Human Induced Pluripotent Stem Cell‐Derived Neural Progenitor Cells Produce Distinct Neural 3D In Vitro Models Depending on Alginate/Gellan Gum/Laminin Hydrogel Blend Properties

Abstract

Stable and predictive neural cell culture models are a necessary premise for many research fields. However, conventional 2D models lack 3D cell‐material/‐cell interactions and hence do not reflect the complexity of the in vivo situation properly. Here two alginate/gellan gum/laminin (ALG/GG/LAM) hydrogel blends are presented for the fabrication of human induced pluripotent stem cell (hiPSC)‐based 3D neural models. For hydrogel embedding, hiPSC‐derived neural progenitor cells (hiNPCs) are used either directly or after 3D neural pre‐differentiation. It is shown that stiffness and stress relaxation of the gel blends, as well as the cell differentiation strategy influence 3D model development. The embedded hiNPCs differentiate into neurons and astrocytes within the gel blends and display spontaneous intracellular calcium signals. Two fit‐for‐purpose models valuable for i) applications requiring a high degree of complexity, but less throughput, such as disease modeling and long‐term exposure studies and ii) higher throughput applications, such as acute exposures or substance screenings are proposed. Due to their wide range of applications, adjustability, and printing capabilities, the ALG/GG/LAM based 3D neural models are of great potential for 3D neural modeling in the future.

The manuscript describes two fit‐for‐purpose approaches to generate 3D in vitro neural models based on human induced pluripotent stem cells. It is shown that stiffness and stress relaxation of the proposed gel blends, as well as the cell differentiation strategy influence the 3D model development. Successful differentiation and intracellular calcium signaling within the 3D microtissues are further shown.

1. Introduction

The human central nervous system (CNS) consists of a complex cellular network in which the microenvironment, like spatiotemporal exposure to signaling molecules and cell‐cell and cell‐matrix interactions, plays a crucial role for its proper development and function.^{[ 1 ]} Given that such interactions are largely absent in 2D cell cultures, the frequent failure in translating in vitro findings into in vivo applications is inevitable.^{[ 2} , ³ , ⁴ , ^{5 ]} The generation of stable and predictive neural cell culture models is central for many fields dealing with, for example, toxicological evaluation, disease modeling, drug development, and regenerative medicine. Therefore, there is a need for more sophisticated models that better mimic the physiology of the human brain.

Substantial efforts have been made to add the third dimension to standard 2D cell cultures. Starting from the beginning of this century, several approaches for the generation of 3D neural models such as neurospheres and hydrogel scaffolds were developed.^{[ 6 ]} Essentially, 3D structure can be achieved by harnessing the self‐organization properties of cells, for example, to drive the formation of neurospheres, or by providing support and structure for cells with hydrogel scaffolds in an engineering‐based manner. Neurospheres are cell aggregates that consist of neural progenitor cells (NPCs) and are cultivated in the presence of growth factors. In the absence of growth factors and when seeded on 2D poly‐D‐lysin/laminin (PDL/LAM)‐coated surfaces, NPC neurospheres migrate and differentiate into neurons, astrocytes, and oligodendrocytes, thereby generating complex networks.^{[ 7} , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ^{13 ]} The dimension of such neurospheres when plated for migration and differentiation is called “secondary 3D.”^{[ 14 ]} NPCs generated from induced pluripotent stem cells (iPSCs), iNPCs, have been previously employed in neurotoxicity testing and modeling of Alzheimer's and Parkinson's disease.^{[ 13} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ^{20 ]} These secondary 3D models were developed to improve the classical 2D‐monolayer neuronal cultures.^{[ 21} , ^{22 ]} However, plated neurospheres do not form a 3D network on 2D substrates. Engineered 3D biomaterial models such as hydrogel scaffolds complement the existing cell models by providing an adaptable, controlled, and consistent extracellular environment.^{[ 23 ]} 3D models augment the complexity of conventional cell cultures, thus rendering them more predictive and physiologically relevant.^{[ 6} , ^{14 ]} However, variability and reproducibility in these systems are still challenging, often due to the nature of the materials. Batch‐to‐batch variations in, for example, matrigel or limited cell‐material interaction in synthetic gels play important roles.^{[ 24} , ^{25 ]}

The interest in hydrogels has increased largely since their discovery in 1960.^{[ 26 ]} These 3D matrices consist of hydrophilic polymers that hold large quantities of water. Due to this high water‐content (>90%), these gels can exhibit tissue‐like properties.^{[ 27 ]} Hydrogels maintain their structure by chemical, physical, or biochemical crosslinking of the polymer chains.^{[ 28 ]} Ideally, extra‐cellular matrix (ECM)‐mimicking hydrogels should support cell survival, growth, differentiation, cell‐cell, and cell‐matrix adhesion, as well as facilitate proper nutrient flux.^{[ 29} , ^{30 ]} In order to allow cellular outgrowth, a certain degree of hydrogel‐degradation is desirable. The material stiffness also plays an important role for the generation of neural models. With a Young's elastic modulus of 0.5 to 50 kPa, the brain is one of the softest tissues in the human body.^{[ 31} , ^{32 ]} As an example, natural polymers such as alginate (ALG) and gellan gum (GG) are inherently suitable surrogates for ECM, due to their high water content and tunable stiffness, as well as their chemical versatility and biocompatibility.^{[ 33} , ³⁴ , ³⁵ , ³⁶ , ^{37 ]}

ALG is a seaweed‐derived marine polysaccharide and one of the most commonly used biomaterials for hydrogel formation. It can be easily gelatinized with divalent cations and is highly biocompatible.^{[ 38 ]} It is composed of D mannuronic acid (M) and L guluronic acid (G) monosaccharide units and forms hydrogels through crosslinking of the G residues with divalent cations.^{[ 34} , ^{39 ]} GG is a natural extracellular polysaccharide produced by the bacterium Sphingomonas paucimobilis, which forms a gel after crosslinking of its double helices with divalent cations such as Ca²⁺ or Mg²⁺.^{[ 33 ]} Although many modifications and blends have been produced from ALG and GG,^{[ 34} , ^{40 ]} no ALG/GG blend has yet been used for the development of 3D neural models. Since ALG and GG gels are biologically inert,^{[ 34} , ^{41 ]} they are often functionalized with native ECM molecules, such as LAM, collagen, or fibronectin.^{[ 1} , ⁴² , ⁴³ , ^{44 ]} LAM has previously been used to support cell survival, network formation, and functional development of neural cultures in vitro.^{[ 1} , ^{45 ]} Moreover, it was reported that ALG‐based hydrogels can mimic the complex mechanical properties of brain tissue.^{[ 46 ]}

The engineering of physiologically relevant structures and the recapitulation of spatiotemporal availability of signaling molecules in neural models is highly challenging. 3D bioprinting offers a promising tool for the generation of such model systems.^{[ 36 ]} Cell‐supplemented biocompatible materials are printed with micrometric precision and may even combine several cell types, thus generating complex cellular networks. The printing process depends on three main variables: 1) the material (biomaterial ink), 2) the cells and 3) biomechanical factors.^{[ 25} , ⁴⁷ , ⁴⁸ , ^{49 ]} Due to their highly tunable rheological properties, hydrogels are suitable for bioprinting applications.^{[ 35} , ^{50 ]} However, neural cells require soft materials and are sensitive to shear stress, which only permits hydrogels with very specific properties.^{[ 51 ]}

Bioinks based on ALG or GG, yet not in combination, have previously been employed to generate bioprinted and non‐bioprinted 3D neural models.^{[ 37} , ⁵² , ⁵³ , ⁵⁴ , ^{55 ]} Combining suitable gel properties for adequate cell‐culture development and functionality, with printability and long‐term integrity of the hydrogels remains a challenge. Although both ALG and GG are promising candidates for the generation of 3D neural cultures, they each lack desirable properties. While GG is not surface adherent, ALG deforms severely upon crosslinking. However, ALG appears surface adherent and is stable over long periods of time, which is important for long‐term cultivation. The GG is printable in low concentrations and its softness appears favorable for neural cultures. To date, there is no gold standard for the generation of hydrogel‐based 3D neural cultures and the full potential of such systems has yet to be exploited.

