Aspiration-assisted Freeform Bioprinting of Tissue Spheroids in a Yield-stress Gel

Bioprinting of cellular aggregates, such as tissue spheroids or organoids, in complex three-dimensional (3D) arrangements has been a major obstacle for scaffold-free fabrication of tissues and organs. In this research, we unveiled a new approach to the bioprinting of tissue spheroids in a yield stress granular gel, which exhibited unprecedented capabilities in freeform positioning of spheroids in 3D. Due to its Herschel-Bulkley and self-healing properties as well as its biological inertness, the granular gel supported both the positioning and self-assembly of tissue spheroids. We studied the underlying physical mechanism of the approach to elucidate the interactions between the aspirated spheroids and the gel’s yield-stress during the transfer of spheroids from cell media to the gel. We demonstrate the application of the proposed approach in the realization of various freeform shapes and self-assembly of human mesenchymal stem cell spheroids for the construction of cartilage and bone tissues.

Three-dimensional (3D) bioprinting within granular gels or suspension baths exhibiting Herschel-Bulkley or Bingham plastic properties has recently become a powerful approach to create complex-shaped anatomically-accurate tissues and organs [1][2][3][4][5][6][7][8][9] . Carbopol microgels have been one of the popular granular gel medium due to its shear thinning and self-healing properties, in which the granular gel transforms from a stable solid state into a flowing fluid phase when exposed to an external stress that exceeds its yield stress 6,10-12 . As the nozzle moves inside the granular gel, the gel locally fluidizes when in contact with the nozzle but then rapidly solidifies after the nozzle has passed thus supporting the bioprinted tissue constructs 1,7 . In most cases of bioprinting in a granular gel, cells are bioprinted while encapsulated within a hydrogel formulation, resulting in limited cell densities. 13 Hence, cellular aggregates, such as tissue spheroids, possess greater promise due to their favorable properties in building native-like tissues [14][15][16] .
Bioprinting of spheroids is an attractive approach, in which spheroids are used as building blocks for fabrication of tissues that mimic the native counterparts in terms of histology and physiology 14,16,17 . Several spheroid bioprinting techniques have been reported. The first technique is extrusion-based bioprinting 18 , in which spheroids are loaded in a syringe barrel and extruded in a delivery gel medium one by one. However, spheroids self-assemble readily in the syringe and are prone to break apart during the extrusion process. Concurrently, support structures need to be 3D printed to facilitate the aggregation of extruded spheroids. An important advance has been made by utilizing the Kenzan method 19 , where spheroids are skewered on a needle array. Since the position of each spheroid depends on the needle size, location and arrangement, freeform (i.e., complex-shaped) bioprinting of spheroids is quite challenging as the spheroid positioning of spheroids along the z-axis (direction parallel to the needles) is not independent in each layer.
Drop-on-demand bioprinting has also been reported to deposit spheroids 20 . In this approach, spheroids are encapsulated within gel droplets. As such, drop-on-demand bioprinting has inherent limitations on the precision of the 3D bioprinting process. To overcome some of major the challenges of current techniques, we recently demonstrated an aspiration-assisted bioprinting (AAB) technique 21 enabling precise bioprinting of spheroids into or onto sacrificial or functional gel substrates. However, freeform bioprinting of spheroids in 3D has been a long-standing problem due to the layer-by-layer-building nature of the existing techniques.
Here, for the first time, we demonstrate the freeform bioprinting of tissue spheroids by precisely positioning them in a self-healing biologically-inert granular gel in 3D allowing for the subsequent self-assembly of the bioprinted spheroids towards fabrication of tissues and organs. We used our previously demonstrated AAB technique to aspirate and pick spheroids and, taking advantage of the Herschel-Bulkey properties of the granular gel receiving the spheroids, we succeded in the direct transfer of spheroids from the cell media and their freeform positioning within the granular gel on-demand. In order to better understand the response of biologics to the bioprinting process, we studied the underlying mechanism explaining interactions between the spheroids and the granular gel during bioprinting. We then explored the potential of our Aspiration-assisted Freefrom Bioprinting (AAfB) technique in building complex-shaped configurations and demonstrated multiple applications, including cartilage and bone tissues throughout this study.

