Inhibition of Na⁺/H⁺ exchanger modulates microglial activation and scar formation following microelectrode implantation

Abstract

Objective.

Intracortical microelectrodes are an important tool for neuroscience research and have great potential for clinical use. However, the use of microelectrode arrays to treat neurological disorders and control prosthetics is limited by biological challenges such as glial scarring, which can impair chronic recording performance. Microglia activation is an early and prominent contributor to glial scarring. After insertion of an intracortical microelectrode, nearby microglia transition into a state of activation, migrate, and encapsulate the device. Na⁺/H⁺ exchanger isoform-1 (NHE-1) is involved in various microglial functions, including their polarity and motility, and has been implicated in pro-inflammatory responses to tissue injury. HOE-642 (cariporide) is an inhibitor of NHE-1 and has been shown to depress microglial activation and inflammatory response in brain injury models.

Approach.

In this study, the effects of HOE-642 treatment on microglial interactions to intracortical microelectrodes was evaluated using two-photon microscopy in vivo.

Main results.

The rate at which microglia processes and soma migrate in response to electrode implantation was unaffected by HOE-642 administration. However, HOE-642 administration effectively reduced the radius of microglia activation at 72 h post-implantation from 222.2 μm to 177.9 μm. Furthermore, treatment with HOE-642 significantly reduced microglial encapsulation of implanted devices at 5 h post-insertion from 50.7 ± 6.0% to 8.9 ± 6.1%, which suggests an NHE-1-specific mechanism mediating microglia reactivity and gliosis during implantation injury.

Significance.

This study implicates NHE-1 as a potential target of interest in microglial reactivity and HOE-642 as a potential treatment to attenuate the glial response and scar formation around implanted intracortical microelectrodes.

1. Introduction

Intracortical microelectrode arrays have been a vital tool in neuroscience research while more recently emerging as bi-directional interfaces for quality-of-life changing interventions in clinical neuroprosthetic applications [1–9]. Intracortical microelectrodes provide real-time acquisition of neural activity with high spatial and temporal fidelity [10, 11]. The ability for microelectrodes to isolate brain activity down to individual neurons provides advantages over non-invasive modalities such as imaging techniques and electroencephalograms. At the bench-side, electrodes used to both record and stimulate at the micron level are crucial for understanding the basic brain circuitry that governs complex neural networks [12–14]. At the bedside, intracortical electrodes are incorporated in a large share of brain-computer interface (BCI) technologies used to restore motor and/or sensory brain functions after injury. Their high spatial resolution and long-term fidelity are critical for the control of neuroprosthetic limbs that could improve the lives of many neuro-disabled patients [15, 16].

However, applications involving microelectrodes implanted within the brain for indefinite periods of time remain a challenge due to severe inflammatory responses [17–19]. Acute damage due to insertion trauma and the chronic tissue response to the implantation of a foreign body can result in biological failure modes in which the local microenvironment at the electrode–tissue interface becomes unfavorable to support neuronal health and activity [20–32]. Carefully targeted insertions can avoid penetrating larger vasculature, but damaging the blood-brain barrier is inevitable due to the vast capillary beds that span the cortex [19, 33, 34]. Plasma protein leakage, mechanical tissue strain due to edema, and damaged cellular debris can promote activation of glial cells and secretion of pro-inflammatory signaling factors [35–37]. The inflammatory response induced from this tissue disruption is believed to engage a cascade of signaling pathways that result in neuronal cell death [19, 31, 33, 38]. Injury from the resulting tissue strain is not limited to the surgery, as volume displacement and tissue micromotion can persist chronically throughout the lifetime of the implant [37, 39–42]. Furthermore, risk of material failures such as corrosion, cracks, and delamination of insulation are heightened in inflammatory biological environments, especially due to pervasive glial activation and encapsulation, which can reduce the recording efficacy of electrodes over time [42–48].

Microglia are a particularly significant contributor to glial scar formation and neuroinflammation within the brain [49–54]. They are ubiquitously distributed across the cortex and therefore one of the primary responders to injury. Under resting conditions, they exist in a morphologically ramified state characterized as stationary cell soma which continuously extend and retract filopodia processes in all radial directions to survey their surroundings for physiological disturbances [55–57]. As immunosurveillance glia, they continuously scavenge the brain for harmful plaques, damaged or dying cells, and biological waste [58–60]. Following brain injury, they transition into an activated state, secreting inflammatory cytokines and phagocytosing debris [58, 59, 61–65]. We have previously demonstrated that microglia undergo morphological changes following microelectrode implantation characterized by processes preferentially extended towards the electrode [21, 49–51]. Over hours to days post-insertion, microglia migrate, proliferate, and form lamellipodia that spread out and encapsulate the surface of the electrode [66]. These reactive microglia can express cytokine signals that activate other microglia, NG2 glia, and astrocytes around the electrode, which promotes a neurotoxic and inflammatory microenvironment at the electrode–tissue interface.

