3D-bioprinted all-inclusive bioanalytical platforms for cell studies


Design principles and considerations

The ability of OOCs systems to mimic cell functions that is representative of organ-level physiology allows for the generation of novel platforms to examine disease mechanisms, progressions, and testing of candidate treatments20,21. However, the design, production, and successful applicability of OOC devices require several critical deliberations including rapid manufacturability of fluidic perfusion systems, development of cell models with sufficient physiological microenvironment relevance, and combined execution of these two into a single cohesive platform. For in vitro drug screening testing systems, for instance, it is critical to examine different therapeutics or their concentrations on the same model population. The incorporation of fluidic handling devices such as micromixers enables the preparation and delivery of different doses of (or combinations of) therapeutics within a chip flow circulation. Development, characterization, and validation of such devices are some of the efforts in this study.

On the other hand, the generation of suitable 3D-bioprinted constructs requires the optimization of multiple parameters simultaneously to achieve the desired resolution for diverse 3D-constructs. 3D-bioprinting involves the precise layering of cells, biologic scaffolds, and growth factors capable of creating 3D-cell constructs for a variety of applications27. The use of 3D-bioprinting requires optimization of parameters such as extrusion pressure, printing speed, and the use of proper nozzles to achieve desired resolutions, which is part of our reported study, as explained in “3D-bioprinting for fluidic handling systems” section. Moreover, the generation of accurate representations of the complicated cell physiological microenvironments, similar to those found in 3D-cell models, is extremely important. This is an especially crucial feature for drug testing platforms to provide 3D-cell models, as it is shown that these models can improve preclinical drug efficacy or failure predictions27. Such 3D-cell models must be able to replicate the intricate cell environment while the platform must encompass arrayed features (for high throughput testing) that can be flexibly designed to various sizes, and compartments to promote the desired environment. The use of 3D-bioprinting technologies lends itself to the flexible printing of both the platforms and 3D-cell structures. In this work, considering these requirements, we adapted 3D-bioprinting technology to develop customized OOC-like platforms along with 3D-cell structures for the analysis of 3D-cell culture models as well as the study of cell-therapeutic responses. Materials also play a critical role in 3D-bioprinting since the material properties can affect the fabrication process and the application of the 3D-printed components. Scientists developed several materials for these applications according to their characteristics such as transparency, printability, viscosity, and flexibility32,33. Transparency of matrices is one of the important features in fluidic handling devices (e.g., microfluidics) since visual observations are usually required in such devices. In biological applications, the biocompatibility of these materials is another critical parameter. Many researchers extensively utilized a variety of polymers and thermoplastic materials for 3D-bioprinting applications. Solidifiable fluids such as photopolymer resins, temperature-sensitive polymers, and ion cross-linkable hydrogels are the most commonly used materials for 3D-bioprinting in recent years25,34,35,36. Polydimethylsiloxane (PDMS), for instance, is one of the most commonly used materials to manufacture microfluidic devices because of its excellent transparency and biocompatibility29,37. Here, we first demonstrate 3D-bioprinted spherical and rectangular cell constructs within 3D-bioprinted well arrays. One should note that the physically defined well arrays can potentially be useful for growing cells onto the scaffold with the ultimate goal of in vivo transplantation to the bone, cartilage, ligament, skin, vascular, neural, and skeletal muscle tissues38. It is also worth noting that the recent studies confirm that cells cultured using well arrays exhibit different enzyme expression levels and drug reactivity compared to culturing in traditional 2D format39. Besides material selection, the use of 3D-bioprinting requires other critical considerations, including limiting induced mechanical stress during 3D-bioprinting of the structures, supplying cells with nutrients during post-printing culture, using suitable bioinks, and monitoring and maintaining printed constructs viability and proliferation process. For instance, extrusion pressures and nozzle sizes should be optimized to not inflict shear stress on cells suspended in viscous fluids and decrease the cells’ survival rates. In this work, we aimed to carefully consider and optimize these requirements, as explained in “3D-bioprintability and viability monitoring of 3D-HCT116-constructs” section. In this study, human colorectal cancer cells were used, since colorectal cancer is the third most prominent cause of cancer-related deaths among women and men40. Though many developments in early detection and treatment have been shown for colon cancer, there is still room for therapeutic improvement, especially for those patients with invasive and aggressive tumors presenting drug resistance41. One of the other main considerations in 3D-bioprinting of cell constructs is maintaining the post-viability of printed cells. Here, we printed and monitored arrays of 3D-bioprinted encapsulated cell constructs, with both spherical and rectangular shape, within 3D-printed PDMS well arrays. The spherical model was selected as they resemble the compact arrangement of tumors and the rectangle shape was selected as they resemble a simple model of a vasculature structure. Experiments showed successful cell viability, as explained in the “Viability analysis of 3D-HCT116-constructs” sub-section. Lastly, to more closely mimic the in vivo environment by replicating the mechanical cues such as fluid flow (i.e., shear stress) that tissues are subject to within the body, we continued the effort by the development of an OOC-like platform. This device is a perfused well-based fluidic handling platform, consists of two main parts: fluidic delivery channels and concave wells. The fluidic delivery channels were designed to subject cell constructs to physiological fluid flow, and at the same time, deliver nutrients. The platform also consisted of a concave well array that held the 3D-human colon cancer cell constructs (HCT116). Next, we aimed to optimize the process to maintain shape and stability at physiological temperature, while at the same time simulating an optimal microenvironment (e.g., adhesion sites) for cells. For this, we used, Gelatin methacryloyl (GelMA), a gelatin-based bioink that can maintain shape and stability at physiological temperature, has proven biocompatibility, and provides mammalian cells with a milieu that resembles some essential properties of their native environment. GelMA also has high content of arginine-glycine-aspartic acid (RGD) useful for cell attachment, and target sequences of matrix metalloproteinase (MMP) that assist in cell proliferation which are desirable attributes necessary for our printed cell structures42,43,44. GelMA HCT 116 cell constructs were printed into the well arrays and perfused with media using the fluidic handling channels. In this platform, printed GelMA HCT 116 cell structures formed ring/toroidal shapes within a day after printing, which are useful structures to model the tubular geometry of the colon, where tumors are usually found attached to the inner wall of the large intestine. Finally, preliminary drug screening investigations were performed with 2D HCT116 cell models within 3D printed PDMS well arrays.