In this study, we chose ALG and GG for their completive favorable properties and present two ALG/GG/LAM blends for the generation of functional 3D neural models. We chose the gel blends according to their gel integrity, surface adherence, cell survival, and neural outgrowth. We characterized the hydrogels via Fourier transformed infrared spectroscopy (FTIR), scanning electron microscopy (SEM), mechanical testing, and degradation studies. We additionally show the printability of the gel blends using an extrusion‐based 3D bioprinter. Pre‐differentiated and non‐pre‐differentiated hiNPC neurospheres were embedded into both 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM hydrogels. The growth and differentiation of these multicellular models into neurons and astrocytes within the gels were assessed via immunocytochemistry (ICC) and the network functionality was verified by intracellular calcium imaging.

2. Results

2.1. Material Characterization

Reproducibility and close characterization of the cellular microenvironment are important factors in cell culture maintenance and application. Physico‐chemical tests are generally used to characterize material properties and to anticipate their behavior and performance under cell culture conditions. Therefore, we characterized the developed hydrogels and used pure ALG and GG as comparative materials. ALG and GG both formed clear and stable hydrogels and blends upon crosslinking with CaCl₂. Pure ALG deformed upon crosslinking (Supporting Information), whereas GG and the 0.3% ALG/0.8% GG blend retained their structural integrity. The 1.5% ALG/0.5% GG blend appeared slightly deformed after crosslinking (data not shown). All gels maintained their structure over at least 4 weeks, when immersed in CINDA medium. ALG and both hydrogel blends reliably adhered to PDL/LAM‐coated plastic and polymer surfaces, as well as to non‐coated plastic surfaces, whereas GG is not surface adherent (Supporting Information).

All hydrogels formed microporous microstructures after crosslinking, as shown by SEM (Figure  1A). Peak shifts from 1406 to 1427 cm^–1 and 1595 to 1627 cm^–1 were indicative for the presence of LAM in all LAM containing samples, corresponding to amine and amide bonds (amide I ≈1621 cm^–1; Figure 1B).^{[ 56} , ^{57 ]} Peaks of GG at 1031 and 1627 cm⁻¹ corresponding to C─O stretching and COO^– asymmetric stretching^{[ 58 ]} confirmed the presence of GG in all GG‐containing hydrogel blends, while the peak at 1627 cm^–1 cannot be well discerned from the amide I peak expected from LAM. In sum, the FTIR results indicated the successful presence of ALG, GG, and LAM in all respective hydrogel blends.

Figure 1.

Physico‐chemico characterization of ALG and GG based hydrogels. A) SEM images of i) 0.9% GG/0.01% LAM, ii) 1.5% ALG/0.5% GG/0.01% LAM, iii) 0.3% ALG/0.8% GG/0.01% LAM, and iv) 1.0% ALG/0.01% LAM hydrogels. B) FTIR analysis of ALG/GG/LAM composite hydrogels. 1.0% ALG serves as reference. C) Hydrogel swelling and degradation of i) 0.9% GG, ii) 1.5% ALG/0.5% GG, iii) 0.3% ALG/0.8% GG, and iv) 1.0% ALG hydrogels assessed for 21 days of incubation in PBS (37 °C; N = 7). D) Qualitative stress relaxation behavior of the different hydrogels. E) Initial elastic modulus and F) stress relaxation time as the time after which 50% of the initial stress (at 15% strain) has been dissipated (N > 3) of the hydrogels. The legend for (B), (D), (E), and (F) is shown on the bottom. Data are shown as mean ± SD. Statistically significant differences among means between the different groups are indicated as *p < 0.05, analyzed using one‐way ANOVA and Bonferroni post‐hoc tests. Abbreviations: GG, gellan gum; FTIR, Fourier transformed infrared spectroscopy; SEM, Scanning electron microscopy; Ref., reference.

All GG‐containing hydrogels show an initial swelling phase of 3–5 days, followed by water release and degradation (Figure 1C). 1.0% ALG/0.01% LAM lacks the initial swelling phase, but also degrades over time. All hydrogels are stable for at least 21 days under cell culture conditions (Figure 1C). Figure 1D,F displays the qualitative stress relaxation experiments, as well as the quantification of the stress relaxation time (t_1/2), as the time point after which 50% of the initial stress (at 15% strain) has been dissipated from the hydrogels. All gels showed a fast stress relaxation (t_1/2 < 20 s) with 0.3% ALG/0.8% GG/0.01% LAM having the shortest (≈7 s) and 1.5% ALG/0.5% GG/0.01% LAM having the longest (≈16 s) stress relaxation time (Figure 1D,F). The addition of 0.3% ALG to 0.8% GG and 0.01% LAM accelerated stress relaxation in comparison to pure 0.9% GG hydrogels, while the reduction of ALG content from 1.5% to 0.3% significantly (*p < 0.05) reduced stress relaxation time of the hydrogels (Figure 1D,F). Figure 1E indicates the gel stiffness between the different conditions, derived from uniaxial compression test.^{[ 59 ]} The initial elastic modulus of the hydrogels revealed 1.5% ALG/0.5% GG/0.01% LAM as the stiffest (≈35 kPa) and 1.0% ALG/0.01% LAM as the softest (≈5 kPa) hydrogel, while 0.3% ALG/08% GG/0.01% LAM hydrogels showed a lower stiffness in comparison to the 1.5% ALG/0.5% GG/0.01% LAM composition (≈20 kPa; Figure 1E).

We showed that the developed hydrogel blends are structurally stable and surface adherent, which is beneficial for long‐term cultivations. Furthermore, all three components are contained within the crosslinked hydrogels, thereby excluding initial washout of ALG, GG, or LAM. The gel blends degrade slowly over time, which is generally thought of as beneficial for cellular outgrowth and migration. The fast stress relaxation times may additionally support this effect. The elastic moduli of the hydrogel blends lay within the range of brain stiffness.