Na⁺/H⁺ exchanger isoform 1 (NHE-1) is a membrane-bound ion transporter that facilitates the 1:1 efflux of H⁺ and influx of Na⁺ and is widely expressed within microglia cells. The primary purpose of NHE-1 is to maintain intracellular pH (pH_i) within microglia, which is involved in cytoskeletal rearrangement during cellular changes in polarity and motility [67–70]. NHE-1 activity is upregulated following LPS-mediated microglia activation, producing a ~2.3 fold increase in the rate of pH_i alkalization [68]. This alkalinizing process is important to prevent acidification during microglia ‘respiratory burst’, which involves the reduction of molecular oxygen (O₂) to superoxide anion (O₂⁻) by NADPH oxidase (NOX). NOX, whose activity is pH-sensitive, therefore depends on NHE-1 to maintain an alkaline environment [71]. The pharmacological agent HOE-642 (commonly known as ‘cariporide’) is a potent and selective inhibitor of NHE-1 activity [72]. Blocking NHE-1 with HOE-642 was shown to hinder microglial activation in the presence of proinflammatory conditions as a result of ischemia in vivo [73]. Despite the positive findings of HOE-642 administration in transient focal cerebral ischemia and reperfusion experiments [73], the effects of HOE-642 treatment on microglial activation have not been assessed following implantation of a microelectrode array.

Here, we use the previously established inhibitor of NHE-1 activity, HOE-642, in an effort to suppress microglial activation and gliosis around implanted intracortical electrodes. We hypothesized that inhibition of NHE-1 using HOE-642 can attenuate microglial activation, migration, and encapsulation of intracortical microelectrodes. To test this hypothesis, we used two-photon microscopy to compare microglial activity in HOE-642 administered animals against untreated control animals up to 72 h following microelectrode implantation. This allowed us to compare the drug treatment and the injection insult together against the standard practice of electrode implantation without any additional injections. HOE-642 administration reduced the radius of microglial activation and encapsulation around the microelectrode. These findings reveal a novel mechanism of microglia-mediated reactivity around intracortical microelectrodes and present HOE-642 as a potential therapeutic intervention to reduce tissue scarring and inflammation post-implantation in order to preserve long-term device fidelity and performance.

2. Methods

2.1. Animals and HOE-642 administration

Transgenic CX3CR1-GFP mice (male, 6–8 weeks old) expressing a green fluorescent protein (GFP) under the CX3CR1 promoter in microglia (Jackson Laboratory, Bar Harbor, ME) were implanted intracortically with Michigan style silicone microelectrode arrays. Mice were separated into individual housing post-surgery to reduce the likelihood of dislodging their implants and headcaps. The treatment group received 0.25 mg kg⁻¹ HOE-642 injected intraperitoneally (i.p.) twice a day, 8 h apart throughout the duration of the experiment, a similar dosing regimen that has been shown to be effective in a rodent glioma model [67]. HOE-642 was initially dissolved in DMSO (1 mg ml⁻¹) prior to dilution with sterile saline for injection. To examine the acute microglia response to implantation, initial injections were administered during anesthesia induction about 2 h before surgery in ‘acute’ mice (figure 1(a)). To examine more chronic microglia responses to implantation, initial injections were administered immediately after surgery in ‘sub-chronic’ mice (figure 1(a)). For brain injuries such as stroke or TBI, drug treatment most often begins after the medical emergency. However, for BCI implantation surgeries, pharmacological interventions may be administered prior to the procedure. Dosage was chosen so as to balance the desire to see potential effects on microglia with the knowledge that high levels of HOE-642 can be toxic to the animals [74]. All protocols involving mice were approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Figure 1.

Experimental setup for two-photon imaging. (a) Timeline of experimental procedures. HOE-642 was administered immediately prior to surgery for acutely observed mice. (b) Schematic of headcap and imaging window installation suitable for sub-chronic in vivo two-photon imaging of a microelectrode implanted in the mouse cortex. The imaging window was sealed with a transparent elastomer and glass coverslip to protect the brain surface while allowing for longitudinal imaging. (c) 2D representative image visualizing microglia (green) and cortical vasculature (red) around an implanted microelectrode array (shaded blue). (d) A microglial cell of ramified morphology. (e) A microglial cell of transitional (activated) morphology.

2.2. Surgical probe insertion

Surgical methods were consistent with previously described experiments [21, 49, 50, 75]. Anesthesia was induced in mice using a drug cocktail consisting of 75 mg kg⁻¹ ketamine and 7 mg kg⁻¹ xylazine injected i.p. A stereotaxic frame (Narishige) was used to secure anesthetized mice throughout the duration of surgery and imaging. During surgery, breathing and toe-pinch responses were monitored to provide feedback on depth of anesthesia, and updates of 40 mg kg⁻¹ ketamine were given hourly or as needed. Animal scalps were shaved prior to being washed repeatedly with betadine and ethanol and then removed along with connective membranes to expose the skull. Small drops of Vetbond (3M) were applied to dry the surface of the skull and increase adhesion between the skull and dental cement headcap. Both experimental mice receiving HOE-642 and control mice were split into acute and sub-chronic surgical preparations. For sub-chronic preparations, two bone screws were embedded in holes over the motor cortex and secured with light-curable dental cement. For both acute and sub-chronic preparations, a ~4 × 4 mm craniotomy was formed centered over the ipsilateral visual cortex. Only a unilateral craniotomy was required to investigate the effects of HOE-642 administration on microglial reactivity following device insertion. Throughout drilling, saline was repeatedly applied to sweep away bone fragments and cool the surgical site to prevent thermal damage to the brain. For acute preparations, a 3D printed cone was cemented in place surrounding the craniotomy to enable imaging into the exposed brain using a water-immersive objective lens. Following opening of the craniotomy on all preparations, non-functional four-shank Michigan style silicon probes were inserted in a rostral direction through intact dura mater at a 30° angle parallel to midline to a depth of approximately 300 μm below the pial surface. Positioning of the probes on the brain surface was carefully chosen to avoid disrupting major blood vessels. In acutely prepared mice, imaging began 1 min before and up to 80 min following implantation. For sub-chronic preparations, an in situ curing silicon elastomer (Kwik-Sil, World Precision Instruments) and a glass coverslip were used to seal the craniotomy after implantation and provide a chronic imaging window for two-photon imaging. Light-curable dental cement was used to secure the imaging window to the skull and build up a well around the side 2 mm in height to allow for the use of a water-immersive objective lens. The process to create a chronic imaging window was necessary to return the mice to their housing units between sub-chronic imaging sessions. As a result, these additional steps prevent imaging in the minutes prior to and immediately following probe implantation, which is why it was necessary to have a separate cohort of mice to gather acute data following insertion. Lastly, ketofen (i.p., 5 mg kg⁻¹) was administered daily up to two days post-operation.