Working mechanism of AAfB
In this study, we further advanced our recently published Aspiration-assisted Bioprinting (AAB) technique 21 to demonstrate the freeform bioprinting of spheroids within a granular gel. Specifically, aspiration forces were used to pick up spheroids from the spheroid reservoir (placed inside the cell media compartment) and transfer them into the granular gel (occupying inside the support gel compartment) one by one (Figs.  1A1-1A7). The spheroids were transferred from the cell media through a highly mobile transition interface into the self-healing granular hydrogel. In general, gels have small elasticity and high viscosity, and their mechanical response is usually described by a viscoelastic model 22 . Here, we present some elementary moment balance arguments leading to the estimate of the minimum aspiration pressure that is needed for a spheroid to be transferred from the media to the gel compartment. With reference to Figs. 1B1-1B2, whether the spheroid was moving through the interface or through the gel, we observed that the spheroid was acted upon by forces due to its interaction with its environment and with the nozzle. If the aspiration pressure fell below a critical value " , the spheroid would separate from the nozzle, typically by pivoting against the trailing edge of the nozzle (trailing relative to the direction of motion). We label the pivot point by T in Fig.  1B1. We denote by $ the magnitude of the resultant force acting on the spheroid due to its interaction with the environment. Referring to Fig.  1B1, we observed that at the critical pivot condition the only forces contributing to the moment about the point T are the resultant of the applied aspiration pressure distribution and the force with magnitude $ . With this in mind, we can then estimate the critical aspiration pressure " by considering the balance of moments about T. Clearly, to proceed to such an estimate we need to know both the values of $ and its direction as well as the state of motion of the spheroid. Since $ represents the resistance offered by the gel to the spheroid's motion, we make the simplifying assumption that, when the spheroid is moving at a constant speed along a horizontal line, the resistance is also horizontal and with a line of action going through the spheroid's center. Under these simplified conditions, the moment balance about T, ∑ ' = 0, yields the following relation: where, with reference to Figs. 1B1-1B2, r is the nozzle's radius and is such that tan = / .
Solving Eq. (1) for " , we obtain " = √$ ; <= ; Next, we need to provide an estimate for the value of $ . This estimate can be complex in that $ is determined by different physics depending on the position of the spheroid relative to the interface between the medium and the gel compartments.
When the spheroid is moving through the gel as a constant speed, it is reasonable to assume that $ = A , where A is the drag acting on a sphere moving at a constant speed in a viscous fluid under laminar conditions. In fact, treating the spheroid as a rigid particle with a radius R (< 450 µm), the relevant Reynolds number 23 is Re = 2ρgelUR/ C , where ρgel is the mass density of the gel, which is assumed to be the same as water (as a matter of fact, the mass density of the spheroids can also be assumed to be that of water: E = FGH = I ), U~2.5 mm/s is the bioprinting speed (also the speed of the spheroid's center-of-mass, and C is the gel's Newtonian equivalent viscosity or zero-shear rate viscosity 44 Pa·s). Under these assumptions, Re is on the order of 10 -6 confirming that the flow around the spheroid during bioprinting is indeed laminar. Under these conditions, we can use the well-known formula A = 6 L , where the value of viscosity L depends on as the gel is shear-thinning 24 .
More complex is the estimation of $ when the spheroid is traversing the interface between the medium compartment and the gel. In this case, we can distinguish four contributions to $ : again the drag exerted on the spheroid by its surroundings ( A ), the resistance provided by the elasticity of the gel below the yield limit as the spheroid is indenting the gel ( M ), the thermodynamic force ( N ) representing capillary effect at the interface, and nonlinear and dynamic terms ( O<A ) , neglecting fluctuations and the rotational effects 25 , as the motion cannot be treated as being steady: Whether in the gel or at the interphase, for simplicity, we will estimate A using the same drag formula mentioned earlier scaled to account for the fact that the spheroid is not completely in the gel (Fig.  1B2): A = Q L (6 + 8 + 2 ), where Q is the contact radius and where the advancing angle α is defined via the relation tan = Q /( Q − ℎ), ℎ being the indentation depth 24 (Fig. 1B2). We feel that his estimate is acceptable in an effort to understand what physics dominates the value of $ . Clearly, the maximum resistance provided by the gel to the spheroid after traversing the interface is A = 6 Q L , as previously discussed. Referring to Fig. 1C1, our experimental rheological characterization indicates that the gel should be modeled as a (shearthinning) Herschel-Bulkey fluid with viscosity: where C is the yield stress, is the shear rate which we estimate as ̇= / for the motion in the gel or as ̇= / Q for the motion through the interface, K is the consistency index, and n is the power-law exponent (n<1 for shear-thinning fluids 26 ). Two fluid property constant can be identified from the power-law shear-thinning regime in Fig.  1C1, n can be obtained by adding one to the slope of the viscosity versus shear rate curve and the consistency index K is equal to the viscosity of the gel when the shear rate is equal to 1. From our experiments, we see that K=44 (Pa·s n ) and n=0.3. Thus, L = 44̇< C.c + τ C /, with a unit of Pa·s.
