Combination of 3D Micro Contact Printing and Scraping Technique for Site Selective Cocultivation Across a Porous Micro Structured Scaffold
Dana Brauer, Anni Peng, Justyna Borowiec, Jörg Hampl, Shannon Prehl, Maren Klett, Merle Johanna Küstner, Frank Weise, Andreas Schober, Sukhdeep Singh
TL;DR
This paper introduces a simple scraping technique combined with 3D microcontact printing to create 3D cell patterns on both sides of a scaffold, enabling complex co-culture models for studying cell interactions.
Contribution
A technically simple scraping technique is introduced to enable site-selective co-cultivation on both sides of a porous microstructured scaffold.
Findings
The combination of 3D-µCP and scraping allows for 3D cell patterning on both the front and reverse sides of a substrate.
A tri-culture model of endothelial, hepatoblastoma, and fibroblast cells was created to mimic the liver tumor microenvironment.
The method supports multi-layered co-culture systems for studying cell-to-cell crosstalk at the microlevel.
Abstract
3D‐microcontact printing (3D‐µCP) technique combines the advantages of microcontact printing and microthermoforming for the fabrication of functional biomaterials with complexity closer to real tissue. Despite its unmet advantages in terms of complexity and processability in a single step, this technique is limited to the front side of the substrate. Considering the advantage of inherent topography patterns on the reverse side of the substrate, an additional degree of patterning can be envisioned. However, selective patterning on the reverse side is challenging due to the fragility of the cell culture on the front side. Herein, a technically simple scraping technique is presented in combination with 3D‐µCP to generate 3D cell patterns on both the front and the reverse side of a micro thermoformed substrate. The technical advancement of 3D‐µCP with the scraping technique offers a…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8| Designed size | L200S200 | |
|---|---|---|
| Front side | height | 38,2 ± 7,8 µm |
| width ridges | 235,7 ± 12,9 µm | |
| width channels | 177,5 ± 11,8 µm | |
| Back side | height | 33,7 ± 3,9 µm |
| width ridges | 273,4 ± 16,8 µm | |
| width channels | 98 ± 16,1 µm |
- —Carl‐Zeiss‐Stiftung10.13039/100007569
- —AiF Projekt10.13039/100018329
- —Bundesministerium für Bildung und Forschung10.13039/501100002347
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topics3D Printing in Biomedical Research · Additive Manufacturing and 3D Printing Technologies · Nanofabrication and Lithography Techniques
Introduction
1
Biological structures generate a variety of interesting functions, which are determined by their complexity and inspired by the hierarchical organization of different cell types at the microscale.^[^ 1 ^]^ Therefore, modern advanced cell cultivation techniques rely on generating the complexity that represents the real tissue environment in vivo. In order to achieve such a complex task, selective cell patterning on different geometrical scales needs to be considered, which can assist the fabrication of so‐called “functional biomaterial,” This innovation utilizes engineered surfaces with defined micro‐ and nanoscale patterns to guide cellular behavior, promoting specific cell arrangements to reach tissue‐like structures.
The ability to imitate the complex 3D architecture of real tissues is essential for in vitro applications such as the investigation of cell–cell interactions, disease modeling, drug discovery, and regenerative medicine.^[^ 2 ^]^ Traditional 2D cell cultures, though effective for some basic applications, lack the spatial complexity and cellular interactions found in vivo, and thus limit their physiological relevance.^[^ 3 ^]^ Consequently, there has been a shift toward advanced 3D culture techniques that better emulate the in vivo environment.^[^ 4 ^]^ One of the most commonly used in vitro systems for studying cellular crosstalk between different cell types are transwell insert. They contain a porous/permeable membrane on which cells grow on both sides and can communicate with each other via paracrine signaling molecules. In contrast to a typical microtiter plate cell culture model, they better mimic the in vivo environment and enable studies of cell function,^[^ 5 ^]^ polarization,^[^ 6 ^]^ cell migration,^[^ 7, 8 ^]^ barrier formation,^[^ 9 ^]^ and invasion, e.g., cancer cell invasion.^[^ 7, 10 ^]^ Transwell inserts have been used to study the intestinal mucosa,^[^ 6 ^]^ the blood‐brain barrier,^[^ 11 ^]^ brain metastasis,^[^ 10 ^]^ the barrier formation of pulmonary alveoli^[^ 12 ^]^ and the coculture of primary hepatocytes (PHH) and liver sinusoidal cells (LSECs)^[^ 5 ^]^ among many others. Advanced Transwell systems contain PDMS layers that are selectively colonized with cells or used to deliver soluble reagents in a controlled manner.^[^ 13, 14 ^]^
It was shown that for a Transwell‐layered coculture of PHH and LSECs, the PHH viability was maintained up to 39 days, compared to 11 days for PHH in well culture.^[^ 5 ^]^ Controlling the heterotypic cell‐to‐cell interactions by mimicking the space of Disse appears to be a key feature for maintaining and enhancing hepatocyte viability.^[^ 5, 15 ^]^ Simple transwell applications are based on the planar growth of cell types separated by a membrane without lateral spatial restriction, which lacks the typical topographical features obvious in the real tissue. In human organs, there are directed arrangements of cell aggregates that are also separated laterally by geometric barriers (e.g., extracellular matrix, membranes) to maintain polarity, differentiation state, and function of the cells.^[^ 16, 17 ^]^ Therefore, in order to create more intricate in vivo‐like tissue morphologies, 3D micro topographies such as channels and cavities must be considered for advancing the Transwell insert systems.
Another method for arranging cells in a specific pattern is micropatterning. Micropatterning techniques are a valuable tool to control cell attachment, shape, spreading, and tissue architecture as a function of the engineered spatial properties of the culture surface.^[^ 18 ^]^ The additional integration of geometrical and mechanical features of the surrounding microenvironment also influences cellular behavior in terms of polarity, proliferation, differentiation state, and functionality.^[^ 19 ^]^ Consequently, such techniques are essential to develop new innovative experimental approaches for the identification of previously unknown mechanisms of cell behavior and cell‐cell interactions in the tissue microenvironment in vivo.
