Abstract
Advances in primary and stem cell derived neuronal cell culture techniques and abundance of available neuronal cell types have enabled <i="">in vitro</i> neuroscience as a substantial approach to model <i="">in vivo</i> neuronal networks. Survival of the cultured neurons is inevitably dependent on the cell culture incubators to provide stable temperature and humidity and to supply required CO<sub="">2</sub> levels for controlling the pH of culture medium. Therefore, imaging and electrophysiology recordings outside of the incubator are often limited to the short-term experimental sessions. This restricts our understanding of physiological events to the short snapshots of recorded data while the major part of temporal data is neglected. Multiple custom-made and commercially available platforms like integrated on-stage incubators have been designed to enable long-term microscopy. Nevertheless, long-term high-spatiotemporal electrophysiology recordings from developing neuronal networks needs to be addressed. In the present work an incubator-independent polydimethylsiloxane-based double-wall perfusion chamber was designed and integrated with multi-electrode arrays (MEAs) electrophysiology and compartmentalized microfluidic device to continuously record from engineered neuronal networks at sub-cellular resolution. Cell culture media underwent iterations of conditioning to the ambient CO<sub="">2</sub> and adjusting its pH to physiological ranges to retain a stable pH for weeks outside of the incubator. Double-wall perfusion chamber and an integrated air bubble trapper reduced media evaporation and osmolality drifts of the conditioned media for two weeks. Aligned microchannel-microfluidic device on MEA electrodes allowed neurite growth on top of the planar electrodes and amplified their extracellular activity. This enabled continuous sub-cellular resolution imaging and electrophysiology recordings from developing networks and their growing neurites. The on-chip versatile and self-contained system provides long-term, continuous and high spatiotemporal access to the network data and offers a robust <i="">in vitro</i> platform with many potentials to be applied on advanced cell culture systems including organ-on-chip and organoid models.
<small="">Export citation and abstract</small> <span class="btn-multi-block"=""> <a href="/wg_8714b7f1589aa0f6c92979708057c4a57/en/es/iopscience.iop.org/export?type=article&doi=10.1088/1758-5090/acb466&exportFormat=iopexport_bib&exportType=abs&navsubmit=Export+abstract" class="btn btn-primary wd-btn-cit-abs-bib" aria-label="BibTeX of citation and abstract"="">BibTeX</a> <a href="/wg_8714b7f1589aa0f6c92979708057c4a57/en/es/iopscience.iop.org/export?type=article&doi=10.1088/1758-5090/acb466&exportFormat=iopexport_ris&exportType=abs&navsubmit=Export+abstract" class="btn btn-primary wd-btn-cit-abs-ris" aria-label="RIS of citation and abstract"="">RIS</a> </span>
1. Introduction
Highly-ordered function of brain circuits is derived from precise organization of neuronal circuits and strictly controlled biophysical properties of neuronal cell microenvironment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib1" id="fnref-bfacb466bib1"="">1</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib3" id="fnref-bfacb466bib3"="">3</a>]. Long-range axonal growth and pathfinding, organized network structure, formation and plasticity of synaptic connections, and changes in the network activity patterns all develop in strictly predictable developmental timelines [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib4" id="fnref-bfacb466bib4"="">4</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib6" id="fnref-bfacb466bib6"="">6</a>]. Underlying cellular and molecular mechanisms responsible for the formation and proper wiring of neuronal circuits during development are not yet well known in modern neuroscience [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib7" id="fnref-bfacb466bib7"="">7</a>]. A massive number of <em="">in vivo</em> studies have been conducted to explore development of neuronal networks structure and function [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib8" id="fnref-bfacb466bib8"="">8</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib12" id="fnref-bfacb466bib12"="">12</a>]. An alternative approach, is to generate simplified <em="">in vitro</em> neuronal culture systems that allow improved control over experimental conditions [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib10" id="fnref-bfacb466bib10"="">10</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib13" id="fnref-bfacb466bib13"="">13</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>].
Despite <em="">in vivo</em> brain tissue, the embedded systems to support nutrients and regulate metabolites, pH and temperature are missing in the <em="">in vitro</em> neuronal networks [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib16" id="fnref-bfacb466bib16"="">16</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib19" id="fnref-bfacb466bib19"="">19</a>]. In conventional cell culture models like 2D cultures on flat substrate and in advanced 3D models like organoids and spheroids, nutrients are supported by frequent media exchange [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib20" id="fnref-bfacb466bib20"="">20</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib21" id="fnref-bfacb466bib21"="">21</a>]. Survival of sophisticated 3D culture models and organoids require an adequate attention to the culture media perfusion parameters and precise control over cellular microenvironment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib22" id="fnref-bfacb466bib22"="">22</a>]. Media exchange, on the other hand, exposes neuronal cells to sudden mechanical and chemical perturbations in their microenvironment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib23" id="fnref-bfacb466bib23"="">23</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib26" id="fnref-bfacb466bib26"="">26</a>]. In addition to the media exchange sessions, native <em="">in vivo</em> regulatory systems need to be replaced by cell culture incubators to provide stable temperature and to control media pH based on CO<sub="">2</sub>-dependent buffering system [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib27" id="fnref-bfacb466bib27"="">27</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. However, constant need to keep <em="">in vitro</em> neuronal networks inside the incubator is often a limiting factor for screening platforms [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib28" id="fnref-bfacb466bib28"="">28</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib33" id="fnref-bfacb466bib33"="">33</a>]. For instance, microscopy imaging and electrophysiology recordings often need to be performed outside of the incubator in room temperature with no humidity and CO<sub="">2</sub> support [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib34" id="fnref-bfacb466bib34"="">34</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib35" id="fnref-bfacb466bib35"="">35</a>]. The influence of ambient conditions on cellular microenvironment can affect experimental results in short-term and network survival in long run [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib36" id="fnref-bfacb466bib36"="">36</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib38" id="fnref-bfacb466bib38"="">38</a>].
Long-term analysis of 2D and 3D constructed <em="">in vitro models</em> is crucial for understanding the developmental features of the growing networks and modeling neurological disorders [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib39" id="fnref-bfacb466bib39"="">39</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib41" id="fnref-bfacb466bib41"="">41</a>]. Different types of on-stage incubators are commercially available that are equipped with self-contained imaging systems to record morphological changes over time or in large sample sizes [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib42" id="fnref-bfacb466bib42"="">42</a>]. Even though, miniaturized on-stage incubators provide an excellent platform for long-term microscopy, these systems still require bulky accessories to provide gas and humidity that makes it difficult to be integrated with screening tools like electrophysiology recording setups [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib43" id="fnref-bfacb466bib43"="">43</a>]. Incubator-independent screening platforms aim to provide long-term continuous access to the network microscopy and electrophysiology data, and to exclude disturbances related to the media exchange and handling [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib44" id="fnref-bfacb466bib44"="">44</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib46" id="fnref-bfacb466bib46"="">46</a>].
