Here, we present a unique, 3D-printable implant for rats, named TD Drive, capable of symmetric, bilateral wire electrode recordings, currently in up to ten distributed brain areas simultaneously.
Intricate interactions between multiple brain areas underlie most functions attributed to the brain. The process of learning, as well as the formation and consolidation of memories, are two examples that rely heavily on functional connectivity across the brain. In addition, investigating hemispheric similarities and/or differences goes hand in hand with these multi-area interactions. Electrophysiological studies trying to further elucidate these complex processes thus depend on recording brain activity at multiple locations simultaneously and often in a bilateral fashion. Presented here is a 3D-printable implant for rats, named TD Drive, capable of symmetric, bilateral wire electrode recordings, currently in up to ten distributed brain areas simultaneously. The open-source design was created employing parametric design principles, allowing prospective users to easily adapt the drive design to their needs by simply adjusting high-level parameters, such as anterior-posterior and mediolateral coordinates of the recording electrode locations. The implant design was validated in n = 20 Lister Hooded rats that performed different tasks. The implant was compatible with tethered sleep recordings and open field recordings (Object Exploration) as well as wireless recording in a large maze using two different commercial recording systems and headstages. Thus, presented here is the adaptable design and assembly of a new electrophysiological implant, facilitating fast preparation and implantation.
The multi-area nature of brain interactions during wake and sleep makes it difficult to exhaustively study the ongoing physiological processes. While approaches such as functional MRI (fMRI) and functional ultrasound (fUS) allow sampling of brain activity from whole brains1,2, they exploit neurovascular coupling to infer brain activity from hemodynamic activity, limiting their temporal resolution2. In addition, fMRI requires the placement of the research subject in an MRI scanner, prohibiting experiments with freely moving animals. Optical imaging of calcium dynamics with single or multiphoton imaging enables cell type-specific recordings of hundreds of neurons simultaneously3. However, head-mounted microscopes such as the Miniscope3, which do allow freely moving behavior, are usually limited to imaging superficial cortical areas in intact brains4. While the diameter of their field of view on the cortex can be in the order of 1 mm, the space requirements of these head-mounted microscopes can make it difficult to target several, especially adjacent, areas. Therefore, to capture multi-area brain dynamics in wake and sleep accurately, extracellular electrophysiology, recorded with electrodes implanted in the brain areas of interest, is one of the methods of choice due to its high temporal resolution and spatial precision5. In addition, it allows the characterization of sleep dynamics in animals compatible with analyses obtained from human EEG, increasing the translational value of this method6.
Classically, studies recording brain activity with extracellular electrodes have employed individual wire electrodes or electrode bundles, such as tetrodes7. State-of-the-art probes such as the Neuropixels probe8 allow targeting several areas simultaneously, given that they are aligned on an axis that allows implanting the probe along that axis without impairing the animal. However, accurate simultaneous recordings of multiple, spatially separated areas still remain challenging, with existing methods being either costly or time-intensive.
In recent years, additive manufacturing methods such as stereolithography have become broadly available. This allowed researchers to develop novel electrode implants that were adaptable to their experimental requirements9, for example, simplified repeatable targeting of multiple brain areas. Frequently, these implant designs are also shared with the academic community as open-source hardware, allowing other researchers to adapt them to their own purposes. The degree of adaptability of specific implants varies both as a result of how the implant is designed and how it is shared. Parametric modeling10 is a popular approach in computer-aided design, in which different components of the design are linked by interdependent parameters and a defined design history. Implementing a parametric approach for designing implants increases their reusability and adaptability10, as changing individual parameters automatically updates the complete designs without the need for complex re-modeling of the design. A consequential necessity is that the design itself is shared in an editable format that preserves the parametric relationships and design history. File formats that only represent geometric primitives, such as STL or STEP, make subsequent parametric modifications of published models unfeasible.
While tetrode hyperdrives11,12,13 enable recordings from dozens of tetrodes, their assembly and implantation are time-intensive, and their quality is largely dependent on the skill and experience of the individual researcher. In addition, they usually combine the guide tubes that direct the recording electrodes to their target location in one or two larger bundles, therefore limiting the number and spread of areas that can be targeted efficiently.