3D-bioprinting for fluidic handling systems

Characterization of rapid manufacturing of fluidic handling systems

The rapid and precise manufacturing of OOC devices remains a significant hurdle in their implementation as novel platforms for in vitro disease studies and therapeutic screening. Here, we used micro extrusion-based 3D-bioprinting and its short turnover times for the customized (i.e., different geometries and dimensions) production of fluidic handling systems, as a component of our OOC devices. To do so, we first performed characterization experiments that are required to find the optimum 3D-bioprinting parameters for Pluronic F-127. Pluronic F-127 is a useful class of synthetic block copolymers that is biocompatible and works well with PDMS45. For initial characterization experiments, we first determined the minimum extrusion pressures required to achieve continuous filament printing for the different nozzle diameters. According to our results, the minimum extrusion pressures required were 90, 105, and 110 kPa for conical nozzles with IDs of 410, 250, and 200 µm, respectively (Fig. 1A). Next, we evaluated the minimum extrusion pressure for different shaped nozzles with the same inner diameter (ID:410 µm) (Fig. 1B). It was found that needle-shaped nozzle required a minimum extrusion pressure of ~ 200 kPa, while the conical nozzle necessitated ~ 90 kPa. Once the minimum extrusion pressures were known for each respective nozzle ID and shape, the pressure was fixed at those values and the effect of printing speed on filament feature widths was studied for conical nozzles with ID of 200 µm (Fig. 1C), 250 µm (Fig. 1D), 410 µm (Fig. 1E), and the needle nozzle with ID of 410 µm (Fig. 1F). From this analysis, we determined that higher printing speeds, yield minimum filament widths of 70, 107, 184, 97 µm for conical nozzle with ID of 200, 250, and 410 µm, as well as, the needle nozzle with ID of 410 µm, respectively. From this data, we are able to determine optimum printing parameters for the desired minimum filament resolution of our printed structures.