To enable continuous multisite recordings from developing neuronal networks some models of perfusion chambers have been developed that fits to the standard multi-electrode arrays (MEAs) electrophysiology setups [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib29" id="fnref-bfacb466bib29"="">29</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib32" id="fnref-bfacb466bib32"="">32</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib48" id="fnref-bfacb466bib48"="">48</a>]. These perfusion chambers have been designed to continuously deliver fresh media containing required nutrients with lower flow rates, and to remove generated metabolites from cellular microenvironment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib49" id="fnref-bfacb466bib49"="">49</a>]. Different choices of materials are available for the fabrication of cell culture perfusion chambers which mainly include glass, silicon, metals and polymers [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib50" id="fnref-bfacb466bib50"="">50</a>]. Among these materials polymers and specifically polydimethylsiloxane (PDMS) are inexpensive, transparent, gas-permeable, elastic, easy to prototype, and biocompatible [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib50" id="fnref-bfacb466bib50"="">50</a>]. To adjust the pH of the culture media in ambient conditions, a gas supply containing 5% CO<sub="">2</sub> is connected to the incubator-independent perfusion chambers that provides bicarbonate buffering system [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib51" id="fnref-bfacb466bib51"="">51</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib53" id="fnref-bfacb466bib53"="">53</a>]. Other groups have used CO<sub="">2</sub>-independent media, that mainly contains 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) pH-buffering system, to overcome the need for continuous gas supply [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib44" id="fnref-bfacb466bib44"="">44</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib54" id="fnref-bfacb466bib54"="">54</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib55" id="fnref-bfacb466bib55"="">55</a>]. For instance, in a perfusion platform designed by Saalfrank <em="">et al</em> CO<sub="">2</sub>-independent cell culture media in a transparent PDMS-based chamber and multi-electrode array (MEA) device enabled continuous extracellular recordings from developing rat cortical and hippocampal circuits over months [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. However, media leakage and formation of air bubbles in some occasions interrupted microscopy images, which later resolved by including an air bubble guiding system in design of the perfusion chamber [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib32" id="fnref-bfacb466bib32"="">32</a>]. In these platforms stable temperature was provided by a built-in temperature controller that has been embedded in MEA amplifier [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib24" id="fnref-bfacb466bib24"="">24</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib58" id="fnref-bfacb466bib58"="">58</a>]. In the present work we designed an upgraded-version of the leakage-resistant perfusion chamber and integrated it with compartmentalized microfluidic device to enable long-term microscopy and electrophysiology from engineered low-density neuronal circuits at sub-cellular resolution.
Based on physical confinements and pattern of deposited chemical cues in the microfluidic compartmentalized devices one can guide axonal growth direction and control the structure of neural circuit [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib59" id="fnref-bfacb466bib59"="">59</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>]. These devices are mainly based on transparent PDMS or glass material that is compatible with microscopy and electrophysiology tools [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>]. Microfluidics have been extensively utilized to construct two- and three-dimensional (2D and 3D) <em="">in vitro</em> neuronal networks [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib59" id="fnref-bfacb466bib59"="">59</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib65" id="fnref-bfacb466bib65"="">65</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib70" id="fnref-bfacb466bib70"="">70</a>]. In these compartmentalized microfluidic devices, physical dimensions of microchannels only allow axons and dendrites to grow inside [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib71" id="fnref-bfacb466bib71"="">71</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib72" id="fnref-bfacb466bib72"="">72</a>]. Thus, it enables to study axonal growth and guidance, its response to the chemical cues and injury [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib19" id="fnref-bfacb466bib19"="">19</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib73" id="fnref-bfacb466bib73"="">73</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib75" id="fnref-bfacb466bib75"="">75</a>] in a physically and chemically isolated environment. Additionally, compartmentalized microfluidics are coupled with MEA devices [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib61" id="fnref-bfacb466bib61"="">61</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib76" id="fnref-bfacb466bib76"="">76</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib79" id="fnref-bfacb466bib79"="">79</a>] to amplify and capture axonal extracellular activity along with its growth morphology [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib78" id="fnref-bfacb466bib78"="">78</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib79" id="fnref-bfacb466bib79"="">79</a>]. Previously, an integrated microfluidic-MEA platform has been used to study axonal propagation velocity and response to focal injury during the cortical network development [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>]. However, recordings and microscopy images were obtained in a limited number of days and only performed as snapshots of activity or network morphology.
In the present work we integrated microfluidic-MEA platform with incubator-independent perfusion system that enables continuous recording and microscopy from growing networks and their extending neurites for more than two weeks. Continuous imaging and recording data provided by such platform is of great value for modern neuroscience community to extract the dynamics of developing neuronal circuits in long-term and at sub-cellular resolution. The system can be adapted to the 2D and 3D cell culture models of human stem cell-derived neurons, to the brain-on-chip and brain organoid models, and to the cultures of non-neuronal cells.
2. Methods
2.1. Fabrication of PDMS microfluidic device
Photolithography and soft lithography steps to fabricate the master mold and PDMS device have been discussed previously [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>]. Briefly, a two-layer SU-8 template including the shallow structures for fabrication of the microchannels and larger structures for the fabrication of reservoirs, was fabricated on a silicon wafer (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(A)). First and second layers (5 <em="">µ</em>m and 100 <em="">µ</em>m, respectively) were fabricated by subsequent spin-coating (Ws-650Sz Spin Coater, Laurell Technologies) and UV-exposure of the SU-8 5 and SU-8 50 through separate chromium masks (Photronics Ltd). Alignment of second mask layer with features of first layer was performed in a mask aligner (MJB4, SUSS MicroTec). Parameters for the baking of coated SU-8 layers (pre- and post-baking) and its development were adjusted as suggested by vendor (MicroChem). The physical dimensions of developed templates were measured using a stylus profiler (Wyko NT1100, Veeco). Before soft lithography and fabrication of PDMS devices we coated the surface of the SU-8 template by trichloro (1H,1H,2H,2H-perfluorooctyl) silane (Sigma 448931) to prevent from the attachment of template to the PDMS devices. Silanization was performed by placing 50 <em="">µ</em>l of silane as a droplet on a glass microscopy slide beside the SU-8 template in a desiccator and continuous vacuum for 24 h.
<strong="">Figure 1.</strong> Fabrication of compartmentalized microfluidic device, its coupling with MEA chips and seeding neurons. (A) Two-layer SU-8 master was molded based on two-step photolithography. First the thin layer of SU-8 25 (5 <em="">µ</em>m) was exposed to the UV light through first layer of chromium mask, then the thick layer of SU-8 50 was coated on top (100 <em="">µ</em>m) and exposed to UV through second layer of mask. (B) PDMS was degassed and poured on master template, a transparency film was placed on the back side, PDMS was cured at 80 °C, transparency film was removed and PDMS device detached and punched. (C) Top view of the microfluidic device and its alignment with MEA chip. Four round through-holes were connected through 100 <em="">µ</em>m height channels to the main reservoir. The main reservoir consists of eight modules (100 <em="">µ</em>m height), each connected to a microchannel with 5 <em="">µ</em>m height and 1000 <em="">µ</em>m length. (D) Loading the microfluidic device with cells. Neurons were added into one of through-holes and were let to settle automatically in the main reservoir modules. Network was formed in the main reservoir modules and their neurites were extended into the microchannels. Each microchannel contain 5 electrodes to record neurites activity at their different lengths. (E) and (F) Dimensions of compartmentalized microfluidic device. (G) Aligned microfluidic device on a standard MEA chip.
Download figure:
Standard image High-resolution imagePDMS pre-polymer and curing catalyst were mixed (10:1), degassed and poured on surface of the template (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(B)). A transparency film used to squeeze extra PDMS from top and make a flat surface at the back side of the PDMS device. After curing the PDMS (30 min at 80 °C) and pealing it out from the template, the transparency film was detached and a through-hole was punched in the place of big reservoirs (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(B)). Fabricated PDMS device included three main compartment types (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>): big round reservoirs (height = 150–200 <em="">µ</em>m) for loading the cells, main reservoir (height = 100 <em="">µ</em>m, width = 300 <em="">µ</em>m) with eight connected modules to reside the network, and eight shallow microchannels (height = 5 <em="">µ</em>m, width = 25 <em="">µ</em>m) for axonal guidance (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(C)–(F)). Microchannels connected the main reservoir to a counterpart reservoir with same height, toward which the axonal growth was guided. PDMS microchannel devices were placed on a clean glass substrate, baked for 2 h at 110 °C, then were washed with pure ethanol to remove toxic un-cured oligomers. At the end, all devices were washed with ultrapure water and dried in an oven and kept clean before use.