The implementation of micro‐patterning techniques requires surfaces with different adhesive properties. This is usually achieved by combining cell‐attracting and cell‐repellent surfaces, addressing a functional characteristic of a bioinspired material. A common method for guided adhesion of cells to defined surface areas is the coating with adhesive protein solutions, such as extracellular matrix proteins like collagen. After cell seeding, the cells adhere exclusively to the cell‐attracting areas, thereby creating adhesion surfaces with defined dimensions. These so‐called micro sheets were used to generate arrays of collective groups of cells with controlled spacing in between them to investigate the cellular responses.
A wide range of studies have been performed, in which the cells were patterned into specific geometrical configurations on solid substrates, e.g. to model vascular architectures,^[^ 20 ^]^ cardiac myofibers,^[^ 21 ^]^ neuronal communication via synapses,^[^ 22 ^]^ neuronal polarization,^[^ 23 ^]^ paracrine signal transduction in stem cell colony differentiation,^[^ 24 ^]^ self‐organization of human embryonic stem cells,^[^ 25 ^]^ cell movements^[^ 26, 27, 28 ^]^ and cancer cell morphology.^[^ 29 ^]^
Micro‐patterning with several cell types can be used to construct tissues at the multicellular level. After structuring substrates with adhesive patterns and applying the first cell type, a second cell type is applied to the remaining areas, which have also been rendered adhesive.^[^ 30 ^]^ Generally, the resulting co‐cultures are patterned on the same side of the substrate.^[^ 18 ^]^ These more complex tissue‐like structures with multiple cell types and defined geometries can mimic 2D planar in vivo morphologies and thus enable the investigation of cell‐cell interactions, usually by microscopic methods.
The underlying micropatterning technologies comprise lithography and nonlithography‐based approaches.^[^ 31, 32 ^]^ In both cases, a stamp is produced that can create the adhesion patterns on surfaces. In the first case, a silicon wafer is coated with uncured photoresist, which, after curing, contains the geometric features that were previously designed using software. With soft lithography, elastomeric polymer stamps (usually made of polydimethylsiloxane) are formed that contain the features of the designed wafer. These stamps are then combined with processes such as microcontact printing^[^ 33 ^]^ or microfluidic structuring. For cell adhesion, microcontact printing stamps are inked with biomolecules like collagen, laminin, fibronectin, lectins or with peptides containing the RGD motif for integrin‐mediated cell adhesion.^[^ 34, 35 ^]^
Nonlithography‐based approaches are based on, e.g., stencil‐like parafilm inserts which shield the underlying area from cell attachment. Using this method, cell patterns with different geometries can be seeded using a multi‐well plate format.^[^ 36 ^]^
Another step toward advanced tissue complexity and 3D architecture is the 3D‐µCP technique.^[^ 34 ^]^ 3D‐µCP combines microcontact printing with a thermoforming process, combining ECM molecular patterning with simultaneous shaping of geometric environmental properties. This allows complex in vivo morphologies to be configured and selectively seeded with cells in a 3D environment. However, despite its versatility, 3D‐µCP is limited to patterning only one side of the substrate, which restricts its ability to fully mimic the complex, multi‐dimensional arrangements seen in natural tissues. To overcome this limitation, we developed a simple scraping technique for the reverse side of a 3D‐µCP microstructured polymer substrate. A polycarbonate film with channels and ridges design is patterned on the front side with collagen and the first cell type in the channels, then rotated, and another layer with cell type 2, completely covering the reverse side, is scraped off the ridges. This creates a double‐sided co‐culture system, whose cells are organized in parallel, laterally displaced channels as 3D strands, separated by a porous membrane. Thus, this technology allows the use of structured areas on the reverse side of a substrate, providing a versatile microtopographic platform that expands the geometric features of various tissue models based on their defined cell size, arrangement, and existing spaces, etc.
Such microstructured membranes enable microscale sequestration (≈100–200 µm), bringing cells into a closer in 3D environment that reflects physiologically relevant communication distances. The geometry permits controlled spatial separation and alignment of different cell types in a regular, repetition‐like arrangement, which is not possible with flat membranes. The platform facilitates bidirectional signaling studies under conditions that better mimic the in vivo micro environment, both in terms of 3D cell culture and the patterned organization of cells, e.g., in the liver.
The versatility of the co‐culture can be increased to a tri‐culture with the simple addition of a third cell type. The biological relevance of this technique was proven with a model of the tumor microenvironment in the liver. A functional biomaterial with a site‐selective coculture of HepG2 hepatoma cells, EA.Hy 926 endothelial cells and NIH3T3 fibroblasts were established, mimicking a situation in the microenvironment of liver tumors, in which cancer cells communicate with endothelial cells and fibroblasts. We found a stimulating effect of the endothelium and an inhibitory function of fibroblasts on the tumor progression of HepG2 cells.
In the presented study, we show that the combination of 3D‐µcontact printing and the scraping technique represents a step forward in creating a functional biomaterial mimicking bio‐artificial tissues with hierarchical organization, thereby opening new avenues in tumor research, regenerative medicine, and tissue engineering.
Results and Discussion
2
Guided Adhesion of EA.Hy 926 Cells on 3DµCP Collagen‐Micropatterned Substrate
2.1
In a previous work,^[^ 34 ^]^ we presented the method of 3D microcontact printing (3D µCP), which, in addition to transferring substances with a resolution in the micrometer to submicrometer range onto the surface of substrates, enables the integration of a third dimension into the structuring process. This means that topographical features, such as 3D micro geometric structures, can be introduced into the substrate simultaneously with an ECM molecule coating. We used a porous polycarbonate film to form microstructured substrates with a channels and ridges design by 3D‐µCP (Figure 1a). It is important to mention that the successful adhesion of the cells in the channels is strongly correlated to the channel width. A channel narrower than 100 µm resulted in a previous study in a reduced performance.^[^ 34 ^]^ In order to ensure reasonable cell numbers in the channels for quantitative analysis (e.g., microscopy, cell counting, and immunofluorescence studies), a stamp design with a width of the lines and spaces of 200 µm was selected.