Other implants14,15 expose the complete skull and allow for the free placement of multiple individual microdrives that carry the recording electrodes. While the placement of independent microdrives16 during surgery time maximizes flexibility, it increases surgery time and can make it difficult to target multiple adjacent areas due to the space requirements of the individual microdrives. In addition, while the implants are open source, they are only published as STL files, making modification difficult.
An example of a drive with a more inherent parametric philosophy is the RatHat17. By providing a surgical stencil that covers the whole dorsal surface of the skull, it allows precise targeting of multiple brain targets without the use of a stereotactic frame during surgery. Multiple implant variations for cannulas, optrodes, or tetrodes are available. However, while the drive is free to use for academic purposes, it is not published open-source, creating a hurdle for researchers to evaluate and use the implant.
Presented in this article is the TD Drive (see Figure 1), a novel 3D-printable implant for extracellular electrode recordings in rats. The TD Drive aims to overcome some of the drawbacks of existing solutions: it allows to target multiple brain areas, mirrored across both hemispheres, with independent wire electrodes simultaneously. Due to its simple design, it can be assembled in a few hours at a relatively low cost by less experienced researchers. The TD Drive is published open-source, in easily modifiable file formats to allow researchers to adjust it to their specific needs. Incorporating a parametric 3D modeling approach from the beginning of the TD Drive's design process allows the parameters necessary to be changed to be abstracted: to change target locations, researchers can simply edit the parameters representing their dorsoventral and anteroposterior coordinates, without the need for re-designing the drive themselves. The files to modify and manufacture the TD Drive can be found at https://github.com/3Dneuro/TD_Drive.
Figure 1: Overview of the TD Drive. (A) Rendering of a TD Drive with a protective cap. (B) Rendering with inner parts shown. The TD Drive features (a) multiple, parametrically adjustable recording locations for fixed and moveable electrode wires, an EIB with (b) a high-density Omnetics connector compatible with common tethered and wireless data acquisition systems, and (c) an intuitive channel mapping optimized for recordings with Intan/Open Ephys systems (see Supplementary Figure 1) and (d) a cap to protect the implant during tethered recordings and when no headstage is connected. (C) A guide stencil on the bottom of the TD Drive facilitates the placement of guide cannulas and serves as a redundant verification of implant locations during surgery. Please click here to view a larger version of this figure.
The implant design was piloted in n = 4, validated in n = 8, and confirmed in n= 8 Lister Hooded rats that performed different tasks. The first 4 animals were used to develop the drive and adjust parameters. Then, a full pilot was run with 8 animals (shown in results). A second cohort of 8 animals was run and included in the implant survival analysis. The implant was compatible with tethered sleep recordings and open field recordings (Object Exploration) as well as wireless recording in a large maze (HexMaze 9 m x 5 m) using two different commercial recording systems and headstages. The two cohorts of 8 were recorded with two different acquisition systems – tethered for longer sleep recordings and wireless for large maze exploration recordings. We can conclude that this simple wire drive allows for long-running experiments with larger cohorts by less experienced researchers to enable sleep stage analysis as well as oscillation analysis in multiple brain areas. This is in contrast to most electrophysiology implants to date, which, due to difficulty and time intensity, allow for smaller animal cohorts and usually need very experienced experimenters. However, with this drive, no individual neuron activity can be recorded; thus, the use is limited to investigations of local field potential (LFP) and summation activity.
The present study was approved by the Dutch Central Commissie Dierproeven (CCD) and conducted according to the Experiments on Animals Act (protocol codes: 2020-0020-006 & 2020-0020-010). Male Lister Hooded rats of 9-12 weeks on arrival were used. The reagents and the equipment used in the protocol are listed in the Table of Materials. See Supplementary Figure 1 and Supplementary Figure 2 for the steps of the drive-building process.
1. Adjusting and creating 3D models and electrode interface board (EIB) data
2. Printing the 3D models and manufacturing the EIB
NOTE: For the present study, a commercially available 3D printer was used to produce the parts (see Table of Materials). When using different printers or outsourcing the production, different, comparable resins for producing the parts might need to be used.