Figure 1

(A) Impact of extrusion pressure on the printed filament width for conical nozzle tips with different inner diameters (200, 250, 410 µm) (B) for conical and needle-shaped nozzle with same ID (410 µm). Impact of printing speed on the filament width for different conical nozzle ID: (C) 200 µm, (D) 250 µm, and (E) 410 µm and (F) needle-shape nozzle ID:410 µm. PDMS printed configurations of (G) 10 mm × 10 mm × 3 mm (L × W × H) well array (H) cross shape (I) and grid well structure. Images of PDMS-based (J) microchannels, (K) concave well connected by channels, and a (L) Y-shaped micromixer derived from printed Pluronic molds. (M) Diagram of the Y-shaped micromixer, showing the positions where mixing between a red and blue solution were optically verified. (N) Images of red and blue solutions mixing for solution flow rates of 1,3, and 5 µl/min and at positions i, ii, and iii which indicate 10 mm,75 mm and 145 mm distance from the inlet respectively.

Characterization and validation of the micromixers

In vitro drug screening systems should have the capability to examine different doses and a variety of therapeutics on the same model population. The incorporation of microfluidic devices such as micromixers allows for the preparation and delivery of different concentrations or (combinations of) therapeutics within the chip flow circulation46. In particular, Y-shaped passive micromixers are simple in design, rapid, and usually high performance. These features make them suitable candidates for a wide variety of ‘‘lab-on-a-chip’’ applications, where a rapid and homogenous mixing process is essential. Here, as a model of study, we designed and validated a low-cost and rapidly prototyped Y-shaped micromixer (Fig. 1L,M). The micromixer was manufactured from Pluronic molds that were printed with the dimensions of 800 ± 50 µm wide, 80 ± 10 µm height, and composed of two inlet streams and one outlet stream (Fig. 1L). The micromixer’s performance was then validated at three different flow rates of 1, 3, and 5 µL/min. The inlet streams were infused with red and blue dye deionized (DI) water solutions. As shown in Fig. 1M,N, at the flow rates < 3 µl/min, the homogenous mixing of the solutions occurs at ~ 75 mm distance (position ii) from the inlet (Fig. 1M). While higher flow rates > 5 µl/min require longer distances, and the homogenous mixture was observed at approximately 145 mm (position iii) from the inlet (Fig. 1M). Also, at low flow rates < 3 µl/min, the flow of micromixer mostly is laminar, and the molecular diffusion is dominant. However, by increasing the flow rate, longer mixing time is required for interfusion of two solutions in mixing zones. The threshold flow rate of proper homogenous mixing in our device was experimentally determined to be 7 µl/min.

3D-manufacturing of high throughout PDMS well arrays

OOCs in vitro drug testing platforms require array-based and multiplex systems in order to achieve rapid and high throughput performance. Here, we demonstrated the direct 3D-bioprinting of PDMS well-based arrays for the analysis of 3D-cell culture models. We are capable of printing a wide range of geometries of PDMS wells, as observed in Fig. 1G–J, including square wells, cross configurations for multidirectional perfusion, and small well grid structure. However, to analyze different printed 3D-cell constructs and cell populations in our studies, three different sizes of 3D-printed PDMS well arrays were designed and directly printed. The design and direct printing of the PDMS well arrays showed the adaptability of 3D-bioprinting for the integration of bioanalytical platforms. Then, 3D-spherical HCT 116 cell constructs were printed and analyzed within 5 mm × 5 mm × 3 mm (length, width, height) well arrays (Fig. 2A), rectangular HCT 116 cell 3D-constructs were printed and analyzed within 10 mm × 10 mm × 3 mm well arrays (Fig. 4D), and 2D HCT 116 models were analyzed within 15 mm × 15 mm × 3 mm 3D-printed PDMS well arrays (B), as explained in “3D-Printability of 3D-HCT116-constructs” and “Preliminary drug screening of SN-38 on 2D HCT116 cell models within 3D-PDMS bioprinted well arrays”. One should note that conventional microfluidic device production approaches usually require the design of multiple non-tailorable masks and multi-steps of photolithography to pattern and generate microstructure molds47,48. Those processes are then followed by PDMS casting, degassing, and curing steps to produce the final PDMS-based microfluidic device. In contrast, our process is a single step, rapid, inexpensive, and tailorable design process. Here there is no need for lithography masks or performing any complicated microfabrication techniques. CAD designs can be given to a 3D-printer to simply, rapidly, and directly print any desired configuration of the PDMS well arrays. Our approach enables rapid and cost-effective production of microfluidic devices with a variety of dimensions, > 30 µm. It makes our process an ideal process for rapid prototyping of microfluidic device integration (with a printing time that is less than 5 min).

