3. The MatriGrid®-Family
3.1. 3D Hepato MatriGrid®
Complimentary to various liver-on-a-chip models
[37][38][39], MatriGrid
®s with typical cavity/container-like morphology were used for culturing human primary Upcyte
® hepatocytes and HepaRG hepatocarcinoma cells to mimic organotypic liver growth. Generally, biopsy-derived primary human hepatocytes (PHHs) are the gold standard for in vitro experiments on liver biology and for studying the hepatoxicity of a wide variety of drugs
[40]. They can be cultured as monolayers only for a limited period of time due to rapid dedifferentiation and loss of the expression of CYP450 enzymes. Using Upcyte
® technology that releases primary hepatocytes from cell cycle arrest by overexpressing the HPV oncogenes E6 and E7 without immortalization, long-proliferating hepatocytes from various donors with stable function were created recently
[41]. Many studies have demonstrated that Upcyte
® hepatocytes are suitable for preclinical drug metabolism and hepatotoxicity investigations
[42][43][44]. Another hepatocarcinoma cell line, the HepaRG cell line, is also convincing due to long-lasting hepatofunctionality, but it has a significant disadvantage, namely, a long differentiation time of over 4 weeks with dimethylsulfoxide
[45][46]. Both cell types were established in MatriGrid
® scaffolds and compared in terms of albumin production. Upcyte
® hepatocytes were seeded with different starting cell numbers in MatriGrid
®s, and albumin production was monitored over a time period of 28 days (
Figure 4). The highest seed cell number resulted in a sharp increase in albumin secretion after 7 days of culturing in MatriGrid
®s. Continuous culturing of the cells for up to 28 days resulted in an almost similar and stable albumin secretion that was independent of the seeding cell number. In contrast, differentiated HepaRG cells produced more than 20× less albumin than the Upcyte
® hepatocytes grown for 28 days in MatriGrid
®s.
Figure 4. (A) Albumin secretion of Upcyte® hepatocytes (donor 422) and differentiated HepaRG cells cultured in MatriGrids® for up to 28 days. Albumin secretion is linked to the cell seeding number. Shown are the mean values ± SD; n = 2 experiments. (B) Upcyte® hepatocytes (donor 10_03) were cultured for 7 days in 2D or in MatriGrid®s, and albumin secretion was measured. Shown are the mean values ± SD; n = 3 experiments. (C) DAPI/F-Aktin/ZO-1 staining detects Bile canaliculi in Upcyte® hepatocytes cultured for 7 days in 2D or in MatriGrid®s (MG). Arrows label ring-like Bile canaliculi in 2D-cultured cells and tube-like Bile canaliculi in MatriGrid®s by detection of the tight junctional marker ZO-1. Bars represent 30 µm.
Additional monitoring of cell numbers and viability revealed gradually increasing cell counts over time and stable viability up to 28 days for Upcyte® hepatocytes (data not shown). MatriGrid® versus monolayer (2D) culturing of Upcyte® hepatocytes for 7 days showed 2-fold increased albumin production due to the 3D environment in MatriGrid® cavities (Figure 4). This impressively demonstrates that the 3D environment provided by the cavity morphology leads to an improvement in the hepatofunctionality of Upcyte® hepatocytes. This result was further supported by labeling Bile canaliculi through ZO-1-F-actin staining in monolayer- and MatriGrid®-grown Upcyte® hepatocytes. Organotypic hepatocyte culture clearly promotes tubular versus ring-like Bile canaliculi labeled by zona occludens protein-1. With these data, it has demonstrate that the organotypic 3D culturing of hepatocytes in MatriGrid®s significantly improves hepatofunctionality and, thus, is more suitable for studies on liver biology and hepatotoxicity than monolayer culture.
3.2. Lung MatriGrid®—An Example of Directed Oligocellular Coculture
Early cell culture models of the lung were mostly based on flat and porous membranes made of, e.g., polycarbonate (PC) or polyethylene terephthalate (PET), on which different combinations of alveolar cells, epithelial cells, and blood cells were cultured
[47][48][49][50]. Khalid et al. have developed a lung-cancer-on-chip platform as a promising tool for the cytotoxicity evaluation of novel drug compounds
[51].
While using well plate inserts in an air–liquid-interface could be applied to these models to mimic specific lung physiology, Huh et al.
