In Vitro Reconstruction of Neuronal Networks Derived from Human iPS Cells Using Microfabricated Devices

Morphology and function of the nervous system is maintained via well-coordinated processes both in central and peripheral nervous tissues, which govern the homeostasis of organs/tissues. Impairments of the nervous system induce neuronal disorders such as peripheral neuropathy or cardiac arrhythmia. Although further investigation is warranted to reveal the molecular mechanisms of progression in such diseases, appropriate model systems mimicking the patient-specific communication between neurons and organs are not established yet. In this study, we reconstructed the neuronal network in vitro either between neurons of the human induced pluripotent stem (iPS) cell derived peripheral nervous system (PNS) and central nervous system (CNS), or between PNS neurons and cardiac cells in a morphologically and functionally compartmentalized manner. Networks were constructed in photolithographically microfabricated devices with two culture compartments connected by 20 microtunnels. We confirmed that PNS and CNS neurons connected via synapses and formed a network. Additionally, calcium-imaging experiments showed that the bundles originating from the PNS neurons were functionally active and responded reproducibly to external stimuli. Next, we confirmed that CNS neurons showed an increase in calcium activity during electrical stimulation of networked bundles from PNS neurons in order to demonstrate the formation of functional cell-cell interactions. We also confirmed the formation of synapses between PNS neurons and mature cardiac cells. These results indicate that compartmentalized culture devices are promising tools for reconstructing network-wide connections between PNS neurons and various organs, and might help to understand patient-specific molecular and functional mechanisms under normal and pathological conditions.


Introduction
The nervous system consists of the central and peripheral systems that are connected with each other, and thus form an electrical signaling network throughout the body. Although each neuron type is differentiated from different stem/progenitor cell pools, interactions between various cell types are well-coordinated both morphologically and functionally. The peripheral nervous system (PNS) is connected to the central nervous system (CNS), and this functional

Device fabrication
The co-culture device was fabricated from PDMS using soft lithography and replica molding technique. For producing the master mold, SU-8 3005 (Microchem) was spin-coated on a φ76 silicon wafer (Matsuzaki Seisakusyo., Ltd.) at 4000 rpm for 60 s to reach a height of 5 μm. The coated wafer was pre-baked at 95°C for 3 min. Then, the wafer was exposed to ultraviolet (UV) light with a UV crosslinker (CL-1000L; UVP) through a custom-made photomask. The photomask was designed to fabricate 20 microtunnels with a width of 50 μm and a length of 3 mm. After UV exposure, the wafer was developed with the SU-8 developer (Microchem), and then it was rinsed with 2-propanol (Wako Pure Chemical Industries). After its development, the wafer was placed in a conventional culture dish (φ100 mm; Corning).
Mixture of the PDMS-prepolymer and curing catalyst (10:1 weight ratio; Silpot 184, Dow Corning) was poured over the fabricated wafer to achieve a thickness of 5 mm. Then, PDMS was cured in an oven at 70°C for 1h. After curing, the PDMS sheet was trimmed using a surgical knife and was released from the master. To prepare the two culture compartments, which were connected by the microtunnel structures, holes were opened with a punch (φ8 mm; Harris Uni-Core; Ted Pella). We verified that each microtunnel was at least 1 mm in length; lengths 450 μm have been reported to allow only axons to pass through microtunnels [15]. The PDMS chamber was sealed with a cell culture dish (φ35 mm; Corning), which was previously coated with 20 μg/ml poly-L-ornithine (PLO; Aldrich). Then 5 μg/ml laminin (Sigma-Aldrich) was poured into the PDMS chamber and was incubated in the chamber until cell plating.

