Bioprinting-Based High-Throughput Fabrication of Three-Dimensional MCF-7 Human Breast Cancer Cellular Spheroids

Cellular spheroids serving as three-dimensional (3D) in vitro tissue models have attracted increasing interest for pathological study and drug-screening applications. Various methods, including microwells in particular, have been developed for engineering cellular spheroids. However, these methods usually suffer from either destructive molding operations or cell loss and non-uniform cell distribution among the wells due to two-step molding and cell seeding. We have developed a facile method that utilizes cellembedded hydrogel arrays as templates for concave well fabrication and in situ MCF-7 cellular spheroid formation on a chip. A custom-built bioprinting system was applied for the fabrication of sacrificial gelatin arrays and sequentially concave wells in a high-throughput, flexible, and controlled manner. The ability to achieve in situ cell seeding for cellular spheroid construction was demonstrated with the advantage of uniform cell seeding and the potential for programmed fabrication of tissue models on chips. The developed method holds great potential for applications in tissue engineering, regenerative medicine, and drug screening.


Introduction
In vitro tissue models such as those based on cellular sphe roids have attracted increasing interest in cellular biology, tis sue engineering, regenerative medicine, and drugscreening applications [1][2][3][4].Cellular spheroids are threedimensional (3D) aggregates of multicells, which usually scale from tens to hundreds of micrometers.It is widely accepted that cells in cellular spheroids behave and respond to changes in mi croenvironment cues in a much more natural manner than those on traditional twodimensional (2D) culture plates [5,6].In addition, cellular spheroids are inherently scaffold free as compared to scaffoldbased 3D models; this property may help cellular spheroids to avoid biocompatibility issues [7].Different types of cells have been engineered into cellular spheroids, including normal [8,9], cancerous [10,11], and stem cells [12,13], as well as a mixture of cell types [14,15].In particular, the introduction of chip technologies has recently inspired the emergence and prominent development of tis sue/organonachip technology, which is believed to hold great promise in biomedical fields [16][17][18].
Various methods have been developed for generating cel lular spheroids by exploring the fluidic and selfassembly abilities of cells [19].A facile method is to culture cells on nonadherent substrates [20], where the cellcell interaction is stronger than that of cellsubstrate, resulting in the aggrega tion of cells.To promote the assembly of cells, gravitational and magnetic forces have been explored, leading to the de velopment of the hangingdrop [21] and magneticlevitation methods [22], respectively.Microwells have also been widely used to confine the growth space of cells with the benefit of forming cellular spheroids with controlled and uniform size [11].Among these methods, microwells have attracted in creasing attention; especially those with concave structures, due to advantages such as easy operation, good controllabili ty, and highthroughput capacity [23,24].However, almost all of the current microwell methods require the manual seed ing of cells after the preparation of a microwell plate, which may be cumbersome and can cause cell loss and nonuniform cell seeding [25].In addition, special templates and a careful molding process are required to obtain well-defined concave microwells, which limits their accessibility.Therefore, there

Research
is still an urgent need for a flexible manufacturing method for fabricating cellular spheroids, particularly those with the potential for engineering tissues on chips.
In this study, we developed a bioprintingbased method for fabricating concave wells and generating cellular spheroids in situ on a chip.A custombuilt bioprinting system was first constructed and applied in order to generate cell-laden sacrificial gelatin arrays.These cellladen gelatin arrays were subsequently used as templates for fabricating concave wells and forming cellular spheroids on a nonadhesive hy drogel substrate.The method is facile and, to the best of our knowledge, represents the first use of biocompatible hydrogels as templates for fabricating concave wells with in situ cellseeding ability for cellular spheroid formation.

Set-up and improvement of the bioprinting system
To get high controllability for the droplet generation of gelatin sol, a pressure assisted valuebased bioprinting system was applied.This system was custom built, and is shown in Figure 1(a).The system was set up in a vertical laminar-flow clean workbench, with the main components being a threeaxis motion stage (KDT180 100LM (XY) and MT10550LM (Z), Feinmess Dresden GmbH, Dresden, Germany), a solenoid valve ejector (Model G100150300, TechElan, Mountainside, NJ, USA), and a digital signal generator (Agilent 81101A, Test Equipment Connection, Lake Mary, FL, USA).All of these components were controlled using a computer.line (PBS) solution at a concentration of 3% (w/v) and gently stirred at 50 °C un til fully dissolved.The sol solution was then waterbathed to 37 °C, transferred to the sampleholding syringe, and printed onto a culture petri dish coated with a polytetrafluoroethene (PTFE) membrane.The PTFE membrane was used because it is physically hydropho bic and chemically inert, as well as easy to use.The temperature of the gelatin solution in the syringe was kept at 37 °C during printing.The assisted printing pressure and pulse width of the signal were adjusted to control the printed volume of the gelatin droplet and thus the size and shape of the gelatin arrays.The petri dish with the printed gelatin arrays was immediately transferred to a refrigerator and cooled at 4 °C for 5 min in order to gel the gelatin precusor.