2.2. Long‐Term Neural Differentiation

We tested the hydrogel blends for their applicability in long‐term cultivation and differentiation of hiNPCs. HiNPC spheres were chopped to 0.1 mm and embedded into 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM blends (Figure  2A). Both ALG/GG blends exhibited satisfying cell compatibility, as indicated by medium to high cell viability in calcein‐AM (live) and ethidium‐homodimer‐1 (dead) double‐stainings (LIVE/DEAD; Supporting Information). Spontaneous neural differentiation and the migration pattern of differentiating cells within the gel blends after 21 days were assessed using ICC staining. Migration of differentiating cells out of the sphere core into the surrounding gel matrix was observed in both gel blends (Figure 2B,C). Outgrowth within the gels occurred in thick bundles over the course of 3 weeks and was subsequently assessed by staining for the neuronal epitope β(III)‐tubulin and the cytoskeletal marker F‐actin. A more detailed characterization of the differentiation pattern revealed differentiation into neurons (ßIII‐tubulin⁺), dopaminergic neurons (TH⁺), and astrocytes (GFAP⁺) in both hydrogel blends, as shown in Figure 2B,C.

Figure 2.

hiNPC differentiation in 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM gels. A) Proliferating hiNPC spheres were chopped (0.1 mm) and embedded into the respective gel (4.9 × 10³ spheres mL⁻¹ gel). The gels were then cultivated in differentiation medium for 21 days and subsequently stained. Confocal images of differentiated spheres in B) 1.5% ALG/0.5% GG/0.01% LAM or C) 0.3% ALG/0.8% GG/0.01% LAM gels are shown. The samples were stained for nuclei (Hoechst, blue), ßIII tubulin (Alexa 546, green), F‐actin (phalloidin, Alexa 488, yellow or grey), dopaminergic neurons (TH, Alexa 546, magenta) and astrocytes (GFAP, Alexa 647, red). Images of the neural network formation are depicted in the upper row far right of (B) and (C). Differentiation into multiple cell types within the gels is shown in the lower two rows of (B) and (C). D) Proliferating 0.3 mm iNPC spheres generated from a disease cell line derived from a patient with Cockayne syndrome B (CSB), were embedded and differentiated as described above. The gels were subsequently stained for nuclei (Hoechst, blue), ßIII tubulin (Alexa 546), F‐actin (phalloidin, Alexa 488, yellow or grey), and astrocytes (GFAP, Alexa 647, red). Left to right: F‐actin, ßIII tubulin, F‐actin incl. network formation, ßIII tubulin, GFAP, ßIII‐GFAP merge. The red circles show the approximate size and location of the originally embedded spheres. Images represent sections through spheres in varying depths or maximum intensity projections of Z‐stacks. Abbreviations: GG, gellan gum; hiNPCs, human induced pluripotent stem cell‐derived neural progenitor cells.

Additionally, a disease cell line of a Cockayne Syndrome B (CSB) patient was cultivated and stained in the same manner as the IMR90 cells, in order to show the suitability of the 0.3% ALG/0.8% GG/0.01% LAM gel blend for disease modeling. We show outgrowth and differentiation of the disease cell line within the respective hydrogel blend.

In a second approach, 0.3 mm hiNPC spheres were allowed to pre‐differentiate as floating spheres for 1 week in differentiation medium, before subsequent embedding into 1.5% ALG/0.5% GG/0.01% LAM and 0.3% Alg/0.8% GG/0.01% LAM hydrogels, where they were allowed to differentiate for another 7 days (Figure  3A). ICC stainings reveal a dense F‐actin and ßIII‐tubulin network, as well as TH‐positive dopaminergic neurons and GFAP‐positive astrocytes within the spheres (Figure 3B,C). However, no significant outgrowth or migration of pre‐differentiated hiNPC spheres into the surrounding gel matrix was observed.

Figure 3.

Pre‐differentiated hiNPCs differentiated in 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM. A) Proliferating 0.3 mm hiNPC spheres were pre differentiated for 1 week and subsequently embedded into the respective gel. The gels were then cultivated in differentiation medium for 7 days and subsequently stained. Confocal images of differentiated spheres in B) 1.5% ALG/0.5% GG/0.01% LAM or C) 0.3% ALG/0.8% GG/0.01% LAM hydrogels. The samples were stained for nuclei (Hoechst, blue), ßIII tubulin (Alexa 546, green), F‐actin (phalloidin, Alexa 488, yellow or grey), dopaminergic neurons (TH, Alexa 546, in magenta) and astrocytes (GFAP, Alexa 647, red). Images of the neural network formation are depicted in the upper row of (B) and (C). The red circles show the approximate. size and location of the originally embedded spheres. Differentiation into multiple cell types within the gels is shown in the lower two rows of (B) and (C). D) As a comparison for the gel differentiation, proliferating 0.3 mm NPC spheres were differentiated as free‐floating spheres without gel‐embedding for 2 weeks and subsequently stained for nuclei (Hoechst, blue), ßIII tubulin (Alexa 546, green), F‐actin (phalloidin, Alexa 488, yellow or grey) and astrocytes (GFAP, Alexa 647, red). Left to right: ßIII tubulin, F‐actin, ßIII tubulin, GFAP, ßIII‐GFAP merge. Images represent sections through spheres in varying depths or maximum intensity projections of Z‐stacks. Abbreviations: GG, gellan gum; hiNPCs, human induced pluripotent stem cell‐derived neural progenitor cells.

As a comparison to the second approach, we differentiated 0.3 mm NPC spheres as free‐floating culture for 14 days and subsequently stained the spheres (Figure 3D). ICC images show an actin and tubulin network.

With the first approach, non‐pre‐differentiated NPCs developed into complex multicellular 3D microtissues within the course of 3 weeks. In contrast, pre‐differentiated spheres did not extend into the hydrogel, but instead formed multicellular inter‐spherical networks in 3D within 1 week. Here, two fit‐for‐purpose applications seem to emerge. Long‐term cultivations of complex cellular models could be valuable for applications that require a higher degree of physiology, such as disease modeling, while the quick generation of complex models in an easy to handle 3D model could be beneficial for, for example, substance screening.

2.3. Calcium Signaling

In order to test the functionality of the neural networks developed in 3D, cellular activity was monitored using calcium imaging. The hiNPCs were embedded in either hydrogel blend and the resulting crosslinked sphere‐laden hydrogels were differentiated for 21 days, before they were employed for calcium imaging (Figure  4A). The 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM gels exhibited an average frequency of ≈1.5 signals per 10 min (mean ± SEM: 1.6 ± 0.2, N = 3, n = 51, n _s = 21; mean ± SEM: 1.5 ± 0.1, N = 4, n = 89, n _s = 28; Figure 4E), the peak amplitude was 0.9 to 1.5% (mean ± SEM: 0.9 ± 0.1, N = 3, n = 51, n _a = 34; mean ± SEM: 1.5 ± 0.3, N = 4, n = 89, n _a = 41; Figure 4F) and the average length of the signals was ≈100 s (mean ± SEM: 102.3 ± 14.6, N = 3, n = 51, n _f = 32; mean ± SEM: 92.1 ± 10.3, N = 4, n = 89, n _f = 35; Figure 4G).

Figure 4.