[52] were the first to report a lung-on-chip model with a uniaxially stretched membrane to mimic the breathing motion of the lung. In a similar approach, Huang et al. reported a hydrogel-based physiologically relevant model of human pulmonary alveoli
[53]. That concept was later improved with a 3D-stained PDMS membrane
[54] and the integration of impedance sensors to monitor cell behavior and membrane movement
[55]. Furthermore, with respect to the alveoli dimensions, a biodegradable and stretchable biological membrane from collagen and elastin was reported to mimic the central aspects of the air–blood barrier
[56]. Comparative to that approach within the MatriGrid
®-family, a Lung-MatriGrid
® based on the shape and size of human lung alveoli was developed using the previously described biotechnical microscale engineering and micro thermoforming technology (
Figure 5).
Figure 5. Biotechnical multiscale engineering of the lung.
The Lung-MatriGrid
® permits the possibility of a 3D oligocellular coculture model of the blood–air barrier with respect to physiological characteristics. The blood–air barrier is characterized by an extremely thin and highly connected layer of epithelial cells that is spread over pulmonary capillaries
[57]. Only a 1–2 µm thin structure supports the passive diffusion of respiratory gases
[58]. To enable such organotypic exposure of the alveolar epithelia cells against ambient air, the oligocellular coculture model is cultured under an air–liquid interface (ALI) condition. This more physiological ALI culturing is realized by attaching the Lung-MatriGrid
® onto the previously described semi-active system (
Figure 5). The barrier itself is realized by the seeding of capillary endothelia cells to the basal side of the scaffold, while alveolar epithelia cells are brought into the alveolar-like cavities on the apical side (
Figure 6)
Figure 6. (A) Scheme of generation of 3D co-culture. Seeding of endothelial cells on the basal side of the MatriGrid®s in a petri dish, transfer to an MTP, and seeding of the apical side with epithelial cells, 24 h incubation as LLI Culture, creation of the ALI culture; (B) MatriGrid®, (C) REM-picture of a sliced MatriGrid®, dimensions of the cavities (red arrows), (D) semi-active system with MatriGrid® on the bottom, (E) 24-well MTP assembled with semi-active systems.
The medium in ALI culturing is only present underneath the basal side of the scaffold supply of the apical cells, which is ensured by the porosity of the scaffold (pore diameter = 2–4 µm, pore density = 106 pores/cm2, thickness = 10–40 µm). Viability and metabolic activity of alveolar epithelial cells (A549) and capillary endothelia cells (EAhy.926) cultured in a lung MatriGrid® are shown to be not compromised in comparison to the culturing in a liquid–liquid interface (LLI) culture over 12 days (Figure 7). Additionally, A549 forms a thick monolayer, with the expression of cell-adhesion molecules (ZO-1, E-cadherin) in the cavities of the Lung-MatriGrid® (Figure 6A). Since the Lung-MatriGrid® can be reversibly separated from the insert system, the cell layer on the scaffold, as well as the cell culture medium, can be evaluated with established methods such as (immuno)-histochemical staining or ELISA to identify changes in cytokine levels or the expression of cell-adhesion molecules due to exposure to, e.g., nanoparticles. Further research is ongoing in the DFG project (DFG 397981139) in cooperation with the Institute of Environmental Toxicology at Martin-Luther-University Halle-Wittenberg to examine the toxicity of BaSO4 nanoparticles on primary cells and cell lines in Lung-MatriGrid®s.
Figure 7. Immunohistochemical stainings of cell-adhesion molecules. (A) Zonula occludens-1 and (B) E-cadherin; (C) comparison of the viability of air–liquid interface (ALI) and liquid–liquid interface (LLI) cultures of A549 on the apical side of the MatriGrid®. Bar represents 50 µm.
3.3. NeuroGrid®—Scaffolds for the Manipulation and Directed Growth of Neurons and Cerebral Organoids
It is well known that different cellular processes, such as attachment, proliferation, directional migration, and differentiation of neurons, are dependent on morphological and biochemical cues in the surrounding surface
[59][60][61]. In this way, the direction and outgrowth of axons and dendrites have been studied by symmetric or asymmetric shapes of trenches
[62][63] or specific microplates
[64] to guide the connectivity between 3D neuronal cell clusters
[65] or to record muscle activity after stimulation of axons in different microfluidic chambers
[66]. In addition, there are indications found that the migration capacity of neural cells depends on the stage of neuronal differentiation
[67]. Additionally, circular 3D PDMS scaffolds have been used for defining spheroid-like neuronal cell agglomerates
[68]. The design of polymeric scaffolds in PC was described in here, which should be used as a proof-of-concept study for the handling and shaping of neurons and neuronal organoids.
