In extrusion-based bioprinting, a mechanical or pneumatic fluid dispensing system is used to force the bioink through the nozzle, resulting in a continuous filament (Cui et al., 2010; Jungst et al., 2016; Moroni et al., 2018a). The computer controls the 3D movement of the printhead in order to print in a layer-by-layer fashion according to the CAD files on a stationary printbed (Cui et al., 2010; Jungst et al., 2016). In mechanical-driven systems, a screw or piston applies the driving force allowing precise control of the extruded volume (Malda et al., 2013; Ozbolat and Hospodiuk, 2016; Wenger et al., 2022). In the former case, rotational mechanical forces are directly applied on the ink by a screw connected to the motor (Gu et al., 2020). In the latter case, the ink is extruded by linear mechanical forces exerted by the piston connected via a guide screw to the motor (Gu et al., 2020). The pneumatic-driven system applies compressed air (5–800 kPa) on the bioink (Cui et al., 2010). This approach has less control of the extruded volume as it depends on the applied pressure as well as on the rheological properties of the ink and the printing set-up (Jungst et al., 2016). Sterilization of the air via a filter is required when the air is directly applied onto the cell-laden ink (Gu et al., 2020).

During the printing process, the cells experience shear, compressive and extensional forces reducing the cell viability (80%–90%) (Chang et al., 2008; Hölzl et al., 2016; Ning et al., 2020; Xu H. et al., 2022). The forces exerted on cells in the pneumatic dispensing system are similar to those in the piston dispensing system (Ning et al., 2020). In both groups, the cells experience shear stress in the nozzle and extensional stress at regions from the needle cartridge to the needle tip (Ning et al., 2020). The screw-based system exerts additional shear stress on the encapsulated cells due to the direct ink-screw contact (Ning et al., 2020). The shear/extensional force is the dominant force causing cell damage and cell death (Paxton et al., 2017; Cidonio et al., 2019b; Boularaoui et al., 2020; Ning et al., 2020). The shear stress can be modified by changing the nozzle diameter/length, nozzle shape, printing pressure, print head speed and ink viscosity (Billiet et al., 2014; Boularaoui et al., 2020; Ning et al., 2020; Schwab et al., 2020). Ning et al. concluded that the screw-based system induces greater cell damage than the pneumatic/piston-based system making the former less suitable for biofabrication (Ning et al., 2020). Despite the risk of cell damage and cell death, shear stress within a specific range (and other mechanical forces) are biophysical cues inducing the differentiation of stem cells into specific lineages (Moehlenbrock et al., 2006; White and Frangos, 2007; Zhao et al., 2007; Dong et al., 2009; Wong et al., 2012; Boularaoui et al., 2020). When bone marrow-derived stem cells (BMSCs) are exposed to fluid flow induced-shear stress, osteogenic differentiation is induced (Yourek et al., 2010). In contrast, Blaeser et al. reported an unaltered mesenchymal stem cell phenotype during microvalve-based bioprinting upon exposure to shear stress below 15–20 kPa (Blaeser et al., 2016). Therefore, additional research is needed to determine the impact of extrusion-based bioprinting on the stem cell phenotype.

Extrusion-based technologies are promising for biofabrication. Similar to drop-on-demand inkjet printing (DoD), multiple nozzles and different inks can be combined into a heterocellular, multi-material construct. A broad range of biomaterials are compatible with extrusion-based bioprinting having a viscosity window ranging from 30 mPa.s up to 6 × 10⁷ mPa.s (Chang et al., 2011). Even higher viscosities are compatible with the printing process when a mechanical dispensing system is used (Habib et al., 2018). The used hydrogels regularly exhibit shear thinning behavior, resulting in a decreasing viscosity with increasing shear rate. Hence, when a pressure is applied during printing, the viscosity drops, allowing a smooth extrusion. Upon deposition, the shear rates drop drastically, resulting in an increasing viscosity and the preservation of the extruded shape (Chimene et al., 2016; Boularaoui et al., 2020). Extrusion-based bioprinting can be applied with bioinks encapsulating high (single) cell densities (∼10⁸ cells/mL) and spheroids, allowing printing of physiological cell densities in a hydrogel scaffold (Murphy and Atala, 2014; Diamantides et al., 2019; De Moor et al., 2021; Shao et al., 2021). Additionally, the speed can range from 2 up to 60 mm/s depending on the used system (Tarassoli et al., 2021).

Challenges associated with extrusion-based bioprinting are related to sedimentation, clogging, lack of reproducibility and (relatively) low resolution. Sedimentation of the encapsulated cells influenced by the ink’s viscosity, the density of cells and the cell-adhesion site distribution results in an inhomogeneous cell distribution (Chen et al., 2019). This is specifically valid when employing low viscosity inks and large printing times. Additionally, the low viscosity results in poor mechanical strength, hence, collapse of a multi-layered structure (Yin et al., 2018; Chen et al., 2019). Conversely, a too high viscosity results in high shear stresses, inducing cell damage and cell death. Hence, the viscosity should be carefully tuned to prevent both sedimentation and cell death/damage. Secondly, clogging caused by the accumulation of cells, particles or solidified material obstructs the ink flow through the nozzle (Shao et al., 2021). A third limitation is the sensitivity of the printing process/parameters to environmental parameters including temperature and humidity as well as batch-to-batch variability (Wenger et al., 2022). While the environmental variations can be excluded by printing in a temperature-humidity controlled room, the batch-to-batch variability requires the identification of working windows of the printing parameters including pressure, nozzle/printbed temperature, print-speed, and layer height, amongst others. Finally, the general resolution is low as compared to other biofabrication technologies (200–1,000 µm) (Hölzl et al., 2016).