Cell culture
Human iPS cells (201B7 line) were provided by the RIKEN Bioresource Center in Japan. iPS cells were plated on culture plates coated with Laminin-511-E8 (iMatrix511, Nippi) and were maintained in mTeSR1 WO 2ME/MV medium (Stemcell Technologies) at 37°C in a 5% CO 2 incubator. The medium was changed every day. When the iPS cells were nearly confluent, iPS colonies were digested into single cells with accutase (Life Technologies), and then cells were passaged or induced as described below.
Differentiation of iPS cells into PNS neurons was performed according to the following protocol. Similar to above, the EBs of iPS cells were cultured in KSR medium containing 2 μM DM and 10 μM SB on day 0. On day 2, the medium was changed to KSR medium containing 3 μM CHIR99021 (CHIR; Cayman Chemical) and 20 μM SB. The KSR medium containing 3 μM CHIR and 20 μM SB was changed on every second day until day 12. On day 12, EBs were dissociated with TrypLE, were plated in PLO/laminin-coated dishes at a density of 1 × 10 5 cells/cm 2 , and were maintained in NDM2, which consisted of DMEM-F12, 1% N2 supplement, 1% NEAA, and 1% P/S. NDM2 was also supplemented with 10 μM forskolin, 50 μg/ml ascorbic acid, 10 ng/ml BDNF, 10 ng/ml GDNF, 10 ng/ml nerve growth factor (NGF; Wako Pure Chemical Industries), and 10 ng/ml neurotrophin 3 (NT-3; Wako Pure Chemical Industries). The medium was changed twice a week.
For co-culture experiments in the PDMS chamber, PNS neurons were first plated into one chamber and were incubated for 30 min. After the adhesion of PNS neurons was confirmed, CNS neurons were plated into the other chamber. Each neuron type was plated at a density of 2-5 × 10 5 cells/cm 2 . The medium in each compartment was changed on every second day.
Differentiation of iPS cells into cardiomyocytes was performed as previously described [16]. Briefly, the EBs of iPS cells were transferred into RPMI-1640 medium (Wako Pure Chemical Industries) containing 2% B27 supplement without insulin (Life Technologies) and 7.5 μM CHIR on day 0. 24 hours later, the medium was changed to fresh medium without CHIR. On day 3, the medium was changed to RPMI-1640 medium containing 2% B27 supplement without insulin and 10 μM IWR-1 (Sigma-Aldrich). On day 5, the medium was changed to fresh medium without IWR-1. From day 7, EBs were maintained in RPMI-1640 containing B27 supplement (Life Technologies), and the medium was changed on every second day. For co-cultures with PNS neurons, contracting EBs after two weeks in culture were used.

PKH labeling
In order to differentiate neurons in the co-culture chamber, each neuron type was labeled with a PKH permanent fluorescent marker [17] before plating into the co-culture chamber. PNS and CNS neurons were labeled with 1 × 10 −6 M PKH26 (red fluorescence; Sigma-Aldrich) and 1 × 10 −6 M PKH67 (green fluorescence; Sigma-Aldrich), respectively. An electro multiplying (EM) charge coupled device (CCD) camera (iXon+; Andor) mounted on an inverted microscope (IX-81; Olympus) was used for detecting the fluorescence signal of PKH dyes. PKHlabeled neurons were also used for the calculation of axon growth rates. Only axonal regions that passed through microtunnels were used for the growth rate calculation in order to exclude non-linear lengths resultant from the circular shape of the culture wells. Images were visualized and analyzed with the ImageJ software (National Institutes of Health; available at http:// imagej.nih.gov/ij/).

Calcium Imaging
Calcium dynamics were visualized with a calcium imaging technique. The cells were labeled with 5 μg/ml Fluo-4/AM (Invitrogen), a calcium indicator dye, for 30 min. After labeling, the medium was replaced with Ringer's solution (148 mM NaCl, 2.8 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES, and 10 mM glucose; pH 7.4). Then, the cells were placed on the stage of the inverted microscope, and fluorescence was detected with the EM CCD camera. In PNS-CNS co-culture experiments, electrical stimulation was used to evoke cellular responses. For stimulation, a pair of platinum electrodes (0.5 mm diameter; Nilaco) was immersed into the solution, and electrical pulses with a constant voltage were delivered to the cells using an electrical stimulator (SEN-3401; Nihon Kohden). Negative phase pulses with 3 ms duration and 10 V intensity were used. A frame rate of 2 frame/s was used for the PNS-CNS co-cultures. For cardiomyocytes, a frame rate of 14 frame/s was used. The recorded signals were analyzed with the ImageJ software.