3. Post-processing of the 3D-printed body
NOTE: Cap and shuttles should not need post-processing. Depending on the quality of the 3D prints, they might need to be lightly sanded or have leftover support traces removed. When sanding and drilling, take care not to break the walls of the drive body. If necessary, clean post-processed parts with isopropanol and, a soft cloth, and/or compressed air.
Figure 2: Rendering of the TD Drive. (A,B) TD Drive (A) without and (B) with a protective cap on a rat skull model. (C) Polyimide guide tubes correctly inserted into each of the six recording sites. (D) An isolated, completed shuttle assembly featuring the guide screw, 3D-printed shuttle, and the soldered brass insert. (E) TD Drive body with two shuttles inserted. Marked in red: (a) countersink holes for the shuttle, (b) shuttle guide, (c) center pedestals of the drive body, (d) guide stencil.(F,G) Important locations on the top (F) and bottom (G) of the drive body that might require post-processing after 3D printing are indicated by a red arrow each. Please click here to view a larger version of this figure.
4. Shuttle assemblies
5. Assembling the drive
6. Preparing the protective cover
7. Preparing the wire electrodes
8. Preparing the ground wire and EEG wires
9. Loading the wire bundles into the drive
10. Drive implant surgery
NOTE: This step briefly outlines the surgical procedures for implanting the TD Drive. A more extensive implantation protocol, including a description of tools, as well as doses and concentrations of drugs, can be found in Supplementary File 1.
11. EIB recovery
Using the instructions provided in the protocol, the TD Drive could be built easily by multiple experimenters. After drive development (n = 4), a full pilot was run with eight animals. An additional batch of eight animals was implanted, and experimental data collection was performed. As data analysis has not been completed on these animals, they have been included in the survival analysis, but not in other analyses (e.g., targeting or histology). Implant surgery was performed 2 weeks after arrival (see Figure 3A for the target locations used in the pilot). The implant was performed with usual surgical procedures and lasted ~3 h. An experienced surgeon performed initial implants and could teach both experienced as well as novice experimenters with 2-3 surgeries to independence.
Figure 3: Implant surgery, sleep data, and broadband activity. (A) Schematic overview showing the target locations for craniotomies (blue circles) and skull screws (green: EEG, blue: GND, gray: anchoring screws, note that two anchoring screws are on the side of the skull). (B) Photograph of implanted animals with a tethered headstage during sleep and wakefulness. (C) Example sleep data from tethered animal PFC (Prelimbic) and HPC (Ca1), divided in REM sleep with theta and NonREM sleep with delta, spindles and ripples. Y axis: microvolt, x axis: seconds. This data can be used for example for sleep scoring or oscillation event detection and analysis (D) Example broadband activity recorded wirelessly in an awake animal (noisy channels on the left were not connected). Please click here to view a larger version of this figure.
Figure 4: Implant durability, delta detections, and histology. (A) Survival plot for the implant for two rounds of longer-running experiments. Of note, on day 85, n = 8 finished the experiment and were planned to be perfused. (B) Data example for stability. Shown is the count of delta detections in the hippocampal channel over recording days (~3 days per week). Each animal showed a normal variation depending on the amount of sleep, but there was no general drift over time in the signal and thus detections. (C) Representative histology showing bilateral targeting for one rat. Left column: left hemisphere, right column: right hemisphere. AP coordinates indicate anteroposterior coordinates of the depicted slice, and arrows point to lesions in targeted areas. Magnification: 1.6x. Please click here to view a larger version of this figure.
