Acta Biomaterialia 10 (2014) 4650–4660
Contents lists available at ScienceDirect
Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat
The foreign body response to the Utah Slant Electrode Array in the cat sciatic nerve M.B. Christensen, S.M. Pearce, N.M. Ledbetter, D.J. Warren, G.A. Clark, P.A. Tresco ⇑ Department of Bioengineering, College of Engineering, University of Utah, 36 South Wasatch Drive, Rm 3100, Salt Lake City, UT 84112, USA
a r t i c l e
i n f o
Article history: Received 19 November 2013 Received in revised form 17 June 2014 Accepted 12 July 2014 Available online 17 July 2014 Keywords: Peripheral nerve Electrode Neural prosthesis Foreign body response Histomorphometry
a b s t r a c t As the field of neuroprosthetic research continues to grow, studies describing the foreign body reaction surrounding chronic indwelling electrodes or microelectrode arrays will be critical for assessing biocompatibility. Of particular importance is the reaction surrounding penetrating microelectrodes that are used to stimulate and record from peripheral nerves used for prosthetic control, where such studies on axially penetrating electrodes are limited. Using the Utah Slant Electrode Array and a variety of histological methods, we investigated the foreign body response to the implanted array and its surrounding silicone cuff over long indwelling periods in the cat sciatic nerve. We observed that implanted nerves were associated with increased numbers of activated macrophages at the implant site, as well as distal to the implant, at all time points examined, with the longest observation being 350 days after implantation. We found that implanted cat sciatic nerves undergo a compensatory regenerative response after the initial injury that is accompanied by shifts in nerve fiber composition toward nerve fibers of smaller diameter and evidence of axons growing around microelectrode shafts. Nerve fibers located in fascicles that were not penetrated by the array or were located more than a few hundred microns from the implant appeared normal when examined over the course of a year-long indwelling period. Ó 2014 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
1. Introduction Microelectrode arrays offer the potential to record from and stimulate nerve tissue, lending themselves to clinical and basic research applications. Such devices have found clinical use as implants to aid the deaf [1], as sacral nerve stimulators for the treatment of bladder incontinence [2,3], as peroneal nerve electrodes for the treatment of foot-drop [4,5] and as deep brain stimulators for the treatment of Parkinson’s disease [6,7]. More advanced prosthetic devices, such as robotic arms for amputees or nerve stimulation for patients with spinal cord injury, are in the investigational stages of development. To date, extraneural or cuff-style electrodes have been the most widely used and best studied type of microelectrode for peripheral nerve applications [8]. However, other types of electrodes that penetrate the nerve fascicle or completely transect the nerve and allow it to regenerate through the electrode also are being developed for neuroprosthetic control [9–11]. The available evidence indicates that both the amount of injury and the foreign body response (FBR) that accompanies the use of these different types
⇑ Corresponding author. Tel.: +1 801 587 9263; fax: +1 801 585 5361. E-mail address: [email protected] (P.A. Tresco). http://dx.doi.org/10.1016/j.actbio.2014.07.010 1742-7061/Ó 2014 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
of electrodes, as well as their containment systems, vary considerably and likely reflect differences in design, materials used, surgical approach and assessment methods. For example, a number of studies have reported no difference in morphometric parameters following implantation of cuff electrodes [12–16], indicating that the implantation of such devices, which do little physical damage to the nerve, can be relatively benign with regard to changes elicited from the FBR with regard to nerve fiber composition. Other studies using extraneural cuff-style electrodes have reported alterations in morphometric parameters, including changes in fiber density [17–19] and fiber diameter distributions [19–21], which are likely due to differences in electrode design and methodological approach. In contrast, the histological changes associated with the FBR to sieve, regenerative and intrafascicular electrodes – electrode types that induce more physical damage upon insertion or application – tend to be greater and include changes in fiber count [22–26], fiber density [26], fiber packing [22] and fiber diameter and/or g-ratio (ratio of axon diameter to the outer diameter of the myelinated fiber) distributions [25–28] as they involve a variable amount of nerve axotomy, depending on the size and design of the electrode used. In general, qualitatively speaking, the FBR to all types of electrodes is associated with enhanced connective tissue deposition surrounding the implant [16,26,28–39] and the presence of macrophages at the biotic–abiotic interface [38–41]. The