2.2. Integration of the microchannel devices with MEA chips and preparation for cell seeding
PDMS devices were autoclaved at 120 °C for 20 min and manually aligned with electrodes of standard MEA device (30/200 ir, Multi-Channel Systems) using a droplet of pure ethanol [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>]. Alignment was performed under light microscope (5<b="">×</b>) to match the position of the electrodes on MEA substrate with position of the main reservoir and microchannels of the PDMS device (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(C) and (E)). Alignment was checked under higher magnifications (10<b="">×</b> and 20<b="">×</b>) to be sure that electrodes in each row matches the microchannels. Each reservoir module was aligned with two electrodes and each microchannel was aligned with five electrodes. Combined MEA-microchannel devices were plasma treated for 2–3 min (Femto, Diener) to hydrophilize the microchannels and surface of the MEA. Microchannels and corresponding MEA surface was coated by adding poly-D-lysine (PDL, 0.1 mg ml<sup="">−1</sup>) mixed with 0.05 mg ml<sup="">−1</sup> laminin (5–10 <em="">µ</em>l) [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib39" id="fnref-bfacb466bib39"="">39</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib80" id="fnref-bfacb466bib80"="">80</a>] through one of the big reservoirs (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f1"="">1</a>(C) and (D)). The coating was dried under the hood and washed 3 times with ultrapure water. Then water was replaced by cell culture media (Neurobasal media mixed with 2% B27, 1% penicillin/streptomycin and 2% Glutamax) and incubated overnight before seeding neurons.
2.3. Preparation and seeding neural cells in microchannel devices
Pregnant Sprague Dawley rats (CD IGS, Charles River) were used for harvesting the cortical neurons from their embryos at E17/18. At conception days 17–18 Rats were anesthetized and sacrificed by cervical dislocation. A previously described standard tissue dissociation protocol was applied to harvest cortices of the embryos after removing the meninges in cold Hank's balanced salt solution (HBSS). The cortices were extracted and dissociated by incubating them for 10 min in 0.25% (w/v) trypsin in HBSS buffer. Trypsin digestion was deactivated by 0.25 mg ml<sup="">−1</sup> soybean trypsin inhibitor along with 0.01% (w/v) DNase (Sigma). Then cells were suspended by sequential pipetting using three fire-polished Pasteur pipettes with decreasing diameters. Cell suspension was centrifuged (200 g, 5 min) and the cell pellets were resuspended in neural culture media ready for seeding.
Old media in MEA ring and big reservoirs of the microchannel device was drained and a total number of 10 000 cells in 5 <em="">µ</em>l was added through one of the big reservoirs that automatically left cells in the main reservoir. Cells were incubated for 15 min and extra cells in the big reservoirs were removed by gentle pipetting (10 <em="">µ</em>l pipette tips). Warm culture media was added to the MEA ring to cover the whole microchannel device in media and then moved to incubator (5% CO<sub="">2</sub>, 37 °C, 95% RH). Custom made PDMS lids were used to seal MEA rings. Neural culture in MEA-microfluidic devices inside the CO<sub="">2</sub>-incubator were let to grow for 12 days before moving them to the perfusion system on a microscope stage.
2.4. Fabrication of the PDMS-based perfusion chamber
A PDMS-based perfusion chamber was fabricated by soft lithography using a perfusion chamber template (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A)). A template composed of four separate components was designed in Autodesk inventor environment and fabricated using computer numerical control milling on Aluminum substrate (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A)). Aluminum template included one external O-ring (outer wall) and one internal O-ring (inner wall) for fabrication of the perfusion chamber walls, a core segment with slope for air bubble guidance and a base plate to assemble all components on it. External O-ring was fabricated with 28 mm inner diameter, thickness of 2 mm and height of 18 mm. Internal O-ring had 19 mm inner diameter, 2.5 mm thickness and 14 mm height. The core segment for fabrication of the air bubble guiding system is a diagonally halved cylinder with its maximum diameter at its base around 15 mm and a height of 10 mm. A round base piece with diameter of 38 mm was fabricated to hold the other three components. All surfaces were polished before being used for the fabrication of PDMS perfusion chamber (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A) and Supplementary data_video 1).
<strong="">Figure 2.</strong> Fabrication of perfusion chamber. (A) Perfusion chamber template was fabricated in four pieces and then assembled together to mold the perfusion chamber. Two O-rings with different heights and inner diameters were used to fabricate the outer and inner wall of the perfusion chamber. A diagonally halved cylinder in the center was used to fabricate the ceiling with slope. A base plate was used to hold 3 components together. Inlet tube was connected to the base and outlet tube was attached to the top of the central part. (B) A single piece PDMS-perfusion chamber was fabricated by pouring degassed PDMS into the template, placing a film on the back to make a smooth surface, curing PDMs in 80 °C, removal of transparency film and base plate and outer O-ring of the template. After loosening the PDMS walls with pure ethanol central part and inner O-ring of template were removed. (C) Integration of perfusion chamber with microfluidic device aligned on a MEA device. Autoclaved and plasma-treated perfusion chamber was pressed gently to the glass ring of MEA to fit the glass ring into the space between inner and outer walls of the chamber. Media was gradually injected through inlet to fill the chamber space from bottom toward the top. (D) 3D design of the perfusion chamber representing the walls and embedded tubing. Middle panel illustrate the filled space with media and right panel shows air bubble guidance through the slope.
Download figure:
Standard image High-resolution imagePrior to the fabrication of the perfusion chamber two polytetrafluoroethylene (PTFE) inlet and outlet tubes (outer diameter 2.1 mm and inner diameter 1.5 mm, Supelco 20531) were attached to the base and top of the core piece, respectively (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A)). Inlet and outlet tubes were collected and kept in one side of the template to keep the central parts of the fabricated chamber accessible for microscopy (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A) and (D)). These configurations also saved space and later helped to integrate the perfusion system with MEA amplifier. PDMS pre-polymer and catalyst (Sylgard 184, Dow Corning) were mixed (10 to 1 ratio), degassed in a desiccator (Cole-Palmer, Germany) and poured into the assembled template with attached tubing [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>]. PDMS left for 15 min in room temperature to get rid of extra bubbles and then a transparency film was placed on top of the PDMS to make a smooth surface for microscopy. PDMS was cured in 80 °C for 40 min, the transparency film was removed and then the perfusion chamber was gently detached from template figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(B). First the base piece was removed and then the O-rings and core part were loosened with the use of pure ethanol. PDMS-based perfusion chamber together with inlet and outlet tubing were released from the template. Outer and inner walls of the PDMS-based perfusion chamber were generated from PDMS-filled spaces between outer and inner O-rings of the template and inner O-ring and core piece of the template, respectively (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(A), (B) and Supplementary data_video 1). In some cases, PDMS material clogged the tubing which was replaced by new inlet and outlet tubing and sealed by extra PDMS. Fabricated perfusion chambers were tested for leakage, cured for 6 h in 80 °C, rinsed in ethanol 70% for 3 h and then washed with ultrapure water and dried in Oven (100 °C, 2 h). To prevent from air bubble attachment to the inner walls of the perfusion chamber we enhanced the hydrophilicity of the inner walls by plasma treatment for 2 min and adding phosphate buffered saline (PBS) solution into the chamber 24 h before the use.
2.5. Preparation of the conditioned media
Cell culture media (Neurobasal medium, 2% B-27 supplement, 1% Penicillin/streptomycin and 2% Glutamax), that was used for seeding neurons and keeping them in the incubator, was conditioned to ambient CO<sub="">2</sub> based on the following steps (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f3"="">3</a>(A)): First media was mixed with 10 mM HEPES (Sigma H0887) buffer in a 50 ml falcon tube and kept in the water bath (37 °C) for 2 h. Then pH of the media was measured and titrated to physiological pH 7.4 using HCl (1 M). Media was mixed again and left for another 2 h at 37 °C water bath. Conditioning the media, pH measurement and titration was repeated until the media reached a steady pH in 37 °C. Then osmolality of the media was checked using an osmometer (Vapro 5520, Wescor). Conditioned media was sterile-filtered using a 0.2 <em="">µ</em>m filter and kept in at 4 °C. Before using the conditioned media for experiments or perfusion system the media was warmed in water bath (37 °C) for 30 min.