2.3. Two-photon imaging

A two-photon scanning laser microscope (Ultima IV; Bruker) was utilized to image CX3CR1-GFP transgenic mice expressing GFP in microglia. The components of the microscope included a scan head, OPO laser, non-descanned photomultiplier tubes (Insight DS+; Spectra-Physics), and 16X, 0.8 numerical aperture water immersion objective lens (Nikon). Mice were injected i.p. with sulforhodamine 101 (SR101) as a vascular contrast agent. The microscope laser was tuned to 920 nm to excite both GFP and SR101, and the resulting fluorescence was captured in green and red channels, respectively. ZT-stack images were taken every minute for the first 80 min following implantation for acute mice. In addition, Z-stack images were taken at 2, 3, 4, 5, and 6 h post-implantation for acute mice and 6, 12, 24, 48, and 72 h post-insertion for sub-chronic mice. Mice were stereotaxically secured while anesthetized with 1%–1.5% isoflurane during all imaging sessions. Stacks covered a span of 412.8 by 412.8 μm (1024 by 1024 pixels) in the horizontal plane with depths of around 300 μm. Images were taken above the top shank and/or below the bottom shank depending on variable visibility due to blood vessels or pial surface bleeding. Only microglia outside the outer shanks and in plane with the probes were included for analysis.

2.4. Data analysis

Z-stacks were analyzed using ImageJ software (National Institutes of Health) [76]. Microglial activation was quantified using a ramification index. Microglia typically reside in a ramified state, but can enter a transitional state with fewer and longer projections directed toward the probe when activated [56]. In the image stacks, microglia in plane with the outer probe shanks were identified and visually characterized as either ramified (1) or transitional (0). The distance of these glial cells from the surface of the probe was also recorded using the ‘Measure’ feature in ImageJ so that they could be binned by increments of 50 μm up to a distance of 250 μm away. For each time point, data was then fit with a logistic regression to show the Bernoulli probability distribution of the state of microglia being ramified or transitional as a function of the distance from the outer shank. The distance at which 50% of microglia are expected to be activated based on this fit was defined as the radius of activation for microglia around the probe [49–51].

In addition to the ramification index, a transitional index (T-index) and directionality index (D-index) were calculated from properties and measurements of the microglia related to the morphological changes seen when activated. The T-index is based on the length of the longest process extending toward (n) and away (f) from the probe, whereas the D-Index takes into account just the number of processes extending toward (n) and away (f) from the probe. In order to define the direction of process extension, a line extending parallel to the edge of the probe and passing through the midpoint of each microglial soma was used to distinguish the hemisphere toward and away from the implant. Both indices were calculated at every time point using the following formula [49–51]:

$$\text{Index=}\frac{(f-n)}{(f+n)}+1.$$ (1)

Note that this formula mathematically limits index values to be zero or positive. When there are more and longer microglial projections in the direction of the probe, n will be larger than f, and the indices will approach zero. When the processes are approximately evenly distributed radially around the cells in both position and length, n and f will be about the same, and the indices will approach one. As such, indices of zero correspond to activated microglia and indices of one indicate a ramified state. However, unlike the binary ramification index, the T-index and D-index also provide some information on the extent of activation from a morphological perspective. Following binning, the data was fit with a custom dual sigmoidal function in MATLAB [49–51]:

$$y(d)=\frac{a}{1+e^{d-d_1/w_1}}+\frac{1-a}{1+e^{d-d_2/w_2}}$$ (2)

where a is amplitude, d₁ and d₂ are shoulder location (in μm), and w₁ and w₂ are shoulder width. Note that this function is constrained between values of zero and one.

ZT-stacks were used to quantify process velocities. Grouped z-projections were done to create 2D image series from 0 to 80 min after implantation. A ‘StackReg’ plugin for ImageJ was applied to correct for motion by aligning the image series [77]. Projections stemming from activated microglia around the probes were identified, and their XY coordinates throughout time were recorded with the ‘Measure’ function. Subtracting consecutive coordinates gave displacement vectors representing the distance traveled by the process end feet each minute, throughout 80 min. Vectors were assigned to be positive if pointing toward the probe and negative if pointing away. Then, the values were averaged over time to calculate microglia process velocity.

For the purpose of tracking soma migration velocities, z-stacks from consecutive sub-chronic timepoints had to be merged to identify the same microglial cells throughout timepoints on the scale of hours. To accomplish this, a ‘TurboReg’ plugin for ImageJ was utilized to determine the x- and y-axis offset between stacks [77]. The ‘Translate’ feature could then be used to align the z-stacks. Depth offset was found by noting the frames at which a static point on the electrode was located in the consecutive stacks. Then, this dimension could be aligned through the creation of substacks with the reference point at the same depth. The shortest distance from the center of the same microglia soma to the edge of the probe was recorded for each timepoint. The migration velocity could then be calculated using the formula:

$$\overline{v}=\frac{\Delta x}{\Delta t}$$ (3)

where $\overline{v}$ is the average velocity of the microglia in the direction of the probe between two imaged timepoints, Δx is the change in distance from the probe, and Δt is the amount of time passed between consecutive imaging sessions.