In vitro differentiation of PNS and CNS neurons from human iPS cells
Each neuron type was differentiated from same human iPS cells. We followed previously established protocols to differentiate human iPS cells (201B7 cell line) into human PNS and CNS neurons in vitro (Fig 2A). Briefly, we induced PNS neurons using neural crest cell induction with CHIR, which inhibits glycogen synthase kinase 3 beta (GSK-3β) signaling [18,19], via the formation of EBs (Fig 2B). The induced cells were re-plated in PLO/laminin-coated dishes, and we continued the differentiation in vitro for at least 2 weeks (Fig 2C). Most differentiated cells expressed the type III intermediate filament Peripherin, which is a marker of peripheral neurons, and TUJ1 in their neurites (Fig 2E). We also confirmed that Peripherin-positive neurons expressed the neural crest markers P75NTR and HNK-1 (S1 Fig). These data indicate that differentiated PNS neurons have characteristics of neural crest-lineage cells. On the other hand, we used a dual SMAD inhibition protocol under non-adherent conditions for the induction of CNS neurons. After 2 weeks of induction, the EBs were re-plated in PLO/laminin-coated dishes, and the cells started to extend neurites almost immediately (Fig 2D). Immunochemical staining revealed that the neurites of most cells expressed TUJ1 at a high level on day 15 ( Fig  2F). However, Peripherin and P75NTR were not expressed, suggesting that these TUJ1-positive neurons were CNS neurons [20]. Thus, we prepared human PNS and CNS neurons in a large quantity in vitro. Co-culture of PNS and CNS neurons in the fabricated culture device PNS neurons re-extend their neurites into peripheral tissues, including internal organs and limbs, in response to neuronal injury. The neurite re-innervates target organs and the CNS to re-establish bidirectional signals. However, the process of re-connection between the PNS and CNS is not fully understood yet. Therefore, we developed an in vitro cell culture device to image and understand the underlying mechanisms of neuronal wiring. Using PDMS, we fabricated a culture chamber with 2 separate wells and tunnels as outlined in Fig 3. Schematic image of the fabricated culture chambers is also shown in Fig 4A. Importantly, no cross-contamination must occur between the wells, and the tunnels need to allow only the growth of extending neurites to the other side. Therefore, connecting tunnels were designed with a height of 5 μm to exclude neuronal cell bodies from the microtunnels [12,15]. After PNS and CNS neurons were fully differentiated, each type of neuron was collected and re-plated in separate wells of the culture chamber. In order to monitor the cell bodies and neurites of each neuron type, cells were labeled with different dyes before re-plating. During co-culture experiments, neurites entered the microtunnels, but the cell bodies of neurons were excluded (Fig 4B). After 2 weeks of co-culture, neurites reached the opposing well and made neuronal connections. Interestingly, mostly the neurites of PNS neurons crossed the microtunnels (Fig 4C). In contrast, only a few CNS neurites passed through the microtunnels. To confirm these observations, we performed TUJ1 and Peripherin immunostaing, which showed that neurites were double-positive for TUJ1 and Peripherin (Fig 4D). PNS neurites that passed through the microtunnels extended at a rate of approximately 0.2 mm/day during the 2 weeks of co-culture. This rate of growth was lower than that in vivo [21,22]. Furthermore, the morphology of PNS neurites suggested that neurites gathered together and made bundles before and during entering the tunnels (Fig 4E), and these bundles passed through the microtunnels together.
Additionally, we examined whether the connected neurons constructed a neuronal network. To reveal the structure of connection more closely, we evaluated Synapsin-1 expression at the locations of neurite arrival in the CNS well. Synapsin-1 is localized in synaptic vesicles, thus its presence is suggesting axon/neurite interactions. In the well containing CNS neurons, TUJ1positive PNS bundles integrated with aggregated CNS neurons, as Synapsin-1 was detected in the terminal of the bundles, not in the microtunnels (Fig 5A and 5B). These data indicate that iPS cell-derived PNS and CNS neurons are able to connect with each other and make a neuronal network in our culture device. In addition, a functional assay was carried out using calcium imaging. Preliminarily studies demonstrated that differentiated CNS neurons exhibit spontaneous activity with kinetics typical of neuronal calcium transients (a rapid onset and decay occurring within 10 s, Fig 5H). Alternatively, PNS neurons rarely exhibited spontaneous activity but did respond to the TRPM8 agonist menthol and the TRPV1 agonist capsaicin (data not shown, [23]), which classically activate peripheral afferent neurons. To monitor the responsiveness of PNS bundles to external stimuli, electrical stimulation was applied to the cells with an electric stimulator. Fig 5C and 5D show phase-contrast and fluorescent images around the microtunnel region on day 39. Fluorescence images show the amount of intracellular calcium before and 1 s after applying the electrical pulse. Fig 5E shows   differences between signals before and after applying the electrical stimulation (each shown individually in Fig 5D). Electrical pulses elevated the intracellular calcium concentrations of PNS bundles. Kinetics of calcium transient onset before and after the electrical stimulation in 6 PNS bundles from different samples are summarized in Fig 5F. These data indicate that the PNS bundles were functional and could be activated artificially by external stimuli. We then analyzed calcium activity in CNS neurons during electrical stimulation of putatively networked bundles from PNS neurons. An electrical pulse was applied 7 times to bundles from PNS neurons with an interval of 30 s (approximate electrical stimulation period was 200 s in total). Calcium activity was first measured in the bundles and then fluorescence intensity changes were measured in the central regions of CNS wells (Fig 5G). Traces of calcium activity in positively identified CNS neurons at 600 s are shown in Fig 5H. As expected, calcium activity in positively identified CNS neurons was only observed following the axonal bundle stimulation. In conjunction with the statistical data presented in Fig 5I, our data confirm the presence of functional cell-cell interaction between differentiated PNS and CNS neurons. Thus, our culture device could serve as a promising tool to culture different types of neurons and to analyze their interactions.

