High-density Multielectrode Array Recordings of Retinal Waves Using An Electrophysiology Platform

Summary

High-density multielectrode arrays (HD-MEAs) are used to study spontaneous retinal waves, which play a crucial role in neural circuit development. This protocol outlines the steps for preparing mouse retinal tissue and performing electrophysiological recordings using HD-MEAs on an electrophysiology platform.

Abstract

Spontaneous retinal waves are a hallmark of retinal network activity during development, playing a crucial role in the formation of the visual system by influencing the refinement of axons, permeability of vasculature, and overall maturation of neural circuits. These waves are commonly studied in ex vivo retinal preparations using multielectrode arrays (MEAs), which enable electrophysiological recordings of large populations of retinal ganglion cell (RGC) activity. MEA-based electrophysiology has become a powerful tool due to its ease of use to rapidly collect high-throughput data, thus making it ideally suited to study retinal activity in a variety of experimental conditions.

In this protocol, we outline the critical steps for preparing retinal tissue for the acquisition of electrophysiological data using a High-Density MEA (HD-MEA) on an electrophysiology platform. The process begins with the careful isolation of intact retinas from neonatal animals under physiological conditions. Once prepared, the retina is carefully mounted onto an HD-MEA chip, which consists of a grid of 26,400 electrodes capable of performing simultaneous extracellular recordings from at least 1,000 RGCs. Recordings can last up to several hours. Ultimately, this methodological approach offers valuable applications in investigating retinal development, disease, and potentially cross-species comparative studies, contributing to broader advancements in neuroscience and vision research.

Introduction

Spontaneous retinal waves are periodic bursts of correlated activity observed in the developing retina before the onset of vision. In mice, the circuits that initiate and propagate retinal waves change rapidly during development, starting embryonically and ending at eye opening (postnatal day 14)¹. As retinal circuits develop, the spatiotemporal properties of retinal waves change dramatically^2,3. Several studies support that those specific spatiotemporal properties instruct the development of the visual system: asynchronous activity between the eyes instructs eye-specific segregation⁴, wave area size instructs the refinement of retinotopic axons in binocular brain regions^5,6, and the propagation direction of waves has been implicated to instruct the direction selectivity circuits in the superior colliculus³. Beyond neural circuits, retinal waves are responsible for the development of the permeability of the blood vasculature⁷. Additionally, it was discovered that the stage II retinal waves control the outgrowth of receptive-field areas and stabilize it in retinal ganglion cells (RGCs)⁸. Critically, aberrant retinal waves during development could be the basis of various neurodevelopmental disorders, and indeed, retinal waves are abnormal in a mouse model of congenital nystagmus³. Given their critical role in early visual system development and implications for disease, there is a need a to record the spatiotemporal dynamics of retinal waves at any age and in any animal model.

Several methods have emerged to study retinal waves. Both single-cell electrophysiology^9,10 and calcium imaging^{11,12,13,14,15} have led to seminal findings about circuits and the function of retinal waves. Here, we focus on multielectrode array recordings (MEA), a method that has been used since the early investigation of retinal waves¹⁶ and has continued to improve as a technique². MEAs enable simultaneous extracellular recording of action potentials from hundreds to thousands of RGCs, allowing researchers to track wave initiation, propagation, and termination similarly to what is possible with calcium imaging. Since MEAs directly record action potential from RGCs, MEAs can be used to study various genetic mouse models or non-model organisms without the added complication of introducing a calcium indicator. Tissue can also be cultured directly in the MEA, enabling very long recording times. MEAs are also highly scalable, with current products enabling up to six active MEAs at once, which could enable rapid assessment of retinas and catalyze drug discovery. Perhaps the biggest strength of MEAs is their ability to sample many cells at a high sampling rate, which enables laboratories across the world to quickly obtain rich datasets but requires expertise in postprocessing.