Calcium signaling in pre‐differentiated and non‐pre‐differentiated hiNPC‐derived neural networks in 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM gels. A) The hiNPCs were chopped to 0.1 mm size and subsequently embedded into the respective hydrogel blend. The hiNPCs were differentiated within the gel for 21 days before slicing and calcium imaging. 0.3 mm pre‐differentiated spheres were not chopped, but sorted and pre‐differentiated for 1 week in the absence of growth factors. Thereafter, the pre‐differentiated spheres were embedded into the respective hydrogel blend and cultivated under differentiating conditions for 1 week. The sphere‐laden gels were then sliced into 250 µm thick slices before calcium imaging. B) Transmission and widefield images showing Fura‐2 fluorescence (after excitation at 380 nm) of a pre‐differentiated neural network in 1.5% ALG/0.5% GG hydrogel. Dashed lines illustrate regions of interest (ROIs) 1‐4, representing cell bodies, as analyzed in (C). C) Corresponding ROIs showing spontaneous calcium signals during a 10 min recording period. Black dotted lines indicate the baseline and grey dotted lines indicate 5 × SD of the baseline. Black lines represent the smoothened traces (Savitzky Golay Filter: 15). Black triangles mark calcium signals; red dots mark the peak amplitudes ΔR/R, red lines mark the Full Width at Half Maximum (FWHM) of two signals in one example trace. D) The pie charts display the percentages of active vs. inactive ROIs in all imaged samples of pre‐differentiated spheres differentiated in 1.5% ALG/0.5% GG/0.01% LAM (black, left), non‐differentiated hiNPCs differentiated in 1.5% ALG/0.5% GG/0.01% LAM (dark blue, middle) and non‐differentiated hiNPCs differentiated in 0.3% ALG/0.8% GG/0.01% LAM (light blue, right). The graphs show E) the average frequency (mean ± SEM) of calcium signals per 10 min, F) the peak amplitude ΔR/R (in %) and G) the FWHM (in s) of each individual calcium signal, respectively. Each color in (F) and (G) represents individual Ns. For illustration purposes, seven data points of pre‐differentiated hiNPCs in 1.5% ALG/0.5% GG, exhibiting amplitudes exceeding 20%, are not shown. Abbreviations: N, number of gels; n, number of cells. ALG, alginate; GG, gellan gum; LAM, laminin; ROI, region of interest; iPSCs, induced pluripotent stem cells; hiNPCs, human iPSC‐derived neural progenitor cells; FWHM, full width at half maximum; pre‐diff, pre‐differentiated.

The 1‐week pre‐differentiated spheres were also embedded in both hydrogel blends and further differentiated for 7 days. Pre‐differentiated spheres in 0.3% ALG/0.8% GG/0.01% LAM gels could not be sliced due to the softness of the gels, precluding calcium imaging. The pre‐differentiated spheres in 1.5% ALG/0.5% GG/0.01% LAM gels were sliced and analyzed as described above. Spontaneous calcium signals were measured, with 74% of the cells being active (Figure 4D). The average frequency was ≈3.2 signals per 10 min (mean ± SEM: 3.2 ± 0.3, N = 3, n = 76, n _s = 56; Figure 4E), the peak amplitude was ≈4.5% (mean ± SEM: 4.6 ± 0.4, N = 3, n = 76, na = 179; Figure 4F) and the average length of the signal was ≈40 s (mean ± SEM: 40.1 ± 3.0, N = 3, n = 76, n _f = 177; Figure 4G). In summary, these results show that in all conditions tested (3/4 models), cells generate spontaneous calcium signals, indicative of physiological activity.

2.4. Bioprinting

To evaluate whether the 3D bioprinting technology can be used to print the developed gel blends, we utilized an extrusion‐based bioprinter. We intended to directly deposit the hydrogels onto the electrodes of a 24‐well multielectroda array (MEA), which could later be used for the electrophysiological measurements of neural 3D networks. A general challenge of the MEA technology is the direct positioning of the cells onto the electrodes, so that electrical activity can be measured. This is especially challenging in 3D, where the cells are potentially far away from the electrodes. Our print pattern was therefore designed as 4 lines, surrounded by a stabilizing square, which lays directly on top of the electrodes. In a preliminary test trial using silicon, this grid was printed with the necessary precision onto 24‐well MEAs. However, the mwMEA plates are manufactured with small deviations in the electrode position from plate to plate, resulting in variable position outcomes of the printed structures in relation to the electrodes (Figure  5A). Therefore, printability of the hydrogels was further examined in standard tissue culture 24‐well plates. Suitable printing parameters were identified for 0.9% GG and both ALG/GG blends and the grid was successfully printed (Figure 5B, supplementary material). Both printed gel blend structures, but not pure GG structures were surface adherent. Pure ALG was not printable. As a proof of principle, we then embedded iPSC‐derived single cell NPCs into the 1.5% ALG/0.5% GG/0.01% LAM gel blend, bioprinted them in the same manner as the cell‐free gels and crosslinked the gels after printing. The cell‐laden gels were then cultivated under differentiation conditions for 3 days and subsequently stained for live and dead cells (Figure 5C). The stainings show ≈62% live cells within the gels. We conclude, that the developed hydrogel blends are suitable for bioprinting applications, even for small scale structures.

Figure 5.

Bioprinting strategy. A) Detailed view of a well from a 24‐well MEA plate (top). The silicon was printed in a square with four horizontal lines onto the electrodes with the currently highest possible accuracy (bottom). B) Representative microscopy images of 0.9% GG, 1.5% ALG/0.5% GG and 0.3% ALG/0.8% GG gels printed into a 24 well plate. Depicted are two representative images of the printing layer of the intended structure. All images were taken immediately after printing. C) Exemplary images of bioprinted single cells in 1.5% ALG/0.5% GG/0.01% LAM. Cells were differentiated for 3 days after printing and crosslinking, and subsequently stained with calcein‐AM for live cells (green) and ethidium homodimer‐1 for dead cells (red). The graph shows the living cells within the gels after 3 days (n = 4 gels). Abbreviations: MEA, microelectrode array; GG, gellan gum; ALG, alginate; LAM, laminin.

3. Discussion and Conclusion

The establishment of complex and robust neural in vitro models and reliable cultivation systems that reflect human physiology becomes increasingly important. Gaining a deeper understanding of cell and tissue function in health and disease will help us to understand pathomechanisms. Moreover, utilization of such in vitro models for assessing the neural responses to toxicants and drugs will improve hazard assessment. 3D models augment the complexity of conventional cell cultures, thus rendering them more predictive and physiologically relevant.^{[ 28} , ⁵⁴ , ⁵⁵ , ⁶⁰ , ^{61 ]} However, there is currently no gold standard for hydrogel‐based 3D neural cultures, as limitations such as the high variability and the general low throughput of 3D models still need to be overcome. Here we suggest two cytocompatible ALG/GG/LAM hydrogel blends for the generation of human iPSC‐based 3D neural models with spontaneous intracellular calcium signals.

Animal models and rodent primary cell cultures have given us great insights into the brain's development, basic function, disorders, and its reaction to substance exposure. However, inter‐species differences suggest a limited predictive power of such models for humans,^{[ 12} , ¹³ , ²² , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ^{66 ]} especially when it comes to disease models.^{[ 67} , ^{68 ]} This leads, amongst others, to high attrition rates for drugs.^{[ 2} , ³ , ^{4 ]} Therefore, we developed our 3D models with widely available and ethically justifiable hiNPCs, that reflect the correct species. These pluripotent cells can be differentiated into multicellular networks containing cells of the neuronal and astroglia lineage. It is, however, not yet possible to obtain fully mature networks in vitro, which needs to be considered, when utilizing hiPSC‐based models.