Inkjet bioprinting, a deposition-based biofabrication technique, entails the precise deposition of cell-laden droplets according to a computer-aided-design (CAD), thereby resulting in 3D cellular constructs (Lorber et al., 2013; Xu et al., 2013; 2019). Continuous inkjet printing and DoD are the main types of inkjet printing (Li et al., 2020). In the former type, a piezoelectric crystal causes the nozzle to vibrate, ensuring a continuous stream of ink through the nozzle (Li et al., 2015). Droplets are continuously formed according to the Rayleigh-Plateau instability, even though the droplets are not contributing to the print (Li et al., 2020). A potential difference between the nozzle and the substrate charges the droplets, enabling their deflection when passing through charged deflectors (Derby, 2010). In this way, the unneeded droplets are separated from the desired ones (Derby, 2010). Subsequently, this captured ink is sent back to the printhead to be re-used (Derby, 2010; Saunders and Derby, 2014; Alamán et al., 2016; Kumar et al., 2021). The final droplet position is regulated by controlling the movement of the droplets and the position of the substrate (Derby, 2010). The use of continuous inkjet bioprinting is limited due to the printer’s complexity (i.e. droplet charging, deflection and recycling system) and the contamination risk if droplets are re-used (Saunders and Derby, 2014; Li et al., 2015; 2020). DoD is another type of non-contact deposition-based printing exclusively producing droplets when the actuator is activated (Romagnoli et al., 2016; Li et al., 2020). The actuator induces a thermally or mechanically generated pressure pulse resulting in picolitre droplets with a 15–100 µm diameter and a long tail rupturing into the primary droplet followed by satellite droplets (Wallace and Grove, 2003; Derby, 2010; Romagnoli et al., 2016; Li et al., 2020). If the droplets are not merged prior to the impact on the substrate, a non-circular impact is caused lowering the resolution and accuracy (Derby, 2010). This phenomenon, called droplet splashing, is also caused by the high-speed droplet impact and should be controlled when printing µm-scale constructs (Li et al., 2020). When the actuator is not activated, the fluid remains within the fluid chamber due to surface tension (Derby, 2010). Thermal and piezoelectric DoD are the most prevalent inkjet techniques (Li et al., 2015).

DoD exhibits potential for biofabrication applications due to its high throughput, non-contact and drop-on-demand printing. The maximal throughput depends on the number of nozzles (up to hundreds) and the ejection frequency (up to 250 kHz) and can be up to 80 mL/h/printhead (Xu et al., 2006; Wijshoff, 2010; Cui et al., 2012a; Li et al., 2020). By using multiple nozzles, diverse bioinks with different cell types can be printed within a single construct. Non-contact printing lowers the contamination risk hence allows in situ printing (Cui et al., 2012b; Li et al., 2020). Moreover, it prevents the deformation of previously deposited structures (Li et al., 2020). Lastly, the computer-controlled drop-on-demand printing allows precise spatial and temporal control. However, its use is limited due to the low viscosity requirement (<10 mPa.s) to avoid clogging and the low cell concentration (∼1 million cells/mL) (Murphy and Atala, 2014; Hölzl et al., 2016; Li et al., 2020). An important limitation due to the low ink viscosity is cell sedimentation resulting in an increase in cell density at the bottom of the printhead and subsequently cell aggregation (Liu et al., 1970; Xu H. et al., 2022). This phenomenon results in a non-uniform cell distribution, unstable droplet formation and nozzle clogging (Lorber et al., 2013; Xu H. et al., 2022). Different solutions have been applied associated with pros and cons including active bioink stirring, bioink manipulation to obtain neutral buoyance and active bioink circulation (Liu et al., 2022; 2023; Xu H. et al., 2022; Liu and Xu, 2024). Additionally, cells adhere to the inner surfaces of the printing set-up by Van der Waals forces, resulting in constriction and clogging as well as a lower cell number with respect to the theoretical number (Dersoir et al., 2015; Sendekie and Bacchin, 2016; Ng and Shkolnikov, 2024).

Both printing methods, stereolithography (SLA) and digital light processing (DLP), project UV- or visible light patterns of the discretized and sliced CAD in a point-by-point and layer-by-layer fashion respectively onto the photo-crosslinkable resin. After one layer is finished, the motorized build platform moves away vertically to allow the uncured resin to flow back whereafter the print head is repositioned to allow crosslinking of the subsequent layer. These printing techniques mainly differ in the way the light is patterned with SLA using raster laser scanning whereas DLP uses either a digital mirror device (DMD) or a liquid crystal display (LCD) projection system (Li W. et al., 2023). Overall, the outcome of the bioprinting processes DLP and SLA in terms of printability, printing time, attainable sample size, resolution, shape fidelity and print stability is mainly affected by the constituents of the bioresin (i.e. photo-crosslinkable polymer concentration and reactivity, cell type and concentration, concentration and efficiency of photo-initiators, -absorbers and/or -inhibitors), the light projection method, sample post-processing and the delivered light dose to the resin through variations in the light intensity and exposure time (Liang et al., 2021; Goodarzi Hosseinabadi et al., 2022; Levato et al., 2023; Li W. et al., 2023).

The LCD/DMD of the DLP can cure an entire layer at once, making the DLP process faster than the point-by-point crosslinking associated with the more conventional SLA process resulting in a printing time in the order of minutes with DLP rather than minutes to hours with SLA to build a 1 cm³ construct (Liang et al., 2021; Levato et al., 2023). Nevertheless, the overall printing time is also largely influenced by the selected sample height and the interplay between the reactivity of the proposed bioresin formulation and the applied optimized light dose per layer (Liang et al., 2021). Moreover, the positive lateral resolution attained with vat polymerization methods relies heavily on the optical voxel size (SLA: laser spot size, DLP: LCD/DMD pixel size), reactivity of the bioresin, degree of light dispersion resulting from the applied cell density, light wavelength and light dose distribution in and around the voxel of interest (Liang et al., 2021; Levato et al., 2014; Li W. et al., 2023). Practically, for both DLP and SLA, this translates into a positive lateral bioprinting resolution of several tens of micrometers (Zandrini et al., 2023). It should, however, be taken into account for DLP that a trade-off exists between the projection area and the pixel size since decreasing the pixel size for an enhanced resolution also results in a reduced projection area due to the inherent build-up of the LCD/DMD light projection system (Li W. et al., 2023). Furthermore, the axial resolution is determined by the movement resolution of the build platform together with the light penetration depth which is inversely correlated with the molar extinction coefficient and concentration of the photo-initiator, the amount of photo-absorber or -inhibitor added to the bioresin and the bioresin viscosity (Ng et al., 2020; Goodarzi Hosseinabadi et al., 2022; Li W. et al., 2023). The addition of photo-absorbers or -inhibitors allows to control the light penetration, delaying the onset of photo-polymerization and hereby improving the resolution through alleviating the mismatch between the light penetration depth and the selected printing layer thickness which should be slightly smaller than the light penetration depth to allow adherence between the different printing layers (Liang et al., 2021; Goodarzi Hosseinabadi et al., 2022; Li W. et al., 2023). The viscosity of the bioresin can also be altered whereby an increase in the density of the bioresin causes a reduction in light penetration depth together with decreasing the risk of encapsulated cell sedimentation (Goodarzi Hosseinabadi et al., 2022; Levato et al., 2023). However, care should be taken that the bioresin viscosity remains beneath a threshold of 10 Pa.s in order to allow it to flow back between printing of two subsequent layers (Ng et al., 2020). The hereby associated risk of encapsulated cell sedimentation can also be prevented through the selection of appropriate photo-crosslinkable moieties within the curable modified polymer in the bioresin that allow for an adequately fast crosslinking rate (Levato et al., 2023).