The primary goal of this protocol is to provide a detailed and reproducible protocol for preparing retinal tissue and performing electrophysiological recordings using a single-well High-density MEA (HD-MEA) system to study spontaneous retinal waves ex vivo. This method ensures high-throughput and long-duration data acquisition, making it suitable for a wide range of developmental and disease-related studies. The new advanced CMOS-based HD-MEA system, which allows >1,000 electrode sites at once, offers several advantages over traditional electrophysiological approaches. Its high-density electrode array (26,400 electrodes) enables precise, simultaneous extracellular recordings from up to 1,012 RGCs, capturing fine-scale spatial activity patterns. The single-well design provides uniform recording conditions¹⁷. Additionally, this system allows for long-term recordings lasting several hours, making it possible to study wave dynamics under various physiological and pharmacological conditions without significant signal degradation¹⁸. By providing a protocol, this study aims to make HD-MEA technology accessible to a broader audience of researchers in neuroscience, ophthalmology, and vision science. This method represents a significant step forward in the study of retinal waves, offering new avenues to study early visual system development, disease mechanisms, and potential therapeutic interventions.

Protocol

This protocol outlines the steps for isolating and preparing retinal tissue from neonatal mice for MEA recordings of retinal waves using Maxwell Biosystems HD-MEA platform. The procedure is designed to preserve physiological conditions, ensuring the retinal tissue remains structurally intact, free from damage, and properly prepared for optimal electrode contact. These experiments were approved by the Vanderbilt Animal Care and Use Program, under protocol number M2200056-00. The mice (1-2-week-old, both sexes) were housed in a 12 h day/night cycle vivarium and fed a regular chow diet.

1. Preparation of retinal tissue

1. Isolation of the retina
   NOTE: See the Table of Materials for details about the materials, equipment, and reagents used in this protocol (also see Figure 1).
   1. Prepare the workspace
      1. Prepare artificial cerebrospinal fluid (aCSF):125 mM NaCl, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, bubbled with carbogen: 95% O₂/5% CO₂ (pH 7.4, osmolarity 290 ± 10 mOsm) and store it at 5 °C. Keep the aCSF on ice while it is being oxygenated for 10 min before dissection.
         NOTE: The CaCl₂ should be added last, after the solution has been bubbled with carbogen for 10 min.
      2. Begin oxygenation of fresh, ice-cold aCSF (for at least 10 min).
      3. Set up dissection tools and microscope.
   2. Euthanize the animal according to institutional guidelines.
   3. Enucleate the eyes using curved spring scissors-
      1. Since postnatal pups have closed eyes until P11-13 (in mouse) and the lids need to be carefully opened, use micro forceps to grip the eyelids and spring scissors to remove them. Carefully take out the eyes using spring scissors, avoiding ocular damage.
      2. Cut the optic nerve and place the eyes in oxygenated aCSF. Use a transfer pipette to gently move the eye between the Petri dishes.
         NOTE: For the rest of the dissection protocol, refresh with oxygenated aCSF every 10 min.
   4. Under a dissection microscope,
      1. Make a small incision in the cornea using a needle while stabilizing the eye with forceps.
      2. Bluntly dissect the cornea by gently tearing along the scleral perimeter.
      3. Remove the iris and lens. If the retina is attached to the lens in young animals, gently separate it using forceps.
         NOTE: In young animals, the retina is sometimes attached to the lens. Use the forceps to gently pull the lens from the retina using one pair of forceps to pull on the lens and the other resting on the outer edges of the retina to keep the retina in place while the lens is being removed.
   5. Orient the retina.
      1. Identify the ventral side using the choroid tissue (see Sondereker et al.¹⁹ for a detailed protocol on orienting) and position it facing the researcher with the RGC layer facing up.
      2. Use a curved scalpel to make a reference cut 45° counterclockwise from the ventral axis.
      3. Using fine forceps, gently separate the retina from the retinal pigmented epithelium (RPE) and sclera.
         NOTE: In younger mice, the RPE easily peels from the retina. In older mice, the RPE is more attached to the retina. If, after removing the RPE from the majority of the retina, the RPE is still anchored to the retina via the optic nerve, use spring scissors to sever that connection. This prevents damage to retinal tissue near the optic nerve.
   6. Neonatal retinas have almost no vitreous body, but older retinas do (appears as a brown filament along the rim of the retina, but the fibers connect to one another within the vitreous humor as well). Using the fine forceps, remove the vitreous tissue that lies above the RGC layer.
      NOTE: This fibrous layer sometimes comes off when the lens is removed. If it did not, it is important to remove as much of it as possible, or it will prevent proper contact of RGCs with electrodes.
   7. Using a scalpel, make four relief cuts to help the retina lay flat in the dish. To know and maintain orientation later on, trim the two corners of the ventral quadrant with a scalpel.