3.1. 3D Outgrowth and Migration of hiPSC‐Derived Neural Cells in Hydrogels

The brain is one of the softest tissues in the human body, with a Young's elastic modulus of 0.5–50 kPa.^{[ 31} , ^{32 ]} Thus, mimicking the brains ECM in vitro is highly challenging.^{[ 69 ]} Hydrogels feature advantageous properties for the 3D cultivation of neural models, such as the high water content (>90%), potential for functionalization, as well as chemical and physical tunability.^{[ 28 ]} Studies demonstrated that soft hydrogels (<1.5 kPa) support neurite outgrowth^{[ 39 ]} whereas stiffer hydrogels promote astrocyte differentiation.^{[ 70} , ^{71 ]} Koivisto et al. embedded hiPSC‐derived neurons into bioamine‐crosslinked GG gels with compressive moduli ranging from 3.9 to 23 kPa and compared them to the rabbit brain (7–10 kPa).^{[ 29 ]} They showed good cytocompatibility and migration of embedded hiPSC‐derived neurons, as well as maturation of neuronal cultures underneath a gel cover. Electrical activity of the neurons was not assessed in their study. Matyash et al. plated primary rat and human neurons on 1% ALG gels with elastic moduli ranging from 20.8 to 0.64 kPa depending on the crosslinker concentration.^{[ 72 ]} This group showed supported neurite outgrowth and increased resistance to oxidative stress of neurons cultured on soft ALGs. Moxon et al. generated even lower percentage GG gels with elastic moduli of ≈10 kPa and applied sonication at various amplitudes to alter the gels properties.^{[ 73 ]} However, very soft hydrogels are difficult to handle and tend to be incompatible with advanced biofabrication techniques, such as bioprinting or fluidic devices. Therefore, we approached the issue by generating hydrogel blends that are within an appropriate range of stiffness and exhibit quick stress relaxation times. 0.3% ALG/0.8% GG/0.01% LAM gels are less stiff than 1.5% ALG/0.5% GG 0.01% LAM gels and exhibit significantly quicker stress relaxation times, which was attributed to the reduced ALG content in those hydrogels. Outgrowth of non‐pre‐differentiated spheres in 0.3% ALG/0.8% GG/0.01% LAM gels seems to occur slightly quicker and in thinner bundles. A significant difference between the gels was observed during calcium imaging of pre‐differentiated spheres. Here, the higher stiffness of 1.5% ALG/0.5% GG/0.01% LAM gels allowed easy slicing, while 0.3% ALG/0.8% GG/0.01% LAM gels were more difficult to handle. The results suggest that, in ALG/GG blends, ALG and GG may sterically hinder each other in ionic crosslinking, suggesting that ALG or GG may act similarly to spacer molecules. Spacer molecules have already been introduced to PET and ALG‐based hydrogels and were shown to control and accelerate stress relaxation of these materials.^{[ 59 ]} Since natural ECM, specifically the brain ECM, are viscoelastic, with time‐dependent mechanical responses to stress, the property of quick stress relaxation times seems not only beneficial, but crucial (reviewed by Madl et al. and Axpe et al.).^{[ 69} , ^{74 ]} We suggest that the quick stress relaxation of our gel blends supports growth and migration of neural cultures, even in stiffer gels.

However, a deeper knowledge on the mechanics of the here identified optimal ALG/GG/LAM hydrogel composition (1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM) should be scope of future research, as others indicated that multimodal mechanical testing is required to gain full knowledge on the complex mechanics of, for example, brain tissue and brain‐mimicing materials.^{[ 31} , ^{46 ]} As a result, further mechanical testing to complement the compressive modulus and stress relaxation data in the present study should be performed in future research, like rheological assessments on storage and loss moduli (G′/G′′) of the gels, to gain knowledge on frequency dependent hydrogel viscoelasticity, as demonstrated earlier for GG hydrogels.^{[ 29 ]} This would allow to better understand potential mechanical cues of the here presented hydrogels for future neural tissue engineering applications.

We present stiffness and stress relaxation measurements derived from uniaxial compression testing. We also confirm slow degradation of the hydrogel blends over time, which additionally benefits cellular outgrowth.^{[ 75 ]} In future studies, we will assess calcium affinity to both ALG and GG, which may elucidate which component preferentially crosslinks through the available calcium ions, and if mutual weakening of the crosslinking is caused by using the here developed blends.

In general, there is no one‐model‐fits‐all approach, when it comes to neural cultures and their various applications. Substance screenings usually require a medium to high throughput approach and very high reproducibility, while disease modeling is content with lower‐throughput systems, but often relies on a higher degree of specificity and complexity. In vitro acute substance exposures commonly range from minutes to a few days, while chronic substance exposures are modeled over several days, however there is no common rule concerning the exposure length yet.^{[ 76 ]} As a result, we developed two different materials and cultivation systems, which can be applied for different purposes.

3.2. Two Protocols for the Differentiation of hiNPC into Neurons and Astrocytes

The first system consists of hiNPCs embedded in the 0.3% ALG/0.8% GG/0.01% LAM gel. This softer hydrogel model combines cellular outgrowth and migration with differentiation into astrocytes and neurons thereby generating complex microtissues with spontaneous calcium signals, as confirmed by ICC and calcium imaging after 3 weeks in vitro. We suggest this model to be valuable for applications requiring a high degree of complexity, but less throughput, such as disease modeling and long‐term exposure studies. We were able to show that also NPCs derived from a disease hiPSC line, that is, from a patient with Cockayne Cyndrome B (CSB), can be differentiated in 0.3% ALG/0.8% GG/0.01% LAM gels to form 3D neural networks. Spheres showed outgrowth out of the sphere and clear differentiation into neurons. The astrocyte staining did not offer conclusive results, which might be due to the disease. This, however, needs to be further evaluated in future experiments yet offers a valuable starting point of cellular analyses of the pathomechanisms of this neurodevelopmental disease. We hereby show, that the gel blend is also suitable for studying patient‐derived cells and that the here developed gel blend is a valuable tool for 3D disease modeling. In the future, more hiPSC lines from healthy and diseased donors need to be tested for their 3D migration and differentiation behavior using 0.3% ALG/0.8% GG/0.01% LAM gel blends.