Gelatin-array templated fabrication of concave wells
Polyethylene glycol (PEG)based hy drogels were used to mold concave wells from gelatin arrays due to their biocompatibility, photocrosslinkability, and cellnonadhesive properties.These hydrogels have also been used by oth ers to fabricate wells for the formation of cellular spheroids [26][27][28].In our experiment, PEGdimethacrylate (PEG DMA, MW 1000, Polysciences, Inc., Warrington, PA, USA) was dissolved in a PBS solution at a concentration of 20% (w/v) and then cooled to 4 °C.After be ing poured onto the petri dish with the gelatin arrays, the PEGDMA solution was exposed to 365 nm ultraviolet (UV) light with a power of 2.9 mW .cm -2 (Mod el XLE1000 A/F, Spectroline, Westbury, NY, USA) for 20 s for gelling.During the photocrosslinking, 2hydroxy 2methylpropiophenone (TCI Shang hai Development Co., Ltd., Shanghai, China) was used as the photoinitia tor at a concentration of 0.1% (w/v).After incubation at 37 °C for 24 h, the gelatin arrays redissolved into a sol phase and concave wells formed on the PEGDMA substrate.

Characterization of gelatin arrays and concave wells
The gelatin arrays were imaged with an inverted fluorescence microscope (Olympus IX 81, Olympus, Irvine, CA, USA) before molding with PEGDMA.

Bioprinting-based fabrication of sacrificial gelatin arrays
Gelatin was chosen for the sacrificial hydrogel arrays due to its biocompatibility and its injectable and reversible gelling abilities under mild temperature condi tions.To fabricate the gelatin arrays, porcineskin gelatin powder (gel strength 300, Type A, SigmaAldrich, St. Louis, MO, USA) was added to a phosphatebuffered sa To image the wells, the PEGDMA substrate was gently de tached from the PTFEcovered petri dish, and swelled to equilibrium in a PBS solution.After being taken out of the PBS solution, the excess PBS solution on the PEGDMA sub strate surface and in the wells was gently removed with blot ting paper.To capture a crosssection of the wells, the PEG DMA substrate was carefully cut through the wells with a razor blade and images were immediately taken with an IX 81 microscope.The images were analyzed with ImagePro Plus (IPP, version 6.0, Media Cybernetics, Silver Spring, Rock ville, MD, USA) to quantify the sizes of the gelatin arrays and wells on the PEGDMA substrate.

In situ cell seeding and cellular spheroid formation
We used MCF7 human breast cancer cells to verify the in situ cellseeding ability of our method for the highthrough put fabrication of cellular spheroids on a chip.Cells were cultured in RPMI 1640 medium (HyClone, South Logan, UT, USA) and supplemented with 10% fetal bovine serum (Gibco Industries, Inc., Big Cabin, OK, USA) in a humidified 5% CO 2 incubator at 37 °C.Before the experiment, the cells were digested with trypsin (EDTA 1×, Mediatech, Inc., Manassas, VA, USA) and centrifuged, and the liquid supernatant was re moved.The cells were then resuspended in the culture medi um.The prepared cellular suspension was then gently mixed with a prefiltered gelatin solution at a final cell density of 5 × 10 5 cells .mL -1 to 1 × 10 6 cells .mL -1 and a gelatin concentra tion of 3% (w/v) at 37 °C.The cellgelatin mixture was then loaded into the syringe for the fabrication of the gelatin ar rays and wells as described in Sections 2.2 and 2.3.The petri dish cover was removed after incubation for 1 h and the cell culture medium was changed daily.Phasecontrast images of the cellular spheroids were observed with an Olympus IX 81.Confocal fluorescence images (LSM700, ZEISS, New York, NY, USA) were observed after the spheroids or aggregates were stained using a live/dead viability/cytotoxicity kit (In vitrogen, Washington, DC, USA).The kit stained live cells green with calcein AM and dead cells red with ethidium ho modimer-1.Cell viability was quantified from confocal slices using IPP.