2. Mounting the retina on the MEA chip

1. Mounting the retina-
   1. Gently transfer the retina from the Petri dish to the MEA chip using the transfer pipette with the tip cut off. Make sure there is enough aCSF in the chip to cover the bottom of the well.
   2. Use the single-hair brush to gently maneuver the retina to be positioned above the electrodes, RGC layer facing the electrodes.
   3. Using the 10 µL pipette, gently remove extra aCSF around the retinal tissue. Tilt the chip at an angle to get rid of extra aCSF.
      NOTE: Never touch the electrodes or amplifiers with the pipette tip. As aCSF is removed, the retina will flatten more. If any part of the retina is folded underneath itself, use the single-hair brush to gently scoop under the desired leaflet to unfold it. To prevent the retina from moving while unfolding any part of the leaflets, a second single-hair brush may be used to apply gentle pressure in the center of the retina, where the optical nerve connects.
   4. Dry the well as much as possible by tilting the chip at a 45° angle and removing the last of the aCSF that pools in the tilted portion of the chip.
      NOTE: This will increase the connection between the retina and electrodes.
   5. Use filter paper to gently dab and dry the edges of the retina. Place the MEA chip in a recording chamber.
2. Enhancing electrode contact-
   1. Use a replaceable tissue holder insert to gently push the retina onto the electrodes, ensuring optimal contact and improving the signal-to-noise ratio.
      NOTE: The tissue holder insert should stop being lowered as soon as it makes contact with the retina (a clear "wet spot" will be visible at the point of contact). Further lowering of the electrode holder risks damaging retinal tissue. Quickly transition from step 2.1.4 to 2.2.1 to minimize the duration the retina remains without oxygenated aCSF; this step is critical and should be completed as swiftly as possible. It is important to move fast when flattening the retina on the chip, however, it is also important that most of the aCSF from the dissection be dried out so that the retina makes good contact with the electrode. Going too fast to get the retina under oxygenation risks not having good contact with recording electrodes, and going too slow risks retinal cells becoming hypoxic and dying. We find that 30-60 s leads to good contact and healthy retinas (i.e., we observe typical retinal waves).
   2. Using the referenced perfusion system, begin filling the MEA chip well with oxygenated aCSF at a flow rate of 2 mL/min.
      NOTE: An open perfusion system is recommended, as the electrical noise remains low and it provides fresh aCSF to the retina. Be sure an in-line heating system is attached to the perfusion setup to warm the aCSF to physiological conditions.
   3. Turn on the in-line heater of the perfusion system to warm the aCSF to 32-34 °C.

3. Electrophysiological recordings

1. Extracellular recordings-
   1. Switch on the recording system.
   2. On the acquisition computer, double-click first on the Server icon and then on Scope to open it.
   3. Click on the Initialize icon in the Scope interface and enter the MEA chip ID to begin extracellular recordings using the MEA system.
   4. Click on the Offset icon in the Scope interface to set up the offset.
   5. Adjust the gain and filter settings to optimize signal detection. Use the default settings of 512 gain and a highpass of 300 Hz; set the sampling rate at 20 kHz.
   6. Select the appropriate recording configuration. The total active area that can be sampled is 3.85 x 2.1 mm². The software lets users choose how to configure the electrode spacing; to follow this protocol, use the default setting that equally spreads out the recording electrodes over the maximum active area, as this samples across as much of the retina as possible. This leads to a spacing of 87.5 µm per recording electrode.
   7. Allow the retina to acclimate to the recording chamber for 60 min.
   8. Begin spontaneous activity recordings from RGCs by navigating to the Assay tab and selecting the Create New Assay icon. Choose the Network assay type, assign an appropriate filename, and click OK to save the configuration. Set the desired duration for the recording by adjusting the Record time (sec) parameter. Enable the Only Spike option to focus on spike data, then click the Run Assay icon to begin the recording session.
      NOTE: For recordings of retinal waves of 10 min and longer, we recommend using the spikes only setting, which saves only the timestamps of action potentials using a built-in spike detection tool instead of the full raw electrophysiological trace per channel. This is because lengthy recording (>10 min) across 1,024 channels can easily result in TB-sized data files that are technically challenging to analyze and drastically slow research progress. We have confirmed that raw data using spikes times only replicates the known spatiotemporal properties of retinal waves: for stage II waves, we observe large waves that cover most of the retina and occur at frequencies that match previous reports (see Figure 2). We do note that if the research project requires more in-depth analysis of spike waveforms, then recording the raw electrophysiological trace is a necessity.
   9. Periodically monitor the ongoing recording to ensure data acquisition is proceeding as expected and the system is functioning properly.