3D‐µCP collagen‐patterned substrate and cell adhesion of EA.Hy 926 cells: A) SEM analysis of channels and ridges substrate; left: front side, right: backside, scale bar: 1 mm B) Localization of EA.Hy 926 cells in 3D‐µCP collagen patterned channels were verified by SEM analysis; scale bar:100 µm; C) staining of the actin cytoskeleton of EA.Hy926 cells, scale bar: 100 µm and C) detection of the endothelial marker CD31, scale bar: 200 µm 48 h after seeding; n = 3 experiments.
A confocal microscope was used to measure the surface dimensions of the front and reverse sides of the thermoformed films. The following dimensions were obtained after thermoforming: 177.5 µm ± 11.8 and 235.7 ± 12.9 µm width on the front side and 273.4 ± 16.8 and 98 ± 16.1 µm width on the back side (Table 1). The channels on the front side were coated with the ECM molecule collagen during the 3D‐µCP to guide the adhesion of endothelial cells only in the channels. We have already demonstrated the successful transfer of collagen into the channels by antibody‐based detection and attenuated total reflection Fourier transform infrared spectroscopy (ATR‐FTIR).^[^ 34 ^]^
Endothelial‐like EA‐Hy 926 cells were seeded onto the front side of the substrate, and guided adhesion of the cells into the channels was monitored 48 h after cell seeding by scanning electron microscopy, staining of the actin cytoskeleton, and labeling the cells with a marker for endothel, CD31 (PECAM‐1). We found that the EA.Hy 926 cells attached mainly to the collagen‐patterned regions (Figure 4B), while forming cord‐shaped 3D aggregates (Figure 1B,C) in the channels with efficiency above ≈90%.^[^ 34 ^]^ The EA.Hy 926 cells were positively stained for the endothelial marker CD31 (PECAM‐1) (Figure 1D), which was used for labeling EA.Hy926 cells throughout the study. The successful performance of the stamping process with subsequent cell seeding, we demonstrated in a previous study.^[^ 34 ^]^ Only a few cells attached after the seeding process to the ridges due to the low surface free energy of unprinted polycarbonate.^[^ 37 ^]^ The cells proliferated in the channels and remained viable up to day 10 of culture.
Cell Removal of HepG2 ‐Cells by Cell Scraping
2.2
It is important to note that 3D micro contact printing technique only enables the cultivation of cells in the channels on the front side. However, the reverse side is still available with its inherent topography for the selective placement of the other cell type, like HepG2 cells. The challenge in this case is the site‐selective inking of the biomolecules in the reverse side channels. The one possible solution is the selective removal of the ink from the reverse side ridges simply by scraping. However, the presence of the EA.Hy 926 cells on the front side channels make the systems sensitive to multiple manipulations. Therefore, a combined scraping of the cells together with the inked biomolecular collagen layer could provide a reliable solution to limit the growth of the cells in the reverse side channels.
Therefore, to create a hepatocyte‐endothelial co‐culture, in which both cell types and inks are present only in the channels on opposite sides of the substrate, the scraping technique was used to selectively remove the hepatocytes from the ridges on the reverse side of the substrate. For this purpose, HepG2 cells were seeded and grown on the backside of the substrate, which had been coated with collagen before, for 24 h. The substrate with the grown HepG2‐cell layer was placed upside down on a glass slide, and a PDMS weight was added. By targeted movement with a micromanipulator, the cells, which were grown on the ridges of the substrate, were scraped off (details are given in the Experimental Section). Scraped cells were removed by a final washing step from the microstructured substrate.
The outcome of the scraping procedure has been observed by actin cytoskeleton staining of HepG2 cells before, directly after, and 24 h after scraping (Figure 2A). Furthermore, SEM images were taken directly after and 24 h after scraping (Figure 2B). Both methods showed that after the scraping process was complete, the HepG2 cells on the ridges had largely disappeared. HepG2 cells were fixed immediately after the scraping procedure and labeled with an antibody against the tumor marker α‐fetoprotein, which resulted in a visible cytoplasmic staining of the tumor marker α‐fetoprotein in the HepG2 cell strands (Figure 2C).
Scraping of HepG2 cells on micropatterned substrate: A) labeling of the actin cytoskeleton before, directly after and 24 h after scraping. scale bar: 100 µm B) SEM images of HepG2 cells directly and 24 h after scraping, scale bar:100 µm; C) HepG2 cells were examined for intracellular α‐fetoprotein (AFP) expression with a specific antibody, scale bar: 200 µm; n = 3. D) 24 h after scraping, brightfield images of 11 scraping trials with HepG2 cells (N = 4; n = 2–3) on foils were investigated. The 9 unscraped HepG2‐populated foils (N = 3; n = 3) were investigated in parallel and served as controls. Cells were counted with the ImageJ analysis software with the cell counting tool. The population with HepG2 cells of the channel area and the right and left ridge area was counted in a rectangle with dimensions of 3× the ridge width (width) and 6× the ridge width (length) (Figure S2, Supporting Information). Shown are the mean values ± SD from 11 trials, which were normalized to the mean value of the unscraped controls, which was set to 100%. The right column represents the mean ±SD cell amount of the right and left area.
However, a few remaining cells were occasionally found on the ridges (Figure 2B), mostly at the edges of the channels (Figure 2B). In addition, it cannot be ruled out that a few cells are swept in the channels by the final rinsing step. We analyzed the success of the scraping procedure by counting the remaining cells on the ridges. Brightfield images of 11 foils were investigated to determine the success of the scraping (removal) performance 24 h after the scraping procedure with HepG2 cells. The 9 Foils with unscraped HepG2‐cells were investigated in parallel and served as controls. The counting procedure is described in the Experimental Section—Determination of Scraping (Selective Removal) Performance.