Co-culture of PNS neurons and cardiomyocytes in the fabricated culture device
Next, we constructed another type of co-culture system using PNS neurons and cardiomyocytes derived from iPS cells. The heart beats autonomously throughout life, but it is also regulated by neuronal stimuli [24,25]. Cardiomyocytes are innervated by PNS neurons, which regulate the beating rate of the heart. Thus, preparing an in vitro co-culture system of PNS neurons and cardiomyocytes, both differentiated from same human iPS cells, would be a promising model system for investigating the mechanisms that regulate heart beating, and for drug discovery in cardiovascular diseases associated with peripheral neuropathy. Therefore, we attempted to establish a coculture using PNS neurons and cardiomyocytes derived from human iPS cells. Following the cell induction method shown in Fig 6A, the EBs of human iPS cells were differentiated into cardiomyocytes. After 2 weeks in vitro, the cell aggregates started to contract autonomously (S1 Movie). We also checked the cardiomyocyte beats with calcium imaging, which revealed that almost all cells exhibited calcium oscillations (Fig 6B and S2 Movie).
To establish the compartmentalized co-culture system quickly, PNS neurons were first replated in the left side of the chamber and were cultured for two weeks. After we confirmed that the PNS bundles passed through the microchannels, differentiated contracting cardiomyocyte aggregates were isolated and transferred in the right side of the chamber. On the right side of the chamber, the bundles innervated the contracting aggregates (Fig 6C and S3 Movie). To confirm the formation of an apparent connection between PNS neurons and cardiomyocytes, we probed co-culture samples for expression of cTnT and syanpsin-1 in co-culture samples. cTnT expression indicated the presence of mature cardiomyocytes in the well, and PNSderived Synapsin-1 was co-localized on these cells (Fig 6D). These data indicate the formation of a neuromascular connection between mature cardiomyocytes and PNS neurons in the PDMS devices. These results suggest that our PDMS chamber device is also useful for establishing a connection between PNS neurons and cardiomyocytes and for examining how PNS neurons control cardiomyocyte functions in response to various neuronal stimuli.