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
major issue is one of safety and biocompatibility, as reports exist where the reaction to such implants can negatively affect nerve function [25]. A few studies have examined the tissue response to Utah electrode arrays that penetrate peripheral nerves and reside over long indwelling periods in large animals. Of these, results indicate that such penetrating arrays can compress individual nerve fibers under the electrode tip after acute implantation, where the compression zone can be over 100 microns below the electrode tip [42]. Whether such sequela transforms into prolonged nerve injury is not known. In a longer term study, where a number of different containment systems and different sized arrays were explored with both functioning and non-functional implants, it was shown that connective tissue is deposited beneath the array, suggesting a potential mechanism for pulling the electrode away from the implanted nerve. In that study it was also reported that nerve fiber diameter distribution was shifted toward smaller diameter fibers; however, the authors only compared data from single animals [43]. To build upon these studies, the foreign body response to the Utah Slant Electrode Array (USEA) encapsulated in a silicone cuff was studied in the sciatic nerve of cats over long indwelling periods. Unlike previous studies, all of the cats were purpose bred and received the same type of recording electrode and containment system placed in one nerve in the same location. Contralateral, unimplanted nerves served as controls. The FBR was assessed at the implantation site and 1 cm above and below it. Using immunohistochemical methods, we examined the degree of activated macrophages in both the epineurial and endoneurial spaces at or just proximal to the implant site. We also examined fascicle area, fiber count, fiber density, fiber packing (the percentage of fascicle area occupied by fibers) and mean g-ratio values, as well as fiber diameter distributions, in the implanted and contralateral nerves using high-power light level microscopy. We found that all implanted nerves were associated with an FBR accompanied by persistent inflammation at all implant surfaces, including the silicon containment system and wire bundles. We also observed that the initial injury of implantation is accompanied by a compensatory response that includes connective tissue deposition and a shift in nerve fiber diameters toward smaller diameter fibers, which appear to restore the total number of nerve fibers and their associated degree of myelination toward normal. 2. Methods 2.1. Animals All studies were conducted with the approval of the Institutional Animal Care and Use Committee at the University of Utah. Fourteen purpose-bred cats, with a mean weight of 3.13 ± 0.14 kg, were implanted with USEAs, with indwelling times ranging from 13 to 350 days. Animals were kept in a common housing area and were allowed to roam freely after recovery from surgical implantation. For analysis over the indwelling period, animals were grouped into four categories: acute (2 weeks; n = 2), short-term (5–6 weeks; n = 4), long-term (10–15 weeks; n = 3) and chronic (22–26 weeks; n = 4). 2.2. Microelectrode array insertion Prior to surgery, animals were anesthetized with an intravenous Telazol injection (10 mg kg 1) that was maintained by isoflurane inhalation (1–2.5%). A single USEA (Fig. 1A) was implanted into the sciatic nerve as previously described, approximately 2 cm proximal to the bifurcation of the tibial and peroneal nerves [43,44]. Briefly, the surgical site was shaved and treated with
4651
betadyne, and an incision was made from the hip to the knee. The biceps femoris and vastus lateralis muscles were separated to expose the sciatic nerve. The array was then positioned on the nerve and inserted using a pneumatically actuated impulse inserter (Cyberkinetics, Salt Lake City, UT). Silicone (kwik-cast, WPI, Sarasota, FL) was applied in two stages. First, the nerve was gently lifted and the two-part mixed silicone was applied under the nerve, after which the nerve was replaced so that the silicone could conform to it. Second, the silicone was poured over the array to form a conforming cuff. Wires exited subcutaneously from the array to a connector that was secured to either the femur or the iliac crest (see Clark et al. [44] for more details on connector securement). Sufficient slack was left in the wire bundle to prevent tethering forces on the array. Fig. 1B shows a picture of a representative implanted system, including the electrode, silicone containment system, wire bundle and connector. Contralateral or unimplanted nerves served as controls.