At the initial stage of contact 27 , M = 4 ℎ or using the same geometric configuration as above, where is the Poisson ratio, which, for an incompressible material like Carbopol, can be taken to be equal to 0.5 28 . Our measurements of ′ are reported in Fig. 1C2. The term M is only considered while the spheroid is traversing the interface and neglected when the spheroid is fully submerged in the gel.
Another term is the thermodynamic interfacial force is experienced when the spheroid is traversing the media-gel interface. The maximum value can be estimated to be: Here, _,-is the surface tension coefficient between the media and gel. >= ? 6 ( 0 +̇− 1 ).
L is the viscosity at printing speed of 2.5mm/s. As a result, " is a function of R, r, U, gel properties ( , , C ). We have not included the terms that are a weak function of E, _,-and , which are negligible compared to the viscosity of the gel. However, while crossing the interface, FE term should also be included in the estimation of Pb.
In order to determine an appropriate gel concentration for AAfB, we preferred to test 0.8, 1.2, and 1.6% concentrations of Carbopol, where such a range was comparable with respect to Carbopol concentration used in a previous study 13  All concentrations showed shear-thinning properties indicated by decreasing viscosity with shear rate (Fig.  1C1), and solid to fluid transition occurred at ~13, 28, and 57% strain for 0.8%, 1.2% and 1.6% Carbopol, respectively. Fig.  1D shows bioprinting positional accuracy with respect to the spheroid size, which was improved with increasing Carbopol concentration such that 1.6, 1.2, and 0.8% Carbopol yielded 19, 37% and 97% positional accuracy, respectively. As shown by the error bars in Fig. 1D, the positional precision for 0.8, 1.2, and 1.6% concentrations were determined to be ~97, 22, and 12 %, respectively. In order to validate the theoretical approach, we performed bioprinting experiments to establish a relationship between r and " . As indicated in Fig.  1E, the theoretical approach was confirmed by the experimental approach and the results were close to each other, particularly for spheroids with smaller radii. Yet, increasing aspiration pressure may lead to deformation of the spheroids 21 . Spheroids needed to be transferred in a safe manner without leading to significant deformations. As its know that external stressors induce considerable damage to cell viability 29 , 1.2% Carbopol concentration was preferred to be used in our further experiments. Tubular cartilage tissues were bioprinted using MSC spheroids following two strategies in order to investigate the effect of the chondrogenic differentiation timeline on the functional and structural properties of bioprinted tissues (Fig.  S2). In the first strategy, which we will refer to as Strategy I, MSC spheroids were maintained in the growth media for three days and then were 3D bioprinted into a tube shape on Day 3. The bioprinted constructs were removed from Carbopol on Day 4 and further maintained in a chondrogenic induction medium for 20 days. In the second strategy, which we will refer to as Strategy II, MSC spheroids were maintained in the growth medium for three days followed by a 19 day culture in a chondrogenic induction medium, and finally bioprinted on The higher the surface tension the better the bioprinting will be due to the spheroids' decreased sensitivity to aspiration forces 21 . In this regard, chondrogenic spheroids had a surface tension that was approximately twice that of MSC spheroids (Fig. 3B2). Furthermore, the surface tension values for both spheroids were within feasible ranges for bioprinting 21 . Finally, we also observed a 2.2-fold increase in the sGAG content (μg/ng DNA) for chondrogenic spheroids as compared to MSC spheroids (Fig. 3B3). We then used these spheroids from Strategy I and Strategy II to bioprint tubular cartilage tissues, following the corresponding culture protocol for each strategy (Figs. 3C1-3C2). The bioprinted tubular shape was preserved during 1-day culture in Carbopol post-bioprinting and after removal of the tissue from the Carbopol (Fig. 3D1) We also demonstrated the bioprinting of bone tissue using osteogenic spheroids as building blocks. Osteogenic spheroids were fabricated in three different groups from MSCs, and their differentiation was characterized in detail (Fig.  S3). In Group 1, MSC spheroids were formed on Day 0 and cultured in a osteogenic differentiation medium for 28 days. In Group 2, spheroids were formed after MSCs were cultured on the tissue culture plate (TCP) for seven days, followed by an additional 21 days in osteogenic differentiation media. In Group 3, MSCs were culture on TCP in osteogenic differentiation media for 12 days before spheroids were formed. Spheroids were then cultured in a osteogenic differentiation medium for additional 16 days. For each group, spheroids