For this reason, it is being investigated whether appropriate modifications of the MatriGrid
® scaffold can enable the directed growth of neurons to reproduce desired morphologies, with the vision of developing flexible tools to mimic the complex hierarchies of neuronal tissue. Various designs of MatriGrid
®s were used to induce the guided growth of embryonic and adult neurons. The set of MatriGrid
®s with function-dependent embossed structures is given in
Figure 8. Besides the fact that neurons are guided easily by microchannels, the geometries of structures were inspired by the size of neuronal fiber bundles, which are in the range of approximately 500 µm in the case of a cortical column
[69][70].
Figure 8. Scaffold designs for guided cell growth; (A) vertical and horizontal trenches; (B,C) different scaled tree structures to mimic the cortical column; (D) NeuroGrid® tool with openings for the 3D MEA.
Primary rat cortical neurons were used to allow directed growth through the structures. In the same way, the culturing of neurospheres from induced pluripotent stem cells (iPSC) was investigated. MatriGrid® cultures are also used to bring the guided neurons into close contact with microelectrodes of 2D and 3D MEAs so that it is easier to capture the neuronal signals. That targeted application of MatriGrid® scaffolds can easily be extended by stacking those scaffolds to create complex 3D models to evaluate real 3D network signals from neuronal cells.
Figure 9 shows the directional growth of neuronal cells that have grown out of neurospheres. The typical radial outgrowth of neuronal cells from a neurosphere can be seen on the unstructured PC foil (Figure 9B), although it is not possible to evaluate the growth of neurons within the trench structure using bright field microscopy (Figure 10A). Cell staining is needed to estimate any guided neuronal cell growth inside the narrow NeuroGrid®-structures. Live–dead staining as well as immunofluorescence staining against neuronal markers microtubulin-associated protein 2 (MAP2) and βIII-tubulin (TUBB3) were performed. In addition, the cell nuclei were stained with DAPI (blue). The trench structure forces a directional growth compared to the unstructured PC foil (Figure 9C,D). Mainly viable (green-stained) cells outside the neurospheres can be seen. Viable cells, on the other hand, are present in the neurospheres. While TUBB3-stained neurons grow along the trenches, MAP2-stained cells do not undergo this directed growth and are also found in the areas between the trenches (Figure 10E). In contrast, on the unstructured PC foil, mainly MAP2-stained cells grow out of the aggregates of the rat cortex neurons, and only a small number of TUBB3-stained cells can be seen (Figure 9F).
Figure 10. Directional growth of neuronal cells within tranches and radial growth on unstructured PC foil. Brightfield microscopy of the (A) channel structure and (B) unstructured PC foil. Live–dead staining of neurospheres on the (C) channel structure and (D) unstructured PC foil (green—viable cells, red—dead cells). Immunofluorescence staining of rat cortex neurons within (E) a trench structure and (F) on unstructured PC foil (blue—cell nuclei (DAPI), red—βIII-tubulin (TUBB3, Alexa Fluor 594), green—microtubulin-associated protein 2 (MAP2, Alexa Fluor 488)). Bar represents 100 µm.