The second system consists of pre‐differentiated hiNPC spheres embedded in the 1.5% ALG/0.5% GG/0.01% LAM gel. These samples were easy to handle for calcium imaging due to their slightly higher stiffness and quickly produced dense intra‐spherical networks, consisting of neurons and astrocytes. This model displays spontaneous calcium signals after only 1 week of differentiation within the hydrogel. However, it does not form microtissues due to the short cultivation time. In comparison to the spheres differentiated in suspension only, the neuronal network of 3D cultivated spheres appears more prominent and morphologically advanced. We believe that the 3D cultivated spheres are easier to handle, especially for staining and calcium imaging, since they do not need additional embedding. Combined with the fast and easy production of these 3D cultures, we suggest this model to be suitable for higher throughput applications, such as acute exposures to chemicals or substance screenings. The cultivation length for both models including the pre‐differentiation phase could be optimized in future experiments, depending on the required complexity and maturity of the networks. With the TH stainings we confirm the presence of dopaminergic neurons not only in secondary 3D,^{[ 13 ]} but also in the gel‐based 3D models. This information is useful for applications concerning dopaminergic neurons, for example, for studying attention deficit hyperactivity disorder of Parkinson's disease.^{[ 77} , ^{78 ]}

3.3. Calcium Signals in Differentiated 3D Models

Spontaneous calcium signaling plays an important role in the regulation of many cellular processes such as gene expression,^{[ 79} , ^{80 ]} neuronal outgrowth,^{[ 81} , ^{82 ]} neuronal differentiation, and other developmental progresses.^{[ 83 ]} Calcium imaging is widely used for investigating calcium signaling in tissue slices. However, in the past calcium imaging has also been used to monitor calcium transients in human differentiated neural aggregates or brain organoids,^{[ 84} , ⁸⁵ , ^{86 ]} as well as in primary human neurospheres differentiated in secondary 3D.^{[ 87 ]} We are, to the best of our knowledge, the first to establish calcium imaging in slices of in vitro 3D‐cultivated neural samples. We detected calcium signals in both differentiation systems. The pre‐differentiated hiNPC spheres in 1.5% ALG/0.5% GG/0.01% LAM gels displayed calcium signals with the highest average signal frequencies and peak amplitudes, plus shorter FWHM. Both hydrogels with non‐pre‐differentiated spheres showed lower frequencies and peak amplitudes plus longer FWHM. We therefore want to enhance the differentiation time in the non‐pre‐differentiated hydrogels to get a higher number of active cells and increased frequency rates. Within one sample we observed calcium signals with varying frequency, amplitude and length. This is in accordance with findings from Gualda et al., who imaged differentiated human midbrain‐derived NPCs.^{[ 85 ]} The observed heterogeneity in our samples may arise partly from the fact that the hiPSC‐derived networks contain several neuronal subtypes and also astrocytes.^{[ 13 ]} The calcium signals will be further analyzed during the future development of the 3D models. Here, application of saline containing high K+ concentration could, for example, be employed to differentiate neurons from astrocytes. Alternatively, sodium signals could be analyzed to probe for glutamate transporters, which are highly expressed by astrocytes. After further development of the method, for example by embedding organoids, one might envision the development of human correlate to rodent brain slices. Additionally, basic exposures during calcium imaging, such as glutamate or bicuculline (GABAAR antagonist) treatment need to be used to further characterize the developing networks. In addition, the proposed application scenario for both models has to be tested in the future.

3.4. Hydrogel Functionalization

ALG and GG are inexpensive materials and their processing for the generation of the ALG/GG gel blends is easy and fast. Although both gels are cytocompatible, no specific cell binding sites are present.^{[ 34} , ^{41 ]} Since cell‐matrix interactions are crucial for cell growth, proliferation, and differentiation of neural cultures, a lot of effort is going into functionalizing biologically inert hydrogels with native ECM molecules, such as LAM, collagen, or fibronectin. LAM is one of the major integrin interactor groups in the brain ECM and promotes cell survival, network formation, and functional development of hiPSC‐derived neural cultures in vitro.^{[ 1} , ^{45 ]} Improved progenitor cell and primary cortical neuron interaction with hydrogels were also achieved by integration of the RGD motif, which is part of the LAM structure.^{[ 52} , ^{88 ]} In our study, we implemented LAM to support the cell‐matrix interaction. This approach is sufficient for short‐term cultures and high‐throughput systems, such as in the 1.5% ALG/0.5% GG 0.01% LAM with embedded pre‐differentiated spheres. As the brain ECM is very complex, additional steps toward improved functionalization are of grave importance, especially when it comes to long‐term cultivations of complex neural models. Therefore, we aim at further enhancing specifically the 0.3% ALG/0.8% GG/0.01% LAM gel blend with additional native ECM molecules such as collagen and fibronectin in future studies. Both additives are widely studied for their effect on cell survival, growth, and adhesion, as well as their ability to provide growth factor binding sites in hydrogels.^{[ 89} , ⁹⁰ , ^{91 ]} However, these ECM molecules have not yet been fully evaluated for their potential in 3D neural modeling. Alternative fibrinogen‐,^{[ 92 ]} hyaluronic acid‐^{[ 71 ]} and chitosan‐based^{[ 91 ]} hydrogels have also been proposed, featuring their individual advantages and disadvantages.

3.5. 3D Bioprinting of Hydrogels

Advanced biofabrication and cultivation methods, such as 3D bioprinting, are up and rising in many in vitro fields like tissue engineering and substance testing, as extensively reviewed by Parrish et al.^{[ 93 ]} They are able to increase the reproducibility and complexity of 3D structures.^{[ 37} , ^{94 ]} As a functional readout besides calcium imaging, the electrophysiological analysis of in vitro neural network is often performed on MEAs.^{[ 30} , ^{95 ]} However, the measurement of 3D models on MEAs is highly challenging, due to the planar structure of the electrodes and difficulties in positioning the cell‐containing material onto the electrodes Here, 3D bioprinting of hiPSC‐derived neural cells onto MEA electrodes might offer a valuable solution, especially for higher throughput applications. Although in this work we did not succeed in producing a ready‐to‐use 3D bioprinted neural model, we contribute to its generation by showing the printabilily and cytocompatibility of the developed ALG/GG/LAM gel blends. Notably, 1% ALG by itself cannot be 3D bioprinted, due to its softness in the uncrosslinked state, and the GG by itself is not adherent to PDL/LAM‐coated surfaces. Hence, the established printability of the gel blends is a big step forward in future brain tissue engineering. In addition, our preliminary and unpublished work demonstrates that hiNPC spheres embedded and differentiated in ALG gels exert electrical activity measured with 24‐well MEA systems (Figure S5, Supporting Information). Hence, our data prepares the ground for future work in CNS tissue engineering using 3D bioprinting that needs to address two issues: 1) missing bioprinting precision arising from variable plate manufacturing needs to be overcome, for example by applying a camera‐based printing approach and 2) more work needs to be performed on systematically optimizing and characterizing the MEA readouts after 3D bioprining. In the future, these data will help to create physiologically‐relevant and reproducible 3D in vitro systems for disease modeling or compound screening.

3.6. Summary and Conclusion

In summary, in our study we present data on ALG/GG/LAM blends, which, amongst others, revealed the proposed favorable property of quick stress relaxation. Although many studies in the field of 3D modeling rely on primary or rodent cell cultures, iPSC plays an important role in the development of 3D models, not only due to species difference, but also due to known, and potentially unknown, varying requirements in culture conditions.^{[ 12} , ¹³ , ²² , ⁶² , ⁶³ , ⁶⁴ , ^{65 ]} Therefore we used iPSC‐based cell culture models, which gain more and more importance in the context of alternative testing methods and the aims of 3R (replacement, reduction, and refinement of animal testing).^{[ 96 ]} We characterized our differentiating cultures via ICC and verified calcium transients within the 3D models. The implementation of the CSB disease hiPSC line supports the applicability of our model. We also showed the suitability of our gel blends for biofabrication purposes, specifically for extrusion‐based 3D bioprinting. With our proof‐of principle study we open the opportunity for the further development of our models in many directions, such as disease modeling, substance screening, as well as basic research concerning the adaptation of standard methods such as calcium imaging, ICC, and MEA measurement.