We found a loss of 21% cells in the channels after the scraping procedure compared to 100% in the channels of unscraped controls (Figure 2D; Table S1, Supporting Information). On the ridge area on the right side, 65% cells, and on the ridge area on the left side, 79% HepG2 cells were scraped off, leading to a mean scraping efficiency of 71,8% (Figure 2D; Table S1, Supporting Information). Mostly, the remaining cells were counted at the interfaces of the channels and ridges of the microstructured substrate.
We observed the cellular constructs of HepG2 cells up to 48 h of incubation after the scraping procedure. HepG2 cell localization was confined to the channels in between (Figure S1, Supporting Information).
Site Selective Coculture of EA.Hy 926‐ and HepG2 Cells
2.3
Both methods, 3D‐µCP and the scraping technique, have now been combined to obtain a co‐culture of EA.Hy 926 and HepG2 cells in the channels on opposite sides of the microstructured substrate, as shown in Figure 3A. For this purpose, EA.Hy 926 cells were seeded in the 3D‐µCP collagen‐structured channels on the front side of the substrate on the first day, followed by the seeding of HepG2 cells on the collagen‐coated back side one day later. On day 3, the HepG2 cell layer was scraped off the ridges, resulting in an assembly of the remaining hepatocytes in cords in the channels of the reverse side of the substrate. Cell localization of EA.Hy 926 and HepG2 cells were analyzed by SEM images (Figure 3B) and double staining against CD31 and α‐fetoprotein directly after the scraping procedure (Figure 3C). The front view image (Figure 3D) of the coculture revealed the attachment of EA.Hy 926 cells and HepG2 cells on the bottom of opposite channels of the microstructured substrate, separated by a porous polycarbonate film. Additional live/dead staining of the respective cocultures of HepG2 and EA.Hy 926 cells showed viable cells, labeled in green with calcein, directly and 24 h after the scraping procedure (Figure 3E). Only a few dead cells (labeled in red) were found 24 h after the scraping process. This indicates that the HepG2 cells in the channels withstand the scraping process well due to their deeper position and their soaking in culture medium.
Site‐selective coculture of HepG2‐ and EA.Hy 926 cells on microstructured substrate realized with 3D µ‐contactprinting and scraping: A)scheme for cocultivation of EA.Hy 926 cells and HepG2 cells. B) SEM images of EA.Hy 926 cells located on the front side and HepG2 cells located at the backside of the microstructured membrane; scale bar = 100 µm; investigated on five foils; C) labeling of HepG2 cells with anti α‐fetoprotein‐Ab and EA.Hy 926 cells with anti‐CD31‐Ab in the coculture; scale bar: 200 µm; n = 3. D) CD31‐ and α‐fetoprotein‐ labeled EA.Hy 926‐ and HepG2 cells in the established coculture with front view image; scale bar: 150 µm E). Cocultures were analyzed for their viability directly and 24 h after the scraping procedure by Calcein/EthidiumDIII homodimer staining, Calcein labels living cells (green color) and Ethidium DIII homodimer labels dead cells (red cells); The cell aggregates with higher calcein intensity are the HepG2 cells after scraping, while the EA.Hy 926 cells in the opposite channels show lower calcein intensity. Only a few dead cells were labeled 24 h after scraping (white arrows); scale bar: 100 µm; n = 2.
It was observed that the intensity of calcein staining continued to increase after 24 h due to the higher cell density (with living cells) in the channels. HepG2 cells in the channels exhibited a more pronounced calcein staining, presumably caused by their increased metabolism as a tumor cell line. In addition, HepG2 cells appeared to grow out of the cell aggregates or showed slight signs of migration when cocultured with EA.Hy 926 cells for 24 h (Figure 3E).
The double‐sided cell‐patterned, microstructured substrate provides a complex microtopographic platform that arranges hepatocytes and endothelial cells in a defined manner with lateral offset, using geometric features such as channels and ridges. Using 3D‐µcontact printing and scraping technique, we established a co‐culture of HepG2 hepatocytes and EA.Hy 926 cells, in which two cell types are arranged as strands or cords in close proximity to each other and are only separated by a porous polycarbonate membrane. With this configuration, cells can communicate via signaling molecules through the membrane, without direct cell‐to‐cell contact.^[^ 12 ^]^ Additional spatial restriction is providing geometrical parameters in 3D space, especially to the hepatocytes, which can show altered proliferation and migration behavior without geometrical restrictions and 3D culturing.^[^ 38, 39, 40 ^]^ Furthermore, hepatocytes and endothelial cells are arranged side by side in the sinusoids of the liver, rather than on top of each other, as was the case, for example, in studies using Transwell systems.^[^ 5 ^]^ This is a unique feature of our system and should better reflect in vivo conditions.
Such cell architectures are common in tissues, with different cell types separated from each other by lumens or the deposition of extracellular matrix. Under healthy conditions, this spatial separation ensures that cells can maintain their differentiated state and thus retain their normal function, e.g., in the liver lobulus.^[^ 41 ^]^
Biological Relevance
2.4
In the liver tumor microenvironment, the cross‐talk between cancer cells and endothelial cells (ECs) plays a critical role in tumor progression, recurrence, cancer stemness^[^ 42, 43 ^]^ as well as tumor angiogenesis.^[^ 44 ^]^ Indirect coculture approaches (working with growth factor‐enriched culture supernatants) showed the role of secreted growth factors like VEGF, EGF, and bFGF in migration and proliferation of cancer and endothelial cells and tube formation of endothelial cells^[^ 42, 45 ^]^.