All animals recovered well and tolerated the implant (Figure 3B). Frontal and retrosplenial electrodes were fixed, but the hippocampal bundles were movable. Hippocampal bundles were implanted at a dorsoventral depth of 2 mm and adjusted to maximize HPC coverage during 2 weeks of surgery recovery, where the signal was checked live during sleep habituation periods. In 7 out of 8 animals, all target sites were reached on at least one hemisphere (Table 2 for hit rates, see Figure 4C for representative histology). Wake and sleep recordings were performed successfully tethered in a recording box, as well as wireless recordings in a larger maze (example data Figure 3C,D). Animals kept the implants for 2 months, when individual animals would start losing them; however, the majority of animals kept the implants until the experimental end day 85-100 after implant (Figure 4A). During this time, the LFP remained stable, as can be seen in an example analysis where delta oscillations were detected (Figure 4B). There was normal variability across time but no systematic drift of the signal in any of the recorded brain areas (including the pyramidal layer of CA1). It is recommended that experiments be ended within 10-15 weeks after surgery. All EIBs could be recovered.
We have applied this implant mainly to measure sleep stages and sleep oscillations in response to learning and other interventions. For example, how oral CBD intake influences oscillation occurrence and coherence across brain areas (see Samanta et al.20).
Parameter | Minimum value (mm) | Maximum value (mm) |
medioLateralSite1 | 0.75 | 2 |
medioLateralSite2 | 1.5 | 5 |
medioLateralSite3 | 0.75 | 2 |
Table 1: Overview of the manually imposed limits on the parameters controlling the mediolateral coordinates of the recording sites.
Hemisphere | PFC | RSC | CA1 pyr. |
Right | 8 of 8 | 5 of 8 | 6 of 8 |
Left | 8 of 8 | 7 of 8 | 7 of 8 |
Table 2: Hit rate for pilot of 8 animals. In 4 out of 8 animals, all electrodes were placed correctly. However, in 7 out of 8 animals, all brain areas were correctly targeted in at least one hemisphere (with the exception of 1 rat missing the CA1 pyramidal layer).
Supplementary Figure 1: TD Drive architecture. (Top) Overview of the channel mapping for the TD Drive when used with an Intan RHD32 headstage. (Bottom) An additional illustration of the wire bundle configurations.Please click here to download this File.
Supplementary Figure 2: Additional pictures of the TD Drive at several stages of the building process. Please click here to download this File.
Supplementary File 1: An example protocol for implantation of the TD Drive. The procedures and target locations are adjusted to the authors' research questions and institutional policies.Please click here to download this File.
Presented in this article is an adaptable implant for bilateral, symmetric multi-area wire electrode recordings for freely-moving rats.
The ability to easily adjust the implant by changing predefined parameters was one of the motivations for the creation of the TD Drive. While aiming to maximize the flexibility for changing parameters, inherent constraints in the relations between them necessarily impose limits to this adaptability. No limits are set by default for the anteroposterior parameters, with these coordinates instead being governed by logical inter-site interactions and the overall size of the drive body. The parameters controlling the mediolateral coordinates of all recording sites are subject to manually imposed limits (see Table 1). Underlying the many parametric options are numerous interactions between the design features. These interactions can become disjointed under certain conditions. In an ideal situation, all possible combinations of parameter values are valid. However, with the more complex design of the TD Drive, it was opted to limit the mediolateral coordinates to within the tested range. Choosing coordinates outside the tested limits is, in principle, possible. However, making such changes is not recommended as the integrity of the design can become compromised, and restoring it while maintaining the out-of-limit coordinates might require CAD modeling experience. This is an inherent result of the trade-off between ease of drive production and flexibility – larger degrees of freedom do make the parametric definition of the drive more complex and can result in overly fine-grained, undesired outcomes (for example, the need to produce different EIBs for small changes in the drive design).
The parameter choices made in this manifestation of the TD Drive are guided by the preferences of the current experimenters. It was opted, for example, to increase the height of the TD Drive's pedestals, improving ease of implantation during surgery at the cost of a slightly higher final implant. However, the implant is still much smaller than drives with individually moveable tetrodes and was well-accepted by the animals. The two pedestals used to anchor the drive ("c" in Figure 2E) are deliberately placed in the center plane, parallel to the midline suture of the rat's skull. This limits recording locations to mediolateral coordinates > 0.75 mm. Often, the presence of the sagittal sinus below the midline imposes a similar limit for craniotomies anatomically. The current design of the TD Drive has been created in Autodesk Fusion. While it is one of the most advanced programs for parametric 3D computer-aided design and, at the time of publication, does provide a free license for academic use, the commercial and cloud-based nature of the program does pose a risk for the free availability of the design. Therefore, porting the design to a true open-source parametric CAD software21, such as FreeCAD, might be necessary for a future iteration.