Applications of the AAfB technique
were collected for analysis purposes on Days 14 and 28.
When the spheroids were compared on Day 28, H&E staining for Group 3 showed considerable more bone matrix deposition as compared to Groups 1 and 2 ( Fig.  4A1-A3). Confocal images of Group 3 demonstrated the strongest expression of OSTERIX, which is a late-stage osteogenic differentiation marker (Fig.  S4). spheroids (spheroids formed after 12-day 2D differentiation followed by 2-day 3D differentiation).
In this group, bioprinted tissues were cultured in the osteogenic differentiation media for 13 days after removal from the Carbopol.
We bioprinted triangle-shaped osteogenic tissues using six spheroids following these three BMP-4 (6.2-fold increase) were significantly higher in Strategy II as compared to those in Strategy I. Our results indicate that the longer the cells are exposed to induction media on 2D, the more pronounced the osteogenic differentiation in spheroids as well as bioprinted tissues.

Discussion
Although extrusion-based bioprinting in granular gels has already been demonstrated in the literature 1,5-7,10,13,30 , its utilization in bioprinting of prefabricated cellular aggregates is quite challenging. Here, we presented a novel approach with the ability to bioprint cellular aggregates such as tissue spheroids in an accurate and precise manner in 3D. In this study, the presented AAfB approach enabled the freeform biofabrication of 3D complex-shaped constructs using spheroids as building blocks: we want to stress that this is not similarly achievable using existing bioprinting methods 15,16,19 . In addition to its strength in positioning of spheroids in 3D, the AAfB approach also made it possible to bioprint spheroids with a wide range of sizes (Fig.  2). While in this study we only utilized spheroids with radii ranging from 150 to 450 μm, we could readily modify the system to enable the bioprinting of spheroids with dimensions ranging from 100 μm up to almost 1 mm.
Bioprinting positional accuracy increased with the concentration of Carbopol. Due to the shear thinning behavior of the granular gel, when the Carbopol concentration was low (e.g., 0.8%), the granular gel liquefied and maintained insufficient viscosity and self-healing properties to hold the bioprinted spheroids in place accurately. The positional accuracy was increased with increasing Carbopol concentration. However, higher levels of aspiration pressure is required in these cases to transfer the spheroids from their initial location to their final placement. The higher aspiration pressure might induce substantial spheroid damage, such as their breakage during transition into the gel or their complete aspiration into the nozzle. Consequently, to exploit the potential of this new technique, it is crucial to determine of optimal gel properties and bioprinting speeds to guarantee the spheroids' accurate placement while preserving their integrity and viability. Thus, Carbopol concentration of 1.2% was preferred to use throughout our experiments.
While it might be convenient to assume that the gel properties are uniform within the entire gel domain, our empirical observation is that the cell medium diffuses into the gel and changes the gel properties accordingly. In particular, we note that the gel properties changed considerably when the bioprinting time was prolonged. This said, we managed to bioprint larger constructs by minimizing the total volume of medium in order to control the issues related to the diffusion induced gel heterogeneity. Clearly, the bioprinting time could be minimized by increasing the bioprinting speed. However, a substantial increase in the bioprinting speed could also result in failure as spheroids could easily get stuck at the medium/gel interface due to the substantial resistance exerted by the gel (Supplementary  Video  2). In this regard, we used 2.5 mm/s as our preferred bioprinting speed, which allowed for a rapid enough assembly of the presented tissue models while remaining safe enough to successfully transfer the spheroids from the cell media to the gel. Meanwhile, we also maintained an amount of