As a proof of concept, we compared site‐selective coculture with direct coculture of HepG2‐and EA.Hy 926 cells regarding the occurrence of signs of sprouting/tube formation of endothelial cells (Figure 4A,B). We observed with staining against the endothelial marker CD31, that only in the direct coculture approach morphological changes became apparent, resembling endothelial cell sprouting (Figure 4B). In contrast, site‐selective cocultivation of both cell types in a close proximity, but without cell‐to‐cell contact, did not induce these morphological changes (Figure 4A). Thus, by comparing the two methods, insights into the regulation of the angiogenesis process can be obtained that are not possible with conventional methods such as indirect co‐culture (use of growth factor‐containing culture supernatants). We confirm with our observations the results of a study of Chiew et al.,^[^ 46 ^]^ in which they reported that in addition to VEGF signaling in endothelial sprouting, a cell‐cell contact between cancer and endothelial cells is indispensable. Furthermore, we have noticed that after cocultivation for up to 48 h, in certain cases, a good number of cells (HepG2‐ and EA.Hy 926 cells) are present on the ridges even after scraping (as seen in Figure 4A) compared to the initial cell localization (Figure 3C,D). This observation is in line with the work of Feng et al.,^[^ 42 ^]^ where they have shown that the growth factors secreted by one cell type can influence the migration/proliferation of the other cell type.^[^ 42 ^]^ However, this fact needs further investigation with respect to the geometrically constrained cell culture environment.
Site‐selective versus direct coculture of HepG2‐ and EA.Hy 926 cells: Cocultures were seeded on microstructured substrate and on coverslips using the same temporal sequence. After 48 h of cocultivation HepG2 and EA.Hy 926 cells were labeled for α‐fetoprotein and CD31. Scale bars were 100 µm (site selective) and 200 µm (direct); n = 2.
Increasing Versatility by Integration of a Third Cell Type
2.5
It is known from the literature that cell‐cell interaction via direct contact between stromal fibroblasts and liver cells strongly modulates the hepatocyte function, like albumin secretion.^[^ 47, 48 ^]^ By embedding neighboring cells in a physiological matrix and producing growth factors, fibroblasts contribute to proper epithelial proliferation, differentiation, and mechanical stability.^[^ 49 ^]^ In another instance Lee at al.^[^ 50 ^]^ has reported the precise control over the hepatocyte spheroid formation when there is a fibroblast layer underneath. Such triculture models are useful to study certain disease models like neuroinflammation, nonalcoholic fatty liver disease,^[^ 51, 52 ^]^ diseased hepatocytes^[^ 53 ^]^ and the investigation of hepatocyte‐fibroblast‐endothel crosstalk.^[^ 54 ^]^ Furthermore, it is known that fibroblasts, through the production of extracellular matrix and soluble factors, can inhibit cancer cell progression.^[^ 55, 56 ^]^ Therefore, it is logical to include the fibroblasts in our cultivation system to replicate the components of the liver tumor microenvironment^[^ 57 ^]^.
We selected a simple approach to further expand the complexity of our system by the implementation of fibroblasts as an additional cell type on the scraped side of the microstructured scaffold (Figure 5A). After scraping the HepG2 cell layer, we cultivated Violet‐BMQC‐tracker‐loaded NIH3T3 fibroblasts on the HepG2 strands and observed the tricultures for up to 48 h. We found a uniform adhesion of the NIH3T3 fibroblasts on the scraped ridges as well as on the HepG2 cell strands (Figure 5B) in the triculture. Fibroblasts kept the tumor cells in their position in the channels, which indicates that the presence of the NIH3T3 fibroblasts may inhibit tumor cell migratory behavior (e.g., caused by soluble factors of endothel) in vitro (Figure 4A). Alkasalias et al.^[^ 55 ^]^ has reported that such an inhibition is strongly dependent on the fibroblast type as well as on cell‐to‐cell contact. Normal fibroblasts inhibit the migration of cancer cells, whereas cancer‐associated fibroblasts (CAFs) are not supporting migration inhibition. Although other studies show the tumor‐promoting effects of CAFs, they may sometimes restrain cancer progression as a host defense mechanism against neoplasia.^[^ 58, 59 ^]^ Therefore, we believe that the potential of our system, with restriction of cell location through geometrical cues, can be explored for studying a cell therapy approach in tumorigenesis.
A) schematic illustration of the triculture of HepG2‐, EA.Hy 926‐ and NIH3T3 cells on microstructured substrate; B) scraped HepG2 cells were labeled for α‐fetoprotein, EA.Hy926 cells in µCP‐channels for CD31 and NIH‐3T3 fibroblasts were visualized via labeling with Violet‐BMQC tracker. Shown are the tricultures directly after NIH3T3 fibroblast addition and after 24 and 48 h cocultivation. scalebar:100 µm; n = 3.
Conclusion
3
Combining the technological potential of 3D microcontact printing with specialized scraping technique allows us to construct the 3D micropatterned cell culture environment across two sides of a thermoformed foil‐based scaffold. The presented system has an immense potential to be explored as a functional biomaterial that is closely mimicking the in vivo cellular arrangement. The proposed technique is offering a significant complexity in terms of cellular arrangements without compromising the practical simplicity. Such a technical simple solution has an immense potential to fabricate functional tissue patches with desirable cellular properties. In the presented case we have designed the cellular patterning inspired from the liver lobe geometry and utilized the cell‐patterned substrate for studying cell‐cell interaction in the tumor microenvironment in the liver. We used HepG2 cancer cells, EA.Hy 926 endothelial cells and NIH3T3 mouse fibroblasts to create co‐ and tricultures with the aforementioned technique. The site‐selective coculture approach with HepG2 and EA.Hy 926 cells without direct cell‐cell contact revealed that endothelial sprouting in tumor angiogenesis not only depends on soluble factors like VEGF, but also on direct cellular contact to cancer cells. We have shown that cancer migratory phenotype is inhibited in the triculture containing NIH3T3 fibroblasts, which indicates that geometrical control over the different cell types can be achieved through a confluent fibroblast layer. Furthermore, it is concluded that our experimental approach reflects an in vivo situation, in which the close proximity of endothelial cells to cancer cells offers a valid 3D tumor model to study the genesis of disease. For example, it is known that early and inefficient tumor vascularization, increases the cancer progression via soluble factors.^[^ 60 ^]^ Our system provides the required geometrical and topographical parameter set for the investigation of minute conditional factors in tumorigenesis studies.