2.3. Euthanasia and tissue preparation Animals were terminally anesthetized using a 10:1 ketamine/ xylazine cocktail administered intravenously at 0.2 ml kg 1. Following anesthesia, animals were transcardially perfused with pH 7.4 phosphate-buffered saline (PBS) followed by 4% paraformaldehyde in PBS. The sciatic nerve and its two main branches, the tibial and common peroneal nerves, were exposed from the latero-posterior thigh by blunt dissection of the biceps femoris and photographed in situ. The main sciatic trunk from each leg was dissected free, post-fixed in the same fixative overnight and stored in PBS with 0.01% sodium azide at 4 °C. Implanted arrays and the associated silicone cuffs were removed from the nerve following the post-fixation period. Following array and cuff removal, nerves were cut in crosssection approximately 1 cm proximal and distal to the implantation site and at 3 cm proximal to the bifurcation on the contralateral nerve to produce nerve segments approximately 2 mm in length (Fig. 1C). The nerve segments were then further post-fixed in 4% paraformaldehyde/2% glutaraldehyde in PBS for 24 h followed by equilibration in 2% osmium tetroxide in 0.1 M sodium cacodylate buffer for 24 h. Sections were dehydrated through a graded series of ethanols, cleared with propylene oxide and infused with Embed 812 (Electron Microscopy Sciences, Hatfield, PA). Sections were cut at 0.7 lm using an Ultracut EMUC6 (Leica, Bannockburn, IL), placed on Superfrost Plus glass slides, and stained with thionin and acridine orange (Sigma–Aldrich, St. Louis, MO). Sections were dried and coverslipped using Cytoseal XYL (Richard-Allan Scientific, Kalamazoo, MI). For immunohistochemistry, nerve sections at the implant site or immediately proximal to the implant were equilibrated in a 30% sucrose solution, embedded in Tissue Freezing Medium (Triangle Biomedical Sciences, Durham, NC) and sectioned using a Leica CM3050 S cryostat at 40–50 lm. Free-floating sections were placed in a blocking solution consisting of 4% (v/v) goat serum (Invitrogen, Carlsbad, CA), 0.05% (v/v) Triton-X 100 and 0.01% (w/v) sodium azide overnight at 4 °C. The sections were then treated with primary antibodies (Table 1) in blocking solution for 48 h at 4 °C, followed by three room-temperature rinses in PBS (2 h per rinse). The appropriate secondary antibodies in blocking solution were added for 48 h at 4 °C and followed by three additional rinses. All sections were counterstained with 10 lM DAPI to visualize cell nuclei. Sections were then mounted on slides and coverslipped using Fluoromount-G (Southern Biotech, Birmingham, AL). Retrieved arrays and the silicone containment system were processed similarly for immunohistochemical analysis using antisera against macrophages.
4652
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 1. (A) Representative scanning electron micrograph of a USEA consisting of 10 rows of 10 electrodes each, with each row ranging in length from 0.6 to 1.5 mm, with 0.1 mm steps between rows. Each electrode is 80 lm wide at the base and tapers to a point, spaced at 400 lm between centers, and project out of a 0.2 mm thick silicon substrate measuring 4.2 4.2 mm. (B) Representative photograph of complete implant system, which includes an array (not visible) with encapsulating green cuff, lead wires and connector. (C) Schematic (not to scale) showing positioning of the array in the cat sciatic nerve and the location of nerve segments processed for quantitative histological analysis located 1 cm proximal and distal from the center of the implant site.
Table 1 List of primary and secondary antibodies used. Antibody
Antigen
Cell Type(s)
Isotype
Concentration (lg ml
NF-200 S-100b b-III Tubulin MAC387
Neuronal 200 kD intermediate filament b-chain of S-100 dimer protein Isotype 3 of b-tubulin L1 or calprotectin protein
Neurons Glial/Schwann cells Neurons Monocytes/macrophages
Rabbit IgG Mouse IgG1 Mouse IgG2b Mouse IgG1
8.3 14.6 7.6 0.5
2.4. Quantitative analysis Images were obtained using ImagePro Plus 4.0 (MediaCybernetics, Bethesda, MD) and a color CCD camera (Photometrics, Tucson, AZ) attached to a Nikon E600 microscope. For immunohistochemistry, sections were imaged at 100 final magnification. For thin sections, whole nerve images were obtained at 200 final magnification, while fascicle images were obtained at 1000 final magnification. For both whole nerve and fascicle imaging, serial overlapping images were taken over the entire nerve or fascicle. Using the photomerge command in Adobe Photoshop CS, serial images were reconstructed to form a mosaic image. The reconstructed images were then processed through a series of contrast enhancement steps in Photoshop to further differentiate myelin from background. Fascicle areas were calculated by manually tracing the inner edge of the perineurium and total myelinated fiber counts were performed manually. All other morphometric measurements were obtained automatically using ImagePro Plus. 2.5. Statistics Comparisons were made between nerve sections taken proximal and distal to the implant, as well as from the contralateral
1
)
Vendor Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich Millipore
nerve, for nerve cross-sectional area, fiber count, fascicle area, fiber density, fiber packing and mean g-ratio values across each animal group using paired t-tests with a Bonferroni correction factor. A comparison was also performed across all groups at all locations using a one-way analysis of variance with a Tukey’s post hoc test. The fiber diameter distributions did not exhibit normal distributions, so they were evaluated using a Wilcoxon signed-rank test by ranking and comparing all bins less than 10 lm (to compare the amount of small diameter fibers) and greater than or equal to 10 lm (to compare the amount of large diameter fibers). p-values below 0.05 were considered significant. All data are represented as average ± SEM. 3. Results 3.1. Macroscopic analysis At the time of dissection, contralateral nerves and their derivative tibial and peroneal branches had a white appearance and were surrounded by a clear epineurial layer (Fig. 2A). In general, the main trunk of the sciatic nerve was elliptical in shape. Implanted nerves were covered with a greater amount of connective tissue, which was primarily present around and within the silicon cuff
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
and around the lead wires (Fig. 2B). The lead wires exiting the encapsulation cuff and continuing subcutaneously to the connector were covered in a sheath of connective tissue whose thickness varied from animal to animal, as well as along the wire bundle within a given animal. The connective tissue was non-adherent to adjacent muscles. In acute cases (2 weeks), the connective tissue exhibited a higher degree of vascularization that was not cleared by perfusion compared with contralateral nerves (Fig. 2C). At all time points, blood products that were not cleared by perfusion were observed under the explanted USEA. Granulation tissue beneath the array was present in longer term and chronic animals (>10 weeks), and increased as the indwelling time increased (Fig. 2D–F). No functional impairment in freely moving activity was observed during the course of the study. 3.2. Microscopic analysis All fascicles were examined at a final magnification of 1000 to examine individual myelinated axons and surrounding cellularity. Contralateral nerves showed no evidence of hypercellularity in the epineurial space or of infiltrating macrophages in the endoneurial space, and showed little or no evidence of axonal turnover. Fiber distribution and myelin thickness appeared uniformly distributed in all contralateral nerves. In contrast, implanted nerves showed varying degrees of cytoarchitectural changes. For proximal sections, animals in the acute and short-term groups showed some evidence of degenerative processes, whereas animals in the long-term and chronic groups showed little evidence of ongoing inflammation (Fig. 3A). Distal sections of implanted nerves displayed infiltration by presumptive macrophages in the endoneurial space, particularly in smaller fascicles. These macrophages appeared to be involved in active phagocytosis of myelin and in clearing cellular debris at all time points (Fig. 3B). Implanted nerves showed signs of Wallerian degeneration characterized by axon degradation, myelin breakdown and the presence of myelin-laden phagocytic cells. Signs of axonal regeneration were also evident as evidenced by an apparent shift in fiber distributions toward small diameter fibers without a change in the total number of myelinated fibers.