A major problem in deriving neuronal signals is the growth of neuronal cells outside the sensor area of the MEA. This circumstance makes planning more difficult and prevents the standardization of the experiments. For this reason, the targeted application of neuronal cells to the electrodes of the 2D and 3D MEAs was tested using MatriGrid® foils (Figure 11A). Normal PC foils and MatriGrid® with a towering tree structure were used for the experiments with 2D MEAs. Special 3D MEA foils with cutouts for the 3D needle electrodes were used for targeted positioning on 3D MEAs. In both variants, neurospheres were pre-cultured on the foils for 7 days and then transferred to the 2D or 3D MEA and cultured further for at least 7 days. The signals were recorded daily after the transfer to the MEA. The neurospheres for the 2D MEA were pipetted as centrally as possible onto the foil or into the tree structures of the MatriGrid®. For the 3D MEA, neurospheres were placed in the structures on the ridges between the cutouts for the needle electrodes. All foils were coated with Geltrex in order to cover the entire foil with cells. Figure 11 shows the behavior of the neuronal cells after application over a culturing period of 9 days. While the neurospheres are intact on the first day after the transfer (Figure 11B), they show gaps during longer culturing, which become larger the longer the foils are cultured on the MEAs (Figure 11C,D). Individual outgrown neuronal cells can be seen within these gaps. However, it cannot be seen whether these have grown on the foil or the MEA (Figure 11D). After a few days, detached cells are present in the medium, both at the edge of the foils placed (Figure 11E) and over the remaining MEA surface (Figure 11F). These do not adhere to the surface of the MEA but form cell aggregates. It could also be observed that the PC film slips on the 2D MEA when changing the medium (Figure 11F).
Figure 11. (A) Principles of targeted application of neuronal cells by MatriGrid®; (B) neurospheres transferred from MatriGrid® to 2D MEA after 1 day; (C) 6 days and (D) 9 days of culturing; detached neuronal cells (E) at the edge of the applied foil and (F) on the outer MEA surface; (G) PC film slipped after medium change.
Figure 12 shows the 3D MEA foil before it was transferred to the 3D MEA (Figure 12A). When the foil was applied, it was transferred to the 3D MEA with the cells facing down. When neurospheres are cultured in the tree structures of the 3D MEA foil, they can be placed directly on the needle electrodes (Figure 12B); when cultured on the ridges between the cutouts, it is also possible to position the neurospheres specifically on the bottom electrodes of the 3D MEA (Figure 12C). It should be noted that individual neurospheres can become detached from the 3D MEA film during transfer (comparison of Figure 12A,B).
Figure 12. The 3D MEA foil for the targeted application of neurospheres on MEA electrodes. (A) Bright-field image of the 3D MEA foil before transfer to the 3D MEA; (B) 3D MEA without 3D MEA foil; (C) 3D MEA with 3D MEA foil and spheroid with contact with the needle electrodes; (D) sketch of a 3D MEA with bioreactor housing and 3D-stacked NeuroGrids; bar represents 500 µm.
Figure 13 shows an example of a 3D MEA measurement of neuronal signals after the targeted application of a 3D MEA foil. Neural signals were measured at the opposing electrodes B-014 and B-020 (comparison of Figure 13A,B). Here, electrode B-020 is a bottom electrode that is opposite the middle needle electrode B-014. The signals from the two electrodes show a high degree of synchronicity (Figure 13C). The background noise of the electrodes is between 5 and 10 µV; mainly negative spikes with a maximum of 20 µV were measured. Both signals contained bursts.
Figure 13. Neuronal signals measured after transferring the 3D MEA film to a 3D MEA (7 days of pre-culturing in a 6-well MTP and 7 days of culturing on the MEA) (A) Electrode array with numbering of the electrodes (B) Measurement from middle of the needle electrode (B-014) (C) Measurement on bottom electrode (B-020).
Besides the use for the directed growth of neurons and a targeted application to 2D or 3D MEAs, the MatriGrid®s can also be used as a handling tool to create more uniform spheroids of neuronal cells. For this purpose, dissociated rat cortex neurons were pipetted into the cavities of the MatriGrid®s. The cavities were previously coated with anti-adherence rinsing solutions. The spheroids precultured and shaped in the MatriGrid® were transferred directly onto 2D MEAs or into structures of other MatriGrid®s (Figure 14). For both approaches, the attachment of the spheroids and the outgrowth of neurons were verifiable. A big advantage of that approach in forming spheroids is the possibility of defining the diameter of the spheroid through the size of the cavities. Because of that, the spheroids are highly adaptable to the purpose they will be used for.
Figure 14. (A) Spheroids generated from cortex neurons in MatriGrid® adhered to 2D MEAs and (B) in the fir-tree structures of the 3D MEA foil. Bar represents 100 µm.
