A mesh microelectrode array for non-invasive electrophysiology within neural organoids

Organoids are emerging in vitro models of human physiology. Neural models require the evaluation of functional activity of single cells and networks, which is best measured by microelectrode arrays. The characteristics of organoids clash with existing in vitro or in vivo microelectrode arrays. With inspiration from implantable mesh electronics and growth of organoids on polymer scaffolds, we fabricated suspended hammock-like mesh microelectrode arrays for neural organoids. We have demonstrated the growth of organoids enveloping these meshes, their cultivation for at least nine months, and could measure spontaneous electrical activity within organoids. Our concept should enable a new class of microelectrode arrays for in vitro models of three-dimensional electrically active tissue.


Introduction
Neural organoids show promise as physiologically relevant in vitro models of the human brain. 1 Electrophysiology is critical to demonstrate functionality of translational models of the central nervous system. 2 Electrical activity has been observed in organoids by adapting existing tools ( Figure 1B-F), 25 however, there remains an unmet need for more appropriate tools and methods specific for electrophysiology of organoids. 3 What is needed for organoid electrophysiology? Scientists may have different demands: Can we extract relevant information using only a handful of electrodes? 4 Can we measure the activity of all neurons in an organoid, 5 and would this be useful? How does the activity of organoids relate to human 30 physiology? 1 Should we consider new ethical concerns? 6 In fact, current knowledge of the functional electrical activity within organoids does not allow these questions to be addressed and thus we suggest a new technical approach to tackle them.
Neural organoids are on a spectrum between conventional cell culture and a living brain. Research tools are available for both ends of this spectrum, yet are not ideal for organoids. In conventional cell 35 cultures, cells adhere onto planar microelectrode arrays (MEAs). Typical in vitro models relate action potentials and bursting activity to cellular or synaptic effects of applied compounds. In contrast, animal models attempt to resolve the activity of many single neurons over an extended period of time, and to relate it to parameters such as behavior.
In either case, microelectrodes may be sparsely distributed to efficiently capture independent activity 40 or may be densely packed to capture all activity at a high resolution. In the simplest in vitro models, network activity can be captured by only a few electrodes, yet many independent samples are needed to measure small effects. 4 An ultimate vision in neuroscience is to record all cells in the brain at millisecond resolution in vivo. 5,7 Capturing the activity of single cells requires close proximity to electrodesand tracking cells over 45 time requires a stable structure. Therefore, an alternative to conventional methods of growing freely floating organoids is needed. Organoids can be grown on 2D-MEAs for several months, 8 but they flatten as neurons migrate and spread on the MEA surface ( Figure 1C). CMOS MEAs can capture activity in tissue at a high resolution and detect activity to a depth of 100 µm. 9 The ability to resolve internal activity similar to electroencephalography (EEG) cannot be assumed in organoids, as EEG 50 measures coordinated activity in structured cell layers. 10 Deeper activity can be recorded by MEAs with protruding microelectrodes ( Figure 1D) and slicing of organoids 11 or implantable neural probes ( Figure 1E). In addition to the technical challenge of implantation, tissue damage limits recordings to a single endpoint. 12 Beyond such electrical methods, optical methods can record superficial activity ( Figure 1F) but with lower sensitivity. 13 55 To move beyond these limited approaches, we were inspired by the growth of organoids on artificial structures 14 and demonstrations of implantable mesh electronics with cellular or subcellular structure sizes. 15,16 We believe that the combination of these approaches should enable guided growth of neural organoids with non-invasive electrophysiology ( Figure 1A).
A similar device has been demonstrated for acute experiments and suggested for organoids. 17 60 Superficial electrophysiology of cardiac spheroids has been demonstrated with a flexible self-rolling (but not mesh) device. 18 In this work, we present a mesh microelectrode array specifically designed for physically suspended growth and long-term electrophysiology of neural organoids ( Figure 1A). With a hammock-like artificial structure, the device suspends organoids far from solid surfaces to enable their unimpeded growth. 65 Our successful recording of spontaneous extracellular action potentials supports the use of such devices for non-invasive electrophysiology of activity within and throughout neural organoids.

Design
In designing the device, we prioritized compatibility with organoid culture methods, minimization of 70 technical complexity and material choices for long-term device stability. We adopted methods previously used for fabrication of perforated MEAs. 19 Microelectrodes (30 µm diameter) were distributed across a suspended region with a diameter of 2 mm. Minimum dimensions of 4 µm (lines and spaces) ensured a high fabrication yield. We focused on a design compatible with 256-channel amplifiers from Multi Channel Systems MCS GmbH (Reutlingen) with minimal complexity of electrical 75 packaging.

Fabrication
Our devices consist of flexible microfabricated meshes assembled into glass-bottomed wells ( Figure  2). The meshes were produced on carrier wafers using cleanroom microfabrication. 19 Electrical paths (Ti/Au/Ti, 400 nm) were insulated on both sides by 6 µm of polyimide (DuPont PI-2611). Metals were 80 deposited by sputter coating and structured by plasma etching. Polyimide (PI) was structured by plasma etching against a silicon nitride hard mask. Microelectrodes (30 µm diameter) of titanium nitride (TiN, 500 nm) were structured by sputter coating and lift-off using a photoresist mask. After release from carrier substrates, we assembled the meshes into wells comprising multiple machined polymer and glass components using EPOTEK 301-2SL ( Figure 2E). Glass components were 85 pretreated with aminopropyltriethoxysilane before gluing.