4. Clean-up procedure

1. Once the recording is complete and the data have been saved, switch off the in-line heater and then turn off the perfusion system.
2. Carefully remove the tissue holder insert and extract the MEA chip. Clean the chip by rinsing it first with 70% ethanol, followed by deionized (DI) water.
3. To prevent clogging of the tubing, flush the perfusion system with DI water for 15 min.

Results

High-throughput recordings and analysis of retinal waves with HD-MEAs
We performed an hour-long HD-MEA recording of spontaneous retinal waves (Figure 2). A raster plot of neuronal activity shows the structured pattern of retinal waves, where each dot represents a detected action potential from an individual electrode (Figure 2A, bottom). Summing activity across electrodes results in a single trace, which makes it easier to visualize the timing of waves over the 1 h recording (Figure 2A, top). A heatmap representing average neuronal activity demonstrates that we can sample over large areas of the retina (Figure 2B). We performed analysis of the HD-MEA recording data to analyze spatiotemporal properties of waves, showing wave initiation sites (Figure 2C), inter-wave interval (IWI) (Figure 2D), wave duration (Figure 2E), wave area (Figure 2F), and a representative vector flow field of one wave (Figure 2G). To demonstrate the robustness of this methodology in capturing neuronal network activity in healthy retinas for longer hours, we also performed a 3 h HD-MEA recording of spontaneous retinal waves (Figure 2H). These results validate the protocol's ability to capture neural activity across large areas of the retina and detect spontaneous retinal waves, demonstrating the reliability of the MEA system in detecting and analyzing neural dynamics in the developing retina.

Figure 1: Workstation for preparation of retinal tissue. (A) Dissection tools used for isolation of the retinal tissue. 1. Two fine-tipped forceps; 2. Needle (30 G x ½); 3. Curved blade scalpel (#10); 4. 3 mL disposable transfer pipette (with end cut off); 5. Single-hair brush (a small paintbrush can be cut so there is just one center hair left); 6. Whatman filter papers (#1), cut into small pieces; 7. Roboz micro Adson forceps (4.75" long, 1 x 2 teeth, 0.5 mm tip); 8. Roboz spring scissors (10 mm cutting edge, 0.15 mm tip width, 3¾" overall length); 9. Petri dish for dissection (35 mm or 60 mm). (B) Dissection microscope used for the isolation of the retinal tissue. See Table 1 for more details. Please click here to view a larger version of this figure.

Figure 2: High-throughput recordings and analysis of retinal waves with HD-MEAs. (A) Bottom: raster plot of action potentials recorded from individual electrodes (channels) over 1 h of HD-MEA recording. Each dot represents an individual detected AP. Top: Overall firing rate across all recorded electrodes over 1 h of HD-MEA recording, calculated by binning APs/s across all recorded electrodes. (B) Heatmap of average retinal activity (APs/s). (C) Wave initiation sites. (D) Inter-wave interval analysis. (E) Wave duration. (F) Wave area. (G) Representative vector flow field of one wave. (H) Raster plot and sum activity for a very long-term HD-MEA recording (3 h) to demonstrate the robustness of the recordings. Spatial resolution: 87.5 x 87.5 µm²/pixel coverage. Temporal resolution: 20 kHz sampling rate. Abbreviations: AP = action potential; HD-MEA = High-density microelectrode array; N = Nasal; T = Temporal; D = Dorsal; V = Ventral; IWI = Interwave interval. Please click here to view a larger version of this figure.