The surgery for the TD Drive can be performed in 2-3 h. The target locations of the wire are marked stereotactically (PFC +3.5AP next to the midline, HPC -3.8AP -/+ 2.5 ML, RSC -5.8 next to the midline), and the screw electrodes, GND, and additional skull screws for stability are placed relative to those locations. While the TD Drive stencil does provide stereotactic locations, imprecisions in the placement of the polyimide tube and the use of less stiff tubing material can introduce small variations in the position of the electrodes. Therefore, it is recommended to drill small craniotomies (instead of guide tube-sized 0.5 mm burr holes) to account for this variability. In this surgery, a single, larger craniotomy for the RSC and HPC targets is drilled. For PFC and RSC, it was chosen to have wire bundles implanted at a fixed depth. The PFC bundle had wires targeted at two different depths to record from the prelimbic as well as the anterior cingulate cortex. The HPC bundles were movable and were built with 3 wires at varying heights to facilitate reaching the Ca1 pyramidal layer as well as the stratum radiatum. The last, shorter wire allowed recording from the PPC. We achieved the best results for targeting the hippocampal Ca1 when the longest electrode wire was moved to the target depth (2 mm ventrally from the brain surface) during surgery time, with only small adjustments in the two weeks after surgery during live-signal checks to accommodate for individual variances and brain swelling after surgery.
A problem with large implants, such as tetrode hyperdrives, in rats, is the chance of the implant stability degrading and animals losing the implant. For the TD Drive, individual failures were observed due to degradation of implant stability after 2 months (3 out of 16 animals). Therefore, the TD Drive is recommended for experiments with an intended maximum duration of 8 weeks. We show that for this time period, the signal is stable – even the precise hippocampal pyramidal layer recordings – and there is no systematic drift or significant wobble affecting the LFP recordings. One factor in achieving this long-term stability is the use of multiple skull screws (see Figure 3A). In certain situations, the amount of used skull anchoring screws can increase the risk of infections22. However, this is likely mostly relevant in head-fixed rats, where the repeated stress of the head fixation on the headplate and connected anchoring screws can result in degradation of the implant that facilitates infections. Another factor that increases the risk of implant failure (and can induce discomfort in behaving animals) is the weight of the implant. A complete TD drive implant with anchoring screws and dental cement weighs around 7 g, with the 3D printed parts and EIB accounting for approximately half of the weight. Due to the low weight of the TD drive (less than 1/3 of other large tetrode hyperdrives), excessive implant weight is unlikely to be a significant factor for failures of the TD drive. Generally, the main factor for stable, infection-free implants is a sterile surgery procedure and good adhesion of the dental cement that is coating the implant to the implant, anchoring screws, and skull23. We only evaluated the TD Drive for the recording of local field potentials and did not attempt to acquire single-unit activity. While we expect the implant to be generally stable enough to do so, following the same units within and between sessions might require optimization of the shuttle stability, for example, by optimizing the shuttle guide rail ("b" in Figure 2E). The addition of moveable shuttles to the other recording sites would allow for additional use of moveable wire electrodes, which are the preferred way of ensuring better long-term signal quality in unit recordings.
With a total assembly time of around 3 h and surgery time of around 2-3 h, the TD Drive offers a compromise between tetrode hyperdrives and simpler, less adjustable multi-wire implants24. With the chosen targets, recordings from 10 brain areas with 6 bundles were achieved. Compared to other implants for non-moveable wire bundles, the symmetric placement of recording sites yields another advantage: if lateralization is not relevant, simultaneously implanting wires in both hemispheres doubles the chance of hitting the correct target and the data yield per animal. In the pilot, 4 out of 8 animals had all 5 target sites (PFC including prelimbic (PRL) and anterior cingulate (ACC) cortex, RSC, PPC, and Ca1 pyramidal layer of HPC) targeted correctly bilaterally, but 7 out of 8 had at least each brain area on one side recorded. Thus, this drive is advisable for those who want a quick and easy-build solution to record LFP that can be applied by less experienced researchers, especially when higher animal numbers are needed, such as in sleep studies. With many high-end tetrode hyperdrives that would allow for the recording of individual neuronal activity, even very skilled and experienced researchers can only build and implant 2-6 implants per year of which many will not successfully reach any brain area of interest. It takes many years of training to achieve higher success rates, and even then, the number of animals that can be recorded efficiently remains low.