Cell and organoid culture
Here, we present results of two types of neural organoids generated from human induced pluripotent stem (iPS) cells. Cells were verified pluripotent and contamination-free, confirmed negative for mycoplasma, and pluripotent stem cells were maintained feeder-free. Detailed characterization of 90 these organoids is ongoing and will be reported elsewhere.
Cerebral organoids (Figure 3) were generated following the protocol published by Lancaster et al. 20 Human iPS cells were obtained in house via OSKM reprogramming. 21 Briefly, at day 0, 7000 cells from single cell suspensions were plated in 96-well U-bottom low attachment plates to generate embryoid bodies. After 6 days in vitro, media composition was changed to induce the formation of anterior 95 neuroectoderm. At day 10 the neuroectodermal aggregates were embedded in Matrigel to provide them with a scaffold that would allow for further differentiation, maturation and growth. We transferred individual organoids onto mesh MEAs by pipetting. For cerebral organoids, low media volumes for the first 5 days on the mesh kept the organoids at the air-liquid interface to facilitate enveloping of the mesh. Later, full media changes covering the entire organoid were performed every 110 other day. Before placing our second type of organoids, MEAs were coated with poly-D-lysine to promote cell adhesion: MEAs were treated with air plasma, incubated with 1 mg/ml poly-D-lysine for 1 h, then rinsed three times with saline. Media was changed every other day and maintained to keep organoids at the air-liquid interface.

115
Spontaneous electrical activity was recorded by a multichannel amplifier (USB-MEA256 from Multi Channel Systems MCS GmbH) at a sampling rate of 40 kHz. Recordings were high-pass filtered (200 Hz second-order Butterworth) and action potentials were identified by threshold detection.

120
The devices (Figure 2) contain four wells with diameters of 7 mm. Each well can support a single organoid on a central spider-web-like mesh with a diameter of 2 mm and 61 microelectrodes. The mesh has four concentric ring filaments (radii of 0.25, 0.5, 0.75 and 1 mm) containing 8, 12, 16 and 24 electrodes, and one electrode at the center of the mesh. The rings are suspended 2 mm above the glass bottom. Eight radial filaments provide structural support and electrical connections to the 125 electrodes. The filaments have a thickness of 12 µm and width of 20 µm. The width of the radial filaments increases to 80 µm to accommodate all electrical connections. Outside of the wells, the electrical paths extend to gold pads at the device perimeter, which are contacted directly by the amplifier headstage.
We designed this device to allow unrestricted growth in all directions. In the suspended plane of the 130 mesh, 85 % of the central 2 mm circle is open space. With its thickness of 12 µm, the artificial structure would occupy only 0.1 % of the total volume of a spherical 2-mm-diameter organoid. The impact on the development of such an organoid should be greatly reduced in comparison to organoids grown on planar MEAs. Suspending the organoid prevents its adhesion on solid surfaces, thereby reducing the need for mechanical shaking. Nutrients and oxygen can be delivered from all sides, 135 although perfusion could improve delivery. No flow was used in the presented results.
We designed our mesh to support microelectrodes in a fixed arrangement for simple placement of organoids and electrode localization even when embedded in tissue. However, the materials used to produce the mesh would allow flexibility or stretching if modified to include serpentine structures. 22 Our microelectrodes are connected by unbroken thin film metal paths rather than complicated electrical 140 packaging. 23 We have observed no problems of leakage or technical instability, even after cultivation of organoids over a period of 9 months.

Organoids
After their placement on a mesh MEA, organoids enveloped the filaments and continued to grow. Figure 3 shows a cerebral organoid which had grown on the mesh for ten days. The decentered 150 placement of this examplealthough unintentionalrevealed how an organoid can grow on filaments of a mesh MEA. Thin regions of tissue reveal the embedded microelectrodes, while thicker regions hide the filaments and microelectrodes. We have cultivated organoids on these devices for longer than nine monthsand counting! Figure 3: Organoids growing on a mesh MEA. This 40-day-old neural organoid showed outgrowth of cells along the polymer filaments after ten days on the mesh. Thin tissue reveals embedded microelectrodes (black circles in B). Images were taken from above.
A limitation of our design is the large open spaces peripheral to the central electrode field. Adhesion of organoids after placing them on meshes is not immediate, but requires the organoids to be returned to an incubator. Even careful handling may cause some organoids to fall off of their meshes. Although readily solved by closing these regions with structural filaments, a user-friendly solution such as well 160 inserts may simplify handling. 25 We have demonstrated the continued growth of organoids after their placement on mesh MEAs. We envision that growth from earlier states (embryoid bodies or stem cells) may be possible. However, this will require additional adaptation of biological protocols.

165
A neural organoid was placed on the mesh at an age of 197 days. Recordings of spontaneous activity were performed 35 days later (day 232). We observed spontaneous activity on most microelectrodes across a distance of 2 mm in the organoid (Figure 4). We present this result as a demonstration of the device's functionality. Signal amplitudes decrease with increasing distance between electrodes and active cells, 9 and our measured amplitudes suggest that active cells are not directly adjacent to the 170 microelectrodes. Time-dependent studies to compare cell types and morphologies with electrophysiology will be possible by histology of organoids containing embedded mesh microelectrode arrays. 16

Conclusions
We have demonstrated biohybrid neural organoids and the first microelectrode array device designed 175 for long-term, non-invasive electrophysiology within neural organoids. With this device, we hope to contribute to a better understanding of functional activity in organoids.
Full benefits of organoid electrophysiology may be obtained by considering both the device design and organoid culture techniques. Methods which place organoid precursors such as embryoid bodies or stem cells may require modified devices. Assembly of organoids of various brain regions or other 180 tissues may be enabled by the development of specific designs. Future implementations may balance the numbers of wells and microelectrodes for targeted applications. Further technical advancements could include stretchable meshes, mechanical or chemical sensors, or three-dimensional meshes.

Conflicts of interest
There are no conflicts of interest to declare. 185 7 References