Results and discussion
The bioprinting system we constructed is shown in Figure 1(a).High controllability of the bioprinting system for pro ducing droplets of nL volume was demonstrated; this control lability can be achieved by changing the pneumatic pressure or the valveopening duration, as shown in Figure 1(b) and Figure 1(c), respectively.The optimized concentration of the gelatin sol solution was found to be ~3% (w/v), below which the gelatin arrays were too soft to be used as templates and above which valve blocking and solution ponding occurred.
The bioprinting system was then applied to produce con trolled gelatin arrays for fabricating concave wells and in situ cellular spheroid formation, as illustrated in Figure 2. First, gelatin arrays of different sizes were created on a hy drophobic PTFEcovered petri dish, as shown in Figure 3(ac).After molding, concave wells were formed on the PEG DMA substrate with almost the same size as the gelatinarray templates, as shown in Figure 3(d-f), demonstrating the high molding fidelity and thus the feasibility of using gelatin ar rays as templates.By tuning the volume of the printed gelatin droplets from 0.1 mL to 0.48 mL, gelatin arrays with a radius ranging from 420 mm to 530 mm and a depth ranging from 230 mm to 550 mm can be easily obtained, as shown in Figure 3(g).This method led to the formation of concave wells with a radius ranging from 480 mm to 530 mm and a depth ranging from 250 mm to 550 mm, as shown in Figure 3(h).The droplet volume can be easily tuned by adjusting either the printing pressure or the pulse width of the digital signal.Since high pressure may lead to cell damage and an unwantedly force ful impact between the gelatin droplet and the substrate, we only tuned the pulse signal width to control the gelatin droplet volume.Wells with a high depth may have the ad vantage of avoiding cell loss during the medium exchange.We expect that wells with a much more controlled size and a higher molding fidelity could be fabricated by optimizing the production of the gelatin arrays, which may be achieved by improving the gelatin droplet hydrophobic angle by use of more hydrophobic substrates.
To prove the in situ cellseeding ability of our method for cellular spheroid formation on a chip, we first mixed MCF-7 cells with a gelatin precursor at a density of 1 × 10 6 cells .mL -1 and printed the mixture at a volume of 0.48 mL per droplet.After molding and incubation at 37 °C for 1 h, the gelatin arrays reversed to the sol phase and the cells, encapsulated in gelatin, were released and deposited onto the bottom of

Research
Engineering Volume 1 • Issue 2 • June 2015 www.engineering.org.cnspheroids on the 7th day, as shown in Figure 4(a).We observed very few dead cells throughout the process, as shown in Figure 4(b).Cellular spher oid formation with a printed volume of 0.23 mL and the same cellseeding density (i.e., 1 × 10 6 cells .mL -1 ) was also studied.Although cellular aggregates were observed after 7 days of culture, these were not as compacted as those in the 0.48 mL group, as shown in Figure 4(c).The observed cellular aggregates did not change notably even after 14 days of culture (results not shown here).The horizontal crosssectional confocal fluorescence image in Figure 4(b) shows that there is a black central domain without stained cells, which may be induced by diffusion limitation of the live/dead solution in compacted cellular spheroids.This central domain is not observed in Figure 4(d), indicat ing that the formed cellular aggregates with a printed gelatin volume of 0.23 mL may be in a disk shape.From Figure 3(f) and Figure 3(h), we know that the fabricated wells with printed gelatin volumes of 0.23 mL and 0.48 mL per droplet have radiuses of 480 mm and 530 mm, and depths of 390 mm and 550 mm, respectively.Thus, it seems to us that well depth may play an important role in the formation of compacted cel lular spheroids.
We further investigated the influence of initial in situ cellseeding density  wells formed on the PEGDMA substrate.During the subsequent culture, the cells proliferated, huddled together, and finally self-assembled into compacted cellular on the formation of cellular spheroids.The results clearly showed that much smaller cellular spheroids were formed at a cellseeding density of 5 × 10 5 cells .mL -1 , as shown in Figure 4(e)-(h), compared to a density of 1 × 10 6 cells .mL -1 , as shown in Figure 4(a)-(d).Therefore, it is possible to control the size of cellular spheroids by adjusting the in situ cellseeding den sity.Figure 4(i) shows that high cell viability was maintained during cellular spheroid formation.In addition, cell loss was effectively avoided with the in situ cellseeding ability.More over, uniform distribution of seeded cells between wells was achieved due to the controllability of the bioprinting system.Morphologically speaking, cellular spheroids can better mimic the structure of organoids and thus the microenviron ment of cancer cells in vivo than can traditional 2D culture plates.It is therefore expected that MCF7 may behave simi larly in cellular spheroids and in vivo.To demonstrate this, further studies including a drug test are needed; these will be our next work.Although we only tested MCF7 cells here, we believe that other cell types can be used, since gelatin has been widely applied for cell encapsulation and since a polyethylene glycol (PEG) hydrogelbased well plate has been successfully used for producing cellular spheroids of various cell types, including embryonic stem cells [29,30].
Several researchers have demonstrated the use of molded or printed gelatin sacrificial elements for fabricating micro fluidic hydrogels.In this work, we employed gelatin arrays as templates for the first time to generate concave wells for cellular spheroid formation.Although other templates, in cluding solid resin mold, SU8, polydimethylsiloxane (PDMS), and ice array, have been developed for fabricating concave wells, these methods suffer from either cumbersome mold preparation or destructive demolding operations [23,[31][32][33].In addition, the post manual cellseeding step may cause cell loss and nonuniform cell distribution.Here, by using a biocompatible and temperaturereversible cellladen gelatin array template, we provide a possible solution to overcome the above limitations.Moreover, the use of a custombuilt bioprinting system may enable us to fabricate cellular spher oids in situ on a chip in a controlled and highthroughput manner, which will be of great help in engineering tissues on chips.