For example, SF isolated from B. Mori cocoons with a methacrylation degree of 67.3% (SFMA) has been used in various concentrations (10–15 – 25 w/v%) in combination with 2 million mouse calvarial pre-osteoblast (MC3T3-E1) cells/mL hereby enabling DLP-based bioprinting of grid-like constructs encapsulating 0.1 million cells (Rajput et al., 2022). The visco-elastic bioinks exhibited compressive moduli ranging from 12 kPa for the 10 w/v% network up to 41 and 96 kPa for the 15 and 25 w/v% networks respectively. The mass loss after 21 days as a measure of the degradation rate comprised 91%, 65% and 49% respectively for the 10, 15 and 25 w/v% network. Interestingly, network stiffening was observed (nevertheless only measured for the 15 w/v% network) over the time course of the degradation due to the SF β-sheet formation in the presence of water leading to a temporal crystallinity increase of the network. The 15 w/v% SFMA-network incorporated MC3T3-E1 cells showing the highest cell area, cell perimeter, aspect ratio and lowest circularity. This aligns well with the findings from Huebsch et al. and Chaudhuri et al. who showed that the force response of the cell strain is dependent on the initial matrix stiffness. This response then determines whether the cytoskeleton-associated adhesion complexes can be assembled (not the case in very compliant substrates), if the cells can generate enough force to deform the network (not the case in very rigid substrates) and ultimately whether matrix reorganization and cell spreading can take place (Huebsch et al., 2010; Chaudhuri et al., 2016). The complex interplay between network degradation and stiffening in this case then further aids the encapsulated elongated cells to deposit their own ECM and further enhance late-stage osteogenic differentiation through the presence of calcium deposits as was confirmed for the 15 w/v% network in culture medium both with and without osteogenic supplements (Caliari and Burdick, 2016; Loebel et al., 2019; Li X. et al., 2023).

The importance of the used cell type was illustrated by Amler et al. who encapsulated various mesenchymal progenitor cells (from alveolar bone (aBSC), fibula bone (fBSC), iliac crest bone (iBSC), iliac crest bone marrow (iBMSC) and periosteum of the mastoid (PMSC)) in 8 w% GelMA networks (DS not specified) through SLA bioprinting (0.1 w% LAP) at a density of 20 million cells/mL (Figure 6A) (Amler et al., 2021). The identification of the most suitable cell type to be used for bioprinting is important in order to obtain a cell type capable of efficiently undergoing osteogenesis with a fast and easy expansion that is obtained with straightforward and minimally invasive harvesting causing low morbidity. Most of the cell types used in that study were obtained through bone or periosteum explantation of the zone of interest. Only the iBMSCs were harvested through fine needle aspiration. Furthermore, in the case of iBMSC and PMSC, two donors were included to take into account the donor variability. The bone-derived mesenchymal progenitor cells were expanded through explant outgrowth, the cells obtained from the bone marrow were directly seeded for multiplication and the periosteal progenitors were seeded after tissue digestion.

Bioprinted construct shrinkage was observed over 28 days due to cellular contraction which can have a major influence on the outcome of the biofabricated construct due to a reduced nutrient delivery since included features like bioprinted channels might be partially blocked due to cellular bridging (Figure 6B) (Soliman et al., 2022). The extent of contraction and its influence are highly depending on the cell type(s) used, the cellular concentration, the design of the construct and the photo-crosslinkable network applied. Nevertheless, in this case, highly viable cellular constructs were obtained (Figure 6C). By day 10, the highest metabolic activity over the 28-day period across all investigated cell types was observed which was correlated to an enhancement of extracellular matrix secretion in the second week causing impaired diffusion of the metabolic activity dye at later time points. Moreover, upon differentiation, the encapsulated cells lost their highly proliferative status hereby clarifying the diminishing metabolic activity trend in the third and fourth week of bioprinted construct cultivation. Gene expression level quantification over the 28-day period and visualization of the amount of calcification after 4 weeks in the constructs allowed for comparing the differentiation level of the mesenchymal progenitor cells from different sources (Figure 6D). The aBSCs were the superior bone-derived progenitor cell type in terms of osteogenic differentiation when compared to cells from fibular or iliac crest bone. Nevertheless, these aBSCs still appeared to be at an early differentiation stage after 28 days with early markers RUNX2, ALPL and COL1A1 being significantly upregulated after 4 weeks, no downregulation of the later secreted protein acidic and cysteine rich (SPARC, encoding osteonectin) marker and only deposition of nodule-like mineralization structures. IBMSCs also showed only deposition of nodule-like mineralization structures but nevertheless already downregulated the SPARC marker as a sign of higher maturity. In contrast, PMSCs showed a high and uniform mineralization signal in combination with downregulated early marker genes, a downregulated SPARC gene and a stable mature secreted phosphoprotein 1 (SPP1, encoding OPN) gene rendering them a clinically relevant cell type for further bioprinting studies given their high proliferation capacity and the fact that they can be obtained in a minimally invasive way. Nevertheless, donor variability should also be taken into account. Here, it was found that IBMSCs show a higher variability compared to PMSCs. However, more extensive research is needed to fully capture the bioprinting outcome of progenitor cells from more sources and different donors.

