Robust neuronal differentiation of human iPSC-derived neural progenitor cells cultured on densely-spaced spiky silicon nanowire arrays

Nanostructured cell culture substrates featuring nanowire (NW) arrays have been applied to a variety of basic cell lines and rodent neurons to investigate cellular behavior or to stimulate cell responses. However, patient-derived human neurons—a prerequisite for studying e.g. neurodegenerative diseases efficiently—are rarely employed due to sensitive cell culture protocols and usually long culturing periods. Here, we present human patient induced pluripotent stem cell-derived neurons cultured on densely-spaced spiky silicon NW arrays (600 NWs/ 100 µm\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$^2$$\end{document}2 with NW lengths of 1 µm) which show mature electrophysiological characteristics after only 20 days of culturing. Exemplary neuronal growth and network formation on the NW arrays are demonstrated using scanning electron microscopy and immunofluorescence microscopy. The cells and neurites rest in a fakir-like settling state on the NWs only in contact with the very NW tips shown by cross-sectional imaging of the cell/NW interface using focused ion beam milling and confocal laser scanning microscopy. Furthermore, the NW arrays promote the cell culture by slightly increasing the share of differentiated neurons determined by the quantification of immunofluorescence microscopy images. The electrophysiological functionality of the neurons is confirmed with patch-clamp recordings showing the excellent capability to fire action potentials. We believe that the short culturing time to obtain functional human neurons generated from patient-derived neural progenitor cells and the robustness of this differentiation protocol to produce these neurons on densely-spaced spiky nanowire arrays open up new pathways for stem cell characterization and neurodegenerative disease studies.


Results and discussion
A top view SEM image of the silicon NW arrays used to culture the human iPSC-derived neurons is shown in Fig. 1a. The about 1 µm long NWs feature a gradually tapered shape resulting in a spiky tip of about 30 nm. The nanostructures are randomly distributed on the substrate with a density of about 600 NWs/100 µm 2 which corresponds to a high NW density compared to other NW substrates used in the field 16 . Random arrangement and random angle orientation are caused by the self-assembling growth process 49,50 . Choosing silicon as the basic element for the substrates not only allows for accessing years of experience in nanofabrication in science and industry but also represents the promising nature of nanoscale silicon for subcellular interfaces 51 .
The protocol to generate the human iPSC-derived neurons used in this study employs proliferating small molecule neural progenitor cells (smNPCs) as an intermediate differentiation step and is based on work from Reinhardt et al. 52. . The smNPCs are maintained in regular well plates coated with Matrigel and are pre-differentiated for 6 days before plating onto the silicon NW arrays or glass coverslips for control (both also coated with Matrigel). First, we investigated the neuronal growth on the NW samples using SEM imaging (Fig. 1b). Overall, the observed neuronal network formation is similar to literature as neurons are growing both separately and in clusters, and neurites build a dense grid in between the neurons 18,22 . The tilted close-up views in Fig. 1c,d display the soma of a neuron and a neurite growing on the NWs. Notably, the cell material is barely in contact with the substrates and only touches the very tip of the NWs. Such a fakir-like settling state on a bed-of-nails was to be expected due to the high NW density 16 . Furthermore, strong interaction with the substrate in form of multiple extensions from the cell soma has been reported before 53 .