3.4. TissGrid®
The need for adapted scaffold structures is not only important in cell cultures based on cell lines or primary cell cultures; the culturing of explants or tissue slices can also benefit from the advantages of a scaffold approach. The scaffold creates an adapted microenvironment for the specific explant and, thus, optimal survival conditions. This is particularly important for longer culturing periods. Flow-induced shear stress has a big influence on cells and tissues
[71][72][73][74][75], either in a positive way, mimicking the effects of vascularization, or in a negative way on sensitive cells and tissue slices, where the stress damages the cells. In a case study, this effect on placenta tissues has been examined, called placenta explants, because drug and particle transport across the human placenta is a deciding factor for fetus development
[76][77]. It is generally known that fluidics also have a major influence on the cellular phenotypes of the placenta
[78]. Therefore, it is desirable not only to culture the explants statically but also to culture them inside microfluidic systems. Placenta explants are sensitive tissue structures that lose their integrity under fluid shear stress; they cannot maintain their viability for an indefinite period. The fluid stress effects on explants under different fluidic regimes was examined. Following the observation that placenta explants are very sensitive, a new scaffold structure was designed, which combined the potential shelter function of a porous cavity with the advantage of better fluidic supply with respect to nutritious flow. The TissGrid
® structure is designed in the following way. A central cylindrical cavity standing on a porous base surface is used to accommodate the explant. This can be easily inserted into the scaffold from above without damage. The cavity is made of microporous transparent polycarbonate film by thermoforming. Microscopic observation of the explant is possible during culturing. Due to the porosity, a very good diffusive supply of cells is possible. In order to achieve an effective flow around the explant cylinder and, thus, high diffusion gradients, bypass openings were inserted at the corners of the base surface. The base surface was also made of porous polycarbonate film, which also allows the microscopic inspection of the samples. To integrate the system into the microreactors described above, the scaffolds were fixed onto a carrier chip. In order to avoid an undesired influence on cell culture, no adhesives should be used to bind the scaffolds. Especially in long-term cultures, substances may leach out of the adhesive. In the application described here, the scaffold parts were, therefore, bonded with solvent. This could be completely removed by appropriate heat treatment in a vacuum. This way, easy storage and handling and good sealing of the scaffold in the bioreactor are possible. A schematic representation of the TissGrid
® is shown in
Figure 15.
Figure 15. (A) Schematic of TissGrid® with the flow path of the fluid; (B) the manufactured TissGrid®.
It was able to show that with the specially designed TissGrid
®s, a flow-through protective structure could be set up, which enables the explants to be supplied with medium/serum flowing past while maintaining viability
[79].
Under conventional culture conditions, without the influence of test substances, it was able to observe relatively stable glucose consumption and stable lactate production in placenta explants in a conventional microtiter plate (MTP) for up to 10 days, which indicates good placental functionality and metabolism. By changing the culture conditions with the help of specially developed fluidic systems (TissGrid
®s in bioreactors
[79];
Figure 16), a placental-active metabolism could even be stimulated, which shows an increased production of estradiol by the syncytiotrophoblast at flow rates of 100 µL/min of the culture medium (unpublished data). The substrate structures were thoroughly tested and led to the specially developed TissGrid
®. In addition, different flow rates were varied in these experiments.
Figure 16. Live–dead assays to determine the viability of placental explants after 14 days of culture under fluidic conditions are shown to the left of the TissGrid® (B) and the MatriGrid® (D). Viable tissue is green; dead tissue is red. The villous structures of the placenta react very sensitively to shearing forces, which leads to a significant reduction in the viability of the placental tissue in the MatriGrid® (C). The protected environment in the TissGrid®, on the other hand, enables excellent regeneration of the placenta explant (A). In the live–dead assay, the original degenerated syncytiotrophoblast (red) can be seen, which has been replaced by a newly formed one (green) on the surface of the explant (A).
In contrast, the relative decrease in estradiol production under static (not perfused) conditions (2D, static plate) and the low flow rates of 10 µL/min indicate the degeneration and decreased function of the explants. Preliminary tests also showed that it was not possible to carry out tests at high flow rates in the structures developed for 3D cell culture (called MatriGrid®s (Figure 16)). The tissue lost its intact structure (Figure 16C). With the help of the TissGrid®s in special MTPs or microbioreactors designed for this purpose (Figure 16A), which are to be standardized, a further time window can be opened up due to the supply of the placenta explants with culture medium.
Besides the culturing of primary cells and even explants in fluidic setups, as described before, long-term experiments bear a high potential for the application of MatriGrid® scaffolds.