4. Experimental Section

Hydrogel Preparation

For the preparation of the 3% (w/v) ALG stock solution, sterile dH₂O (100 mL) was pre‐warmed to 37 °C, ALG (3 g; #71238, Sigma) was added and the solution was steered overnight at 60 °C and 750 rpm to fully dilute the ALG. Finally, the ALG solution was autoclaved and stored at 4 °C until further use. 1% (w/v) ALG was prepared by diluting the 3% (w/v) ALG stock solution with sterile dH₂O or with sterile dH₂O containing a 0.1% (w/v) LAM stock solution in tris‐buffered saline (TBS; #L2020, Sigma), which resulted in a 1% ALG/0.01% LAM solution. For the preparation of the GG stock solution, 10% (w/v) sucrose (#84097, Sigma) was dissolved in dH₂O and sterile filtered (0.2 µm filter, Stericap Plus). A total of 5 mL sterile sucrose solution was heated to 80 °C (MKR23, Hettich Lab Technology). The GG (#G1910, Sigma) was then added to the sucrose solution and fully dissolved at 80 °C for 45 min at 600 rpm, to create 1% or 1.5% (w/v) stock solutions. The GG was subsequently sterile filtered through a pre‐heated filter. 0.9% (w/v) GG was prepared by diluting the 1% (w/v) stock solution with sterile dH₂O or 0.1% (w/v) LAM stock solution in TBS. The latter resulted in a 0.9% GG/0.01% LAM solution. GG was prepared in a 10% sucrose solution to reduce the osmotic pressure on the cells, as suggested by Koivisto et al. 2017.^{[ 29 ]} For both ALG/GG blends, the ALG stock solution was pre‐warmed to 37 °C before addition. The 1.5% ALG/0.5% GG blend was prepared by mixing 3% (w/v) ALG stock solution to 1.5% (w/v) GG stock solution (pre‐heated to 45 °C) and toping up the respective volume with sterile dH₂O to yield the final concentrations. By additionally supplementing the sterile dH₂O with 0.1% (w/v) LAM stock solution in TBS, a final blend of 1.5% ALG/0.5% GG/0.01% LAM was obtained. Similarly, 0.3% ALG/0.8% GG blends were prepared by mixing the 3% (w/v) ALG stock solution to the 1% (w/v) GG stock solution and toping up the volume with sterile dH₂O. The addition of 0.1% (w/v) LAM stock solution in TBS resulted in a 0.3% ALG/0.8% GG/0.01% LAM blend. Both blends were cooled down to 37 °C before the addition of LAM or sphere‐laden LAM. The hydrogels were crosslinked with CaCl2 (0.09 m; #1023780500, Merck) for 5 min.

FTIR

To assess the chemical composition of the hydrogel blends, 1.5% ALG/0.5% GG/0.01% LAM and 0.3% ALG/0.8% GG/0.01% LAM hydrogels were analyzed using FTIR. The 0.9% GG and 1% ALG gels served as controls. The gels were frozen (−80 °C) and freeze‐dried using a lyophilizer (LD‐12 plus, Martin Christ). Attenuated total reflectance (ATR‐FTIR) spectra were recorded with an infrared spectrometer (Nicolet 6700, Thermo Scientific, USA).

SEM

The hydrogels were washed using phosphate buffered saline (PBS +/+; 14040133, Gibco) and fixed using 3% (v/v) glutaraldehyde and 3% (v/v) paraformaldehyde solution (in 0.2 m sodiumcacodylate buffer, pH 7.4, all Sigma Aldrich) for 1 h at 22 °C (room temperature), respectively. Next, the samples were dehydrated by an ascending ethanol series (30, 50, 70, 75, 80, 85, 90, 95, and 99.8%) for 30 min each and subsequently critical point dried by liquid CO₂ exchange using a critical point dryer (EM CPD300, Leica). SEM images were recorded at 1 kV using secondary electron (SE) detection with an Auriga CrossBeam unit (Carl Zeiss Microscopy GmbH).

Mechanical Testing

To assess the mechanical properties of the hydrogels, cylindrical hydrogel specimens were fabricated using custom made cylindrical silicone molds (diameter = 10 mm, height = 5 mm). Unconfined compression tests were performed using an Instron 5967 mechanical testing setup (Instron GmbH) using a 100 N load‐cell and a compression deformation rate of 1 mm × min^–1. The samples were compressed until reaching 15% compressional strain. Initial bulk gel stiffness was derived as the slope in the linear elastic deformation regime of stress–strain diagrams derived from stress relaxation experiments and was evaluated as the slope of the linear elastic region between 5% and 10% strain. Next, the samples were held at 15% strain and the stress was recorded over time. Stress relaxation of the hydrogel cylinders was monitored, and the stress relaxation time t_1/2 was quantified as the time after which 50% of the initial stress at 15% strain was dissipated in the hydrogels (n > 3), as described earlier.^{[ 59 ]}

Degradation Study

The degradation of 0.9% GG, 1% ALG, 1.5% ALG/0.5% GG, and 0.3% ALG/0.8% GG gels was assessed by measuring the hydrogel mass over time. Therefore, six 50 µL gels per condition were separately deposited in 60 mm petri dishes (BD‐Falcon) and cross‐linked with CaCl₂ (0.09 m) for 5 min. Hydrogels were stored in PBS (+/+) at 37 °C and 5% CO₂ between measurements. The weight of the hydrogels was measured with an analytical scale (A200S, Olympus) every other day, by carefully removing the surrounding liquid completely, without drying out the gel itself.

Cell Culture and Maintenance

The cultivation and neural induction of hiPSCs was adapted from Hofrichter and Nimtz et al.^{[ 13} , ^{19 ]} Briefly, hiPSC‐IMR90 lines were obtained from WiCell and maintained under feeder‐free conditions on Matrigel‐coated 6‐well plates (LDEV‐free, #354277, Corning) in mTeSR1 medium (#05850, StemCell Technologies), at 37 °C in a humidified atmosphere of 5% CO₂. The medium was changed on 6 days per week by completely removing and replacing the medium with fresh mTeSR1 (2 mL) culture medium. On the sixth day of feeding, mTeSR1 (4 mL) was added, to substitute for the feeding‐free seventh day.

The neural induction of hiPSC cultures was performed by incubating the cells with ROCK inhibitor (1 µm; Y‐27632, #1254, Tocris Biosciences) in mTeSR1 medium for 1 h at 37 °C and 5% CO₂. Subsequently, the cells were washed with PBS including Penicillin/Streptomycin (PAN‐Biotech) and neural induction medium (NIM; 1 mL) was added. Colonies were then fragmented with a StemPro EZPassage Disposable Stem Cell Passaging Tool (Thermo Fisher Scientific) and transferred into Poly‐HEMA‐coated 6 cm dishes (#P3932, Merck) with NIM (5 mL). ROCK inhibitor (10 µm) was added for at least 24 h. The medium was changed every second day. On day 7, spheres were collected and transferred into new Poly‐HEMA‐coated 6 cm dishes with NIM and hFGF (10 ng mL⁻¹; #233‐FB, R&D Systems) and the spheres were cultured for another 14 days. On day 21 the generated hiNPCs were transferred into new Poly‐HEMA‐coated 10 cm dishes with NPC proliferation medium and hFGF (20 ng mL⁻¹). Cells were fed every second day with NPC proliferation medium and chopped to 0.2 mm when exceeding a size of ≈0.5 mm, or when clumping occurs (McIlwain Tissue Chopper, Ted Pella). All media compositions are listed in Supporting Information.