<strong="">Figure 3.</strong> Preparation of the conditioned media and assembling the perfusion system with microscopy and electrophysiology setups. (A) Neurobasal media was mixed with HEPES buffer and conditioned to the ambient CO<sub="">2</sub>. Mixed media was kept at 37 °C for 2 h. The pH was measured and adjusted and media returned to 37 °C. The pH measure, titration and conditioning were repeated until a stable pH level was achieved. Conditioned media was sterile filtered and stored at 4 °C until it was used in perfusion system. (B) The whole integrated setup. MEA amplifier was fixed to the microscopy stage. Perfusion syringe was installed on support beam of the perfusion setup. After filling the inlet tube, it was locked by a T-piece Luer. Already filled perfusion chamber with MEA and microfluidic devices was mounted on MEA amplifier and connected to the inlet and outlet tubes. Perfusion was started and monitored for one hour. Afterward imaging and electrophysiology recording were started. Perfusion was stopped only for changing the syringe (once every five or six days), with no interruption in electrophysiology recording and microscopy imaging. Inlet and outlet tubing were grounded to avoid possible noise in recordings.
Download figure:
Standard image High-resolution image2.6. Integration of perfusion chamber with MEA-microchannel devices
MEA devices with aligned microfluidic structures and growing neuronal networks at 11 DIV were used for perfusion system experiments (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(C)). Male Luer lock injection sites (World Precision Instruments, 14034-40) were fitted to the endings of inlet and outlet tubes. A rubber stopper with turn-over flange (Carl Roth) was fitted to the end of the male luer lock to seal the tubing. Custom made PDMS lids were removed and 700 <em="">µ</em>l of the media inside the MEA ring was drained. The inner and outer walls of the perfusion chamber were pressed to the MEA ring to fit the MEA glass ring in to the space between inner and outer walls of the chamber and seal it. A syringe containing the already prepared conditioned media was connected to the inlet through the septa at the top of the rubber stopper, and was used to inject culture medium in lower speed to slowly fill the perfusion chamber from base to the top (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(C) and (D)). During media injection MEA was tilted to let the chamber ceiling slope to guide the air toward the outlet. Media was injected until the whole chamber and outlet tube is filled (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>(C) and (D)). Inlet and outlet of the perfusion chamber were tightly sealed with luer-locks. Then, integrated MEA-microchannel devices with perfusion chamber were moved to the incubator for 24 h.
2.7. Assembling the perfusion system
At 12 DIV, MEA-microchannel devices and already installed perfusion chamber were moved outside of the incubator to the microscopy stage and were connected to the perfusion system. Custom-made perfusion system contained a stepper motor [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib49" id="fnref-bfacb466bib49"="">49</a>], syringe pump, support beam, inlet and outlet extension tubing, inlet tube air bubble trap, and a UV light bulb to disinfect the waste collecting tube at the end of the outlet tube (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f3"="">3</a>(B)). Generated force by stepper-motor (rotation = 1.8 degrees per step) was transferred through a micrometer screw gauge (MW-Import) to a 5 ml syringe plunger. Finger flange and tip of the syringe was tightly fixed to the custom-made syringe holder and base of the plunger was fixed to the screw gauge and supporting beam [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. The input of the stepper motor integrated circuit (SAA1042) was controlled by a timer circuit (LM555) that commanded the stepping intervals of the stepper motor. Perfusion rates of the conditioned media were adjusted around 10 <em="">µ</em>l per hour. Before entering the perfusion chamber, the conditioned media was passed through a T-piece Luer to trap air bubbles (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f3"="">3</a>(B)). The outlet of the perfusion chamber was connected to the outlet tubing of the perfusion system to collect the used media in a sealed container. Waste tubing was disinfected by installing an ultraviolet light emitting diode (UVA LED; 405 nm, Conrad Electronics). Perfusion was started after all connections were installed properly. Reliability of the perfusion rate was measured by monitoring the injected volume every 12 h (<em="">n</em> = 5 MEAs) for the period of two weeks.
2.8. Time-lapse microscopy of the growing neural network
A 10<b="">×</b> objective with sufficient field of view covering the whole network area (1.5 mm<sup="">2</sup>) including all electrodes, main reservoirs and all microchannels was used to obtain microscopy images (Zeiss, Axiovert 200). MEA-chips were placed in the MEA amplifier which has been tightly screwed to the already fixed custom-made breadboard stage. This helped to reduce the spatial shifts between time-lapse frames that is a common issue for time-lapse and large-magnification imaging. A digital camera (Canon, G2/G9) was connected to the microscope and controlled remotely (Breeze Systems, PSRemote) to obtain images from the whole network every 10 min. Microscopy light was only switched on during the time-lapse imaging periods for 30 s to prevent from light-induced media toxicity and degradation of the buffers. Image sequences were imported to the Fiji-ImageJ software, cropped, and adjusted for brightness and contrast. Images were stacked in Fiji-ImageJ and stabilized using StackReg pluging for recursive alignment of the stack of the images [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib81" id="fnref-bfacb466bib81"="">81</a>]. Finally, time-lapse images were imported to the Fiji-ImageJ and were saved as video files with 120 frames per second (Supplementary data_video 2). To track axonal growth in microchannels during the 12 days of perfusion period, 82 frames with 200 min intervals were selected. Image sequence was stacked as one file and was stabilized using StackReg pluging. Elongation of selected branches of axons were tracked using manual tracking plugin in ImageJ based on movement of their growth cone between subsequent frames. In the manual tracking window, we used overlay lines option to show the axonal growth trajectory inside the microchannels. Video file of the tracked axons was saved with three frames per second.
2.9. Immunofluorescence staining
Immunofluorescence staining was performed on cortical cultures at DIV 21. These cultures have been prepared in microfluidic devices aligned on 18 mm coverslips. Protocol for PDL-laminin coating, PDMS device alignment and cell seeding was identical to MEA cultures. As described previously [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>] PDMS devices were detached gently before fixation. Neuronal cultures were rinsed with 1% PBS and fixed using paraformaldehyde (4%) plus sucrose solution (4%) for 10 min. We used 0.2% Triton X-100 in 1% PBS treatment to permeabilize the membrane (10 min incubation time), and then cultures were incubated for 30–45 min in blocking buffer containing goat serum (2%) and bovine serum albumin (3%) in 1% PBS. Then networks were incubated by primary antibodies for 60 min at room temperature, were washed 3<b="">×</b> with 1% PBS, and were incubated with secondary antibodies for 1 h in room temperature. Following primary antibodies were used: rabbit polyclonal anti-<em="">β</em>-tubulin-III (T2200, Sigma) and mouse monoclonal anti-glial fibrillary acidic protein (G3893, Sigma). Fluorescent-conjugated secondary antibodies including Alexa Fluor 488 (goat anti-mouse) and Alexa Fluor 633 (goat-anti-rabbit) both purchased from Life Technologies. Fluorescence images from stained cortical networks were obtained using an inverted confocal microscope (Nikon A1+) and Optronics Microfire microscope camera (2-megapixel, MBF). High resolution mosaic images (20<b="">×</b>) were stitched together in ImageJ software (Grid/Collection stitching algorithm) to illustrate the morphology of whole modules.
2.10. Electrophysiology recordings of developing neural network
Extracellular activity of neurons captured by planar electrodes of the MEA chip with 32 K sampling rate using a Multichannel Systems amplifier (MEA60-Up, MCS). MEA chips were placed on a base plate of the amplifier that contains a built-in heating element and a temperature sensor (Pt-100) and was controlled by an external system (TC02, MCS). Before connecting the perfusion system to the perfusion chamber, integrated MEA-microchannel device and perfusion chamber was placed on amplifier, the chamber and tubes positioned properly to allow fixing the amplifier pins with pads of the MEA device (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f3"="">3</a>(B)). As mentioned earlier amplifier has already been screwed to the breadboard of the custom-made stage to reduce the spatial displacement of the images during the microscopy. To reduce the noise in the recorded data, first perfusion tubes were shielded by a grounded wire mesh sock, then inlet and outlet of perfusion chamber were grounded, and finally a grounded custom-made metal cap was placed on top of the perfusion chamber to shield it. A hole in the center of the metal cap allowed microscopy imaging.