Importantly, while line‐and‐space features couple ridge and channel widths between front and reverse surfaces, the platform is not limited to such mirrored geometries. Alternative designs—such as hexagonal pits yielding channel networks, hemispherical features producing pyramidal ridges, or square features forming triangular or pyramidal structures—demonstrate that distinct environments can be engineered on opposing sides of the same scaffold. This design flexibility expands the utility of the µplatform, enabling tailored microenvironments for diverse co‐culture and tissue‐modeling applications.
Experimental Section
4
PDMS Stamp Fabrication
For the preparation of the stamp master (silicon wafer) standard photolithographic methods were applied.^[^ 32 ^]^ A silicon wafer with the following features was used: lines with a width of 200 µm and a spacing of 200 µm were etched to a depth of 40 µm. PDMS replicas were then produced based on the silicon wafer. For this purpose, a mixture of PDMS components in a ratio of 1:10 (curing agent to prepolymer, Sylgard 184, Dow Corning, USA) was vigorously stirred and degassed in a vacuum. The mixture was poured over the entire wafer and cured on a hot plate (70 °C, 30 min). After drying for 24 h at room temperature, the PDMS was peeled off the master mold, cut to the required size, cleaned, and prepared for use. Before final use, the stamps were treated for 1 h at 200 °C in an oven to remove moisture and residual monomers.
3D‐µ Contact Printing (3DµCP) of Porous Polycarbonate Membrane
To achieve simultaneous topographical patterns (lines and spaces) into the membrane and chemical modification with the ECM molecule collagen, the micro‐structured process 3D‐µCP (Figure 6) was applied according to a previously published protocol^[^ 34 ^]^ with some minor modifications.
Fabrication of the 3D‐µCP‐collagen patterned scaffold: 1) Collagen was layered on a glass slide and dried for 1 h followed by rehydration. 2) The PDMS stamp was pressed on the moistened slides for 15 min to transfer the collagen on the stamp. 3) The collagen‐coated stamp was placed onto PC foil with FEP film underneath. 4) 3µ‐CP‐collagen patterned scaffolds were formed using a microthermoforming machine 5) FEP foil was removed.
As the source membrane, a heavy ion etched polycarbonate (PC) foil (it4ip, Louvain‐la‐Neuve, Belgium) with a thickness of 50 µm and a pore size of 1.4 µm was used. Prior to thermal processing, the PDMS stamps and glass slides were cleaned by immersion in ethanol for 15 min in an ultrasonic bath, followed by an additional 15‐min ultrasonic bath in distilled water. After drying overnight, the cleaned stamps were ready for use. Polycarbonate (PC) films were cut into 3 cm × 5 cm rectangles, and the positions of the desired structures were marked on the back of the cleaned slides.
To apply the collagen coating, 50 µL of a 320 µg mL collagen solution was placed on the top of each slide and dried under a fume hood for 1.5 h. In parallel, the PDMS stamp was treated with oxygen plasma (PVA TePla 200) with 150 W for 120 s to achieve hydrophilicity of the surface. To rehydrate the collagen, the slides were placed in a humidifying chamber with a PariBoy nebulizer for 7.5 min, ensuring a constant flow of moist air. The PDMS stamps were then positioned on the moistened slides and gently pressed down for 15 min to facilitate collagen transfer. After carefully removing the stamps, the visible collagen imprint indicated successful transfer. The moist stamps were returned to the humidity chamber for an additional 3 min and then carefully placed onto the pre‐cut PC slides. To complete the process, a 50 µm thick poly‐ (tetrafluoroethylene‐co‐hexafluoropropylene) (FEP) foil was placed underneath PC foil to protect the pores during processing and serve as a force transducing layer. Microstructures were formed at 158 °C and 50 bar for 30 s in a microthermoforming machine (WLP 1600S, WICKERT Presstech, Landau). The patterned PC membranes were stored at 4 °C.
Analysis of Topographical Patterning by a Laser Scanning Digital Microscope
Samples were sputter‐coated with a thin layer of platinum and examined using a laser scanning digital microscope (LSM) (Olympus LEXT OLS4100). Fabricated microstructured substrates with the designed dimensions were analyzed with regard to the height and width of lines and spaces (Table 1). Data were collected from three independent samples, with measurements taken from at least three channels in each sample. All values are presented as the mean ± standard deviation.
With nanoindentation measurements and Attenuated total reflectance Fourier transform infrared spectroscopy (ATR‐FTIR), the local mechanical properties of the structures and the presence of collagen on the PC substrate after 3D‐µCP were proven earlier. Details of the results can be found in the ref. [34]
Cell Culture
The human hepatoblastoma cell line HepG2 (ATCC, LGC Promochem, Wesel, Germany) was cultivated in Minimum Essential Medium (Sigma), supplemented with 10% fetal calf serum, 100 U mL^−1^ penicillin, 100 µg mL^−1^ streptomycin, and 2 mm L‐glutamine. The human endothelial‐like EA.hy 926 cells (ATCC, Rockville, MD) were cultivated in Dulbecco's Modified Eagle's Medium (Sigma), supplemented with 10% fetal calf serum, 1% penicillin/streptomycin, 2% L‐glutamine, and 1% sodium pyruvate. NIH3T3 fibroblasts (ATCC, LGC Promochem, Wesel, Germany) were cultivated in DMEM containing 10% fetal calf serum, 1% penicillin/streptomycin, and 1% glutamin. Medium in all cell cultures was changed every 2–3 days. The cells were split according to a standard protocol. Triculture experiments were performed using the EA.Hy 926 culture medium.