Preparation of Sphere‐Laden Hydrogels

Proliferating hiNPC spheres were pooled and chopped to a diameter of 0.1 mm. The spheres were then resuspended in CINDA differentiation medium and counted in a Nageotte chamber (Marienfeld). The desired number of spheres (4.9 × 10³ spheres mL⁻¹ gel) was centrifuged at 2500 rpm for 5 min and resuspended in 0.1% (w/v) LAM in TBS (final concentration 0.01% (w/v)). The cell‐LAM suspension was then added to the respective gel blend and distributed by careful pipetting. For the generation of pre‐differentiated spheres, 0.3 mm proliferating spheres were sorted into a small dish with CINDA medium.^{[ 13 ]} Subsequently, one sphere per well was added to a Poly‐HEMA coated 96‐well U‐bottom plate (Sarstedt). The spheres were pre‐differentiated in CINDA medium for 1 week at 37 °C in a humidified atmosphere at 5% CO₂ and subsequently embedded as described above.

ICC

Samples were fixed with 4% paraformaldehyde (#P6148, Sigma Aldrich) for 30 min at 37 °C, followed by three PBS (+/+) washing steps, 5 min each. Gels were then pre‐blocked for 2 h with 10% (v/v) goat serum (#G9023, Sigma Aldrich) in 0.1% (v/v) Triton X 100 (#T8787) in PBS (PBS‐T). The desired primary antibodies were diluted in 10% goat serum in PBS‐T. Subsequently, the antibody solution was added to the samples and incubated at 4 °C overnight. Samples were washed three times for 1 h with PBS (+/+). Then, the relevant secondary antibodies and Phalloidin‐Alexa488 (1:70, A12379, Life Technologies) were added to 1% Hoechst (33258, Sigma Aldrich) and 2% goat serum in PBS (+/+). The samples were again incubated at 4 °C overnight. Finally, the samples were washed three times for 1 h with PBS (+/+) and stored in PBS (+/+) until they were imaged with the confocal laser scanning microscope TCS SP8 (Inverse DMi8CS, Leica Microsystems). Maximum intensity projections of recorded Z‐stacks were constructed using Fiji Image J 1.52p.^{[ 97 ]} All antibodies are listed in Supporting Information.

Calcium Imaging

Non‐pre‐differentiated hiNPCs were embedded in 50 µL gels and cultured in 96‐well plates in differentiation medium for 21 days. 1‐week pre‐differentiated spheres were embedded in 100 µL gels and cultured in a 96‐well plates in differentiation medium for 1 week. Gels were subsequently transferred into small Petri dishes containing standard artificial cerebrospinal fluid (ACSF, containing in mM: 130 NaCl, 2.5 KCl, 2 CaCl₂, 1 MgCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, and 10 glucose, bubbled with 95% O₂ and 5% CO₂; pH: 7.4 & osmolarity: 305–310 mOsm/l). Subsequently, gels were placed in a bath filled with ACSF and cut into 250 µm thick slices using a vibratome (HM650 V, Thermo Fisher Scientific). For ratiometric calcium imaging, slices were transferred into standard ACSF which contained 15 µm Fura‐2 AM (Molecular Probes, Invitrogen) and loaded for 30 min at 36 °C in a humidified incubator at 5% CO₂/95% O₂, followed by a washing step in ACSF for 30 min. Fura‐2 loaded slices were then placed in a recording chamber, fixed with a grid, and continuously perfused with ACSF. Experiments were performed at room temperature (20–24 °C).

Calcium signals were recorded using a widefield epifluorescence imaging system based on an Eclipse FN‐PT upright microscope (Nikon), equipped with an Orca FLASH 4.0 camera (Hamamatsu Photonics) and a 40×/0.80 LUMPlanFI water immersion objective (Olympus). Imaging was controlled by the software NIS‐Elements AR 4.5 (Nikon). Fura‐2 was alternately excited at 357 nm (insensitive) and 380 nm (sensitive wavelength) at an imaging frequency of 10 Hz. After background correction, the ratio of the fluorescence emission (R) obtained from individual regions of interest representing cell bodies (ROIs) was calculated using the software NIS‐Elements AR 5.0 (Nikon). Measurements were analyzed using OriginPro 2020 (OriginLab Corporation) and Microsoft Excel 2016 (Microsoft Corporation). Background‐corrected traces of the Fura‐2 ratio of the individual ROIs were normalized to their initial baseline (as determined during the first signal‐free 30 s of measurement), baseline corrected and a smoothing filter (Savitzky‐Golay: 15 Points) was applied. Cells were considered “active,” if they exhibited calcium signals with peaks ≥ the fivefold standard deviation relative to their individual baseline fluorescence. For each ROI and calcium signal, the average frequency (number of signals during a 10 min recording), the peak amplitude ΔR/R (in %), and the Full Width at Half Maximum (FWHM, in s) were analyzed.

3D Bioprinting

A computer‐aided design model for the bioprinting of hydrogels onto the electrodes of a 24‐well multiwell microelectrode array (MEA) plate was designed in Autodesk Inventor Professional 2020. The printing was carried out with an EnvisioTEC 3D‐Bioplotter (Manufacturers Series), using a nozzle with 200 µm inner diameter (7018417, Nordson EFD). The printing parameters were optimized and ideal parameters were identified. The respective parameters for printing of 0.9% GG and the gel blend 1.5% ALG/0.5% GG and 0.3% ALG/0.8% GG are listed in Supporting Information.

Statistical Analyses

The material characterization data were statistically analyzed in GraphPad Prism 9.0, using one‐way ANOVA and Bonferroni post‐hoc tests. Significant differences among means between the different materials were indicated as *p < 0.05. Data derived from calcium imaging measurements were tested using a Mann–Whitney–U‐test. “N” represents the number of hydrogels, “n” the number of single cells. P values were represented as follows: * p < 0.05, ** p < 0.01, *** p < 0.001.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

The author and co‐author contributions are allocated in accordance with CRediT. E.F.: conceptualization, funding, supervision, writing—review & editing; A.R.B.: supervision, writing—review & editing; C.R.R.: supervision, writing—review & editing; C.M.S.: writing—review & editing; F.B.: investigation; I.L.: conceptualization, supervision, writing—review & editing; T.D.: investigation, visualization, resources, writing—review & editing; L.P.: investigation, methodology, resources, visualization, writing—review & editing; J.K.: conceptualization, investigation, methodology, project administration, resources, visualization, writing—original draft.

Supporting information

Supporting Information

Kapr J., Petersilie L., Distler T., Lauria I., Bendt F., Sauter C. M., Boccaccini A. R., Rose C. R., Fritsche E., Human Induced Pluripotent Stem Cell‐Derived Neural Progenitor Cells Produce Distinct Neural 3D In Vitro Models Depending on Alginate/Gellan Gum/Laminin Hydrogel Blend Properties. Adv. Healthcare Mater. 2021, 10, 2100131. 10.1002/adhm.202100131

Data Availability Statement

Research data are not shared.