In summary, the TD Drive presents an easy and fast-to-build wire drive with 6 bundles that can be easily adapted to contain different recording sites and other implants such as cannulas and fibers.
The authors have nothing to disclose.
The authors would like to thank Angela Gomez Fonseca for the inspiration to develop the drive and all the students who ran pilot experiments with the animals, Milan Bogers, Floor van Ravenswoud, and Eva Severijnen. This work was supported by the Dutch Research Council (NWO; Crossover Program 17619 "INTENSE").
0.5 mm drill bit | McMaster | 2951A38 | |
1.27 mm pitch interconnected SIP/DIP socket (Mill-Max) | Mouser Electronic | 575-003101 | For essembling and connection of EEG & GND screws |
5 minute epoxy | Bison | Commercially available | regular off-the-shelf epoxy |
cyanoacrylate glue | Loctite | Super Glue-3 | |
EEG wire | Science Products GmbH | 7SS-2T | |
Electrode wire | Science Products GmbH | NC7620F | |
Ethanol | LC | For standard pre-operative sterilization procedure of drive | |
Fine forceps (5) | FST | 91150-20 | For wire bundle preperation and handling |
Form 3B | Formlabs | 3D printer used to 3D print the self-printed parts of the TD drive | |
Gold pins (small) | Neuralynx, Inc. | 9885 | Attachment of electorde wires to EIB board |
Ground wire | Science Products GmbH | SS-3T/A | |
High-density connector | LabMaker GmbH/Omnetics | A79026-001 | |
Lister Hodded rats | Charles River Laboratories | Crl:LIS | we used male rats, 9-12 weeks of age at arrival |
M1 brass insert | AliExpress | Commercially available | https://aliexpress.com/item/33047616164.html |
M1 tap | McMaster | 2504A33 | |
M1x16 screw | Bossard | 1096613 | |
M1x3 stainless steel screws | Screws and More | 84213_14985 | |
M2.5×5 polyimide screws | Screws and more | 7985PA25S_50 | |
mineral oil | McMaster | 1244K14 | |
Nail polish | Etos | Commercially available | For color coding EEG and GND wires |
painter's tape | Gamma | Commercially available | For wire bundle preperation |
Pin vise | McMaster | 8455A16 | |
plotting paper | Canson | Commercially available | For wire bundle preperation |
polyimide tubes | Amazon / Small Parts | TWPT-0159-30-50 | AWG, 0.0159" ID, 0.0219" OD, 0.0030" Wall, 30" Length |
RHD 32-channel headstage with accelerometer | Intan Technologies, LLC | C3324 | For tethered recordings in the sleepbox |
RHD 3-ft (0.9 m) standard SPI cables | Intan Technologies, LLC | C3203 | From commutator to headstage |
RHD 6-ft (1.8 m) standard SPI cables | Intan Technologies, LLC | C3206 | From OpenEphys box to commutator |
Slip Ring with Flange | Adafruit | 1196 | Commutator: 22 mm diameter, 12 wires |
Solder flux | Griffon S-39 50 ml | Commercially available | For soldering EEG & GND screws |
soldering paste | Amazon | B08CBZ5HC5 | |
stainless steel M2 nut | McMaster | 93935A305 | |
Tethered recording setup | OpenEphys | Acquasition Board | |
Wireless recording logger | SpikeGadgets | miniLogger 32 | For wireless recordings in the task |
Wireless recording setup | SpikeGadgets | Main Control Unit (MCU) incl. breakout board and RF transceiver | For wireless recordings in the task |