medium sufficient to support the growth and viability of the bioprinted tissues. Also, we kept the bioprinted tissues in the Carbopol gel for only one day as one day was sufficient to induce partial fusion of spheroids and provide necessary structural integrity. Although we attempted to minimize the issues that could be encountered due our modeling did not include any considerations on the deformability of the spheroids, which is of primary concern when assessing viability. The future enhancement of the proposed bioprinting technique, especially when trying to tune the gel's physical properties to achieve an optimal printing accuracy and viability, can benefit from a more sophisticated analysis of spheroid motion mechanics.
In our attempts preceding this study, we aspirated and lifted spheroids using a glass pipette with a radius of about 40 μm (see our recently published work 21 ). We encountered problems in the use of a pipette when we transitioned spheroids in the gel domain, i.e., as we crossed the medium/gel interface. As depicted in Supplementary Video 3, spheroids were prone to bounce at the pipette tip because of insufficient aspiration forces against the drag force, which was due to the reduced exposure area of aspiration. In addition, as the pipette enlarged significantly toward its upper portion, we observed other issues such as substantial damages to the gel along with slower and diminished healing. Because of these reasons, we switched to metallic straight nozzles with a larger nozzle radius (inner radius of 100 μm). As long as we bioprinted spheroids with a radius of at least 150 μm at least, the metallic straight nozzles proved sufficient to perform the presented bioprinting work. This said, smaller nozzle tips or even pipette tips could still be utilized for bioprinting of spheroids with radii smaller than 50 μm.
In this study, chondrogenic spheroids were bioprinted into a tubular arrangement using two different strategies in order to understand the role of MSC or chondrogenic spheroids in successful bioprinting of tubular cartilage tissues. We identified considerable differences between MSC and chondrogenic spheroids in term of biological, structural, and mechanical properties. In particular, MSC spheroids shrank in size while chondrogenic spheroids grew over time, which could be due to the significant deposition of chondrogenesis-related extracellular-matrix (ECM) deposition, which in turn yielded higher surface tension and sGAG content in chondrogenic spheroids. Bioprinting of chondrogenicly differentiated spheroids generated tissues with improved chondrogenic properties and shape fidelity.
In bioprinting of osteogenic spheroids, three different strategies were designed to study the role of monolayer versus 3D induction on successful formation of bone tissue with controlled morphology. The results indicated that longer culture period in monolayer improved the construct fidelity as evidence by the result of Strategy III (Fig. 4E). This could be due to the increase exposure of MSCs to osteogenic differentiation media (REF) or improved osteogenesis of MSCs due to the substrate stiffness of TCP 33,34 . As MSCs in 3D spheroid culture had limited integrinmediated adhesion with respect to TCPs, we observed enhanced bone formation at the gene and protein level 33 . It is also known that, osteogenicly differentiated MSCs could have limited proliferation, which might reduce the fusion and compaction of osteogenic spheroids in bioprinted bone tissues. Along with successful bioprinted outcomes, we also encountered some failures which we believe can be overcome with improved gel properties and pH control, or the use of other yield-stress gels.
In sum, we presented a highly effective approach in 3D bioprinting and positioning of tissue spheroids by explaining the interplay between the bioprinting process and gel yield-stress. Such a platform enabled us to pattern tissue spheroids with a high degree of geometric complexity, which will have tremendous applications such as, but not limited to, tissue engineering and regenerative medicine, disease modeling, drug screening, and biophysics.