Finally, microglial encapsulation of the probes was quantified as the percent surface coverage of microglia (GFP) signal [21, 50]. First, the interactive stack rotation plugin for ImageJ was used to re-slice and rotate the z-stacks. The plane was rotated 30°, corresponding to the angle of probe insertion, making the entire surface of the probe visible within a single frame. The probe surface and the tissue volume up to 20 μm above were separated into a substack before undergoing a sum projection, creating a single projected image. A binary mask of this image was made using IsoData threshold method in ImageJ [78]. The outline of the probe was manually drawn onto the mask, and the ratio of nonzero pixels representing thresholded GFP signal within the outline to the total number of enclosed pixels as found through the ‘Measure’ function was taken as the percent surface coverage. This same process was used to quantify both acute and sub-chronic surface coverage data.

2.5. Statistical analysis

Logistic regressions were applied to ramification index data, and a custom dual sigmoidal function was fit to T-index and D-index data as described above. A two-way analysis of variance (ANOVA) was used to assess significance in surface coverage variation between different timepoints and mice groups. Bonferroni corrected Welch’s t-tests were used to compare surface coverage between HOE and control mice at individual timepoints. A p-value of 0.05 was taken as the criteria for differences to be significant.

3. Results

In order to assess the impact of HOE-642 treatment on microglial activation surrounding intracortical microelectrode probes, non-functional silicon probes were implanted within the visual cortex of mice expressing GFP under the microglia-specific CX3CR1 promoter. Among the mice treated with HOE-642 (n = 10), five were examined 0–6 h post-implantation (‘acute preparation’) while five were examined 6–72 h post-implantation (‘sub-chronic preparation’). A separate cohort of acute (n = 5) and sub-chronic (n = 5) mice that were not administered HOE-642 served as a control group (figure 1(a)). A distinct acute experimental preparation was required to capture acute changes in process velocity prior to and immediately following device insertion. However, both HOE-treated and control groups were compared only between animals that received the same surgical preparation in order to carefully control for these experimental factors while evaluating our hypothesis that HOE-642 decreases microglia activity after microelectrode implantation. Differences in microglia process extension, soma migration, morphology, and encapsulation of the probes were tracked longitudinally via two-photon microscopy (figures 1(b) and (c)). In healthy (uninjured) cortex, microglia are evenly distributed and periodically extend and retract processes in all radial directions to survey their environment for the presence of debris and chemical signaling molecules [56]. This resting-state morphology in microglia is defined as ‘ramified’ (figure 1(d)). In contrast, disturbances such as inflammatory signaling molecules, foreign bodies, debris, tissue strain, and plasma leakage from injured blood vessels trigger microglia to activate and undergo structural changes [58]. Long microglial processes extend towards inflammatory stimuli while processes on the opposite side of the cell retract, resulting in a ‘transitional’ morphology (figure 1(e)).

3.1. Microglia process extension after implantation is unaffected by HOE-642 administration

To examine the effects of HOE-642 on microglia process extension, ZT-stacks taken from 0 to 80 min post-insertion in acute mice were used to quantify microglia process velocities. Within minutes following electrode implantation, nearby microglia processes oriented and migrated toward the probe surface. Representative images showing the extension of processes demonstrate a few characteristics of their movements (figure 2(a)). Within the first 30 min, multiple microglial end-feet are observed moving quickly toward the probe (figure 2(a); white triangles). Process velocities ranged from 0.5 to 1.5 μm min⁻¹ within these first 30 min, consistent with a previous study [50]. Eventually, end-feet come into contact with the surface of the probe and slow down, at which point there is a downward trajectory in process velocity (figure 2(b)). At later time points, fewer processes exhibit movement leading to more stationary fluorescence, which is visualized in yellow as the overlap of earlier (red) versus later (green) time points. Process velocities for HOE-treated (n = 33 processes) and control mice (n = 28 processes) were similar throughout the duration of implantation and a likelihood ratio test revealed no group-wise differences in process velocities between HOE-treated and control mice over time. Therefore, these findings demonstrated that NHE-1 is not involved in the speed of microglial process extension following electrode implantation.

Figure 2.

Microglia extend processes in the direction of the microelectrode at velocities unaffected by HOE-642 treatment. (a) Microglial process extension around an implanted probe (outlined blue) over time in a HOE-treated mouse. Consecutive timepoints are labeled in red (preceding) and green (subsequent). Yellow (overlap of red and green) indicates no spatial movement of cells or processes between consecutive time points. Immediately following insertion, microglia processes move toward the electrode as indicated by layers of green end-feet followed by layers of red end-feet (white triangles). Multiple processes extend toward the probe surface until they initiate contact and stop moving (yellow overlap). (b) Velocity of actively migrating microglia processes in the direction of the electrode over time demonstrate active migration until 40–50 min after insertion, at which point most processes make contact with probe surface and stop moving. The velocities of microglia processes in HOE-treated (blue solid) and control (black dashed) animals follow similar trajectories, suggesting that HOE-642 does not influence rate at which processes migrate toward the electrode. Data is plotted as mean ± SEM.

3.2. HOE-642 administration does not influence microglial migration around implanted electrodes

Following process extension, microglia mobilize toward the probe to encapsulate the electrode surface and form a glial scar. To examine the effects of HOE-642 on cellular migration, microglia soma were tracked using z-stacks taken over time (figures 3(a) and (b)). The distance from the center of the cell soma to the nearest edge of the probe over consecutive timepoints was used to calculate the velocity of cell body migration in the direction of the implant (figures 3(c) and (d)). Consistent with previously published results, microglia soma began to migrate 12–24 h post-insertion and traveled at a relatively constant speed before contacting the probe surface. The velocity at which microglia migrated was relatively comparable between HOE-treated (n = 87 cells) and control mice (n = 132 cells) over time (p = 0.495, two-way ANOVA; figure 3(c)). Furthermore, there was no indication of a HOE-642 specific effect on soma migration with respect to distance from the probe surface (p = 0.443, two-way ANOVA; figure 3(d)). Despite an apparent difference in cell body velocity as a function of distance between groups, especially at 100 μm away from the probe surface, statistical analyses did not reveal a significant difference between reported values (figure 3(d)). These results suggest that HOE-642 did not impact the speed at which microglia soma migrated, and velocity was consistent for microglia of varied spatial distance from the probe surface between HOE-treated and control mice.

Figure 3.

HOE-642 administration does not influence microglia migration velocities following microelectrode implantation. (a), (b) Merged 2D images of microglia at 12 h (red) and 24 h (green) migrating toward an implanted microelectrode (shaded blue). Processes and cell soma of activated microglia migrate toward the surface of the probe in both control (a) and HOE (b) mice. Note the less activated morphology of microglia in HOE-treated mice compared to the control. (c), (d) Microglia migration begins around 12–24 h post-implantation. Microglia cell body velocity between HOE-treated and control mice remains fairly constant with respect to time (c) and distance (d) from the probe suggesting that, similar to microglia process extension, HOE-642 does not influence rate of cellular migration following microelectrode implantation. Data is plotted as mean ± SEM.

3.3. HOE-642 administration reduces morphological activation of microglia around implanted electrodes

Given that HOE-642 did not significantly impact process velocity, we next examined the effect on morphological activation of nearby microglial cells. In order to determine the effects of HOE-642 on morphology, microglia were first categorized as either ramified (1) or transitional (0) by observing the orientation of their processes on either side of a parallel line bisecting the cell body (figures 4(a) and (b)). Ramified and transitional values of microglia were then averaged within 50 μm bins to produce a spatial and temporal profile of microglial activation (ramification index) up to 250 μm away from the outer edge of the probe (figures 4(c)–(e)). Averages were taken on the level of individual microglia rather than per animal because different numbers of cells were quantified in different image stacks. This was due to variable visibility in the two-photon images caused by dynamically changing blood vessels and limited bleeding on the pial surface. The fitted curves represent the probability of microglia being ramified as a function of distance from the probe. By 2 h post-insertion, cells within 100 μm from the implant were almost exclusively activated (transitional) while distal microglia (>100 μm) were more likely to retain normal morphologies (ramified). Over time, these ramified microglia transitioned into an activated state with increasing distance from the probe surface (probability curves shifted further to the right). Examining the distribution of microglia ramification between groups revealed that microglia activation in mice treated with HOE-642 (solid line) was shifted to the left (less spatial activation) compared to the control group (dashed line) (figures 4(c)–(e)). During acute implantation (0–6 h), ramification in HOE-treated mice (n = 255 cells), while slightly left-shifted, did not demonstrate a significant difference compared to control mice (n = 153 cells) over time (p = 0.133, two-way ANOVA). Following sub-chronic implantation (6–72 h), ramification between HOE-treated and control mice became increasingly deviated, demonstrating significantly less microglial activation in HOE-treated (n = 445 cells) compared to control mice (n = 372 cells) over time (p < 0.0001, two-way ANOVA). We define the radius of microglia activation as the spatial distance from the probe at which microglia demonstrate a 50% probability of ramification, consistent with previously published studies [49, 50, 66]. Therefore, at 2 h post-insertion, the radius of activation for mice treated with HOE-642 was 63.4 μm compared to 79.0 μm for control mice. By 6 h post-insertion, this activation radius increased to 102.4 μm for HOE-treated animals versus 125.8 for control animals (figure 4(c)). Similar trends in activation curves were found in relation to treatment group and time following sub-chronic implantation (figures 4(d) and (e)). The radius of activation was determined to be 108.8 μm for HOE-treated mice versus 133.6 μm for control mice at 6 h. By 72 h, activation radius was 177.9 μm for HOE-treated animals versus 222.2 μm for control mice. The small differences in values in acute and sub-chronic mice at the overlapping timepoint of 6 h could be due to sampling variability or differences in surgical preparations. Statistical analysis between acute and sub-chronic preparations at overlapping 6 h timepoints were non-significant for both HOE-treated (p = 0.8689, Welch’s t-test) and control (p = 0.4468, Welch’s t-test) groups. In the sub-chronic mice, the rigid glass coverslip seal could affect the pressure exerted on the surface of the brain. For acutely prepared mice, exposing the brain to O₂ for 6 h with an open craniotomy could also affect how the tissue responds. Whether or not the preparations contributed to the slight difference in reported metrics between acute and sub-chronic groups, any differences between the HOE-treated and control mice other than the drug treatment were carefully controlled. Taking the difference in activation radii between experimental and control groups at the same timepoints revealed that there was an increasing separation in spatial ramification between control and HOE-treated mice, with HOE-treated mice demonstrating a smaller radius of microglia activation (figure 4(f)). These data indicate that HOE-642 administration reduced microglia activation, containing it to a smaller distance compared to control mice over all implantation time points examined.

Figure 4.

HOE-642 treatment reduces morphological activation of microglia in a spatiotemporal manner. (a), (b) Representative two-photon images from control (a) and HOE-treated (b) mice depict more ramified (R) and less transitional (T) microglia closer to the probe at 12 h post-implantation following HOE-642 treatment. Red dotted lines show the axis separating processes extending off the hemispheres toward and away from the probe surface (shaded blue). (c), (d) Average ramification indices between HOE-treated (circles and solid lines) and control (asterisks and dashed lines) mice at different time points post-insertion is shown with fitted logistic regression curves across distance away from the probe surface. Acutely implanted mice (c) were measured from 2 h to 6 h post-insertion (3, 4, and 5 h data omitted from plot for clarity). Sub-chronically implanted mice (d) and (e) were evaluated from 6 to 72 h. Plots were separated into 6–24 and 48–72 h to ease readability. For a given spatial bin, ramification generally decreased with time, indicating progressive microglia activation. Activation radii were defined as the distance at which there was a 50% probability for ramification (0.5 ramification index). HOE-642 administration visibly shifted the ramification curve leftwards (less activation) compared to controls indicating a reduction in microglia reactivity after implantation. (f) Difference in activation radii, defined as the distance at which ramification is 50% (0.5) at a given timepoint, between HOE-treated and control groups in sub-chronic mice over time. Error bars are not shown because values are not averages, but rather results from logistic regression analyses. Positive differences in activation radii suggest a greater radius of activation in control mice compared to HOE-treated mice.

The degree of microglia activation was further evaluated using a transitional (T) and directionality (D) index based on the length or number, respectively, of leading (towards) versus lagging (away) microglia processes (figures 5(a)–(f)). Index values closer to 1 indicated an equal length or number of processes facing toward versus away, whereas indices closer to 0 were an indication of greater length or number of processes preferentially oriented toward the implant. Values greater than one suggested a preferred orientation of processes away from the probe and could be the result of population variability or directed process extension toward an opposing blood vessel that may have been damaged due to tissue strain or inflammation [20, 24, 79]. In general, per a given spatial bin, HOE-642 administration resulted in more equal length or number of leading versus lagging processes (T-index and D-index closer to 1) compared to controls, indicating a reduced degree of microglia activation around the implant throughout the entire observed implantation period. Interestingly, statistical results revealed a significant interaction effect between treatment groups and binned distance from the implant during sub-chronic, but not acute, observation (T-index: p = 0.0016, D-index: p = 0.0282, two-way ANOVA). Although treatment with HOE-642 generally decreased the degree of microglial activation around the implant, there was a notable exception in the first 50 μm from the probe at 48 and 72 h post-implantation. These observations may be related to the fact that microglial soma begin to migrate 12–24 h post injury as noted earlier and published previously [49, 66].

Figure 5.

HOE-642 induces changes in number and length of processes directed towards and away from implants. Morphological T- and D-indices between HOE-treated (circles and solid lines) and control (asterisks and dashed lines) mice. (a)–(c) T-index calculates degree of activation depending on length of most prominent leading versus lagging microglia process with respect to the electrode. (d)–(f) D-index calculates degree of activation depending on number of leading versus lagging microglia processes with respect to the electrode. Values near one indicate ramification (equal in length or number of leading vs. lagging processes) and values near zero suggest activation with preferred orientation toward the electrode. Acutely implanted mice were observed from 2 to 6 h post-insertion (3, 4, and 5 h data omitted from plot for clarity). Values are fitted with custom dual sigmoidal functions and error bars indicate standard error. For a given spatial bin, T- and D-indices generally decreased with time, indicating a greater degree of microglia polarization. HOE-642 administration shifted T- and D-index curves upwards (more ramified) compared to controls suggesting attenuation microglia activation in response to electrode implantation.

3.4. HOE-642 administration reduces microglial encapsulation of implanted electrodes

Following microglia activation and migration, lamellipodia which make contact with the probe began to engulf the implant surface in a glial sheath. To evaluate the influence of HOE-642 on the extent of lamellipodia encapsulation, tissue directly above the probe shank (20 μm z-projection) was analyzed to determine percent of microglial (GFP-positive) surface coverage over time. Thresholded images of HOE-treated and control mice at 24 h post-implantation show an apparent reduction of microglial coverage over the surface of the probe (figures 6(a) and (b)). During acute implantation (2–6 h post-insertion), surface coverage in HOE-treated mice was significantly reduced compared to control mice (figure 6(c); two-way ANOVA, p < 0.05). The largest difference between groups was observed at 5 h post-implantation (50.7 ± 6.0% in control mice vs. 8.9 ± 6.1% in HOE mice). During chronic implantation, microglia surface coverage was also reduced in mice treated with HOE-642 compared to control mice only at 6 h post-implantation (figure 6(d); two-way ANOVA, p < 0.05). These differences were diminished during the sub-chronic period on account of an increase in surface coverage in HOE-treated mice over time. Significantly reduced surface coverage within 6 h post-insertion suggests that HOE-642 has a specific effect on lamellipodia encapsulation during acute implantation. The reduced encapsulation of probes at sub-chronic time points suggests that HOE-642 also affects the accumulation of microglia following migration toward the site of electrode implantation. However, the gradual increase in surface coverage of implanted devices in HOE-treated mice during the sub-chronic phase suggests that the efficacy of the drug to prevent microglia encapsulation diminishes over time.

Figure 6.

HOE-642 administration attenuates severe microglial encapsulation of intracortical electrodes. (a), (b) Thresholded microglial (GFP-positive) signal up to 20 μm above the surface of the implant (blue outline) in control (a) and HOE-treated (b) mice demonstrate an extensive reduction of microglia encapsulation in the HOE-treated mouse 5 h post-implantation. (c), (d) Percent of microglial surface coverage between HOE-treated and control mice at acute (c) and chronic (d) time points. HOE-642 treatment significantly lowered surface coverage for both acute (c) and chronic (d) mice (two-way ANOVAs; p < 0.001). Note that surface coverage in HOE-treated mice increased relative to control mice during the sub-chronic phase resulting in no significant difference in surface coverage between the two groups, suggesting a drop in efficacy of chronic HOE-642 administration. Statistical differences in surface coverage between HOE and control mice at individual timepoints are shown (Bonferroni corrected Welch’s t-tests; * p < 0.05). Data is plotted as mean ± SEM.

4. Discussion

Intracortical microelectrode arrays hold significant potential as advanced tools for neuroscience research and rehabilitation, yet their clinical and basic science applications are currently limited due to immune responses to a foreign body. Biological inflammation following implantation is perpetuated by the activation and encapsulation of glial cells, which is believed to contribute to neuronal cell death and a decline in device performance. Microglia are one of the first responders to device insertion and are sensitive to many signaling molecules secreted after implantation of a foreign body [56, 58, 80, 81]. NHE-1 is an Na⁺/H⁺ exchanger that can be inhibited following HOE-642 administration. Expression of NHE-1 is increasingly associated with proinflammatory microglia activation [68]. In this study, a HOE-642 injected group was compared against an un-injected control group representing current standard surgical practices of microelectrode implantation. This was done to test the hypothesis that HOE-642 reduces microglia activation around implanted microelectrodes compared to untreated control animals. Note that this is distinct from the hypothesis that HOE-642 prevents activation of microglia as compared against a contralateral non-implanted control, which was not the focus of the present investigation. Inhibition of NHE-1 activity with HOE-642 did not significantly impact microglia process or soma velocities following electrode implantation. However, HOE-642 administration did inhibit changes in morphology associated with microglia activation in a spatial- and time-dependent manner. These observations suggest that NHE-1 might serve a greater role preventing morphological microglial activation rather than impairing the motility of activated cells. Using longitudinal two-photon microscopy within the mouse cortex, we show that inhibiting NHE-1 with HOE-642 effectively reduced microglial activation and device encapsulation following microelectrode implantation.

4.1. HOE-642 administration reduces microglia activation and glial encapsulation following microelectrode implantation

Selective inhibition of NHE-1 via HOE-642 administration attenuated morphological microglia activation following electrode insertion, yet the mechanisms as to why are not clearly defined. NHE-1 localizes at the leading edge of fibroblasts, epithelial cells, and in microglial lamellipodia, suggesting it plays a role in cell shape and motility [70, 82]. Changes in the structure and morphology of microglia are typically dependent on the intracellular rearrangement of cytoskeletal proteins, specifically actin filaments, to facilitate protrusion or retraction of membranous processes [83]. Cofilin is one actin-binding protein that is responsible for both the resupply of actin monomers via disassembly of existing actin chains and generating additional actin-binding sites by severing filaments at the leading edge. Cofilin activity is inhibited when bound to phosphotidylinositol-4,5-bisphosphate (PI(4,5)P2) in a pH-sensitive process [84]. Therefore, cofilin depends on the efflux of H⁺ ions from NHE-1 in order to disassociate from PI(4,5)P2 and facilitate actin reorganization [85] and the inability to reorganize following NHE-1 inhibition could explain the observable reductions in glial activation after HOE-642 administration. NHE-1 itself anchors cytoskeletal actin filaments to the plasma membrane via its c-terminal binding to ezrin/radixin/moesin proteins [70, 80, 86]. However, it is unclear whether HOE-642 interferes with only NHE-1-mediated ion transport, cytoskeletal anchoring, or both, given that specific mutations of Nhe1 impairing ion transport still demonstrate the ability to regulate cytoskeletal organization [86]. Therefore, further studies are required to determine the specific mechanisms by which NHE-1 inhibition via HOE-642 attenuates microglia activation around intracortical microelectrode arrays.