Electrophysiology recording from the network started 60 min after running the perfusion system. Because continuous recording of raw electrophysiology signals produces large file sizes, only detected action potentials (APs) were saved. To extract and save APs, an online AP detection threshold was adjusted to −5.5 standard deviation from noise level. AP waveforms were saved in 10 min fragments and later merged together in Neuroexplorer software for further analysis. Data between 12 DIV and 24 DIV including 1 million seconds of recording time was used for analysis. Using the firing rate histogram algorithm in the Neuroexplorer we extracted the average frequency in each electrode at 60 s, 12 min, 1 h, 24 h and a week and used it for analysis of temporal changes in network activity dynamics. Rate histogram that measures the firing rate <em="">vs.</em> time were used to continuously measure the overall network activity, network activity of sub-cellular compartments (axons) and as well activity of distal or proximal parts of axonal branches. To measure the network burst activity first the fragmented data were concatenated and imported into the Neuroexplorer software. Then we used burst detection feature of surprise algorithm to measure bursts activities over time. Bursts were detected based on following criteria: a minimum surprise of 4, a minimum burst duration of 20 ms and at least four action potentials per burst [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib82" id="fnref-bfacb466bib82"="">82</a>]. Measured bursts data and their corresponding timestamps in individual electrodes were exported into Microsoft Excel file. Average burst frequency and burst duration were calculated at different temporal scales (60 s, 12 min, 1 h, 24 h and a week) and plotted for whole period of the 12 days on perfusion system.
2.11. Measuring the stability of pH and osmolality of the conditioned media
To probe the functionality of perfusion chamber and conditioned media we measured the effect of ambient conditions on pH and osmolality of conditioned media. To measure pH changes we used 15 MEAs with perfusion chambers. Five MEAs (<em="">n</em> = 5) were filled with non-conditioned media and had no perfusion, 5 MEAs were filled with conditioned media (<em="">n</em> = 5) with no perfusion and five remaining MEAs were filled with conditioned media then were connected to the continuous perfusion (<em="">n</em> = 5). All MEAs were kept in a 37 °C heating plate for 12 days and pH was measured every 12 h. To measure the effect of evaporation from perfusion chamber on osmolality of the conditioned media we compared MEAs sealed with custom-made PDMS lids (<em="">n</em> = 5) with MEAs that had perfusion chamber (<em="">n</em> = 5). Custom-made PDMS lids only had outer wall with no bubble guiding system, inner wall and inlet or outlet tubes. MEA chambers were filled with conditioned media and kept on heating plate 37 °C for 12 days. Osmolality was monitored every 12 h.
3. Results
The space inside the PDMS-based perfusion chamber accommodated around 1.76 ml of the culture media. The main goals of including two PDMS walls in the perfusion chambers were to prevent from the media leakage and to reduce the media evaporation during long-term perfusion period. The glass ring of the MEA was embedded into the spaces between PDMS walls; therefore, the internal pressure pressed the inner PDMS wall to the glass ring and prevented from the leakage, which was common issue in previous versions of the PDMS chamber with single-wall [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. Such tight sealing also helped on reducing the air bubble formation inside the chamber. To test the functionality of perfusion system we studied following parameters before using it for long-term neural network cultures: stability of the temperature, perfusion rate, and consistency of pH and osmolality in conditioned media over time in ambient conditions.
3.1. Perfusion parameters
To test the consistency of the temperature provided by MEA amplifier we placed five MEAs with filled perfusion chamber on the amplifier. Recorded temperature readings showed subtle fluctuation but temperature remained between 37.08 °C and 37.37 °C for 12 days (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(A)). Perfusion rate, volume of injected culture media (<em="">µ</em>l) per 12 h, remained constant with average value of 111.6 ± 0.48 <em="">µ</em>l and maximum and minimum of 116.89 <em="">µ</em>l and 106.32 <em="">µ</em>l, respectively (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(B)). Osmolality of the conditioned media was compared between perfusion chambers with single wall and double wall (<em="">n</em> = 5 per group, figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(C)). To this end, first both chambers were filled with conditioned media and then left on a heating plate (37 °C) and the osmolality was checked every 12 h for 12 days. For each measurement session 50 <em="">µ</em>l was injected through inlet and 50 <em="">µ</em>l was collected through outlet. Our data showed that without active perfusion osmolality increased from 274.45 ± 1.6 mOsmolkg<sup="">−1</sup> to 319.77 ± 11.76 mOsmolkg<sup="">−1</sup> in single-wall chamber (p < 0.01, figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(C)), and from 274.059 ± 1.95 mOsmolkg<sup="">−1</sup> to 279.57 ± 2.63 in double-wall chamber (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(C)). We measured the pH of the conditioned and non-conditioned Medias in double-wall perfusion chambers for 12 days. In groups without active perfusion measurements were performed once a day by draining 500 <em="">µ</em>l of old media and replacing it with 500 <em="">µ</em>l of fresh media. The pH of un-conditioned media without active perfusion increased from 7.47 ± 0.04 to 8.57 ± 0.23 during 12 days, which was significantly higher than the increase in pH of the conditioned media (7.53 ± 0.11–7.98 ± 0.21, <em="">p</em> < 0.01; figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(D)). Changes in the pH of conditioned media with active perfusion (500 <em="">µ</em>l per day) was lower than conditioned media without perfusion (7.46 ± 0.02–7.68 ± 0.10, <em="">p</em> < 0.05; figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f4"="">4</a>(D)).
<strong="">Figure 4.</strong> Probing perfusion parameters in long-term. (A) Temperature readings every 12 h in five perfusion chambers with conditioned media and continuous perfusion. (B) Perfusion rate was recorded based on the volume of perfused conditioned media every 12 h (<em="">n</em> = 5). (C) Osmolality of conditioned media in single-wall perfusion chamber (olive) and double-wall perfusion chamber (green), with no perfusion. Single-wall perfusion chamber did not have the inner wall. Data recorded every 12 h (<em="">n</em> = 5 per group). Data of the same group were compared between different time-points based on Repeated measures ANOVA. *<em="">p</em> < 0.05 and **<em="">p</em> < 0.01 <em="">vs.</em> initial time point in single-wall perfusion chamber. (D) pH of media was probed every 12 h in double-wall perfusion chambers filled with non-conditioned (gray) or conditioned media (green) with no perfusion, and in double-wall perfusion chambers filled with conditioned media with active perfusion (red). Data of each group were analyzed between different time-points in accordance to the Repeated measures ANOVA. *<em="">p</em> < 0.05 and **<em="">p</em> < 0.01 and ***<em="">p</em> < 0.001 <em="">vs.</em> initial time point in non-conditioned media group, and #<em="">p</em> < 0.05 <em="">vs.</em> initial time point in conditioned media group with no perfusion.
Download figure:
Standard image High-resolution image3.2. Network morphology
At 12 DIV time-lapse microscopy started after assembling the combined MEA-microfluidic-perfusion chamber on MEA-amplifier and microscopy stage and running the perfusion. Neuronal networks in the main reservoir had already been formed at 12 DIV and neurites had been elongated up to 450 <em="">µ</em>m of microchannel length (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f5"="">5</a>). Time-lapse images from the reservoir modules revealed that neuronal cell body remain in the main reservoir, however a continuous neuronal cell body displacement was observed in all days (Supplementary data_video 2). This neuronal movement was along with neurites tension and their movements in reservoir modules and microchannels (Supplementary data_video 2, and figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f5"="">5</a>). Immunofluorescent images also showed that cell bodies of the neurons and glial cells only localized to the reservoir compartment while neurites extend into the microchannels (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f6"="">6</a>). A small percentages of the cells represented larger and faster displacements compared to network forming neuronal cells (Supplementary data_video 2). Fluorescent images from stained networks revealed that these cells are glial cells (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f6"="">6</a>). Time-lapse images of glial cells showed that these cells with shorter branches underwent a detectable cell division in some cases (Supplementary data_video 2). Microscopy images from microchannels showed a persistent axonal growth and elongation during 12 days under perfusion (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f5"="">5</a> and Supplementary data_video 3). These axonal branches reached to the counterpart reservoir around 20 DIV (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f5"="">5</a> and Supplementary data_video 3).
<strong="">Figure 5.</strong> Network morphology in a test culture under continuous perfusion with conditioned media. Images from whole network (left) and magnified view of two modules (right) at different time points after perfusion. All images were taken by a 10<b="">×</b> objective. Scale bar, 200 <em="">µ</em>m. Green and yellow arrows illustrate the morphological changes in neurite branches inside two adjacent microchannel and show axonal elongation into the counterpart reservoir over days. Development of the network morphology was shown in Supplementary data_video 2 and an example of tracked axonal growth is shown in Supplementary data_video 3.