Figure 2
figure2

(A) Image of 3D printed structures composed of bioink and HCT116 in forms of spheres. (B) Bar plot showing the maintained cell viability within spheres geometry constructs for 7 days. The viability alters by ~ 19% of the day 1 viability. (C) Fluorescent image representatives of stained spherical HCT 116 cells-bioink constructs, on day 1,4, and 7, where green are live cells and red are dead cells. (D) Image of 3D printed rectangular structures composed of bioink and HCT116 cells. (E) Bar plot showing the maintained cell viability within rectangular constructs for 7 days. The viability alters by ~ 12% of the day 1 viability. (F) Fluorescent image representatives of rectangular stained HCT 116 cells-bioink constructs, on day 1,4, and 7.

3D-bioprintability and viability monitoring of 3D-HCT116-constructs

3D-bioprintability of 3D-HCT116-constructs

3D-models have shown to better predict the success of drug treatments in preclinical trials, due to the improved biological microenvironment relevance compared to 2D-culture models. In this work, we demonstrated the 3D-bioprinting of 3D-construct models of HCT116 cells. HCT116 cells were selected as a model for colon cancer tumors. Despite the advancements in colon cancer treatment, therapeutic drug investigations for those patients that present drug resistance is still necessary. For the HCT116 models, two different geometries were generated, a spherical and rectangular structure (Fig. 2A,D). These models were homogeneously printed from a mixture of HCT116 cells and Cellink bionk at the ratio of 1:10. The bioink, which is composed of alginate and nanofibrillar cellulose allowed for HCT 116 cells to be in an environment that closely resembles their native ECM in the human body. The Cellink bioink also provided a representation of possible in vivo drug transport (delivery), as anti-cancer drugs often must permeate through a mixture of tissue and ECM to reach the tumor. Mixing cells with the bioink and printing structures without any trapped-air-bubbles is usually a critical issue in bioprinting. One should note that even a small amount of trapped-air-bubbles in the cells-bioink mixture may affect the bioprinting parameters. Our printing process was optimized until no bubble formation in the printed cell mixture was observed (Nozzle 410 μm, extrusion pressure 4 kPa, printing speed 100 mm/min).

Viability analysis of 3D-HCT116-constructs

Previous studies in 3D-bioprinting have shown there is a large range of cell survival24. The cell survival rates depend on the level of shear stress that cells are subjected to during extrusion49. Subsequently, monitoring post-printing cell viability is crucial to gauge the success of different structures printing. In this part of the work, we sought to validate and monitor the viability of bioprinted spherical and rectangular 3D-structures within bioprinted PDMS well arrays. The viability of HCT 116 cells within spherical and rectangular constructs was monitored and imaged for 1, 4, and 7 post-printing, as shown in Fig. 2. For viability assessment (Fig. 2B,E), fluorescent images for live and deal labeled cells were imaged at three z-distances and for three different structures to obtain the best results (Fig. 2). All cell viability data are presented as mean values ± standard deviation. The cell viability percentage of spherical 3D-cell constructs was measured as 80.1 ± 4.1%, 67.8 ± 7.5%, and 64.7 ± 7.5% for 1, 4, and 7 days post-printing, respectively. Similarly, the rectangular HCT 116 3D-cell constructs were examined for 1, 4, 7 days post-printing, as shown in Fig. 2E,F. The average cell viability for these constructs was measured as 76.3 ± 9.2%, 64.4 ± 7.6%, and 67 ± 7.3% for 1,4, and 7 days post-printing, respectively.

Inclusive fluidic handling system along with 3D-HCT116-constructs

Preliminary experiments consisted of 3D-cell constructs within 3D-printed PDMS wells. An open well system allows for the simplification of construct manipulation and observation during the initial testing stage. Nevertheless, well-based systems lack the replication of mechanical forces such as fluid flow (i.e., shear stress) that tissues are subject to within the body. In these experiments, we demonstrated that we are able to take all our initial results to generate a process for the production of an inclusive OOC-like device with a 3D-HCT116 culture model using a 3D-bioprinter. This device can be described as a concave well-based microfluidic platform with connecting microchannels (Figs. 1K, 3I). The composition of the microfluidic platform is made up of two parts: microfluidic channels and concave wells. The channel’s width, length, and depth are 800 µm, 30 mm, and 300 µm respectively. The concave well diameter is 1.5 mm. The microfluidic channels enable the application of physiological fluid flow onto cell constructs while refreshing nutrients. The concave well array was designed to hold the 3D-HCT116 cell constructs. Both the channels and wells were similarly fabricated from 3D-printed Pluronic ink molds to which PDMS was then casted onto the molds to form the final structures of channels and wells (Fig. 3A–I).