Natural polymers are ideally suited for cell encapsulation with long-term survival yet, are limited in attaining high-resolution bioprinting with sufficient construct shape fidelity (Levato et al., 2021). Therefore, Lim et al. added 1 wt% GelMA (DS 60%) to 10 wt% methacrylated poly (vinylalcohol) (PVAMA) in combination with 0.2 mM/2 mM Ru/SPS (tris-bipyridylruthenium (II) hexahydrate/sodium persulfate) photo-initiator, 1 wt% Ponceau 4R photo-absorber and 5 million human BMSCs/mL for DLP-based bioprinting of highly defined cell-interactive constructs (Lim et al., 2018). The addition of GelMA resulted in similar physicochemical properties, positive and negative resolutions down to 50 µm yet resulted in a significantly higher compressive modulus as compared to pure 10 wt% PVAMA. Also, supplementation of the modified gelatin allowed enhanced long-term encapsulated cell survival up to 14 days and a qualitatively higher ALP production after 7 days thanks to the fact that the bioprinted stem cells were able to sense the surrounding network resulting from the cell-interactive groups present in the gelatin backbone.

By increasing the natural polymer content, Levato et al. succeeded in bioprinting highly defined complex cold water fish gelatin-based constructs exhibiting complex channels with a perfusable lumen (diameter <200 µm) (Figure 7A) (Levato et al., 2021). The lower hydroxyproline content in gelatin from ichthyic origin resulted in lower melting point triple helices with thermal stability at room temperature and decreased mechanical properties when compared to other types of gelatin (from porcine or bovine sources) making it a suitable candidate for the low-viscosity biofabrication technique DLP. Low-temperature soluble (LTS) bioresins consisted of either a methacryloyl- (DS 90%, LTS-GelMA) or a norbornene- (DS 85%, LTS-GelNB crosslinked with PEG4SH) modified gelatin in combination with the aforementioned photo-initiator and -absorber. The effect of the step-versus chain-growth crosslinking mechanism is nicely illustrated upon determining the resolution where the 10 w/v% LTS-GelMA resin showed the best approximation of the 50 µm positive resolution (non-significant difference with LTS-GelNB). This was in contrast to the 5 w/v% LTS-GelNB network significantly outperforming the LTS-GelMA resin in reaching a closer CAD-CAM mimicry of 100 µm negative resolution despite the comparable compressive moduli, penetration depth and critical energy. Nevertheless, the step-growth ink’s crosslinkability decreased after 30 min likely due to loss of reactivity because of thiol-persulfate redox reactions even in the absence of light hereby limiting the production of larger structures extending in the vertical direction. Therefore, bioprinting was only considered for the 10 w/v% LTS-GelMA resin encapsulating 10 million equine BMSCs/mL which were able to undergo osteogenic differentiation. A higher alkaline phosphatase activity and more extensive calcium deposition could be observed when the encapsulated cells were exposed to osteogenic medium as compared to hyperthrophic or chondrogenic media underlining the importance of the supplemented biochemical cues on the final outcome of the construct (Figure 7B).

Levato et al. further also successfully investigated the use of a second porogen phase (1.6 w/v% PEG with a molar mass of 300 kDa) to create an emulsion bioresin which is immiscible with the combined LTS-GelMA resin (15 w/v%) as a means of enhancing the permeability towards nutrients and metabolic waste products through porogen removal after incubation in a hydrated environment (Figures 8A–C) (Levato et al., 2021). This void-forming behavior was further evaluated through the addition of 3.33 w/v% dextran (molar mass of 500 kDa) to 10 w/v% GelMA (DS not specified) by Tao et al (Tao et al., 2022). Constructs with and without dextran were then DLP-bioprinted in combination with rat bone MSCs (concentration not specified). The void-forming constructs exhibited a significantly decreased compressive modulus, faster degradation and an enhanced diffusion leading to an enhanced proliferation over a 5-day period, an increased migration over 10 days and higher cellular spreading at day 7. The increased permeability also resulted in an enhanced YAP nuclear expression in contrast to the control where the lower YAP signal mainly remained in the cytoplasm. This resulted in significant upregulation of the early RUNX2 and ALP markers on day 7 and day 14 followed by a significant increase in the late OSX marker after 2 weeks. The observed osteogenesis might have arisen from both the enhanced nutrient and metabolic waste product diffusion of the highly metabolically active stem cells as well as the increased ability of the encapsulated cells to deposit their own matrix (Caliari and Burdick, 2016; Loebel et al., 2019; Li X. et al., 2023). They could even show that 8 weeks in vivo implantation of the DLP-bioprinted constructs in a cranial defect in Sprague-Dawley rats gave rise to a more gradually calcified bone integrated within the host bone. Interestingly, when rat DPSCs (concentration not specified) were incorporated into the void-forming phase (3.33 w/v% 500 kDa dextran), the in-situ birth of stem cell spheroids could be observed in the remaining 10 w/v% GelMA (DS not specified) matrix (Zhu et al., 2025). These spheroids showed enhanced proliferation, in vitro osteogenic differentiation (Figure 8D) and in vivo endodontic tissue regeneration capability as compared to rDPSC-encapsulating 10 w/v% GelMA controls without a porogen phase.

Two-photon lithography (TPL) is a laser-scanning technique that relies on the non-linear bridging of the excited state energy gap through simultaneous absorption of two photons (Ovsianikov et al., 2012; Greant et al., 2023). The probability of two-photon absorption scales with the square of the incident light intensity and is inversely proportional to the fourth power of the distance from the laser focal plane (Ovsianikov et al., 2012; Lay et al., 2020). Hence, by adjusting the laser power, this effect can be exploited in a highly localized volume in the focal spot (<1 μm³) to allow light-based photo-crosslinking, -grafting, -degradation or -ablation (Ovsianikov et al., 2012; Greant et al., 2023; Levato et al., 2023). Overall, the outcome of TPL in terms of printability, printing time, attainable sample size, resolution, shape fidelity and print stability is mainly affected by the constituents of the bioresin (photo-crosslinkable polymer concentration and reactivity, cell type and concentration, photo-initiator concentration and efficiency), the optical set-up, sample post-processing and the delivered light dose to the resin (Ovsianikov et al., 2012; Lay et al., 2020; Greant et al., 2023; Levato et al., 2023).