Download figure:
Standard image High-resolution image
<strong="">Figure 6.</strong> Immunofluorescence image from a cortical network in a microfluidic device on a glass coverslip at 21 DIV. (A) Images show network morphology in four modules of the microchannel device including neuronal soma and neurites (magenta), glial cells (green) and nuclei (gray). Nuclei, glial cells and neurons were stained with DAPI, glial fibrillary acidic protein (GFAP) and <em="">β</em>-III-tubulin, respectively. Scale bar, 200 <em="">µ</em>m. (B) Magnified view of a selected area of A in yellow rectangle representing neurons, glial cells and nuclei in a reservoir module. Scale bar, 50 <em="">µ</em>m.
Download figure:
Standard image High-resolution image3.3. Network electrophysiology
Continuous network electrophysiology recordings started alongside the microscopy at 12 DIV. While two electrodes were recording from each network module at its reservoir region, five subsequent electrodes were recording from same axonal branches at their different length inside a microchannel. We measured the network activity as frequency of recorded APs per electrode at different parts of the network. This included recorded activity from residing neuronal networks at main reservoir modules, and from axons in the microchannels. Overall network activity tended to increase gradually from 0.01 Hz at 12 DIV to 0.09 Hz at 20 DIV (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>), after which a sharp and steady increase in the network activity was observed (0.35 Hz; figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>(A)). Recorded data were used to determine the network activity dynamics at different spatial resolutions (ranging from whole network to a network module, to particular axons, and axonal distal or proximal segments), and at different temporal resolutions (weeks, days, hours, and minutes and seconds), figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>(B). This allowed to differentiate the activity at different compartments of the network with sufficient temporal resolution. For instance, between 14 DIV and 16 DIV a particular network module showed two sharp activity peaks which have been obscured in overall network activity (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>(B)). Tracking the activity in the corresponding microchannel of the studied module revealed that at 15 DIV this raise in activity only belongs to the proximal electrodes, or proximal axons, while it is absent in distal parts of the microchannel (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>(B)). These data can be exploited to monitor axonal growth-related activity profile in electrodes of the microchannels. Temporal zooming to a 10 min window of activity at 20 DIV showed that trend of activity, fluctuations in AP frequency per second, is different between whole microchannel, whole axonal branch, and its distal parts, axonal distal segments (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a>(B)). Therefore, temporal and spatial dissociation of the network activity information in the current platform reveals detailed dynamics of developing network activity at sub-cellular and sub-seconds resolutions.
<strong="">Figure 7.</strong> Tracking network activity with different temporal and spatial scales. (A) Overall network activity at different temporal scales from weeks (black), to days (red), hours (green), minutes (yellow) and seconds (blue). Action potential frequency at each time period was averaged in all electrodes of MEA and represented as one data point. Time has been represented in seconds. (B) Network data at different spatial and temporal scales. Action potential frequency data collected from whole network (<em="">n</em> = 64 electrodes), from one network module (<em="">n</em> = 8 electrodes), from specific axonal branches inside a microchannel (<em="">n</em> = 5 electrodes), and only distal segments of those axonal branches (<em="">n</em> = 2 electrodes) over weeks, days, hours, minutes and seconds (up to bottom panels). Time scales are represented as million seconds. In each graph data points represent average AP frequency calculated in electrodes recording from specific part of the network at specific time point. Black asterisk illustrate activity that is missing in distal electrodes (or distal segment of that particular axons) and black triangle shows the activity that present in whole network.
Download figure:
Standard image High-resolution imageBurst activity appeared from 17 DIV and increased by culture age over time. Burst frequency at 17 DIV was around 0.001 bursts per second but it reached to a maximum of 0.055 bursts per second at 22 DIV (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f8"="">8</a>(A)) and declined slightly during the next two days (0.041 bursts per second at 24 DIV). Temporal zoom into the burst frequency data at 12 h of 21 DIV shows that burst frequency fluctuated between 0.01 Hz and 0.1 Hz (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f8"="">8</a>(A)). Burst duration represented a constant increase by culture age and reached from 23.49 ms at 18 DIV to 116.47 ms at 24 DIV (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f8"="">8</a>(B)). Same to AP frequency data the system allowed to study dynamics of the burst activity features in different temporal scales and illustrate the changes in burst activity over time (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f8"="">8</a>).
<strong="">Figure 8.</strong> Tracking the development of burst features at different temporal scales. (A) Burst frequency data collected from whole network at different temporal scales from weeks (black), to days (red), hours (green), minutes (yellow) and seconds (blue). Burst frequency at each time period was averaged in all electrodes of MEA and represented as one data point. Time has been represented in million seconds. (B) Changes in burst duration with network development. Each data point represents average burst duration in all electrodes at different temporal scales from weeks (black), to days (red), hours (green), minutes (yellow) and seconds (blue).
Download figure:
Standard image High-resolution image3.4. Tracking axonal growth morphology and activity
Combined microscopy and electrophysiology at subcellular level enabled tracking axonal growth morphology inside microchannels and corresponding changes in axonal growth-related activity from sequence of electrodes inside the microchannel (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a> and Supplementary data_video 3). During axonal elongation axons first passed the proximal electrodes and later reached to the distal sections (electrodes) of the microchannel (Supplementary data_video 3 and figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(A) and (B)). Tracking the individual axons revealed that even with presence of axons in the proximal electrodes of the microchannel at 12 DIV, only a sparse activity was recorded on these electrodes (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(C)). At 15 DIV with further elongation of axonal branches, frequency of APs increased in the proximal electrodes and an activity with lower frequencies was appeared in the distal electrodes (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(C)). From 18 DIV to 23 DIV further elongations of the axonal branches inside the microchannel was associated with appearance of strong activity in distal electrodes of the same microchannel (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(C)). Axonal growth-related activity was analyzed by measuring the cumulative activities in individual electrodes of the microchannel (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(D)). The cumulative activity in each electrode was measured as AP frequency data in 6 h intervals. The cumulative activity in three proximal electrodes of the microchannel increased between 48 and 72 h of perfusion (14 DIV and 15 DIV, 0.72 Hz), while cumulative activity on distal electrode increased after 96 h of perfusion (around 16 DIV, 0.74 Hz) and on the most distal electrode after 192 h of perfusion time (around 20 DIV, 0.3 Hz, figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(D)). These spatiotemporal changes in the recorded activity of electrodes were correlated with axonal growth and elongation from proximal to distal electrodes inside the microchannels (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f9"="">9</a>(C) and (D)).
<strong="">Figure 9.</strong> Tracking axonal growth morphology and activity in microchannels. (A) Axonal branches inside a microchannel at 23 DIV. Electrode indexes have been mentioned beside the electrodes. (B) Magnified view of red rectangle in A, showing a tracked axonal branch between 12 DIV and 23 DIV (tracked line was represented in blue color). (C) Time-lapse images and activity profile of axonal growth at different DIVs. Red arrows show the position of axon growth cone at each DIV. More details of tracking axonal growth have been included in Supplementary data_video 3. Activity traces (green vertical lines) from a selected window of recording (5 min) in five subsequent electrodes of a microchannel at different DIVs. Each green line represents a detected action potential. The height of the green lines represents the amplitude of the action potentials. (D) Cumulative action potential frequency during 12 days perfusion period in five electrodes of a microchannel. Data were averaged at 6 h intervals in each electrode and represented as cumulative activity over time. Color code from dark green to olive shows the recordings from the most proximal electrode (dark green, electrode 26) to most distal electrode (olive, electrode 22) over time.
Download figure:
Standard image High-resolution image
4. Discussion
The major goal of the present work was to integrate the incubator-independent perfusion chamber with compartmentalized microfluidic device to achieve long-term continuous electrophysiology recording with sub-cellular resolution from developing cortical networks. This makes it possible to continuously monitor network functional dynamics at its different modules, at single branches of axons and even in particular axonal segments. Leak-free perfusion chamber with air bubble guidance system and conditioned media to ambient CO<sub="">2</sub> enabled un-interrupted microscopy and electrophysiology readouts from developing neuronal networks over two weeks with sufficient temporal resolution. Compartmentalized microfluidic device, on the other hand, allowed tracking the morphology and activity of the whole network with sub-cellular resolution.