Conclusions
This work demonstrated the fabrication of concave wells molded from a hydrogel array with in situ seeding of cells for cellular spheroid formation on a chip.The use of a custom built bioprinting approach endows the method with high controllability and a highthroughput capacity.The integra tion of bioprinting and in situ cell seeding with gelatin may enable highly programmed fabrication of cells/tissues on chips, and thus holds great promise for tissue engineering, regenerative medicine, and drugscreening applications.

Figure 1 .
Figure 1.Bioprinting platform.(a) Image of the bioprinting platform; (b) mean droplet volume as a function of pneumatic pressure, and the valve-opening duration is 5 ms; (c) mean droplet volume as a function of valve-opening duration, and the pneumatic pressure is 0.05 MPa.Error bars represent standard error (n = 5).

Figure 2 .
Figure 2. Schematics of concave-well fabrication and in situ cell seeding for cellular spheroid formation.(a) Cover the petri dish with a film layer of hydrophobic PTFE; (b) print gelatin and cell solution onto PTFE-covered petri dish; (c) cell-encapsulated hydrogel array forms when cooled at 4 °C; (d) pour the cold PEG-DMA solution onto the cell-encapsulated gelatin array and (e) expose it to UV light for crosslinking; (f) turn it over and incubate at 37 °C to liquefy the gelatin and release cells onto the well bottom; (g) cellular spheroids form during further culture.

Figure 3 .
Figure 3. Controlled fabrication of hydrogel concave wells.(a) Overall view of printed gelatin arrays of different sizes on a PTFE-covered petri dish; (b) top-view microscope image of the gelatin arrays; (c) side view of two typical gelled gelatin droplets; (d) overall view of concave wells formed on PEG-DMA; (e) top-view microscope image of PEG-DMA wells; (f) side view of the cross-section of the concave wells; (g) controllability of the sizes of gelatin arrays; (h) controllability of the sizes of concave wells.Scale bars in (b), (c), (e), and (f) are 0.5 mm.Error bars in (g) and (h) represent standard deviation (n = 5).

Figure 4 .
Figure 4. Cellular spheroid formation in hydrogel concave wells.(a) Phase-contrast images of cellular spheroid development and (b) calcein and ethidium bromide staining of cells after 7 days of culture in wells made of 0.48 μL gelatin droplets and with a cell-seeding density of 1 × 10 6 cells•mL -1 ; (c) phase-contrast images of cellular spheroid development and (d) calcein and ethidium bromide staining of cells after 7 days of culture in wells made of 0.23 μL gelatin droplets and with a cell-seeding density of 1 × 10 6 cells•mL -1 ; (e, f) phasecontrast images of cellular spheroid formation after 7 days of culture in wells made of (e) 0.48 μL and (f) 0.23 μL gelatin droplets, with a cell-seeding density of 5 × 10 5 cells•mL -1 ; (g, h) live/dead fluorescence images corresponding to (e) and (f), respectively; (i) cell viability during cellular spheroid formation.Scale bars in (a)-(h) are 200 μm.