Cross-sectional images prepared by FIB milling and CLSM were used to support the previous statements regarding the cell's interaction with the substrate (Fig. 2a-e). Figure 2a displays an overview image of the neuron after FIB milling. The close-up views on the cell soma and a neurite in panel b and c clarify that the cell membrane is hovering approximately 1 µm above the substrate bottom which corresponds to the average NW length. Comparable behavior of cells cultured on high NW densities has been reported before by using FIB-SEM imaging 54 . A similar conclusion can be drawn from z-stacks prepared by CLSM imaging. Figure 2d displays a top view of a neuron cultured on a NW substrate. The cell membrane has been labeled using a membrane stain (false-colored in red). The dashed line indicates the position of the cross section in the z-x plane shown in Fig. 2e  www.nature.com/scientificreports/ imaging an uncoated NW sample (supporting Fig. S1). A corresponding image of a neuron on a control substrate is shown in the supporting material (Fig. S2). The cross-sectional reconstruction confirms the close contact of the cell membrane to the NWs and a fakir-like settling state which is in accordance with literature that studied cells on high-density NW arrays with CLSM 45 . Additionally, we tested for focal adhesions (FAs) to check for interactions with the substrate after long-term culture (11 days after plating). In this regard, we see no influence of the substrate compared to control which demonstrates a generally equivalent culturing environment for our human iPSC-derived neurons (Fig. S3, paxillin positive area (FAs) normalized to the area covered with cells (f-actin/cytoskeleton via phalloidin)) and is in agreement with literature for high NW densities 26 . Yet, cell culture substrates with Si nanoneedles have been reported to reduce the formation of FAs in short-term culture (24 h) of mesenchymal stem cells 55 . Nonetheless, the close contact between cell and NW, and the conical shape of silicon nanostructures are able to facilitate cell transfection 29 and delivery of biological payloads 30 , respectively, as shown in recent work using basic cell lines and mesenchymal stem cells. We believe that comparable applications could also be adapted to our neurons to introduce new pathways in stem cell research.
In the next step, we analyzed the cellular outgrowth using (immuno-) fluorescent labeling and widefield microscopy. Six days pre-differentiated smNPCs were transferred to the NW substrates and snapshots of the neuronal development were taken 1, 7, 11, and 15 days after seeding. Figures 3a,b display exemplary images of Hoechst 33342-stained cell nuclei 1 and 15 days post plating. The total cell number increased significantly over time since terminal neuronal differentiation is a gradual process and part of the cell population is still proliferating. We quantified the increase of the cell numbers and compared the results to planar control (Fig. 3c). The number of cells approximately quadruples within 11 days until it reaches a plateau. The plateauing cell numbers on NWs are slightly smaller in comparison to the control but only the final cell number on NWs-in mean about 2000 and 2500 cells per image, respectively-is significantly different. However, a reduced cell number is overall in agreement with literature reporting on reduced cell proliferation on high-density NW arrays 25 . A visual inspection of the cell spreading on NWs compared to control depicts a considerable increase in cluster formation on control substrates (supplements: Fig. S4). Quantification of the cell clusters using a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm validates the intuitive conclusion and shows that the share of single cells is significantly higher on NW substrates compared to control (e.g. about 35% and 13% at 1 DIV, respectively, Fig. 3d). One cause for cluster formation is cell movement and thus, hindered cell movement on the nanostructures is indicated by the increased share in single cells on NWs which is sustained at 7, 11, and 15 DIV. Impaired cell mobility has been described before in literature for cells cultured on NW arrays 19 . The absolute Statistical significances were calculated using a Mann-Whitney U test: * P < 0.05, **P < 0.01, ***P < 0.001, cf. www.nature.com/scientificreports/ shares of cells in clusters, however, reduce over time on both types of substrates due to continuing proliferation which generates additional cell clusters. The mean cell viability of about 60% on NW arrays is stable over time (cf. Fig. 3e) which is in good agreement with literature where high-density NW arrays have no detrimental effect on cell viability 17 . The slight increase in cell viability with further culturing time is reasonable since new vital cells are produced during proliferation. Compared to the control, the cell viability on NWs is lower since fewer new vital cells are produced as described earlier. In addition, we studied the neuronal differentiation of the smNPCs into neurons by IF microscopy. Figure 3f illustrates an exemplary recording of neurons 15 days post plating labeled with an anti-microtubule-associated protein 2 marker (MAP2, green) to identify neuronal phenotype and Hoechst as counterstain (blue). Similar micrographs from a control substrate and 1, 7, and 11 DIV post plating on NWs are available in the supporting material (Fig. S5). As priorly shown via SEM, an intricate neuronal network is developed on the NW arrays. Next, we determined the share of MAP2 positive vital cells 1, 7, 11, and 15 days post plating shown in Fig. 3g. Within this period, the ratio of MAP2 positive cells on NWs increases from about 20% to 60%, which is higher than on control substrates (approx. 50%). The higher percentage in neurons on NW samples could be explained by the slightly reduced proliferation of remaining smNPCs producing new undifferentiated cells. As a result, the smNPCs not only differentiate equally on NWs but also the NWs have a supportive effect to increase the share of terminally differentiated cells. Positive influence on differentiation has also been described in the literature where NW arrays enhanced differentiation e.g. for human embryonic stem cells toward definitive endoderm or neuron-like differentiation of mesenchymal stem cells 42,43 . Furthermore, we quantified the neuronal networks between MAP2 positive cells to evaluate the neurite health (supporting Finally, the electrophysiology and functionality of the neurons were tested using the patch clamping technique (Fig. 4a-e). The overall healthy state of the neurons cultured on NWs is validated by stable resting membrane potentials (RMPs) of -51.4 ± 6.1 mV shown in Fig. 4a. This value is similar to the control (no statistical significance, Mann-Whitney U test: P > 0.05, cf. Table S2) and normal or even above average compared to other human iPSC-derived neurons 45,52,56,57 . Moreover, the cell sizes including neuronal extensions on NWs and control are equal which is indicated by consistent membrane capacitances (MCs) of 11.2 ± 3.7 pF and 11.1 ± 4.7 pF on control and NW substrates, respectively (also Fig. 4a, no statistical significance, P > 0.05, cf. Table S2). In addition, the MCs are in accordance with literature 56,58,59 . The accurate interplay of sodium and potassium ion channels is approved by recordings of the membrane currents at different voltages (-70 to +30 mV, 10 mV steps) which show characteristic early-inward and late-outward currents (Fig. 4b). The peak currents normalized to the MC were plotted against the applied voltage (Fig. 4c) and are comparable to the control measurements and literature 4,52,59 . Neuronal identity is confirmed by recordings of action potentials (APs) evoked by current injection shown in Fig. 4d. AP amplitude (67.0 ± 8.6 mV), AP threshold (− 20.4 ± 4.0 mV), afterhyperpolarization (AHP, 15.1 ± 5.9 mV), and AP duration (3.3 ± 0.8 ms) match the control (Fig. 4e, no statistical significance, all P > 0.05, cf. Table S2) and are in agreement with other human iPSC-derived neurons in literature 56,57,59 .
In summary and conclusion, the human iPSC-derived neurons cultured on the densely-spaced spiky silicon NW substrates show excellent electrophysiological properties within only 20 days of differentiation despite the challenging substrate topology. The large-scale formation of neuronal networks on the NWs which is equal to control allows for novel approaches to study network properties under artificial constraints. Note, that the share of neurons is not only equal to control but even increased by 20% and thus would enhance the efficacy of future studies. The spiky shape of the nanowires renders the substrate suitable for the delivery of biomolecules or genes: such applications have already been shown using basic cell types like human umbilical vein endothelial cells (HUVECs) or human mesenchymal stem cells (hMSCs) 21,30,55 and establishing more advanced cell types in the community would help to increase the clinical impact of the field. Moreover, the broad availability of silicon and many years of experience in both science and industry make the material a great candidate for large-scale and high-throughput purposes in neurodegenerative disease studies or pharmacological drug screenings. Thus, we believe that our findings are an excellent starting point for further studies using biological metamaterials featured by high aspect ratio nanostructured surfaces 60 .

