** < 0.01, and p*** < 0.001.

Skin repair was morphometrically evaluated by taking the images of wounds every other day starting from Day 12 (after removing the Tegaderm) to Day 28 (Figure 6b). On Day 12, a wound closure of 82% was observed; which reached to 100% by Day 20 (Figure 6c (Left)). Further, 67% of re-epithelization was observed on Day 12 (Figure 6c (Right)). Overall, wound closure and re-epithelization rates were slightly lower than those for the skin-only group discussed previously. However, the appearance of the wound images in composite tissue was similar to that of skin-only defects. Three or 6 weeks after the surgery, bone regeneration was evaluated, where we achieved substantial bone regeneration in 6 weeks and obtained ≈2.3-fold bone regeneration BV/TV (%), ≈1.6-fold normalized BMD (%), and ≈1.4-fold bone coverage area (%) compared to those in 3 weeks (Figure 6d). To evaluate the improvement in skin repair and bone regeneration over time, we performed histological assays at Weeks 3 and 6 (Figure 6e,f). H&E images demonstrated that the skin was well-reorganized in 3 weeks, and the epidermis and dermis layers were completely regenerated in 6 weeks (Figure 6e). The bone defect showed increased bone density at Week 6 compared to Week 3. MTS showed that collagen infiltration in skin was remarkably increased in Week 6 (Figure 6e,g). The epithelial layer became thinner and reached to the thickness of the native skin through KT-10 immuno-histology^[63,73] (Figure 6f–h). Bone tissue formation further proceeded from Week 3 to 6, where the regenerated bone at Week 6 expressed stronger RUNX2 signal (Figure 6f–i). Additionally, we observed vascularization in the reconstructed composite tissue (Figure S18, Supporting Information); however, the CD31 staining was weaker compared to that in bone- and skin-only studies.

In this study, we developed a new defect model, and for the first time, investigate the repair of composite CMF defects via IOB combining multiple different bioprinting techniques. IOB of composite tissues demonstrated that skin repair in skin-only and composite defects displayed a similar healing trend and we observed complete skin repair in 4 weeks in both cases. On the other hand, bone healing in composite defects displayed inherently poor regeneration compared to that in bone-only defects. This could be due to several reasons. For example, the presence of the periosteum might contribute to bone regeneration in bone-only defects as previous studies reported that progenitors residing in periosteum contribute to cranial regeneration.^[74] In addition, the hypodermis compartment of skin has blood vessels,^[75] which could feed the underlying periosteum and cranium tissue in the bone-only defect model; thus, the composite defect model may not facilitate similar level of vascularization. For future work, induction of stronger vascularization and reconstruction of other crucial layers, such as periosteum and adipose layer of hypodermis, would be critical to fully restore the composite skin defects while improving healing outcome.^[76]

The clinical implication of our model is highly novel as there is currently no approach to address both skin and bone repair at the same time in clinics. The complete reconstruction of a full-thickness cranial defect is approached by the surgeon in terms of bone and soft tissue. Taken independently, the defect “covered” with skin/soft tissue only does not provide protection from impact. These patients must guard against falls or excessive contact by wearing helmets. Compliance can be challenging especially in children with devastating consequences. Conversely, reconstruction with bone only is not biologically stable. Exposed bone or bone construct is at constant risk of infection which when crossing the Dura is life threatening. The presented approach eliminates the infective risk to the patient while providing the structural protection needed to safely function in society. Although the presented approach may not be directly applicable to all types of tissue deficits and large injuries, there are cases, such as aplasia cutis congenita, where our approach could bring significant advantages. The applicability of our approach in this patient population would be a significant advance in that time to wound closure would be significantly quicker while offering immediate protection to the underlying brain and sagittal sinus. Surgical correction of craniosynostosis by definition has cranial defects at the conclusion of the surgery. The majority of cranial defects are cephalad and do not have any muscle in proximity. The periosteum is also not able to cover these defects due to the bony expansion. It takes 1–2 years in most cases to have spontaneous bony ingrowth to cover these exposed areas. For example, recently there was a report on an infant, who had a 9 × 10 cm defect of skin and underlying bone.^[77] These wounds are typically treated with conservative measures, such as dressing changes, over a prolonged time period until healing is achieved. It is easy to appreciate our findings in this dilemma, where the underlying entity has significant intrinsic regenerative potential. However, our approach could lead to a decreased time to healing, significantly lower the potential morbidity of open wounds, and alleviate the family of cumbersome dressing changes. Further, this condition has a mortality rate as high as 55% when there is a sentinel bleed from the wound. This risk is alleviated once durable coverage has been achieved.

Although the presented defect model is in critical size (for bone), for segmental defects, structural and mechanical support may also be critical to support bone regeneration^[78] and protection of the brain temporarily.^[79] Reconstruction of full-thickness composite defects is usually performed with myocutaneous free flaps, which provide a biologic barrier to infection and provide the bulk of muscle to fill the defect. In such a situation, there is no bony reconstruction used and the muscle sits directly over the Dura. In such a scenario, the weight placed on the brain is clinically insignificant. Postoperatively, there are no clinical ramifications of the bulky muscle and skin flap directly over-lying the brain in these defects. In fact, many older patients opt to never go on to bony reconstruction once the flap is healed and they would then be a candidate. Our 3D bioprinted composite tissue puts a similar mechanical stress on the brain.

3. Conclusion

In this work, we presented IOB of hard, soft, hard/soft composite tissues for CMF reconstruction. In this regard, we demonstrated the preparation of bioinks for both bone and skin regeneration by incorporation of allogenic cells and biomaterials. For future clinical translation of the presented technology, the bioinks should be produced from human autologous cells, compatible materials at the clinical grade. Additionally, the incorporation of artificial intelligence and human-machine system interface with fully-automated defect scanning and bioprinting ability can improve the outcomes, which can further advance the translational studies as well as providing upscaling to repair of larger defects in a rapid manner.

4. Experimental Section

Experimental section can be found in the Supporting Information.