Cell Seeding of EA.Hy 926 and HepG2 Cells on the Microstructured Substrate
The micropatterned polycarbonate structures were placed in a sterile cell culture hood and sterilized using UV light for 15 min. The sterilized scaffolds were then inserted into a specially designed frame apparatus (Figure 7A) that allows for cell co‐cultivation on both sides of the thermoformed scaffolds. The frame apparatus was presterilized in 70% ethanol for 15 min prior to use. After positioning of the scaffolds, a second part of the device was used to secure the samples from above, and the entire setup was placed into a 6‐well plate. Afterward coculture of EAHy 926‐ and HepG2 cells was established (Figure 7B). The top of the scaffold was coated with 25 µL of a type I collagen solution (320 µg mL^−1^) and incubated for 1 h to allow later the cell adhesion of HepG2 cells. After incubation, the collagen solution was carefully aspirated, and the frame apparatus was rotated on the backside to enable access to the other side of the scaffold. The 35 µL of cell suspension containing 3 × 10^4^ EA.hy 926 cells was seeded on that side of the scaffolds. The cells were allowed to adhere in the micropatterned channels for 3 h at 37 °C and 5% CO_2_ in the cell incubator. Subsequently, the frame apparatus with scaffolds was carefully rotated again and gently immersed in 3 mL of DMEM so that no air bubbles would collect under the scaffold surface. EA.hy 926 cells were in contact with the medium, while the outer side of the scaffolds stayed dry. Scaffolds with EA.hy 926 cells were incubated for 24 h. On the next day 35 µL of cell suspension containing 7 × 10^4^ HepG2 cells was seeded on the collagen‐coated side of the scaffolds, followed by an additional 3 h of incubation. Finally, the frame structures were completely covered with medium and incubated for 24 h.
Cell seeding of the coculture using a frame apparatus: A) frame apparatus for integration of three microstructured substrates B) cocultivation seeding procedure using the frame apparatus (described in the main text).
Selective Cell Removal—Scraping Technique
The setup for selective cell removal, hereafter referred to as “scraping,” is shown in Figure 8. In this procedure, the microstructured substrate with the seeded cells is placed on a tape‐covered sterile glass slide in 50 µL culture medium after taking out from the frame apparatus. The side of the scaffold intended for cell removal (in this case, the outer surface with the HepG2 cell layer) faces down and touches the glass slide. A PDMS weight, approximately the same size as the scaffold and weighing 100 mg, is then positioned on the sample (Figure 8A). The scraping process is initiated by piercing the PDMS weight with the tip of a micromanipulator. The micromanipulator is used to move the PDMS weight and attached sample along the orientation of the patterned lines on the scaffold. Friction between the glass slide and the raised areas of the scaffold causes cells to be selectively removed from these areas. This scraping motion is performed at a speed of 130 µm s^−1^, covering a distance of 1.8 mm in the forward and backward direction.
Schematic representation of the scraping technique: A) Experimental setup for targeted cell removal: the PDMS weight and the underlying sample are moved with the tip of a micromanipulator on a sterile glass slide, resulting in scraping of a HepG2 cell layer; a magnifying camera serves for magnification of the procedure. B) cell localization of EA.Hy 926 and HepG2 cells before and after scraping.
After scraping, the sample is carefully rinsed to remove detached cells. Final cell localization is shown in Figure 8B. Depending on the subsequent steps, the scaffold is either reinserted into the frame for further processing or fixed directly for immunofluorescence analysis.
Determination of Scraping (Selective Removal) Performance
To determine the success of the scraping (removal) performance, brightfield images of 11 scraping trials (4 independent experiments with two to three replicates) on foils were investigated 24 h after the scraping procedure with HepG2 cells. 9 unscraped HepG2‐populated foils (three independent experiments with three replicates) were investigated in parallel and served as controls.
For cell counting, a rectangle was created with a width of three times the ridge width and a length of 6 times the ridge width of the investigated foil. The rectangle was always positioned in the center of the image (around a channel and half of the ridges on both sides of the channel) (Figure S2, Supporting Information). For counting the cells, ImageJ analysis software with the cell counting tool was used. The number of HepG cells in the channel and on the left and right ridges in the rectangle was counted.
To calculate the scraping efficiency, the percentage of cell coverage of both areas was determined for every trial by referring to the mean value of all nonscraped, cell‐covered areas, which were set to 100%. The mean value and standard deviation from all trials were calculated.
Direct Coculture
The above‐described seeding procedure (2.5 and 2.6) was used for the establishment of the double‐sided coculture on the microstructured scaffold. The same seeding procedure was used for the direct coculture, except that collagen coating was not used, to observe the effects of direct cell‐cell contact. For microscopic analysis using various staining methods, co‐cultures were fixed after a 48 h incubation period and further processed.
Loading NIH‐3T3 Fibroblasts with the CellTracker Violet‐BMQC
A confluent cell layer of NIH3T3 cells was carefully washed with PBS and subsequently loaded with the cell tracker Violet‐BMQC (#C10094, Thermofisher, Darmstadt, Germany) in serum‐free DMEM at a final concentration a 10 µm for 30 min at 37 °C in a CO_2_‐incubator. At the end of the incubation period, the solution was removed, and normal serum‐containing DMEM medium was added to the cells. The cells were cultivated for 3 h in normal NIH3T3 culture medium at 37 °C and 5% CO2 and then detached by trypsination. 100.000 tracker‐loaded fibroblasts were used for seeding on microstructured substrate in triculture experiments.