Given the highly localized focal volume, TPL achieves subdiffraction minimum feature sizes within the order of 10^–7 m (Ovsianikov et al., 2014; Lay et al., 2020; Greant et al., 2023). Given that for raster-scanning techniques, speed scales with volume, this results, together with the reported resolution, in a printing time in the range of hours to create a 1 cm³ bioprinted construct (Levato et al., 2023). Nevertheless, an increase in the number of lasers or light beams has already been applied to augment the writing speed while still enabling high resolution (Levato et al., 2023). In order to avoid overheating (except in situations where photo-ablation is desired) with the high intensity femtosecond lasers, the photo-reactivity of the applied bioresins should be high in combination with a high transparency at the used wavelength hereby circumventing linear absorption and/or irradiation blockage (Ovsianikov et al., 2012; Lay et al., 2020). Next to this, the viscosity should be adequate (>10 Pa.s) to prevent cellular sedimentation as well as to avoid structure deformation during the printing process (Lay et al., 2020; Levato et al., 2023).

Two-photon ablation (TPA) has been used to create an interconnected cell network (1 µm diameter) hereby mimicking the native, late-stage osteocyte lacunar-canalicular microarchitecture (Gehre et al., 2024). In order to enhance the ablation efficiency and to create a human BMSC-compatible ablation energy dose (100 J/cm²), a two-photon photo-sensitizer (0.5 mM sodium 3,3′-((((1E,1′E)-(2-oxocyclopentane-1,3-diylidene)-bis(methaneylylidene))-bis(4,1-phenylene))-bis(methyl-azanediyl))-dipropionate, P2CK) was added to photo-crosslinked, cell-encapsulated GelMA networks (5 w/v% GelMA with DS of 56%, 0.05% LAP, 2.5 million cells/mL, 365 nm, 3 J/cm²). The 3D ablated networks were successfully colonized by day 7 with long protrusions exceeding 40 µm and the establishment of cellular functional contacts through gap junctions (Figure 9). Moreover, it was shown that the embedded cells preferentially used the confining channels over the ability to spread through proteolytic remodeling within the constraining GelMA network. This also affected the ALP activity after 7 and 14 days with a slightly higher ALP activity for patterned networks versus the non-ablated control. This study nicely illustrates that approaches for stimulating encapsulated cell spreading for enhanced osteogenesis are not limited to the permissive character of the applied bioresin but are also heavily influenced by the presented topography on the cell level.

Computed axial lithography (CAL), tomographic volumetric printing (VP) or volumetric additive manufacturing (VAM) where light energy is delivered to a 3D volume instead of a point (e.g. SLA, TPL) or a plane (e.g. DLP), allows to overcome the limited throughput and the constrained geometric capabilities evoking the need for support or sacrificial materials associated with more conventional layer-by-layer biofabrication approaches (Shusteff et al., 2017; Kelly et al., 2019; Loterie et al., 2020; Bernal et al., 2022; Thijssen et al., 2023; Jing et al., 2024). The accumulated 3D dose distribution on a resin container whose rotation is time sequenced with the light projection, results from the superposition of 2D cross-sectional intensity-modulated image projections from multiple angles hereby allowing to locally reach the solidification threshold of the resin according to the specified input design model of the desired object (Kelly et al., 2019; Loterie et al., 2020). Overall, the outcome of tomographic volumetric bioprinting in terms of printability, printing time, attainable sample size, resolution, shape fidelity and print stability is mainly affected by the constituents of the bioresin (photo-crosslinkable polymer concentration and reactivity, cell type and concentration, concentration and efficiency of photo-initiators (and -inhibitors)), the light projection optics and computation, the delivered light dose to the resin and the post-processing method applied (Shusteff et al., 2017; Bernal et al., 2019; Bernal et al., 2022; Kelly et al., 2019; Loterie et al., 2020; Rizzo et al., 2021; Madrid-Wolff et al., 2023; Thijssen et al., 2023; Jing et al., 2024).

The ability to construct prints volume-wise allows several orders of magnitude faster printing speeds - compared to layer-by-layer biofabrication techniques – requiring a printing time in the order of seconds to build a 1 cm³ construct allowing improved scalability and enhanced encapsulated cellular viability and functionality (Bernal et al., 2019; Kelly et al., 2019; Rizzo et al., 2021; Levato et al., 2023). The resolution of volumetric constructs depends on the viscosity and the reactivity of the resin, potential presence of scattering elements in the resin, the pixel size and the magnification of the light modulating projection system, the spatial coherence of the light source and the tomographic dose reconstruction accuracy (Bernal et al., 2019; Kelly et al., 2019; Loterie et al., 2020; Madrid-Wolff et al., 2023). The positive resolution of the technique is limited to >40 µm whereas the lowest reported negative resolution achieved with VAM is around 100 µm (Rizzo et al., 2021; Bernal et al., 2022; Cianciosi et al., 2023; Madrid-Wolff et al., 2023). The components of the volumetric photoresin should be optimized to have a high reactivity but low absorbance, allowing for moderate light attenuation so that the solidification threshold can be reached across the entire build volume (Kelly et al., 2019; Rizzo et al., 2021; Thijssen et al., 2023). Furthermore, scattering in volumetric resins either polymerization-, dispersed resin additive- (encapsulated cells, spheroids or organoids) or embedded macroscale object-induced, should be minimized through the use of refractive index adjusting agents, by adjusting the optical set-up or through incorporating this effect within the computational reconstruction (Bernal et al., 2022; Madrid-Wolff et al., 2022; Thijssen et al., 2023). Moreover, sedimentation should be limited through either dose optimization algorithms or the modulation of the resin viscosity which should surpass a value of 10 Pa.s (Loterie et al., 2020; Rizzo et al., 2021; Thijssen et al., 2023). Finally, after exposure, care should be taken that, given the short printing times, low conversion of the printed construct might heavily impact the print stability requesting the need for post-curing (or crystallization-inducing processes) (Thijssen et al., 2023).