Double-wall perfusion chamber was fabricated as a single piece of transparent PDMS mold. With small modifications in the design, it can be used for other cell culture substrates with different dimensions and configurations. The perfusion chamber and corresponding inlet and outlet tubing were designed compact enough to fit the available space inside and outside of the MEA chip and limited space of MEA amplifier. To achieve an improved and steady perfusion with no leakage and air-bubble formation we updated an already developed single-wall perfusion chamber design [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. Incorporated internal and external walls in the design of the perfusion chamber allowed to seal the MEA ring from inside and outside (figure <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f2"="">2</a>). An air bubble trapping slope [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib32" id="fnref-bfacb466bib32"="">32</a>] guaranteed the removal of the possibly formed air bubble, that is a common issue in many microfluidic perfusion systems [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib83" id="fnref-bfacb466bib83"="">83</a>]. It affects fluid flow and microscopy imaging and damages neuronal cells [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib32" id="fnref-bfacb466bib32"="">32</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib84" id="fnref-bfacb466bib84"="">84</a>]. Previously, different approaches have been tried to avoid air bubble formation including the incorporation of bubble blocking microchannels [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib84" id="fnref-bfacb466bib84"="">84</a>], hydrophobic bubble capture [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib85" id="fnref-bfacb466bib85"="">85</a>], ultrasonic degassing [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib86" id="fnref-bfacb466bib86"="">86</a>], and many other methods [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib83" id="fnref-bfacb466bib83"="">83</a>]. In the present work we hydrophilized the PDMS chamber inner walls and trapping slope using oxygen plasma treatment right before using the chamber. An air bubble trap was also included in the chamber inlet. These together with air bubble guiding system helped to avoid the accumulation of bubbles in the perfusion system. Moreover, minor media evaporation in our bubble-free perfusion system confirmed by stable osmolality of the cell culture media at long-term.
In this study we used HEPES-treated cell culture media to avoid dependency of our perfusion system to the CO<sub="">2</sub> supply. To maintain neuronal cell culture for long-term, the culture media should be continuously exposed to 5% CO<sub="">2</sub> concentration in a CO<sub="">2</sub> incubator [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib87" id="fnref-bfacb466bib87"="">87</a>]. In cell culture medias that contain 26 mM sodium bicarbonate (NaHCO3), like Neurobasal media, 5% dissolved CO<sub="">2</sub> in media controls the physiological pH through the bicarbonate buffering system [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib88" id="fnref-bfacb466bib88"="">88</a>]. CO<sub="">2</sub>/HCO3<sup="">-</sup> buffer is considered as a critical buffer system for extracellular solutions in the body fluids [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib89" id="fnref-bfacb466bib89"="">89</a>]. However, synthetic buffers like HEPES offer strong and stable buffering capacities for wider ranges of experimental applications like performing long-term measurements in ambient conditions outside of the CO<sub="">2</sub>-incubator [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib90" id="fnref-bfacb466bib90"="">90</a>].
Previous works on developing the incubator-independent cell culture system have been mainly designed for continuous microscopy readouts from cultured cells for hours or days [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib28" id="fnref-bfacb466bib28"="">28</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib30" id="fnref-bfacb466bib30"="">30</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib33" id="fnref-bfacb466bib33"="">33</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib35" id="fnref-bfacb466bib35"="">35</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib43" id="fnref-bfacb466bib43"="">43</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib91" id="fnref-bfacb466bib91"="">91</a>]. For instance, stage-top mini-incubators have been designed to supply CO<sub="">2</sub>, temperature and humidity to the perfusion chambers assembled on the microscope stage [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib43" id="fnref-bfacb466bib43"="">43</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib45" id="fnref-bfacb466bib45"="">45</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib51" id="fnref-bfacb466bib51"="">51</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib53" id="fnref-bfacb466bib53"="">53</a>]. To capture neuronal network activity either on-stage incubators are modified to accommodate the MEA amplifiers or custom-developed perfusion chambers are integrated with MEA electrophysiology setup [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib45" id="fnref-bfacb466bib45"="">45</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib51" id="fnref-bfacb466bib51"="">51</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib53" id="fnref-bfacb466bib53"="">53</a>]. Combining on-stage incubators and corresponding gas-supply tubing with MEA amplifier and acquisition tools require adjustment of the microscopy stage to allow proper microscopy [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib45" id="fnref-bfacb466bib45"="">45</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib51" id="fnref-bfacb466bib51"="">51</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib53" id="fnref-bfacb466bib53"="">53</a>]. For continuous delivery of cell culture media in perfusion chambers, in addition to the gas supplying system perfusion tubing should also be incorporated in the design. For instance a perfusion chamber designed by Kreutzer <em="">et al</em> was composed of glass and PDMS materials and contained gas-in and gas-out tubing connected to a gas-supplying equipment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib47" id="fnref-bfacb466bib47"="">47</a>]. An alternative approach to exclude bulky gas-supplying equipment is to use CO<sub="">2</sub>-independent medium [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib44" id="fnref-bfacb466bib44"="">44</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib54" id="fnref-bfacb466bib54"="">54</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib55" id="fnref-bfacb466bib55"="">55</a>]. In a previous work, continuous perfusion of the HEPES-treated CO<sub="">2</sub>-independent culture media enabled to grow hippocampal and cortical networks for months outside of the incubator [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. Randomly formed networks within a single-wall autonomous perfusion chamber [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib49" id="fnref-bfacb466bib49"="">49</a>] were imaged and recorded continuously for the whole length of the experiment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. In the present work, HEPES-treated media underwent iterations of conditioning and pH and osmolality controlling steps to reach a stable pH in ambient settings. With help of tightly sealed chamber and continuous perfusion the conditioned media experienced subtle pH drifts for two weeks of experiment.
Long-term uninterrupted imaging and electrophysiology recordings in a prior study [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>] and in the present work provided temporally detailed information about network formation and its activity dynamics (figures <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f7"="">7</a> and <a xmlns:xlink="http://www.w3.org/1999/xlink" href="#bfacb466f8"="">8</a>). This allowed to inspect network data at different temporal windows from seconds to weeks. Therefore, the system enables extracting both short-term and long-term features of network morphology and activity dynamics. In previous works, neuronal networks were generated randomly on MEA substrate without controlling the network structure. Thus only network-wide data were accessible [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>]. Network structure can be shaped using compartmentalized microfluidic device and their physical configurations [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib56" id="fnref-bfacb466bib56"="">56</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>]. If aligned with MEA electrodes, these devices allow to guide axons into the soma-free microchannels to be studied in an isolated microenvironment [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib71" id="fnref-bfacb466bib71"="">71</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib92" id="fnref-bfacb466bib92"="">92</a>]. This also allows to amplify and capture extracellular signals from thin axonal branches [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib93" id="fnref-bfacb466bib93"="">93</a>], that improves spatial access to the network data. Here, integrated compartmentalized microfluidic device with perfusion setup permitted continuous recording and microscopy at different spatial scales from whole network to separate sub-network modules, to specific axonal bundles, to specific segments of axonal branches.