Triculture
Directly after scraping of the HepG2 cell layer in the site‐selective coculture approach, 100.000 Violet‐BMQC‐loaded NIH3T3 fibroblasts in 25 µL culture medium were added on top of the microstructured substrate to the HepG2 cell strands and incubated for 30 min. Afterward 1 mL culture medium was added on the tricultures. The triculture was either fixed directly after fibroblast seeding with 4% PFA or further cocultivated up to 48 h. After fixation, the tricultures were microscopically analyzed for the cell tracker Violet‐BMQC (NIH3T3 fibroblasts), AFP (HepG2 cells), and CD31 (EA.Hy 926 cells) (see below).
Immunofluorescence Staining of CD31 and Alpha Fetoprotein
Cells on microstructures were fixed in 4% paraformaldehyde (PFA) for 15 min and treated with 0.25% Triton X‐100 for 15 min at room temperature (RT). After rinsing with PBS, the cells were blocked with 5 % BSA in PBS for 1 h at 4 °C. The cells were then incubated with the following primary antibodies overnight: mouse antihuman CD31/PECAM‐1 antibody (RB01, R&D Systems, Minneapolis, USA) for EA.hy 926 cells and rabbit antialpha fetoprotein antibody for HepG2 cells (cat # PA5‐16658, Thermofisher, Darmstadt, Germany). After three rinses with PBS, the cells were incubated at RT for 3 h with species‐dependent secondary antibodies: Alexa Fluor 647‐labeled goat anti‐mouse antibody and Alexa Fluor 488 labeled goat anti‐rabbit antibody (Thermofisher, Darmstadt, Germany). Cells were mounted in Fluoromount G containing DAPI (00‐4959‐52, Thermofisher Scientific, Darmstadt, Germany). Images were captured with an OLYMPUS laser scanning microscope FV1000 (Olympus, Germany). The z‐axis series of fluorescent images were reconstructed in 3D using Imaris software (Bitplane AG, Zurich, Switzerland).
Live/Dead Staining
For live/dead staining, cells on microstructured substrates were incubated with the live/dead cellular staining kit II (PromoCell GmbH, Heidelberg, Germany) according to the manufacturer's instructions.
The cell culture medium above the sample was removed, and the sample was rinsed once with PBS. Subsequently, 500 µL of the staining solution (containing Calcein and Ethidium DIII homodimer) was added per well, and the samples were incubated at 37 °C for 30 min at 37 °C in an incubator (protected from light). After incubation, the samples were placed directly on a slide in PBS, covered with a coverslip, and placed in the LSM. Images were taken with an OLYMPUS laser scanning microscope FV1000 (Olympus, Germany). The Z‐stack fluorescent images were analyzed using the Imaris software (Bitplane AG, Zurich, Switzerland).
Actin Staining
After fixation and permeabilization (like described above), cells were incubated with rhodamine phalloidin‐solution (ActinRed555, R37112, Thermofisher Scientific, Darmstadt, Germany) for 30 min in an incubator at 37 °C. After two rinses with PBS, cells were mounted with Fluoromount G containing DAPI (00‐4959‐52, Thermofisher Scientific, Darmstadt, Germany). Images were taken like described above.
SEM Analysis
Morphology of the cells inside the microstructured substrates was analyzed with SEM analysis. Cells were fixed using 2,5% glutaraldehyde at 4 °C for 1 h and washed two times with A.bidest. Samples were dried, sputtered with a thin platinum layer, and examined by scanning electron microscopy (SEM Hitachi S 4800‐II, Hitachi High‐Technologies Europe GmbH).
Statistics
To calculate the scraping efficiency, the percentage of cell coverage of the channel area and the left and right ridge area around the channel was determined for every cell removal trial (11 in total) by transforming all single values to the mean value of all nonscraped, cell‐covered areas, which was set to 100% (unscraped control). Subsequently, the mean value and standard deviation from all trials were calculated.
Conflict of Interest
The authors declare no conflict of interest.
Supporting information
Supporting Information
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Y. Wang , S. E. Naleway , B. Wang , Bioact. Mater. 2020, 5, 745.32637739 10.1016/j.bioactmat.2020.06.003PMC 7317171 · doi ↗ · pubmed ↗
- 2C. Jensen , Y. Teng , Front. Mol. Biosci. 2020, 7, 33.32211418 10.3389/fmolb.2020.00033 PMC 7067892 · doi ↗ · pubmed ↗
- 3E. C. Costa , A. F. Moreira , D. de Melo‐Diogo , V. M. Gaspar , M. P. Carvalho , I. J Correia , Biotechnol. Adv. 2016, 34, 1427.27845258 10.1016/j.biotechadv.2016.11.002 · doi ↗ · pubmed ↗
- 4M. Ravi , V. Paramesh , S. R. Kaviya , E. Anuradha , F. D. Solomon , J. Cell. Physiol. 2015, 230, 16.24912145 10.1002/jcp.24683 · doi ↗ · pubmed ↗
- 5Y. B. Kang , S. Rawat , J. Cirillo , M. Bouchard , H. M. Noh , Biofabrication 2013, 5, 045008.24280542 10.1088/1758-5082/5/4/045008 PMC 3935322 · doi ↗ · pubmed ↗
- 6N. J. Darling , C. L. Mobbs , A. L. González‐Hau , M. Freer , S. Przyborski , Front. Bioengin. Biotechnol. 2020, 8, 2020.10.3389/fbioe.2020.00992 PMC 748734232984279 · doi ↗ · pubmed ↗
- 7C. R. Justus , M. A. Marie , E. J. Sanderlin , L. V. Yang , Methods Mol. Biol. 2023, 2644, 349.37142933 10.1007/978-1-0716-3052-5_22PMC 10335869 · doi ↗ · pubmed ↗
- 8J. Pijuan , C. Barceló , D. F. Moreno , O. Maiques , P. Sisó , R. M. Marti , A. Macià , A. Panosa , Front. Cell Dev. Biol. 2019, 7, 107.31259172 10.3389/fcell.2019.00107 PMC 6587234 · doi ↗ · pubmed ↗