Preparation of the granular hydrogel
To prepare the granular gel, 0.8, 1.2 and 1.6% (w/v) Carbopol ETD 2020 NF (Lubrizol Corporation, OH) were dispersed in human chondrogenic or osteogenic differentiation media (Cell Applications, CA) under sterile conditions. NaOH was added drop-wise to the Carbopol-dispersed gel to adjust the pH to 7.4, which facilitated the maximum swelling of Carbopol and its biocompatibility. The GH was then homogenized using a vortex blender for 20 min, centrifuged at 1000 x g for 15 min, and incubated at 37°C and 5% CO2 before further use. Cells were cultured in MSC growth media during spheroid formation, and the medium was changed every three days. MSC spheroids were differentiated into chondrogenic and osteogenic lineages for different applications using human chondrocyte and osteoblast differentiation media, respectively (Cell Applications, CA).

Rheological analysis of the granular gel
Rheological measurements of the granular gel were performed using a MCR 302 rheometer (Anton Paar, VA) using a 25 mm diameter parallel-plate geometry. A Peltier system was employed for temperature control. Amplitude tests were applied to determine viscous and elastic properties of the gel at a constant frequency of 1 Hz and a strain range from 0.01 to 100% at a constant temperature of 25 °C.

AAfB process
For the bioprinting setup, we utilized our previously developed AAB system 21 , a square petri dish was used to hold the gel and cell culture media (Fig.  1A), where a Polydimethylsiloxane (PDMS) slice was used to transfer and confine the gel in order to obtain a vertically-oriented interface.
Spheroids were placed in a reservoir which was submerged in the tissue-specific media. When the reservoir was transferred into the Petri dish, tissue specific cell media was filled to cover the remaining area in the Petri dish. A 27G needle (Nordson, OH) was used to pick the spheroids from the reservoir and to transfer them from cell culture media into GH, with a speed of 2.5 mm/s. Two microscopic cameras (one for each side of the Petri dish) were used to visualize the bioprinting process in real time. In order to validate the theoretical results, MSC spheroids with a wide range of radius (from 150 to 400 micron) were bioprinted into 1.2% Carbopol medium (GH).

Accuracy and precision measurement of bioprinting
In where •‚=FG• and •‚=FG• represent X and Y coordinates of the target position, respectively, ƒ and ƒ are the position of the measured values in X-and Y-axis, respectively, and is the sample size. Precision was represented as the square root of the standard deviation.