Figure 3
figure3

(A) Schematics of printed Pluronic molds and resulting (B) PDMS casts for concave wells and channels. (C) Image of printed Pluronic molds used to fabricate PDMS concave wells and channels (scale bar: 2 mm). The process of (D–E) 3D printing GelMA-HCT116 structures within concave wells, (F–G) assembling the microfluidic platform, and (H) media perfusion of GelMA-HCT116 structures. (I) Photograph of the concave well-based microfluidic platform (scale bar: 2 mm). (J–L) Image representatives of three different toroidal formed structures of the 3D-bioprinted GelMA and HCT116 cell mixture. Live HCT 116 cells within the constructs were labeled with Calcein AM. Image representatives show the toroidal GelMA and HCT116 constructs with (J) smaller, (K) larger inner cavity, and (L) small cell island formed within the inner cavity of the ring.

Unlike the previously presented 3D-constructs, cell structures here were 3D-bioprinted from GelMA and HCT116 cells. Once printed, toroidal structures of GelMA HCT 116 cell structures were achieved (Fig. 3J–L). These toroidal structures (especially if stacked) have the potential to model the tubular geometry of the colon. They mimic tumors that are found attached to the inner wall of the large intestine. The microchannels were then added to the well array substrate where the GelMA cell structures were perfused with media. The simplified well-based perfusion design we demonstrated here can potentially be redesigned to add more channels, valves, and features that replicate human physiology such as cell–cell interactions, or delivery of gradient growth factors.

Preliminary drug screening of SN-38 on 2D-HCT116 cell models within 3D-PDMS bioprinted well arrays

3D- PDMS printed well arrays were used to execute initial drug toxicity studies of 7-Ethyl-10-hydroxycamptothecin (SN-38) on 2D-HCT116 cell models. SN-38 is a drug used for colon cancer, which has the effect of an apoptotic inducer, topoisomerase I inhibitor. In this work, we used the PDMS well arrays to treat an array of HCT 116 cell populations to two concentrations of 20 µM and 200 µM of SN38 as well as maintain an array of control cell populations (Fig. 4B). Cell viability measurements after 48 h of drug treatment indicated that control cell populations have the viability of 90%, while cell populations treated with 20 µM of SN38 have a viability of 57%, and those treated with 200 µM of SN38 have a viability of 48% (Fig. 4A). Figure 4C shows the image representatives of fluorescently labeled HCT116 cell, and it observed that the control population remains adhered to the surface while cells treated with increasing SN38 concentration detach from the surface, leaving behind a less dense cell population. For the data presented here 3 different measurements were taken and are presented as mean values ± standard deviation. The one-way analysis of variance (ANOVA) determined statistically significant differences between the means of control’s cell viability and the addition of drugs with different concentration (20 μM and 200 μM), where statistical significance was shown as *p < 0.0001 for both treated populations.

Figure 4
figure4

(A) Bar plot showing the cell viability within 2D-constructs after 48 h of drug treatment. The control cell populations showed a viability of 90%, cell populations treated with 20 µM of SN38 showed a viability of 57%, and cell populations treated with 200 µM of SN38 showed a viability of 48% after 48 h. (B) Schematic and image of 3D-printed PDMS well arrays with HCT 116 cultured cells. A total of three rows of wells were used, where one row (consisting of three replicate wells) was designated as a control population (no drug), a second row consisted of HCT116 cell population treated with 20 µM of SN38, and a third row consisted of HCT116 cell population treated with 200 µM of SN38. (C) Fluorescent image representatives of NucBlue stained HCT116 cells for control population (left), HCT116 cells with 20 µM of SN38 (middle), and HCT116 cells with 200 µM of SN38 (right) are shown, where control population shows a more cohesive adhered monolayer, while those treated with SN-38 presented significant loss of monolayer distribution with large sparse gaps. One-way ANOVA indicated that statistical significance of *p < 0.0001 for both treated populations (20 and 200 µM of SN38) compared to control population.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *