Platelets and Hemostatic Proteins are Co-Localized with Chronic Neuroinflammation Surrounding Implanted Intracortical Microelectrodes

Abstract

Intracortical microelectrodes induce vascular injury upon insertion into the cortex. As blood vessels rupture, blood proteins and blood-derived cells (including platelets) are introduced into the ‘immune privileged’ brain tissues at higher-than-normal levels, passing through the damaged blood-brain barrier. Blood proteins adhere to implant surfaces, increasing the likelihood of cellular recognition leading to activation of immune and inflammatory cells. Persistent neuroinflammation is a major contributing factor to declining microelectrode recording performance. We investigated the spatial and temporal relationship of blood proteins fibrinogen and von Willebrand Factor (vWF), platelets, and type IV collagen, in relation to glial scarring markers for microglia and astrocytes following implantation of non-functional multi-shank silicon microelectrode probes into rats. Together with type IV collagen, fibrinogen and vWF augment platelet recruitment, activation, and aggregation. Our main results indicate blood proteins participating in hemostasis (fibrinogen and vWF) persisted at the microelectrode interface for up to 8-weeks after implantation. Further, type IV collagen and platelets surrounded the probe interface with similar spatial and temporal trends as vWF and fibrinogen. In addition to prolonged blood-brain barrier instability, specific blood and extracellular matrix proteins may play a role in promoting the inflammatory activation of platelets and recruitment to the microelectrode interface.

Graphical abstract

Implanted microelectrodes have substantial potential for restoring function to people with paralysis and amputation by providing signals that feed into natural control algorithms that drive prosthetic devices. Unfortunately, these microelectrodes do not display robust performance over time. Persistent neuroinflammation is widely thought to be a primary contributor to the devices’ progressive decline in performance.

Our manuscript reports on the highly local and persistent accumulation of platelets and hemostatic blood proteins around the microelectrode interface of brain implants. To our knowledge neuroinflammation driven by cellular and non-cellular responses associated with hemostasis and coagulation has not been rigorously quantified elsewhere.

Our findings identify potential targets for therapeutic intervention and a better understanding of the driving mechanisms to neuroinflammation in the brain.

Introduction

Intracortical microelectrodes allow direct communication with neural tissue by measuring the underlying cortical activity or modulation of brain circuitry through electrical stimulation [1–5]. Compared to other cortical interfaces, such as EEG or ECoG arrays, tissue-penetrating intracortical microelectrodes offer direct access to neurons throughout cortical layers for stimulating or recording individual neurons [6–13]. Recent brain-computer interfaces (BCIs), using penetrating intracortical arrays, have shown clinical viability of rehabilitation and restoration in patients with motor and sensory impairments [5,14–17].

Although there is a significant push toward using BCIs for developing clinical therapies, the performance of implanted intracortical microelectrodes to communicate with nearby neurons rapidly decays over time [18–24]. The decline in functional performance is in part due to a triggered and persistent foreign body response, initially driven by the invasiveness of the implantation procedure [20–24]. Upon craniotomy and insertion, the neurovascular is damaged, leading to disruption of the blood-brain barrier (BBB) [25–30]. Consequently, BBB damage allows for the infiltration of pro-inflammatory cells and blood proteins [20,25,31,32]. Neuronal and glial bodies within the implant insertion path are also damaged, contributing to a focal lesion and local neurodegeneration [22,33–35]. In addition to cellular debris accumulation, cytokines are released from apoptotic and inflammatory cells, resulting in the activation of resident microglia and infiltrating macrophages from the damaged surrounding vasculature [25,36]. Glial activation in turn recruits immune cells that sustain local neurodegeneration as a response to cytotoxic compounds, such as reactive oxygen species (ROS), promoting acute and chronic neuroinflammation [33,37,38]. Over time, a sustained glial scar is developed within the immediate boundaries of the intracortical microelectrode, acting as physical barrier from the native neural tissue [39,40]. While many studies have proposed changes in substrate material, shank dimensions and size, or physical surface modifications, the signal quality of recorded neural activity continues to decline over time [7,13,18,41–44].

Severed cortical blood vessels further exacerbate neuroinflammation as a significant influx of bloodborne cells and blood proteins are rapidly introduced to the microelectrode interface [25,26]. In particular, platelets are generally known for driving hemostasis and wound healing after vascular trauma [45]. Platelet and blood protein fouling to biomaterial surfaces influence biological responses, including thrombosis, complement activation, and immune recognition, and they remain a significant challenge for tissue integration [46–50]. In vitro studies have suggested that these blood proteins can be neurotoxic and impact neuronal viability [51–53].

Previous tracer studies have suggested that the foreign body response leads to a prolonged BBB instability allowing for the leakage of blood constituents into the cortical extracellular space [21,24,54]. The presence of blood proteins such as albumin and immunoglobulin G (IgG) persist at the microelectrode interface and encompass its immediate perimeter [21,24,34,35]. Vascular leakage is caused by dysfunction of either endothelial junctions (paracellular pathway) or through the endothelial cells (transcellular pathway). Although both pathways are impacted after the initial device insertion, plasma and its constituents are regulated through passive diffusion through these intercellular junctions, adherens and tights junctions [55]. As the BBB remains chronically unstable and allows for blood proteins to continuously diffuse into the brain parenchyma, there is an opportunity to investigate platelet physiological function around cortical implants and its impact to the leaky BBB. Although platelets’ primary function is to prevent vascular leaks and promote wound healing, under inflammatory conditions activated platelets have been suggested to promote leukocytic migration through modulating endothelial junctions [32]–[36], [42]–[43]. Furthermore, recent studies highlight that chronic inflammation may be perpetuated through activation and recruitment of inflammatory cells through expressed platelet-derived CD40L and secretory products [57–59].

In this study, we profiled the spatial-temporal relationship of platelet and common proteins that participate in promoting hemostasis (von Willebrand Factor, fibrinogen, and type IV collagen) following intracortical implant-induced neuroinflammation. Here, we quantified common blood protein and activated platelet markers with respect to traditional biomarkers of neuroinflammation, activated microglia, and astrocytes, to identify an overlooked aspect of the neurodegenerative neuroinflammatory response to implanted intracortical microelectrode. Characterization of the role that platelets and their associated hemostatic proteins play in microelectrode performance may identify targets for localized immunotherapy or strategies to dampen the neuroinflammatory response.

Methods

Non-functional Multi-shank Silicon Microelectrode Fabrication

Many studies have used single shank silicon shanks to study biological interactions with intracortical microelectrodes [7,20,33,60,61]. However, Utah electrode arrays are the most-widely used penetrating intracortical implants for long-term human trials in BCI applications. The multi-shank arrays induce multiple injury sites to the underlying neurovascular unit compared to single planar shanks [5,14–16]. In mimicking a multi-stab injury model while reducing complexity such as a 3D electrode array, we opted for a multi-shank planar microelectrode approach.

Multi-shank planar microelectrodes were microfabricated similar to a prior study using standard cleanroom processes (Case Western Reserve University) from a 350 μm thick silicon-on-insulator (SOI) wafer with a top silicon thickness of 15 μm and a buried oxide thickness of 1.5 μm [62]. Wafers were spin-coated with a positive photoresist (Shipley S1813, Dow Chemical, Midland, MI, USA). After photoresist exposure to define final probe geometry, deep reactive ion etching was performed. The wafers were then soaked in solvent to remove the photoresist residues and in diluted 10:1 hydrofluoric acid overnight to lift-off the silicon shanks before they were triple rinsed in distilled water. Individual microelectrodes were released from the silicon wafer carefully using forceps and had final dimensions of 3250 μm × 120 μm × 15 μm (length, width of individual shank, and thickness), as shown in Fig. S1A. Microelectrodes were dipped into 70% ethanol for 5 minutes prior to storage in sterilization bags with marked colored-changing indicators. Sterilization bags underwent ethylene oxide gas sterilization and degassed for a minimum of 12 hours prior to implantation.

Animal Use and Surgical Procedures

All animal procedures complied with the ARRIVE guidelines and were approved by the Institutional Animal Care and Use Committee (IACUC) at the Louis Stokes Cleveland Department of Veterans Affairs Medical Center. Male Sprague Dawley Rats (225–250 g) were ordered from Charles River Laboratories, Wilmington, MA, and housed for up to 2 weeks prior to surgery. Surgical procedures followed established protocols in the lab and will be briefly described here [30,61,63]. Anesthesia was induced at 3–4% isoflurane and maintained at 1.5–2% during surgery. After anesthetic induction, rats were then mounted in a stereotaxic frame via ear bars and incisor bar. Cefazolin (16 mg/kg) and meloxicam (1 mg/kg) were subcutaneously administered to prevent infection and for pain management. Marcaine (0.25%, 0.2 mL) was subcutaneously injected at the incision site prior to hair removal and cleaning with betadine and isopropyl alcohol. A 1-inch midline incision of the scalp was made to expose the skull. Bilateral craniotomies at a diameter of 1 mm were performed over the somatosensory cortex of each hemisphere (2 mm lateral to midline, 3 mm posterior to bregma) using a dental drill at 15,000 RPM at 2–5 seconds intervals with a saline wash in between [30]. The dura mater was reflected using a dura pick, followed by manual insertion of a non-functional multi-shank silicon microelectrode, approximately 2 mm into the brain. Saline was added to exposed cortical tissue to maintain moist environment after microelectrode insertions while maintaining a dry skull area. Silicone elastomer (Kwik-Sil, World Precision Instruments) was applied to completely cover both craniotomies to seal implanted probes. Cold-cure dental acrylic (AM-Systems) was applied over the silicone elastomer package to form a stable cement base to the skull. The incision site was sutured with 5–0 monofilament polypropylene suture.

Platelet and Hemostatic Protein Immunohistochemistry (IHC)

Rats (n=12) were equally assigned to terminal endpoints of 1-, 4-, and 8-weeks for immunohistochemistry tissue processing. At terminal endpoints, animals were anesthetized via intraperitoneal injections of ketamine (160 mg/kg) and xylazine (20 mg/kg), followed by transcardial perfusion with phosphate buffer solution and 10% neutral buffered formalin. Brains were processed for immunohistochemistry.

Immunohistochemistry and Tissue Processing

Brains were post-fixed in 10% neutral buffered formalin for 24 hours after extraction, followed by stepwise gradient sucrose cryoprotection at 10%, 20% and 30%. Cryoprotected brains were frozen on dry ice and transversely cryo-sectioned at 20 μm thickness. As a result of slice-to-slice variability, we strived to include a representative sampling of tissues from multiple depths. In our established lab protocols, random depth sampling is performed; in other words, within each set of stained tissue from a given brain, we include a minimum of 3 slides, each containing 3 slices randomly selected from a multitude of depths.

Slice depth was estimated based on tracking sequential slices collected in our cryostat. The OCT-embedded block was mounted on the specimen holder and trimmed at 20–100 µm section thickness until the surface of the brain was reached, and the approximate location of the hole was identified. This location was marked as the “zero” depth and used as reference for calculating subsequent slice depths. Up to 3 sections were collected on each slide. Serial sections were collected on different slides to produce slides with tissues representing variable depths. Due to inherent variability and human subjectivity in the process, the precision of our slide depth estimate is likely on the order of +/− 100–200 microns.

Mounted slides were washed and permeabilized with 0.1% Triton-X 100 solution, followed by blocking for non-specific antibody adsorption through incubation in 4% goat serum solution. Slides were then incubated in primary antibodies overnight at 4°C. Unbound primary antibodies were washed with 0.1% Triton-X 100 solution, followed by 2-hour incubation of secondary antibodies. Antibodies used for immunohistochemistry were summarized in Table 1. Primary and secondary antibody control slides were run to confirm the lack of non-specific staining.

Table 1.

Summary of Antibodies used for IHC.

Target	Antigen	1° Antibody	Dilution	2° Antibody	Dilution
Platelets	CD41	Invitrogen PA5–79527	1:250	Invitrogen A11037	1:1000
Platelets (Activated)	CD62p	BioLegend 148392	1:250	Invitrogen A11029	1:1000
von Willebrand Factor	von Willebrand antigen 2 / pro-vWF	Proteintech 11778–1-AP	1:250	Invitrogen A11037	1:1000
Type IV Collagen	Type IV Collagen	Invitrogen MA1–22148	1:250	Invitrogen A11029	1:1000
Fibrinogen	Fibrinogen Alpha/ Beta Chains	Invitrogen MA1–26074	1:250	Invitrogen A11029	1:1000
Microglia (Activated)	CD68	Sigma-Aldrich MAB1435	1:200	Invitrogen A11029	1:1000
Astrocytes	GFAP	Dako Z033429–2	1:500	Invitrogen A11037	1:1000
Cellular Nuclei	AT regions, DNA	N/A	N/A	Invitrogen D3571	1:3000

Immunohistochemistry Analysis

Slides were imaged with an automated digital slide scanner, Axio Scan.Z1 (ZEISS) with a 20x/0.8 plan-apochromat objective (ZEISS). Fluorescent markers and brightfield images were captured with a Hamamatsu Orca Flash 4.0 (Hamamatsu). In determining imaging parameters for this study, 2 slides (containing at least 3 sections of tissue) from the 8-week time point were randomly selected and stained for histological markers: CD41, CD62p, vWF, type IV collagen, fibrinogen, CD68, and GFAP. Imaging parameters for light source and exposure times were adjusted for respective antibodies to account for the lowest energy settings needed to excite fluorophores. Afterwards, established imaging settings for respective antibodies were applied for all slides for consistent imaging results. These respective tissue sections used to establish imaging parameters were removed from data analysis. Gain settings (0 dB) were kept consistent for all tissue sections. Specific imaging parameters including light source, light source intensity, illumination wavelength, and fixed exposure times are listed for respective antibody targets, Table 2.

Table 2.

Imaging Parameters for Respective Antibody Targets.

Target	Light Source	Light Source Intensity (%)	Illumination Wavelength (nm)	Exposure Time (ms)
Platelets	LED 567nm	100.0	577–604	250
Platelets (Activated)	LED 475nm	100.0	450–488	100
von Willebrand Factor	LED 567nm	100.0	577–604	150
Type IV Collagen	LED 475nm	100.0	450–488	150
Fibrinogen	LED 475nm	100.0	450–488	150
Microglia (Activated)	LED 475nm	100.0	450–488	80
Astrocytes	LED 567nm	100.0	577–604	75
Cellular Nuclei	LED 385nm	100.0	370–400	1
Brightfield Tissue	TL LED Lamp	1.50	na	2

ZEN 3.1 Blue Edition (ZEISS) software was used to export 16-bit images as .TIFF files at a conversion factor of 0.325 μm per pixel for respective fluorescence (AF488, AF594, and DAPI) and brightfield channels. Image export settings were set to not include contrast adjustments or tonal value corrections for reliable raw image analysis. Images were then analyzed with a previously written MATLAB (Mathworks, Inc) script [64]. In brief, the region-of-interest (ROI) were drawn around a given individual explant hole. The edges of the explant holes were verified from multiple channels (including brightfield) to ensure user-defined ROIs were devoid of tissue and accurately reflect the edges of the tissue. Remaining explant holes were excluded from the fluorescent intensity analysis of up to 100 μm. Concentric rings, based on user-defined ROI around the explant holes, were created throughout the image at fixed bin sizes to measure the fluorescence intensity as a function of distance from the explant region. Raw immunofluorescent intensity of each respective channel was normalized to the background signal at 550 μm away from the implant area as shown in Equation 1, where ${\overline{\text{I}}}_{\text{normal}}$ is the mean normalized intensity, ${\overline{\text{I}}}_{\text{raw}}$ is the mean raw intensity values for a given distance bin, ${\overline{\text{I}}}_{\text{background}}$ is the mean background intensity, and $\text{K}$ is the normalization constant. The normalization factor was set as either 0 or 1 and was determined if histological markers were non-natively expressed (e.g., CD68) or had continual expression throughout the cortical tissue (e.g., GFAP), respectively. Mean and standard error of means for each respective fluorescent channel were reported at bins of 50 μm. ROIs, exclusion zones, and fluorescence intensity analysis were repeated for remaining explant holes. As a result of the inter-shank spacing being <200 µm, and potentially causing confounding interpretation at greater distances, we have limited our statistical data interpretation up to the 150 µm bin.

$${\overline{I}}_{normal}={\overline{I}}_{raw}/{\overline{I}}_{background}-K$$ (1)

Statistics

Cellular responses for each implanted shank of the non-functional silicon microelectrode probes (up to four) were treated as dependents to variety of factors including subjects, cortical hemispheres, implanted shanks, and cortical tissue depth. We explored the differences in biomarker levels across time points independently for various distance measures using linear mixed models with subject-specific random intercepts to adjust for intra-individual correlation (depth and hemisphere). Each observation was nested at each depth and side within each animal. Overall, we looked at the data from 12 different animals, 4 for each time points (1-week, 4-week, and 8-week). Data were collected from four different holes from each side of the brain at different depths. This resulted in a total of 244 observations for each distance bin and for each blood protein marker. The number of observations at different depths varied for each animal. Full models were adjusted for side of the brain and depth. Multiple pair-wise comparison between each distance measures were performed using Tukey’s method. Analyses were performed using SAS Enterprise Guide, version 8.3 (SAS Institute Inc.). Furthermore, a general linear model was built in Python 3.9 to assess statistical significance between time points for respective 50 μm distance bins followed by Tukey’s HSD with Bonferroni correction. Statistical significance represented as: * = p<0.05, ** = p<0.01, and *** = p<0.001.

Results

Immunohistochemistry for type IV collagen and CD41+ platelets

To investigate the impact of microelectrode implantation and residing in cortical tissue on the trauma and integrity of nearby vasculature, we examined the immunoreactivity for type IV collagen, Fig. 1A. Type IV collagen is exclusively found in basement membranes and is widely distributed throughout the basal lamina [65–67]. Basal lamina is the layer of extracellular matrix secreted by the epithelial cells, on which the epithelium sits [65]. Here we detected chronically sustained type IV collagen response up to 8-weeks after the initial implantation, respectively. When compared to the background tissue (550 μm away from the microelectrode interface), type IV collagen responses remained high in immunoreactivity for distances of up to 150 μm from the microelectrode interface at 1-week (p<0.01) and 4-weeks (p<0.001) and up to 50 μm at 8-weeks (p<0.001), Fig. 1B. Peak response of type IV collagen was observed at 50 μm for all considered time points and had a reduction of 0.29-fold when comparing responses at 1-week and 8-weeks. This decline in response was observed for up to 150 μm (p<0.001). No significant differences in type IV collagen response were observed between responses at 1-week vs 4-weeks when comparing all distances at respective 50 μm bins. In addition, there was a reduction in type IV collagen response at distances of up to 150 μm when comparing responses at 4-weeks and 8-weeks, suggesting a potential transition during the scarring process (p<0.01).

Fig. 1. Sustained immunoreactivity of CD41+ platelets and type IV collagen at the microelectrode interface for up to 8-weeks with quantification of normalized immunofluorescent intensity as a function of distance from the microelectrode interface.

A) Representative co-staining for platelets and type IV collagen across 1-week, 4-week, and 8-week time points. DAPI (nuclei) and brightfield microscopy images are included as a reference; scale bar = 50 μm. B) Normalized type IV collagen immunofluorescent intensity showed persistent response up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). C) Normalized CD41+ platelet immunofluorescent intensity showed persistent response up to 8-weeks. ***p<0.001; **p<0.01; *p<0.05. All errors reported as SEM.

Type IV collagen acts as a ligand for platelet recruitment to inflamed vascular tissue [66,68]. Tissues stained with type IV collagen antibodies were also co-stained with the platelet marker CD41 to review co-localization, Fig. 1A. A separate set of tissues stained with CD41 were also co-stained with CD68 for activated microglia and was included in the histological analysis for CD41, Fig. 4A. Increased congregation of platelets are generally expressed in nearby damaged vasculature. When compared to the background tissue, CD41+ platelets remained high in immunoreactivity for distances of up to 150 μm from the microelectrode interface for all considered time points (p<0.01), Fig. 1C. No significant differences in CD41+ response were observed between responses at 1-week vs 4-weeks, 1-week vs 8-weeks, or 4-weeks vs 8-weeks when comparing all distances at respective 50 μm bins. DAPI staining showed stained proteins of interest may concentrate locally with surrounding cells and may participate in the scar-formation at chronic time points. Brightfield images allow for inspection of scar tissue and potential blood clots and may indicate distinct localization of proteins of interest and platelets.

Fig. 4. Sustained immunoreactivity of activated microglia (CD68) at the microelectrode interface for up to 8-weeks with quantification of normalized immunofluorescent intensity as a function of distance from the microelectrode interface.

A) Representative staining for activated microglia across 1-week, 4-week, and 8-week time points. DAPI (nuclei) and brightfield microscopy images are included as a reference; scale bar = 50 μm. B) CD68 immunofluorescent intensity showed persistent activated microglia phenotype up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). ***p<0.001; **p<0.01; *p<0.05. All errors reported as SEM.

Immunohistochemistry for CD62p+ platelets and von Willebrand Factor

P-selectin (CD62p) is an extracellular marker expressed on platelets after activation and enables platelet-leukocytic interactions [69,70]. Here, we stained cortical slices of varying depths for P-selectin to investigate if cortical-introduced platelets are in an inflammatory condition, Fig. 2A. When compared to the background tissue, CD62p+ activated platelets remained high in immunoreactivity for distances of up to 150 μm from the microelectrode interface for all considered time points (p<0.05), Fig. 2B. Peak CD62p+ response was observed at 50 μm from the microelectrode interface for all time points and had a reduction of 0.45-fold when comparing responses at 1-week and 8-weeks. This decline in CD62p+ response was observed for up to 150 μm (p<0.01). In addition, there was a reduction in CD62p+ response at distances of 50–150 μm when comparing responses at 1-week and 4-weeks (p<0.05). A reduction in CD62p+ response was observed between responses at 4-weeks and 8-weeks at distances up to 50 μm from the microelectrode interface (p<0.05).

Fig. 2. Sustained immunoreactivity of CD62p+ platelets and vWF at the microelectrode interface for up to 8-weeks with quantification of normalized immunofluorescent intensity as a function of distance from the microelectrode interface.

A) Representative co-staining for activated platelets and vWF across 1-week, 4-week, and 8-week time points. DAPI (nuclei) and brightfield microscopy images are included as a reference; scale bar = 50 μm. B) Normalized CD62p+ immunofluorescent intensity showed persistent activated platelet phenotype up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). C) Normalized vWF immunofluorescent intensity showed persistent response up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). ***p<0.001; **p<0.01; *p<0.05. All errors reported as SEM.

Cortical slices were further co-stained for von Willebrand Factor (vWF), Fig. 2A. vWF response result of damaged endothelial layers and promotes platelet and leukocyte recruitment and infiltration to inflamed tissue. When compared to the background tissue, vWF remained high in immunoreactivity for distances up to 150 μm from the microelectrode interface at 1-week (p<0.05), up to 100 μm at 4-weeks (p<0.001), and up to 100 μm at 8-weeks (p<0.05), Fig. 2C. Peak vWF response was observed at 50 μm for all considered time points and had a reduction of 0.49-fold when comparing responses at 1-week and 8-weeks. This decline in vWF response was observed for up to 150 μm (p<0.01). No significant differences in vWF response were observed between responses at 1-week and 4-weeks when comparing all distances at respective 50 μm bins. Additionally, a reduction in vWF response was observed between responses at 4-weeks and 8-weeks at distances of up to 100 μm (p<0.05).

Immunohistochemistry for fibrinogen and immune-activated astrocytes

hronic blood-brain barrier dysfunction impairs neuronal health, while the influx of blood proteins promotes and sustains inflammatory conditions. Fibrinogen was investigated as a blood protein of interest as this protein promotes platelet activation and aggregation, Fig. 3A. When compared to the background tissue, fibrinogen remained high in immunoreactivity for distances up to 150 μm from the microelectrode interface at 1-week (p<0.001) and 4-weeks (p<0.05), and up to 50 μm at 8-weeks (p<0.001), Fig. 3B. Peak fibrinogen response was observed at 50 μm for all considered time points and had reduction of 0.23-fold when comparing responses at 1-week and 8-weeks. This decline in fibrinogen response was observed for up to 150 μm distance from the microelectrode interface (p<0.01). In addition, there was a reduction in fibrinogen response at distances from 50–150 μm when comparing responses at 1-week and 4-weeks (p<0.01). A reduction in fibrinogen response was observed between responses at 4-weeks and 8-weeks at distances up to 150 μm from the microelectrode interface (p<0.05).

Fig. 3. Sustained immunoreactivity of fibrinogen and activated astrocytes (GFAP) at the microelectrode interface for up to 8-weeks with quantification of normalized immunofluorescent intensity as a function of distance from the microelectrode interface.

A) Representative co-staining for fibrinogen and activated astrocytes across 1-week, 4-week, and 8-week time points. DAPI (nuclei) and brightfield microscopy images are included as a reference; scale bar = 50 μm. B) Normalized fibrinogen immunofluorescent intensity showed persistent response up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). C) Normalized GFAP immunofluorescent intensity showed persistent response up to 8-weeks with highest response at 50 µm from the microelectrode interface when compared to background tissue (p<0.001). ***p<0.001; **p<0.01; *p<0.05. All errors reported as SEM.

Cortical slices stained with fibrinogen were further co-stained for activated astrocytes (GFAP) to see potential spatial relationships, Fig. 3A. When compared to the background tissue, activated astrocytes remained high in immunoreactivity for distances up to 150 μm from the microelectrode interface for all considered time points (p<0.001), Fig. 3C. Peak GFAP response was observed at 50 μm for all time points and had an amplification of 2.14-fold when comparing responses at 1-week and 8-weeks. This increase in GFAP response was observed for up to 100 μm distance from the microelectrode interface (p<0.001). In addition, there was an increase in GFAP response at distances up to 150 μm when comparing responses at 1-week and 4-weeks (p<0.05). An increase in GFAP response was observed between responses at 4-weeks and 8-weeks at distances up to 50 μm from the microelectrode interface (p<0.05).

Immunohistochemistry for immune-activated microglia

Persistent influx of blood proteins and platelets may promote recruitment and activation of immune cells within the cortical insertion site [45,48,50,59,71,72]. Activated microglia (CD68) have been regarded key contributors to neuroinflammation for implanted intracortical microelectrodes, Fig. 4A. When compared to the background tissue, activated microglia remained high in immunoreactivity for distances up to 150 μm from the microelectrode interface at 1-week (p<0.05), up to 100 μm at 4-weeks (p<0.01), and up to 150 μm at 8-weeks (p<0.05), Fig. 4B. Peak CD68 response was observed at 50 μm for all considered time points and had a reduction of 0.6-fold when comparing responses at 1-week and 8-weeks. This decline in response was observed for distances up to 150 μm from the microelectrode interface, (p<0.01). Additionally, a reduction in CD68 response was observed between responses at 4-weeks and 8-weeks at distances of up to 100 μm (p<0.05). No significant differences in CD68 response were observed between responses at 1-week and 4-weeks when comparing all distances at respective 50 μm bins.

Discussion

In this study, we proposed a conceptual framework of vascular trauma after microelectrode implantation and that the initial insertion trauma may chronically influence the prolonged neuroinflammatory response, Fig. 5. Our framework synthesizes substantial effort from numerous studies published over the years which have all pointed to vascular injury being a central challenge to microelectrode longevity [20,34,73,74].

Fig. 5. Schematic of thrombo-inflammation induced by implanted intracortical microelectrodes.

A) Microhemorrhage occurs as blood vessels rupture during device insertion. B) Schematic of spontaneous platelet activation and its contributions to thrombo-inflammation. Before injury (1), bloodborne cells, including platelets and circulating proteins are trafficked through blood vessels. Upon vascular damage (2), endothelial cells become inflammatory and release compounds such as von Willebrand Factor (vWF) that promotes platelet adhesion to injured blood vessels via exposed type IV collagen. Activated platelets (3) initiate hemostasis and promote thrombosis through interactions with vWF and fibrinogen. Furthermore, activated platelets promote leukocytic (monocyte) adhesion and infiltration to the inflamed environment. Infiltrating blood compounds and bloodborne cells adhere to the microelectrode surface and may influence promote neuroinflammation through interactions and further activation of microglia and resident astrocytes. C) Example immunohistochemistry of chronic glial scarring (GFAP, red) against an implanted non-functional multi-shank silicon microelectrode with infiltration of blood protein (fibrinogen, green) around the 2^nd and 3^rd implant region at 8-weeks post-implantation; scale bar = 100 μm. Blood proteins had the highest response at 1-week implants and subside over time as nearby blood vessels begin to vascular repair. However, blood proteins may remain actively leaking into the cortical extracellular matrix as a function of blood-brain barrier instability as shown by their persistent presence.

Vascular trauma is unavoidable during microelectrode insertion. Although, large surface blood vessels can be avoided, microhemorrhaging is inevitable due to expansive network of capillaries, arterioles, and venules that extends throughout the cortex [75,76]. When vascular trauma is triggered, platelets are recruited to induce hemostasis and coagulation, preventing considerable accumulation of fluid and blood byproducts from entering the cortical extracellular matrix. Although platelets have been well-documented in the periphery, their physiological function is not well-known in brain pathology – including neuroinflammation [45,77,78]. In our study, we demonstrated the chronic presence of known inflammatory markers around insertion sites of our non-functional multi-shank microelectrodes which are involved in the vascular trauma and hemostatic pathway including platelets, von Willebrand Factor, fibrinogen, P-selectin (CD62p), and type IV collagen.

Local neurovasculature remains in a thromboinflammatory state

Insertion of intracortical microelectrodes leads to severing of ruptured blood vessels and an inflow of blood components into the brain extracellular space. Vascular damage causes endothelium activation and leads to expression of adhesion molecules, such as von Willebrand Factor and P-selectin, which promotes tethering and adhesion and recruitment of platelets and leukocytes. Furthermore, endothelium dysfunction leads to introduction of plasma FVII/FVIIa which binds to exposed tissue factor, expressed on pericytes and predominately by astrocytes, suggesting another method for continuous platelet activation [79,80]. Endothelium dysfunction can be sustained as an effect of continuous production of reactive oxygen species within the inflamed cortical space [26,37,55].

We have confirmed the presence of platelets around each shank of our implanted multi-shank probes after 1-week with persistent accumulation up to 8-weeks via immunohistochemistry for CD41+ platelets, an integrin complex on that promotes platelet aggregation, Fig. 1C [81]. Platelets are well-documented to adhere to biomaterial surfaces and act as a mediator for cellular recognition and attachment sites [71,82,83]. Platelets are not present in the cortical extracellular matrix unless vascular damage occurred, Fig. S3. Interestingly, no statistical difference was found between all observed time points when comparing respective 50-μm bins suggesting no significant change has occurred in terms of CD41+ platelet accumulation around the microelectrode interface. However, our results showed significant accumulation of platelets after microelectrode implantation for all considered time points at distances up to 150 μm (p<0.01). This local accumulation of platelets may be influenced by nearby vasculature that are more reactive during the first few weeks of injury [84–86]. Further, it has been suggested that the recording performance of intracortical microelectrodes are heavily dependent on the health of individual neurons located at distances <200 µm from the implant [22,87,88]. Recovery of individual neurons located in this boundary are heavily influenced by the initial insertion-related damage and the prolonged inflammation from multiple cellular and non-cellular sources, including a sustained influx of active platelets and blood proteins. Many studies suggest that persistent BBB leakage heavily influence neuroinflammation and is a primary driver in poor microelectrode recording performances [20,21,23].