Physical properties of MSC and chondrogenic spheroids
The chondrogenic differentiation of MSC spheroids was started on Day 3, and the radius of spheroids were measured by an EVOS ® microscope (Invitrogen, MA) until Day 24. Surface tension of spheroids was also measured according to the protocol described in our previous published work 35

Gene expression of osteogenic spheroids using quantitative real-time polymerase chain reaction (RT-qPCR)
In order to investigate the effect of different strategies on the osteogenesis of spheroids, 3 groups were designed. In Strategy I, spheroids were prepared from MSCs and cultured for 28 days in osteogenic differentiation media. In Strategy ii, MSCs were cultured in monolayer for seven days, followed by fabricating and culturing spheroids for 21 days in osteogenic differentiation media. In Strategy III, MSCs were cultured in monolayer for 12 days, followed by fabricating and culturing spheroids for 16 days in osteogenic induction media. For all groups, the total induction period in monolayer culture and in the form of spheroids was kept 28 days in total.
For testing of bone-specific gene expression using RT-qPCR, single differentiated spheroids per sample were homogenized in TRIzol reagent (Life Technologies, CA), followed by adding 0.2 mL chloroform per 1 mL TRIzol reagent and centrifuging the mixture at 12,000 x g for 15 min at 4˚C.
The upper aqueous phase with RNA was transferred and RNA was then precipitated by adding 0.5 mL isopropyl alcohol per 1 mL TRIzol reagent, followed by centrifuging at 12,000 x g for 10 min, at 4˚C. Subsequently, the precipitated RNA was rinsed twice by 75% ethanol, air-dried for 10 min and dissolved in 50 µL diethyl pyrocarbonate (DEPC)-treated water. RNA concentration was measured using a Nanodrop (Thermo Fisher Scientific, PA). Reverse transcription was performed using AccuPower® CycleScript RT PreMix (BIONEER, Korea) following the manufacturer's instructions. Gene expression was analyzed quantitatively with SYBR Green (Thermo Fisher Scientific, PA) using a QuantStudio 3 PCR system (Thermo Fisher Scientific).
Bone-specific genes tested included OSTERIX (Transcription factor Sp7), COL-1, OCN (osteocalcin), BMP-4 (Bone Morphogenetic protein-4) and BSP (Bone sialoprotein). The reader is refereed to Table  1 for the gene sequences. Expression levels for each gene were then normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The fold change of hMSCs spheroids after formation on Day 2 was set as 1-fold and values in osteogenic groups were normalized with respect to that of the group.

Bioprinting of osteogenic tissues
In order to investigate the effect of different osteogenic strategies on formation of bioprinted bone tissue, 3 groups were designed. In Group 1, spheroids were prepared using MSCs and cultured with osteogenic induction media for 14 days. In Group 2, MSCs were cultured with osteogenic induction in monolayer for seven days, followed by fabricating and culturing spheroids for seven days with osteogenic differentiation media. In Group 3, MSCs were cultured with osteogenic induction in monolayer for 12 days, followed by fabricating and culturing spheroids for two days with osteogenic induction media. After bioprinting of osteogenic spheroids, the excess amount of Carbopol was gently removed (without affecting the structural integrity of the bioprinted constructs) in order to maximize the diffusion of cell media to better support the growth of the tissue. For all groups, the total differentiation period in monolayer culture and in the form of spheroids were kept 14 days in total. Triangle bone structure were then bioprinted and cultured for 14 days in osteogenic differentiation media for a total of 28 days culture for all groups.
RT-qPCR of bioprinted tissues were conducted as described in Section 3.11. H&E staining was carried out to visualize the morphology as described in Section 3.7.

Statistical analysis
All values were presented as mean ± standard deviation. Multiple comparisons were analyzed by using one-way analysis of variance (ANOVA) by Post-hoc Tukey's multiple-comparison test was used to determine the individual differences among the groups. Differences were considered significant at *P < 0.05, **P < 0.01, and ***P < 0.001, and ****P < 0.0001. All statistical analysis was performed by Statistical Product and Service Solutions software (SPSS, IBM, USA).