Discussion
Here, we reported the manufacturing of a co-culturing device, which was used for constructing neuronal networks between PNS and CNS neurons derived from iPS cells. The compartmentalized device made from PDMS could be a useful experimental platform for monitoring neurite/cell interactions in various cell types, for screening chemicals, or to investigate the effectiveness or safety of drugs. Neurites passing through the microtunnels were extended primarily from the PNS neurons towards the chamber containing the CNS neurons. These observations are consistent with those obtained in vivo. It is well known that PNS neurons can extend their axons faster and more far than CNS neurons do. However, the growth rate of PNS axons was approximately 0.2 mm/day in our experiments, which was lower than that in humans (1 mm/day) based on clinical studies [21,22]. A possible explanation for this difference might be that growing conditions are different in vivo and in vitro. Furthermore, PNS bundles in the CNS chamber were exposed to CNS culture medium, which did not contain NGF or NT-3. NGF and NT-3 have been reported to support peripheral nerve regeneration [26,27]. Thus, the absence of these growth factors might affect the growth of PNS bundles. Lack of supportive cells in the PNS chamber might also explain this difference in neurite growing. Schwann cells, which support and myelinate PNS neurons, emerge from human PSCs within one month [28,29]. However, mature astrocytes and oligodendrocytes in CNS require an induction for 2-3 months [30][31][32] compared with neurons that require one month from human PSCs. Thus, the cellular environment providing support for neurons might not be present in the CNS chamber. Optimizing coculture conditions to enhance neurite growth should be considered in further investigations. Furthermore, our device might be useful for identifying chemical compounds and growth factors that modulate neurite growth, especially if we can reconstruct the outgrowth of defected neurites, such as those derived of patients with neuropathy.
Our microtunnel structures only allow the isolation axons, but not dendrites, from their cell bodies [12,15]. Hence, PNS bundles originating from PNS neurons, which enter the CNS chamber, are thought to be axons. Accordingly, we did not find synapses within the bundles themselves ( Fig 5A). Furthermore, PNS bundles exhibited reproducible evoked responses to electrical stimulations. These functions of PNS cells in vitro are consistent with those described in a previous study [14]. As these dynamics are easy to regulate, the system might be suitable for investigating neuronal signal-based inter-organ communication in vitro.
Currently, dopaminergic neurons [33], motor neurons [34], and/or other specific types of cells from various organs [35] can be differentiated from human iPS or ES cells. In addition, direct cell conversion techniques are useful for obtaining a purified population of cells [36,37]. Our in vitro co-culture model system is suitable for examining morphological and molecular mechanisms underlying tissue development, homeostasis, and diseases using specific types of neurons.

Conclusions
We co-cultured PNS and CNS neurons derived from human iPS cells in a microfabricated device. We validate methods and protocols for future research on the integration of PNS and CNS neurons to form network structures. In addition, we showed that the co-culture device can be applied for the integration of PNS neurons and other types of cells, such as cardiomyocytes. These data suggest that the compartmentalized PDMS co-culture device is a promising tool for investigating the network-wide integration of PNS neurons, and might help to understand the details of neuronal network mechanisms and the underlying functional mechanisms in normal and pathological conditions.