Gehlen et al. successfully exploited volumetric bioprinting to print vascularized constructs targeting osteogenesis by encapsulating 3 million human BMSCs/mL within a 5% GelMA (DS 57%) perfusable construct (Gehlen et al., 2023). 5% GelMA was chosen since an enhanced osteogenic differentiation was observed as reflected by the increased relative gene expression of the osteocyte marker gene podoplanin (PDPN) compared to denser 10% GelMA networks. Despite the low mechanical properties associated with the 5% network, this network might have allowed for enhanced diffusion of nutrients and waste products together with a more active spreading of the encapsulated cells while depositing their own pericellular matrix which has also been found to be a determining factor in osteogenesis (Caliari and Burdick, 2016; Loebel et al., 2019; Li X. et al., 2023). Further decreasing the concentration with or without the addition of unmodified gelatin was not considered since this could limit the handleability of the volumetrically printed construct. Next, the effect of cellular crosstalk in an encapsulated co-culture of endothelial (0.6 million human umbilical vein endothelial cells (HUVECs)/mL) and stem cells (3 million human BMSCs/mL) was evaluated on osteogenic gene expression in comparison to encapsulated stem cells on their own. The authors observed significantly upscaled early osteogenic markers for the monoculture whereas, for the co-culture, significantly increased osteoblastic markers, an enhanced ALP gene expression and activity and higher early osteocytic markers were seen. However, up to 6 weeks, no calcium deposits were observed through micro-CT in both mono- and co-cultures together with the absence of the mature osteocytic marker sclerostin (SOST) which was correlated to the need for enhanced maturation. Duquesne et al. applied in this regard a stiffer 5 w/v% GelNBNB/GelSH (DS 176/72) matrix as compared to 5 w/v% GelMA (DS 95) (Figure 10A) as volumetric printing bioink (Duquesne et al., 2025). Encapsulated human DPSCs (1 million cells/mL) exhibited enhanced late-stage osteogenic differentiation markers (mineralization, Figure 10B and SOST-expression; Figure 10C) when encapsulated within perfusable step-growth crosslinked, volumetric bioprinted constructs (Figure 10D).

The tissue engineered construct should allow mechanical stability after implantation at the defect site up until the moment the newly formed bone can gradually take over this role (Preethi Soundarya et al., 2018). Hence, to reduce fibrous tissue formation and stimulate callus bridging, mechanical discontinuities should be prevented at the scaffold-bone interface (Prasadh and Wong, 2018). However, large variations are observed in specific target mechanical values since these highly depend on the anatomical defect and its different loadings, in addition to age, gender and possible co-morbidities of the patient (Velasco et al., 2015). Nevertheless, when comparing the order of magnitude of target values (e.g. Young’s modulus: 10⁷–10¹⁰ Pa) versus mechanical properties reported for photo-crosslinked natural polymeric hydrogels (e.g. Young’s modulus: 10³–10⁵ Pa), it becomes clear that reinforcement strategies are paramount towards further clinical translation (Velasco et al., 2015; Cidonio et al., 2019a; Alcala-Orozco et al., 2020; Wang J. et al., 2024).

Following a biomimetic strategy, natural polymer-based hydrogels have in this regard been combined with a ceramic phase. The concentration, distribution, size, aspect ratio, charge and chemistry of this reinforcing phase determine whether natural hydrogel crosslinking is maintained and/or whether additional (physical and/or chemical) bonds are being created. Based on this multi-factorial outcome, the mechanical properties are altered. A more than two-fold increase in elastic modulus was observed by Yu et al. through the addition of xonotlite (5 wt%) to GelMA (10 w/v%, degree of substitution (DS) not specified) thanks to the presence of attractive forces between the polymer network and the nano-fillers (Yu et al., 2024). The same strengthening effect was observed upon addition of nano beta-tricalcium phosphate (0, 1, 3 or 5 w/v%) to GelMA (5 w/v%, DS not specified) and alginate (1 w/v%) by Zhang et al. as long as the scattering ceramic fraction was not too high to decrease the crosslinking degree (Zhang et al., 2024). The latter effect on the crosslinking degree was also reported by Sun et al. when graphene oxide nanosheets (1 mg/mL) were incorporated within photo-crosslinked gelatin- and silk-based networks (5 w/v% GelMA, 3 w/v% SFMA and 5 w/v% GelDA DS not specified) hereby effectively decreasing the mechanical properties (Sun X. et al., 2023). By first mixing calcium phosphate nanoparticles with gelatin and subsequent methacrylation (final concentrations and DS not specified), Bhattacharyya et al. succeeded in creating a more controlled size, aspect ratio and distribution of the particles leading to improved mechanical properties as compared to conventional methods which involve first the methacrylation of gelatin followed by nanoparticle mixing (Bhattacharyya et al., 2022). Choi et al. reported on a silane modification of silica nanoparticles (10 wt%) exhibiting strong repulsive forces preventing aggregation and allowing good dispersibility and an improved Young’s modulus when introduced within GelMA networks (15 wt%, DS not specified) (Choi et al., 2021). When increasing the whitlockite/hydroxyapatite nanoparticle ratio (25%–100%) within gelatin- and alginate-based networks (7 w/v% GelMA DS 81%, 4 w/v% alginate and 0.5 w/v% gelatin), Ghahri et al. observed a decreased compressive modulus due to the repulsion of the negatively charged surface of whitlockite with the carboxylic acid groups of the natural polymer network decreasing the chemical and ionic crosslinking degree (Ghahri et al., 2023). Finally, Zhu et al. reported on the covalent attachment of bioactive glass particles (concentration not specified) to a gelatin- and alginate-based network (final concentrations of gelatin and oxidized alginate (oxidation degree 30%) not specified) as an effective means to increase the compressive modulus (Zhu et al., 2022).