Furthermore, these introduced platelets may be active in phenotype, as suggested by CD62p+ immunohistochemistry, and persisted in chronic injuries at distances of up to 150 µm from the microelectrode interface (p<0.01), Fig. 2B. Although there is a reduction in the CD62p+ activated platelet response as time progresses from 1-week to 8-weeks, baseline activity was not reached when compared to background tissue (550 µm). We expect that platelets passively localize to the implant sites through leaky vasculature caused by the primary insult as well as the prolonged inflammatory state near the electrode insertion site. We also expect, based on known platelet physiology, that activated platelets release pro-inflammatory and pro-hemostatic signaling molecules from their granules, which in turn may promote active recruitment of immune cells. An increased degree of platelet accumulation and aggregation near the injury site may be expected by virtue of paracrine and vascular signaling that induces additional platelet activation which promotes endothelial platelet “rolling” and extravasation. Future studies may be required to better understand the dynamic localization of platelets near the microelectrode site, as our study focused on analyzing an endpoint condition rather than tracking platelets in real-time.

Both CD62p+ activated platelet and vWF responses chronically persist up to 100 µm and may act as a promoter for activated platelet and leukocytic recruitment to the insertion site (p<0.05), Fig. 2B, 2C. In some cases, there was an observable morphology of potential trafficking of vWF and platelets, Fig. S4. Previous studies have shown hemostatic proteins including vWF and fibrinogen adsorb to surfaces of biomaterials and act as scaffolds to direct immune-regulatory functions such as opsonization and phagocytosis [47,50,82,89]. Furthermore, recent literature has shown vWF and fibrinogen present in other disease models where chronic BBB instability is present [68,89–91]. Our results suggest that platelets and blood proteins associated with platelet adhesion and aggregation may be of high interest in further understanding the relationship between the neurovascular coupling and persistent neuroinflammation after microelectrode implantation.

Initial vascular damage does not fully resolve and leads to chronic impairments to the blood-brain barrier

Disruptions to the underlying neurovasculature affects the normal hemodynamics and the molecular trafficking to and from the native cortical tissue [72,90,92]. Blood-brain barrier (BBB) instability impedes the wound healing process and recovery after microelectrode insertion and promotes a self-perpetuating inflammatory response [25,26,34]. Type IV collagen is an extracellular matrix protein located in the basement membrane and promotes platelet adhesion when exposed during vascular trauma. Prior studies have shown increased type IV collagen response in local ischemic conditions, which are similar to resulting tissue conditions after intracortical microelectrode implants [67,93]. Peak response of type IV collagen occurred at distances up to 50 μm from the microelectrode interface for all observed time points with substantial consolidation by week 8 (p<0.001), Fig. 1B. No significant differences were observed between responses at 1-week and 4-week when comparing respective 50-μm interval distances. Type IV collagen response remained significantly higher for up to 150 μm when comparing responses between 1-week and 8-weeks, suggesting vascular remodeling may have occurred around the cortical implants (p<0.001). Studies indicate vascular remodeling and formation of leaky blood vessels occur within 2–3 weeks after the initial injury [84,85]. In addition a prior study indicated that endothelial cellular debris around the microelectrode may resolve within 24 hours after implantation [94]. By week 8, type IV collagen response subsides and may participate in the scar formation around the implant at approximately 50 µm from the implant (p<0.001). This is consistent with other publications that suggest that the collagenous scar condenses over the first 8 weeks following electrode implantation [95–98]. However, these previous studies examined type I collagen in contributing to the scar formation, while we are reporting type IV collagen, which is usually found in the blood vessels. This device encapsulation acts as an electrochemical barrier where effective charge transfer around the implant is hindered, thus reducing the overall stimulation or recording performances of intracortical microelectrodes [39,99–101]. Recent studies have also suggested that type IV collagen deposition may originate from scar-forming fibroblasts or activated astrocytes [102,103]. Future investigations to the origins of the type IV collagen deposition and the degree of impact at the microelectrode interface may lead to novel approaches to counter the inevitable scar tissue development.

Like vWF, fibrinogen is not commonly expressed in the cortical extracellular matrix unless introduced through vascular impairments. Once introduced into the cortical extracellular matrix, fibrinogen activates nearby glial cells and triggers a cascade of inflammatory events, influence neurodegeneration and vascular breakdown, inhibits tissue repair, and promote glial scarring [32,48,89,104–106]. Furthermore, fibrinogen-induced neuroinflammation has been linked to neurologic disease models such as Multiple Sclerosis, Alzheimer’s Disease, and ischemic stroke [65], [83]–[84]. As the BBB remains in a dysfunctional state where regulation of molecules is impeded, blood proteins, in this case fibrinogen, may continue to pass through the once selective membrane [21,32,72]. In our study, fibrinogen deposition accumulated up to 150 µm from the microelectrode interface and contracted to 50 µm when comparing the responses between 1-week and 8-weeks (p<0.001), Fig. 3B. In contrast, there was no difference at this 50 µm interval when comparing responses between 1-week and 4-weeks. Rather, the reduction in fibrinogen response was observed from distances of 50–150 µm (p<0.01). Our results were similar to previous studies that have shown peak fibrinogen response occurring at earlier stages of cortical injuries after vascular disruptions [32,109].