In a previous work integrated compartmentalized device and MEA chip was kept in an incubator and used to record axonal activity for 5–10 min every 3–4 days [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>]. Microscopy images were also obtained in a limited time points during network development [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib40" id="fnref-bfacb466bib40"="">40</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib64" id="fnref-bfacb466bib64"="">64</a>]. These, inevitably, short-term and interrupted recordings and microscopy have already shared a significant contribution to the <em="">in vitro</em> neuroscience field [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib39" id="fnref-bfacb466bib39"="">39</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib94" id="fnref-bfacb466bib94"="">94</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib97" id="fnref-bfacb466bib97"="">97</a>]. Therefore, in these systems the missing parts of the network activity and morphology dynamics are estimated based on snapshots of short-term recorded data [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib31" id="fnref-bfacb466bib31"="">31</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib95" id="fnref-bfacb466bib95"="">95</a>]. Our new system enabled to constantly record individual axons activity and image their morphology that allowed to visualize the activity profile of the growing axons inside the microchannels. This data is of great value for linking the function to the structure of the developing neuronal circuits [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib2" id="fnref-bfacb466bib2"="">2</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib15" id="fnref-bfacb466bib15"="">15</a>]. Delay in axonal elongation to the distal parts of the microchannels represented by delayed appearance of activity in the distal electrodes around 20 DIV. Regarding the highly dynamic and intricate nature of the <em="">in vivo</em> and <em="">in vitro</em> neuronal function [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib2" id="fnref-bfacb466bib2"="">2</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib98" id="fnref-bfacb466bib98"="">98</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib99" id="fnref-bfacb466bib99"="">99</a>], obtaining robust data from these networks is always demanding [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib59" id="fnref-bfacb466bib59"="">59</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib63" id="fnref-bfacb466bib63"="">63</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib100" id="fnref-bfacb466bib100"="">100</a>]. We believe that continuous readout with sub-axonal resolution will benefit the field by enhancing the reliability of the collected data.
In the present work we used PDMS material for prototyping the perfusion chamber and as well for the fabrication of the microfluidic device. Regardless of its advantages like easy and inexpensive microfabrication and its transparency, one should consider that PDMS material possess disadvantages like adsorption of organic small molecules [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib101" id="fnref-bfacb466bib101"="">101</a>]. To address this issue, perfusion chambers can be fabricated using less-porous materials like glass [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib102" id="fnref-bfacb466bib102"="">102</a>]. Regarding the elasticity of thin PDMS-based microfluidic devices, they are more desired when a reversible bound with MEA electrodes is required, while glass-microfluidics needs to be embedded irreversibly during fabrication of the MEA devices.
There are advanced options available for multisite recording from neuronal networks like high-density MEAs (HD-MEAs) with thousands of electrodes and smaller electrode pitches [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib96" id="fnref-bfacb466bib96"="">96</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib103" id="fnref-bfacb466bib103"="">103</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib106" id="fnref-bfacb466bib106"="">106</a>]. However, in the present work we used standard MEAs because they are light-permissive and tailored well into the transparent perfusion chamber and microfluidic device environment. In addition, the surface of the standard MEAs is flat which allows tight sealing between microfluidic device and MEA substrate and therefore enhanced control over axonal growth direction. Even though, HD-MEAs with flat substrate are not yet available [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib106" id="fnref-bfacb466bib106"="">106</a>], these MEAs can exponentially enhance the spatial resolution of the recorded data [<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib104" id="fnref-bfacb466bib104"="">104</a>, <a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib107" id="fnref-bfacb466bib107"="">107</a>–<a xmlns:xlink="http://www.w3.org/1999/xlink" class="cite" href="#bfacb466bib109" id="fnref-bfacb466bib109"="">109</a>].
5. Conclusion
In the present work we designed and fabricated a transparent double-wall PDMS-based perfusion chamber with air bubble trapping system and integrated it with cell culture media conditioning concept to preserve cultures in ambient environment independent of bulky cell culture incubator and gas supply equipment. This allowed monitoring the network morphology and activity at desired temporal resolutions from seconds to weeks. By integrating a compartmentalized microfluidic device with our perfusion platform and guiding neurites into the microchannel, network activity was available for recording at different spatial resolutions from whole network to sub-axonal levels. As a proof of principle, we show how this integrated system allows to record activity dynamics of the elongating axons inside the microchannels. Our self-contained long-term system with enhanced temporal and spatial access to the network data offer many potentials to obtain robust information from <em="">in vitro</em> neuronal cultures. This versatile platform can be easily modified to serve advanced <em="">in vitro</em> culture models like organ-on-chip and organoids where continuous access to the long-term morphological and functional dynamics of the model needs to be addressed.
Acknowledgments
The author acknowledges scientific inputs by Professor Axel Blau regarding the perfusion system.
Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.
Conflict of interest
Author declares no conflict of interest.
Funding
The author acknowledges funding from Volkswagen Foundation (Freigeist-A110720). This study was partially developed with the financial support from Italian institute of technology (iit), through FLAG-ERA JTC 2015-Graphtivity project.
Ethical statement
All experiments involving the preparation neuronal cell culture were carried out in accordance with the guidelines established by the European Communities Council (24 November 1986) and were approved by the National Council on Animal Care of the Italian Ministry of Health.
<a href="https://content.cld.iop.org/journals/1758-5090/15/2/024102/revision2/bfacb466supp1.mp4?Expires=1728192435&Signature=tLWPwvQDL7ihwVCOG86UmRAVpNyye6apMnOldlgl5yigJN1V-zMcGoUU9ILgs4aNSewCwJgfwnCDIq9RSUZnyIuRqWT32geppUWQ36HUXKD6L3gQw9FGlGd455SsPWA81mCVOmN~zr6cVqHVoDr3wkJoDyLdVmYGbycfYcZzFmk4GB91nZND6lrHHBfbtDZEn7z7FEtKWSBprVYiBxE~jCQOgzRZGwW6gG6wk9aGKBzTfUDVN5ev44R21iBjx5BRCzoM0N5MCL-tHS~kbFSm17u1smvR3Eq-iT4xQ8g8~vHZ8zboZ7uAcv2gUfqItl4NfgLSFK~s4Qo3GgZ40of8Jg__&Key-Pair-Id=KL1D8TIY3N7T8" id="bfacb466supp1"="">Supplementary data_video 1</a> (1.2 MB MP4) Video 1_ 3D animation of the perfusion chamber template (left) and a PDMS/based double-wall perfusion chamber.
<a href="https://content.cld.iop.org/journals/1758-5090/15/2/024102/revision2/bfacb466supp2.mp4?Expires=1728192435&Signature=cruxftLWwoaMhRBwaTDsK3naK8XocdsitawwqizhOiRO2KKSWZjXqcL1~zgaMZhDbFEOBpohjLlWL5xyv8kD9Z1qrNEkOEyeSDScRM3J82zJynAfwgmXjdElvwjvSldadO~haX~FZJp4p3SP~sz1VeAq-JU3o1yOp7pxhklGRjoq0vNdspj0pUxYP21rUGDZF4cdX1CWxQ5fPFzIkwxWUmIWQDGs17tI0v3KHH6scFOYr1qY02SYOfHx5Qj-epJQkHK42DBQzLeoPWbjgkD1G~4ucAujfdKtaKoWtOcYFokdNOJOEj2cS-6g2HWUEwyS7-8nQbNNPh9G3ZXoSR4BJw__&Key-Pair-Id=KL1D8TIY3N7T8" id="bfacb466supp2"="">Supplementary data_video 2</a> (6.0 MB MP4) Video 2_ Time-lapse video of a cortical network morphology during two weeks of perfusion period
<a href="https://content.cld.iop.org/journals/1758-5090/15/2/024102/revision2/bfacb466supp3.mp4?Expires=1728192435&Signature=i19DfpmMBz1AnFZdolxMikVO~eNmw4XviuJlwxnCKspCmiIhl0~eiaZur2q2ptmcb5EhBa9SWKZ1omJiAWWiyaV~AskAuDsq4VissYRQmfEaAZUJSu8yU2nGXqQLHRbTAgH5IbOy56nVnZX~2c7LcNEGA4qakcji1Mt5OrxVnn2i~BC-vLVoNTVMO8jZGAvX~DoXdAYXOSGMldwbkVtGl-tidNg4WAipl5JZr4QQ2yQ~xQTW4urgTKr2I~fT5Cm8SzG-Oydkli2sJQhPveG0nPoJAQ5Ufqyx7pr76V3V4n930eNbv1ZKhs4O3MrVwBShOnyTdZampDU4iMe7RkKYMw__&Key-Pair-Id=KL1D8TIY3N7T8" id="bfacb466supp3"="">Supplementary data_video 3</a> (6.4 MB MP4) Video 3_ Axonal branches were tracked in two microchannels.