Despite promising results, the maximum Young’s modulus obtained for bioprinted ceramic/photo-crosslinkable natural polymer composites is situated around 10⁶ Pa which is not sufficient considering the target value range (vide supra) (Liu et al., 2024). Therefore, more emphasis has to be placed on multi-material and/or multi-technique strategies that allow the combination of a mechanically performant macroscopic system with adequate cellular niches for optimal stimulation of bone healing. Cui et al. combined in this regard fused deposition modeling/fused filament fabrication (FDM/FFF) of poly (lactic acid) with SLA of GelMA (10 wt%, DS not specified) to achieve perfusable tissue engineered constructs with a Young’s modulus around 10⁸ Pa (Figure 12) (Cui et al., 2016). Moreover, multi-material extrusion-based scaffolds of magnesium-reinforced (20 wt%) poly (ɛ-caprolactone) and poly (lactic-co-glycolic acid) were combined with strontium-reinforced (1.5 μg/mL) GelMA (5 wt%, DS 60%) and GelMA (concentration and DS not specified) respectively to finally achieve a Young’s modulus around 10⁷ Pa (Alcala-Orozco et al., 2022; Rodrigues et al., 2024). All reported mechanical properties are nevertheless heavily depending on the specific printing design and hence further research is needed to elegantly combine and spatiotemporally balance the mechanical reinforcement fraction, the bioprinted part and adequate porosity allowing for tissue ingrowth. Moreover, mechanical testing parameters should be more standardized to allow better comparison.

Before neurovascularized bone ingrowth can occur, immunological signaling will largely determine the tissue response (Du et al., 2024). Chronic pro-inflammatory (M1 polarization) signaling will lead to a tissue repair impediment and fibrosis development whereas immunomodulation to a pro-regenerative (M2 polarization) environment after hours to days enables to initiate optimal bone healing (Du et al., 2024; Duquesne et al., 2025). Despite its role as the tissue engineering gold standard, the use of GelMA within bioinks for extrusion-based, volumetric or DLP-based bioprinting resulted in the expression of M1-associated markers both in vitro and in vivo between 7 and 21 days (Du et al., 2024; Liu et al., 2024; Yu et al., 2024; Duquesne et al., 2025). Interestingly, shifting to step-growth crosslinking chemistry gave rise to overall lower levels of pro-inflammatory cytokines at later time points underlining the need to validate more bioinks relying on step-growth photo-crosslinking chemistry (Duquesne et al., 2025). Moreover, the addition of manganese and strontium to GelMA-based bioinks allowed immunomodulation towards an M2 type and the subsequent secretion of cytokines related to tissue regeneration, hereby effectively stimulating osteogenic differentiation in vitro and bone healing in vivo (Figure 13) (Du et al., 2024; Liu et al., 2024; Yu et al., 2024). Yet, more studies are needed to further understand the immunomodulatory role towards bone healing and to implement this knowledge in biomaterial design.

Nerves and blood vessels play important roles in bone development, homeostasis and regeneration (Hankenson et al., 2011; Marenzana and Arnett, 2013; Li Z. et al., 2019; 2025). During bone regeneration, the fracture will be firstly innervated which is a crucial step in the formation of the ossification center (Li Z. et al., 2019). The nerves will release neurotransmitters, neuropeptides, neurotrophins regulating the bone regenerating micro-environment (Sun et al., 2020; Sun et al., 2023 W.). Next, the bone defect will be vascularized allowing the provision of nutrients, oxygen and growth factors and the removal of waste products as well as the recruitment of osteoprogenitor cells (Hankenson et al., 2011; Marenzana and Arnett, 2013; Sun W. et al., 2023). Despite the fact that delayed or absent vascularization and innervation result in impaired fracture healing, hydrogel-based approaches targeting neuro-vascularized bone regeneration are lacking (Hankenson et al., 2011; Li Z. et al., 2019; Meyers et al., 2020). The different cells involved in bone formation, innervation and vascularization require different micro-environments for optimal proliferation and differentiation evoking the need of a scaffold heterogenous in biophysical and biochemical properties (Wan et al., 2020; Li et al., 2025). However, the developed bone scaffolds to date are often lacking this multi-tissue focus.

Fortunately, bioprinting facilitates the fabrication of multi-material and hetero-cellular scaffolds with complex architecture and heterogenous biophysical/biochemical properties targeting multiple tissue type regeneration (Li Q. et al., 2023). Below, representative examples targeting neuro-vascularized bone regeneration through bioprinting will be reviewed. In the first set of examples, solely stem cells were selected. Li et al. encapsulated Laponite loaded with nerve growth factor (NGF, 20 mg/mL) and rat BMSCs (10 million cells/mL) within a mixture of GelMA (5%, DS not specified) and alginate methacrylate (AlgMA, 2%, DS not specified) (Li et al., 2022). The reversible binding of NGF with Laponite evokes a slower release of the growth factor. Subcutaneous implantation of bioprinted constructs revealed an improved osteogenic differentiation through calcitonin gene-related peptide (CGRP) release of sensory neurons stimulated by Laponite and NGF. Additionally, the most pronounced innervation and vascularization were detected using immunofluorescence (based on CGRP, cluster of differentiation 31 (CD-31) and alpha-smooth muscle actin (α-SMA), day 14) and ultrasound imaging (day 14) in the experimental group (Figures 14A,B). Finally, the positive effect on bone regeneration was validated in a cranial defect model after 8 weeks of implantation using µ-CT, hematoxylin and eosin staining, and Masson’s trichrome staining (Figure 14C). Using a similar strategy, mesoporous silica nanoparticles were loaded with propranolol (PRN) and CGRP, causing the sustained release of PRN, CGRP and Si ions and thus an improved osteogenesis and angiogenesis within the bioprinted construct (Guo and He, 2023). Both studies prove that loading growth factors or therapeutic agents into nanoparticles represents a promising approach to achieve (more) controlled release. Another system in which osteogenesis was elegantly combined with angio- and neurogenesis exploited DLP-printing of 10 w/v% GelMA (DS not specified) to encapsulate 50 million human dental pulp-derived stem cells/mL microspheroids (Qian et al., 2023). The researchers showed that compared to 2D cell seeding onto 10 w/v% photo-crosslinked GelMA sheets and tissue culture plate, the 3D microspheroids showed equivalent osteogenic (odontogenic) differentiation (through dentin matrix acidic phosphoprotein 1 (DMP1) and dentin sialophosphoprotein (DSPP) expression) but significantly higher angio- (through VEGFα and angiopoietin 1 (ANGPT1) expression) and neurogenesis (through growth associated protein 43 (GAP43) and microtubule associated protein 2 (MAP2) expression) underlining the importance of dimensionality and cellular concentration influencing the biological outcome of a printed construct.