Discussion

The protocol described here provides a reproducible and high-throughput method for preparing retinal tissue and performing HD-MEA recordings, offering a robust method to study retinal network activity. HD-MEA technology offers significant advantages over traditional electrophysiological and imaging techniques, particularly in capturing high-throughput data. HD-MEA delivers real-time, millisecond-scale accuracy recordings of spontaneous retinal wave dynamics, enabling precise wave initiation, propagation, and synchronization characterization. Additionally, HD-MEA surpasses patch-clamp methods by allowing simultaneous recordings from thousands of retinal ganglion cells (RGCs), providing a view of network activity. This scalability, combined with its ease of use and long-term recording stability, makes HD-MEA an invaluable tool for developmental studies. The protocol can be adapted for the MEA recording for rat and primate retinas with a few extrapolations. Lastly, the high-throughput nature of HD-MEA recordings has potential for drug discovery, where its high electrode density facilitates precise monitoring of drug-induced changes in retinal activity.

One of the most critical steps in this protocol is the proper isolation and mounting of the retina. Ensuring the retina is intact and free from tears or holes is essential for obtaining high-quality recordings²⁰. Perform the dissection and preparation quickly to maintain tissue viability. Carefully inspect the retina under a dissection microscope to make sure the retinal surface is clean and smooth: the RGC layer should be clean and free of debris to facilitate good electrode contact. Use fine forceps to clear debris or leftover vitreous if needed. The removal of the vitreous humor is particularly important, as residual vitreous can impede electrode contact and reduce the signal-to-noise ratio (SNR) of the recordings²¹. Confirm retinal integrity by checking for tears or holes under the microscope (except for the potential small hole where the optic nerves used to be). Avoid excessive stretching or folding of the tissue, and do not touch the RGC side of the retina. Handle the retina gently using fine forceps or a single-haired brush to avoid damage. Ensure the retina remains flat and intact, as damage can affect wave propagation and recording quality. Make sure the retina is flat on the electrodes; if activity is observed only on the edges, it indicates poor electrode contact with RGCs. Additionally, mounting the retina with the ganglion cell layer (GCL) facing the electrodes is crucial for capturing RGC activity. For retinas with strong curvature, the use of filter paper to flatten the tissue can improve electrode contact and recording quality²².

Maintain the retina in oxygenated aCSF with 95% O₂ and 5% CO₂ at all times to prevent hypoxia at physiological temperatures and ensure stable recordings over long periods²³. Use freshly prepared aCSF and adjust pH to 7.4 (by bicarbonate and oxygenation) and osmolarity to ~290 ± 10 mOsm. Maintain the tissue in cold aCSF during dissection and at 32-34 °C during recording. Use a perfusion system to continuously supply oxygenated aCSF during MEA recording. Monitor the retina for signs of degradation (e.g., swelling, discoloration, as well as the stable spike waveforms, consistent wave frequency²) and replace the aCSF periodically to maintain tissue health. The Maxwell HD-MEA system comes with a tissue holder that is placed over the retina to keep it on the electrodes, but uneven pressure causes activity only on the edges. Careful maneuvering of the tissue holder insert might be helpful. Make sure the vitreous is completely removed and the retina is flat on the electrodes. In case of poor electrode contact and low signal quality, check the oxygenation and temperature of the solution to be sure that the retina is healthy and intact. Check for broken/loose connections with ground wires, clogged tubing, unsuitable perfusion rate, and dirty pins on the MEA before starting the experiment. Clogged tubes, connectors, and dirty pins on the MEA can be washed using 70% ethanol and dried in advance.

Of note, these experiments can be done in light or dark conditions with negligible differences in spatiotemporal properties of retinal waves. Much of the classic work helping us understand retinal waves was done using single photon imaging, thus under light conditions^9,10,12. The spatiotemporal properties of retinal waves from these light recordings were replicated in MEA recordings in the dark² and in vivo recordings^13,14,15. As mentioned above, photoreceptors do not make functional synapses with bipolar cells until P10, and stage II and III retinal waves do not respond to ambient light¹¹. Given the minimal impact of light on retinal waves, the advantages of being able to set up these HD-MEA systems on benchtops and not light-controlled rooms decreases the barrier of entry.