Methods
Nanowire growth and sample preparation. The NW growth was previously described in Misra et al. and Zhang et al. 49,50 . In short: The NWs were grown on top of ZnO:Al coated silicon wafers. In the plasmaassisted VLS growth process, the growth of NWs is mediated by Sn nanoparticles, and the density of Sn nanoparticles can be controlled either by varying the initial thickness of the metal or by changing the parameters of H 2 plasma (to reduce the oxide layer and to form tiny catalyst droplets). Therefore, prior to loading into the PECVD chamber, Sn layers with nominal thicknesses from 1 nm to 5 nm were thermally evaporated on the substrates. Then, the Sn layers were exposed to H 2 plasma for 5 min at 180 °C, to transform them into separate Sn droplets. The chamber pressure, the RF power density and the flow rate were fixed at 600 mTorr, 50 mW/cm 2 , and 20 sccm, respectively. Next, 6 sccm of silane (SiH 4 ) and 1.8 sccm of trimethylboron (1% TMB diluted in H 2 ) were introduced and the substrate temperature was raised to 400 °C to initiate the growth of NWs activated by the plasma, followed by a H 2 plasma etching to remove the remnant of Sn droplets. Typically, NWs grew around  . Differentiation was initiated by patterning medium (basic N2/B27 medium supplemented with 100 µM AA, 0.5 µM SAG, 1 ng/mL GDNF (PeproTech, Rocky Hill, NJ, USA), 1 ng/mL BDNF (PeproTech)) for 6 days and terminated in maturation medium (N2/B27 basic medium supplemented with 100 µM AA, 2 ng/ mL GDNF, 2 ng/mL BDNF, 1 ng/mL transforming growth factor-β 3 (TGF-β 3, PeproTech), 100 µM dbcAMP (Sigma-Aldrich). The cells were kept at 37 °C with 5% CO 2 in a humidified atmosphere. 1×10 6 smNPCs were  www.nature.com/scientificreports/ seeded to Matrigel-coated 6-well plate wells and kept in patterning medium for 6 days in vitro (DIV). Cells were split 1:5 using accutase (Sigma-Aldrich), seeded into 12-well plates containing the NW and control samples, and cultivated in maturation medium until the experiment. SEM imaging, confocal microscopy, and patch clamping were conducted 18-20 days after initiation of the differentiation. IF imaging was performed 1, 7, 11, and 15 days after plating on the NW and control samples. Data was collected from four independent differentiations from passages 22/24 and 24/26 from two thawings of vials containing the smNPCs. Two differentiations per experiment were used. Viability assay. Viability was estimated in the CellProfiler pipeline by discrimination of bright nuclei and medium bright nuclei after Hoechst 33342 stain for the representation of dead and living cells, respectively. The procedure results in similar values with no statically significant differences (Mann-Whitney U test) compared to a test using Calcein AM and Propodium Iodide shown in the supporting information (Fig. S7).
Scanning electron microscopy and focused ion beam milling. High-resolution images and cross sections were produced using scanning electron microscopy and focused ion beam milling with a Zeiss Crossbeam 550. Cells were prepared by fixation (4% Formaldehyde in DPBS, 10 min) and dehydration via a stepwise ethanol exchange. The samples were then critical point dried using Tousimis' Autosamdri-815 and a 20 nm gold layer was sputter-coated to avoid charging effects.
Electrophysiology. The patch-clamp setup consisted of an upright Nikon Eclipse FN1 microscope (objective: Nikon CFI TU Plan EPI ELWD 50× N.A. 0.60/W.D. 11.00 mm) and a HEKA EPC 10 USB patch-clamp amplifier with a red star headstage for trace recording. Patch-clamp pipettes (diameter: approx. 950 nm, resistance: 5-7 M ) were manufactured from borosilicate glass capillary blanks (GB150T-8P, Science Products) using a Sutter P-2000 pipette puller. Polishing was conducted with a CPM-2 microforge (ALA Scientific Instruments). Samples were rinsed three times and covered with bath solution (140 mM NaCl, 2.4 mM KCl, 1.3 mM MgCl 2 , 2.5 mM CaCl 2 ,10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 10 mM D-glucose, pH 7.4) and pipettes were filled with pipette solution (125 mM potassium gluconate (K-gluconate), 10 mM NaCl, 1 mM Triethylene glycol diamine tetraacetic acid (EGTA), 4 mM magnesium ATP (MgATP), 10 mM HEPES, 10 mM D-glucose, pH 7.4). Patch clamping was conducted at room temperature. The Patch Master V2x80 software was used for data processing. A Bessel low-pass filter was set to 2.9 kHz and the capacitance and seriesresistance were automatically compensated. Electrophysiological parameters were recorded and determined as follows. RMP: in current-clamp mode by keeping cells with zero current. MC: directly determined by the patchclamp software. For APs, the first AP of an AP train at rheobase current was analyzed-AP height: threshold to peak. AHP: threshold to minimum after AP. AP threshold: membrane potential where the second derivative www.nature.com/scientificreports/ becomes different from zero (determined with Origin). AP duration: full-width half-maximum of the AP. Number of cells analyzed: n(NWs) = 7, n(control) = 10.