Despite the inability for HOE-642 to reduce the rate at which microglia processes extend during acute implantation, it significantly reduced surface coverage of lamellipodia (figure 6, p < 0.05). This suggests that NHE-1 has a more pronounced role in the morphological remodeling of microglia end-feet, lamellipodia formation, and the motility and growth of lamellipodia sheaths than it does for process migration. Since soma migration does not occur until 12–24 h post-insertion, this significant difference in surface coverage within the first 6 h post-insertion is reflective of filopodia process coverage (known to be mediated by NHE-1 activity) rather than cellular encapsulation. The ability for HOE-642 to reduce microglia surface coverage appeared to diminish over sub-chronic implantation periods (figure 6). It is important to consider that the half-life of HOE-642 is approximately 4 h [87]. However, a major difference between this study and previous stroke and TBI studies evaluating the efficacy of HOE-642 is the presence of a foreign body response to an implanted device. It is possible that this persistent foreign body response and the spacing of drug delivery explains why the effects of HOE-642 diminished from 24 to 72 h. Future research should either explore a three dose per day dosing schedule or a controlled drug release mechanism to investigate if the early attenuation of probe surface coverage can be sustained over a chronic implantation period. Furthermore, differences in both microglia morphology and encapsulation as a result of HOE-642 administration likely reflect differences in microglia gene expression [88, 89]. Pharmacological inhibition via HOE-642 or selective removal of microglia NHE-1 results in reduced secretion of proinflammatory cytokines IL-1β, IL-6, and TNF-α while increasing expression of anti-inflammatory factors TGF-B and IL-10 following ischemia [68, 69]. Therefore, future studies should additionally further explore the relationship between microglia cellular structure and morphology with intracellular homeostatic mechanisms, gene expression, and inflammation following inhibition of NHE-1 after device implantation.

4.2. Microglia process extension and cell body migration following microelectrode implantation are unaffected by HOE-642

Inhibiting NHE-1 via HOE-642 administration following device implantation did not affect microglia process extension or soma migration (figures 2 and 3). Interestingly, a previous study demonstrated that inhibition of NHE-1 using HOE-642 altered the rate of lamellipodia protrusion and motility of microglia in response to the chemoattractant bradykinin [80]. However, these effects were observed in vitro in primary microglia cells cultured in an environment lacking a proper extracellular matrix (ECM), whose differences in substrate adhesion and an absence of natural ECM cues are likely to influence cellular migration [90–92]. It is important to note here that the velocities of process extension and soma movement were quantified only in microglia that were actively extending filopodia or migrating toward the electrode implant. This suggests that filopodia and lamellipodia migration, once initiated, are potentially mediated by alternative mechanisms not governed by NHE-1. For example, durotaxis is the process in which cells mobilize in response to stiffness gradients within their local microenvironment, which are more likely to operate through distinct mechanotransduction signaling pathways [93, 94]. It has been suggested that microglia become activated and migrate as a result of tissue stiffening that occurs around neural implants [90, 95]. Therefore, alternative microglia signaling proteins and pathways should be investigated that are involved more specifically in the motility of processes and cell bodies after implantation injury.

4.3. Future directions

Therapeutic interventions that modulate microglial activation have been previously studied in an effort to improve electrode–tissue integration and chronic device performance. Similar to the effects of HOE-642 reported in this study, the anti-inflammatory dexamethasone also attenuates microglia activation around acutely implanted probes [51, 96]. Use of melatonin or minocycline have both demonstrated improved recordability of neuronal activity alongside reduced microglia activation [97–100]. From a biomaterial perspective, neuroadhesive L1CAM coatings have been shown to attenuate attachment of microglia and reduce neurodegeneration around neural probes, resulting in improved recording longevity [50, 101]. These prior studies therefore suggest that the modulation of microglia activation via HOE-642 administration has the potential to enhance the recording performance of implanted microelectrode arrays. However, unlike HOE-642, the mechanism of action of these drugs on microglia function are non-specific. A greater understanding of how these different interventions act on specific cellular functions are required in order to improve the design and development of future pharmacological and biomaterial therapies.

Future efforts should also explore the role of controlled drug delivery systems to continuously administer HOE-642 at the appropriate dosage for the duration of the acute inflammatory tissue response period. Prior delivery systems released drugs from polymer coatings on devices, conductive polymer coatings or porous metal electrode sites via electrical stimulation, resorbable polymeric coatings, hydrogels, and microdialysis infusion pumps [102–114]. Sustained, continuous release at the therapeutic concentration window within the injury microenvironment should improve efficacy and minimize activation of microglia through the acute inflammatory period. Furthermore, because the current research explores a phasic systemic injection 8 h apart, twice a day, it is possible that microglia slowly activate over days when the pharmacokinetics of the HOE-642 drops below the minimum effective concentration. Future studies should also examine if reducing microglial activation in the inflammatory period attenuates microglia and macrophage activation during the chronic phase. For example, if inflammasomes are cleared from the tissue before a continuous HOE-642 drug perfusion system is depleted, it may be possible to inhibit microglia without the need for drug replenishment. The current work highlights the importance of drug delivery systems for both therapeutic treatment and investigative research.

5. Conclusion

This study assessed a novel therapeutic approach to reducing microglial activation and encapsulation following microelectrode implantation within the mouse cortex. HOE-642, an inhibitor of NHE-1, was found to attenuate the morphological changes in microglia associated with device-induced glial activation. While microglia processes and soma migration velocities were not affected by HOE-642 treatment following implantation, microglia activation and encapsulation were significantly attenuated compared to untreated controls. The radius of microglial activation around implanted probes was reduced acutely from 133.6 μm to 108.8 μm at 6 h post-insertion and more chronically from 222.2 μm to 177.9 μm at 72 h post-insertion following HOE-642 treatment. Furthermore, HOE-642 administration significantly reduced the percentage of microglial surface coverage from 50.7 ± 6.0% to 8.9 ± 6.1% at 5 h post-insertion and from 27.0 ± 2.9% to 20.1 ± 3.8% at 72 h compared to controls. Overall, HOE-642 treatment significantly attenuated the microgliosis observed after the initial insertion and throughout the sub-chronic implantation of intracortical microelectrode arrays. This study introduces HOE-642 as a potential novel intervention to reduce the immune response and subsequent glial scarring following intracortical microelectrode implantation.