Despite its advantages, this method has some limitations. The biggest limitation is that this technique can only be used to perform ex vivo recordings, not in vivo. Thus, this technique is unable to resolve whether brain states are important for retinal activity during development. Another limitation is that although HD-MEAs can sample across large parts of the retina, they are limited in their ability to record from deeper retinal layers, such as bipolar or photoreceptor cells²¹. Finally, the high cost of HD-MEA systems and the need for specialized expertise in data analysis may limit their accessibility²².

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
3 mL disposable transfer pipette (with end cut off)	Fisherbarnd	13-711-9CM	Assists in moving retina between petri dishes.
Artificial cerebrospinal fluid (aCSF)			Maintains physiological conditions for retinal tissue; composed of NaCl, KCl, NaH2PO4, NaHCO3, CaCl2, MgCl2, glucose, and bubbled with carbogen.
CaCl2	Fisher Chemicals	C79-500	To prepare artificial cerebrospinal fluid (aCSF)
Carbogen supply (95% O2, 5% CO2)			Used to oxygenate aCSF and maintain tissue viability.
Curved blade scalpel (#10)	Integra	4-110	Used to cut tissue with precision.
Dissection microscope (stereoscope)	Zeiss	Stemi 508	Essential for visualizing and handling retinal tissue.
glucose	Fisher Chemicals	BP350-1	To prepare artificial cerebrospinal fluid (aCSF)
In-line heater	Multichannel System	TC02	Warms aCSF to 32-34°C for optimal conditions.
Ismatec Perfusion System	Ismatec	ISM4208	Maintains continuous flow of oxygenated aCSF.
KCl	Fisher Chemicals	P271-500	To prepare artificial cerebrospinal fluid (aCSF)
KH2PO4	Sigma Aldrich	P5504-100g	To prepare artificial cerebrospinal fluid (aCSF)
MaxOne Recording Unit	Maxwell Biosystems	MX1-BRD	Interface between MaxOne Chip and System
MaxOne System	Maxwell Biosystems	MX1-SYS	Core system for MEA-based electrophysiological recodrings.
MaxOne Tissue Holder with a 3-axis micromanipulator and replaceable inserts	Maxwell Biosystems	MX1-HLD	Ensures precise placement of the retina on the MEA chip.
MEA chip (MX1-S-CHP, MaxWell Biosystems)	Maxwell Biosystems	MX1-S-CHP	High-density microelectrode array for recording neuronal activity.
MgCl2	Fisher Chemicals	M33-500	To prepare artificial cerebrospinal fluid (aCSF)
NaCl	Fisher Chemicals	S271-1	To prepare artificial cerebrospinal fluid (aCSF)
NaHCO3	Fisher Chemicals	S233-500	To prepare artificial cerebrospinal fluid (aCSF)
Needle (30 G x ½)	BD Biosciences	305106	Helps make incisions in cornea.
Neonatal animals (P1-P14; mouse)			Model organisms such as mice, rats, or others used for retinal studies.
PC with MaxLab Live Scope software	HP	Z4	Used for data acquisition and analysis of recordings.
Petri dish (35 mm or 60 mm)	Pyrex	3483E12	Used as a workspace for dissection.
Roboz micro Adson forceps (RS-5232, 4.75” long, 1 x 2 teeth, 0.5 mm tip)	Roboz	RS-5232	Specialized forceps for fine dissection.
Roboz spring scissors (RS-5671, 10 mm cutting edge, 0.15 mm tip width, 3¾" overall length)	Roboz	RS-5671	Precision scissors for cutting delicate tissue.
Single-hair brush			Small paintbrush modified to a single hair for handling delicate tissue.
Two fine-tipped forceps	Roboz	RS-5060	Used for delicate tissue handling.
Whatman filter papers (#1), cut into small pieces	GE Healthcare	1001-042	Used for handling and drying retinal tissue.