Bloodborne platelets may influence neuroinflammation after microelectrode insertion

Traditional neuroinflammation markers, GFAP and CD68, were used to stain for activated astrocytes and microglia, respectively, Fig. 3C, 4B. GFAP significantly increased over time within distances of up to 100 µm from the microelectrode interface when comparing responses between 1-week and 8-weeks, suggesting consolidation of the glial scar (p<0.001). A recent publication has suggested that GFAP response may be upregulated beyond 500–550 µm, which was chosen as our normalization background interval [110]. To determine the magnitude of the impact of this decision on our analysis, we performed a sensitivity analysis and considered different background normalization distances (ranging from 300–650 μm), Fig.S7. While it may have been possible for us to select 600–650 μm as a normalization interval, we determined it would not have made a significant difference in our findings. Therefore, we maintained the 500–550 µm normalization range, consistent with our prior work [20,23,61–63,111]. CD68 persists chronically for up to 150 µm from the microelectrode interface while declining in response when comparing responses at 1-week and 8-weeks, suggesting neuroinflammation around the implant has not subsided (p<0.01). Our results were similar to previously reported studies that show a decline in activated microglia response as astrocytic glial scar develops over time [34,111].

Accumulation of platelets persists at the microelectrode interface after the initial insertion-induced vascular damage, Fig. 1C. Sustained platelet presence may be influenced by a prolonged dysfunctional BBB and inflammatory chemokines released by activated endothelial and immune cells [42], [52], [85]. Our results show that these platelets could be inflammatory in nature, thus may largely influence the chronic neurodegeneration and neuroinflammation around implanted intracortical microelectrodes, Fig. 2B. The sustained presence of proteins such as vWF, fibrinogen and type IV collagen provide the necessities for a possible continuous platelet recruitment and activation at the inflamed cortical tissue. Recent studies showcase that platelets may play a larger role in neurovascular inflammation through influencing resident glial cells such as astrocytes and microglia, mediating infiltration of circulating blood constituents, and modulating endothelial function that separates the two compartments [59,69,70,113,114]. Our results show that platelets and their hemostatic associated proteins have a sustained presence within the proximity of implanted microelectrodes for up to 8-weeks. Further studies are necessary to investigate if cortically introduced platelets and their associated hemostatic proteins are a result of continuous infiltration through a leaky BBB or may be remnants from the initial vascular damage caused by the microelectrode insertion.

There were limitations to this study including variable degrees of vascular trauma for collected cortical slices, leading to variable cellular and non-cellular responses, Fig. S5. Prior publications have suggested that superficial depths may be more prone to vascular damage [75,76]. In this study, we investigated whether there was a substantial influence of depth on our tissue response outcomes, Fig. S6. While we observed significant variability based on slice location, the trends were not overwhelmingly directional in demonstrating a heighted response at deeper or more superficial locations. Areas with significant platelet accumulation may be a response to vascular trauma within vicinity of the implant. As platelets are introduced to the cortical space as a result of vascular trauma, vascularity is an uncontrollable factor when looking at spatial implant location (depth and/or axial plane), hemisphere, or animal [75,76]. These lesions could further promote higher responses of noncellular responses, such as vWF, fibrinogen, and type IV collagen, and cellular responses, such as astrocytes (GFAP) and activated microglia (CD68). A time-matched stab injury model may yield supporting evidence the observed persistence of platelet and their associated hemostatic proteins at the microelectrode interface were driven by a prolonged BBB disruption and foreign body response [34,115]. Antigen-retrieval to optimize the immunohistochemistry methods may allow for better visualization of type IV collagen. Given that we were interested in histology within proximity to the implant, all stains including type IV collagen were still observed. Furthermore, collected tissues from the 1-week cohort may contain existing hemorrhage that has yet to resolve and were not included in the analysis as proteins of interests residing in this area were usually washed away during the staining process, Fig. S2. Using optogenetics and in vivo imaging methods such as multiphoton imaging may avoid histology-related complications and enable live tracking of these blood proteins and platelets and their cellular interactions throughout the cortical layers without further perturbing the tissue.

Conclusion

Platelets are produced daily in the bone marrow with a circulating lifespan of 8–10 days. Despite their typical transient nature, we observed that they persistently localized around the microelectrode interfaces throughout the course of our 8-week study. This suggests that platelet or platelet fragments are remaining bound to the device after the initial breach in the BBB, or that new platelets are increasingly trafficking to the microelectrode interface. Aggregated platelets remain in an active phenotype and are modulated by extracellular proteins such as type IV collagen, vWF and fibrinogen, suggesting potential evidence of their influence to chronic neuroinflammation. Additionally, the wide variability of platelet accumulation throughout the cortex is likely in response to the inconsistency in vascular trauma after shank insertion. Our results indicate the need in developing new electrode designs and methods to avoid vascular trauma during microelectrode implant process. Further, studies focusing on platelet trafficking and the connection to declining microelectrode recording performance are underway.