Besides using solely mesenchymal stem cells, also neural cells and/or endothelial cells were encapsulated to replicate better the cellular composition of the various tissues present within bone. Firstly, the beneficial effect of mesenchymal stem cell–neural cell co-culture in treating bone defects was illustrated by Zhang et al. who extrusion bioprinted sequentially two GelMA (DS not specified, 6 w/v%) bioinks supplemented with calcium silicate (CS, 2%) nanowires encapsulating either rat BMSCs (2 million cells/mL) or Schwann cells (2 million cells/mL) (Zhang et al., 2022). After 4 and 8 weeks implantation of extrusion bioprinted constructs in a cranial defect, the experimental group revealed the most pronounced stimulation of osteogenesis (based on µ-CT and immunofluorescence staining of OCN and OPN) and neurogenesis (based on immunofluorescence staining of CGRP and neurofilament). This stimulation was attributed to the synergistic effect of the CS nanowires releasing bioactive ions including Ca and Si ions, and the neural-bone cell co-culture. Since Schwann cells regulate the proliferation and osteogenic differentiation of stem cells via the release of exosomes, an alternative bioink encapsulated BMSCs and Schwann cells’ exosomes (Jones et al., 2019; Wang T. et al., 2023). In addition to the beneficial effect on osteogenesis and neurogenesis, subcutaneous implantation of the extrusion bioprinted constructs revealed after 14 days a robust blood flow inside the constructs based on ultrasound imaging (Wang T. et al., 2023). Secondly, combining mesenchymal stem cells and endothelial cells in one construct is a promising strategy when targeting vascularization. The endothelial cells can be either exploited to generate large vessels through lining of engineered, hollow features or to generate microvasculature. Those hollow structures have been generated through inclusion of non-covalently crosslinked biomaterials or rapidly degradable covalently crosslinked hydrogels. Shen et al. extrusion bioprinted a porous GelMA scaffold (5 wt%, 5 million bone MSCs/mL) of which the pores were initially filled with PLA-PEG-PLA (10 wt%) encapsulating rat aortic endothelial cells (RAOECs, 5 million cells/mL) (Shen et al., 2022). Within a short time frame (∼1 h), PLA-PEG-PLA was dissolved and resulted in an improved seeding efficiency of endothelial cells compared to conventional seeding. Both in vitro and in vivo experiments revealed an improved effect of both the experimental seeding approach and the co-culture on the formation of new bone and vascularization. Endothelial cells lining the inside of an engineered vessel can also be achieved via multi-axial bioprinting of GelMA bioinks containing specific cells stimulating angiogenesis or osteogenesis as the outer shells, and gelatin as the inner shell (Zhang et al., 2024). Zhu et al. reported enhanced vascularization connected to the host vasculature thanks to co-culture spheroids consisting of HUVECs and human DPSCs (concentration not specified) formed after the introduction of a void-forming phase (3.33 w/v% 500 kDa dextran) in GelMA (10 w/v%, DS not specified) (Zhu et al., 2025). A final alternative strategy entails the inclusion of rapidly degradable GelMA in the inner core of the construct. Byambaa et al. extrusion bioprinted the construct’s inner core of rapidly degradable VEGF-conjugated GelMA (5 w/v%, DS 34%) encapsulating HUVECs and hBMSCs, and an outer core of GelMA (10 w/v%, DS 94%) grafted with a gradient VEGF concentration encapsulating hBMSCs and nano silicate particles (Figure 15A) (Byambaa et al., 2017). The co-culture of MSCs and HUVECs lined the created channel, with the MSCs differentiating into smooth muscle cells, which accelerates the formation and maturation of a vascular network (Figure 15B). Both VEGF and the co-culture positively influenced vasculogenesis by stimulating capillary network formation and endothelial cell spreading. MSCs encapsulated in the outer region differentiated towards osteoblasts due to the presence of encapsulated silicate nanoparticles and VEGF. In this example, the half-life of the growth factor was enhanced by covalently attaching it to the gelatin backbone. Next to grafting, also the small molecule drug fingolimod (1,000 ng/mL) can be used to obtain longer half-life in order to stimulate migration, proliferation and capillary-like structure formation of endothelial cells and thus stimulate angiogenesis (similar to 100 ng/mL VEGF) (Yang et al., 2022). In a final example, multiple 3D bioprinting platforms were used (FFF of poly (lactic acid) and SLA of 10 wt% GelMA (DS not specified)) in which human MSCs were first seeded on the poly (lactic acid) scaffold followed by SLA bioprinting of a co-culture of human MSCs in combination with HUVECs (Cui et al., 2016). The successful combination of FFF with SLA allowed to mimic bone at different hierarchical scales thereby showing the ability to spatially control bioactive factor arrangement, cellular organization and mechanical loading. The use of dynamic perfusion of the construct in combination with the presentation of biochemical cues (growth factors BMP-2 and VEGF) highly stimulated both osteogenesis (in terms of ALP activity, collagen type I synthesis and mineralization) and angiogenesis (VEGF expression).

Interestingly, when rat DPSCs (concentration not specified) were incorporated into the void-forming phase (3.33 w/v% 500 kDa dextran), the in-situ birth of stem cell spheroids could be observed in the remaining 10 w/v% GelMA (DS not specified) matrix (Zhu et al., 2025). These spheroids showed enhanced proliferation, in vitro osteogenic differentiation and in vivo endodontic tissue regeneration capability as compared to rDPSC-encapsulating 10 w/v% GelMA controls without a porogen phase.

To conclude, while constructs composed of photo-crosslinkable natural polymers utilizing single bioprinting technologies offer significant advantages in generating an appropriate osteoid niche to allow osteogenic differentiation, the current focus should be extended towards constructs combining multiple material and/or multiple printing techniques. This multifaceted approach is essential to achieve functional and scalable constructs enabling in vivo bone regeneration. Such constructs would not only support osseous tissue formation but also vascularization and innervation, as well as meet macroscopic mechanical target values. Additionally, given the advantages of step-growth crosslinkable bioinks for cell encapsulation, including their promising immunomodulatory properties, a paradigm shift from conventional chain-growth crosslinkable bioinks to step-growth crosslinkable bioinks is of paramount importance. Although excellent papers have been published addressing various elements of this complex problem, current constructs fail to meet all necessary requirements.