4653
Nerve sections that were taken just proximal to the implant site under the silicone cuff showed an inflammatory reaction around the periphery of the nerve which included connective tissue deposition, presumptive macrophages and foreign body giant cells (Figs. 3C–E and 4A). To further analyze this reaction, tissue from animals in the short-term, long-term and chronic groups was examined using immunohistochemical techniques. Using cellspecific markers for macrophages, we found a pronounced inflammatory reaction at the periphery of the nerve which correlated to the reaction seen in thin sections (Fig. 4C and (D). We also examined the presence of macrophages at the implant site. In cross-sections, we observed evidence of electrode penetration deep within fascicular structures (Fig. 4A). Associated with fascicular penetration, we also found evidence of axonal disruption, including axonal sprouting near the electrode tracks (Fig. 4B). Longitudinal sections revealed an inflammatory reaction near the base of the array (Fig. 5A) that followed the electrode tracks down to the tips. The macrophage concentration around the tips located in the endoneurial space always appeared to be reduced compared with the tips located in the epineurial space (Fig. 5B). There was also positive staining for macrophages in the endoneurial space away from the electrode tips (Fig. 5B). Longitudinal sections also showed disruption of axons and their associated myelin due to the presence of electrodes. Several axons appear to have been severed, whereas many appeared to curve around the electrodes, substantially increasing the width of the fascicles (Fig. 5C and (D). Nerve fibers just outside the reactive zone appeared normal. Retrieved arrays were examined grossly and using immunohistochemical techniques. No adherent tissue was observed on explanted arrays. Fluorescent imaging revealed adherence of cells on the device surface, both at the base and along the electrode shafts, with many of these cells being MAC387 positive. We evaluated several morphometric parameters, including fiber counts, fascicle areas, fiber densities and packing, mean g-ratio values and fiber diameter distributions. Several differences were found, including the fascicle areas in the acute group, which showed an increase in area in distal sections compared with proximal sections (Table 2). Fiber density also showed a significant difference in the long-term group when comparing distal sections
Fig. 2. Representative images of the appearance of the implanted nerve. (A) Contralateral perfused nerve at similar point of implantation (arrow). (B) Representative implantation site from the short-term group. Connective tissue build-up has occurred around the silicone and wires (arrows). (C) Nerve from the acute group. Even after complete exsanguination, the connective tissue surrounding the nerve shows a high degree of uncleared blood vessels, signifying an ongoing angiogenic process of unconnected blood vessels (arrow). (D) In the short-term group, the tissue beneath the array appears firm, clearly showing each electrode penetration profile. (E) In the longterm group, the tissue beneath the array exhibited evidence of blood products, particularly around the edges of the array footprint. (F) At the longest time point of 350 days post-implantation, the tissue under the array showed signs of peri-prosthetic granulation tissue and appeared mushy (arrow).
4654
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 3. (A) Representative 1000 image of the endoneurial space in the long-term and chronic groups proximal to implant. Macrophages are not typically present in the proximal space after the initial degradation of downstream axons due to Wallerian degeneration. This section shows little evidence of fiber diameter or g-ratio distribution changes, although quantitative data from this group suggests that there is a shift in fiber diameter distributions towards smaller diameter fibers compared with contralateral sections (see Fig. 6). (B) Representative 1000 image of the endoneurial space in all groups distal to the implant. Numerous foamy macrophages can be seen actively degrading axons and their associated myelin sheaths (arrows). There appears to be preferential degradation of large diameter fibers. (C) Epineurial space showing hypercellularity in tissue adjacent to silicone cuff. (D) Higher magnification view from (C). Several different cells types are present, although individual cell boundaries are often not distinguishable. Cells are occasionally visible in the vasculature, even after perfusion. (E) Foreign body giant cells are present in the hypercellular zone encapsulating presumptive myelin debris (arrow). Scale bars: A, B, E = 50 lm; C, D = 100 lm.
Fig. 4. (A) Cross-section of nerve from the chronic group at the implant site, labeled with DAPI (blue) and NF200 (green), showing penetration of the electrodes deep within the fascicular structure. An outline of a USEA, modified from a photograph, is shown for reference. From the USEA outline, electrode penetration is estimated to be 1.1 mm in this figure (or the sixth row of the array). (B) Cross-section of a nerve from the long-term group at the electrode nerve interface, labeled with MAC387 (green) and b-III tubulin (red), showing macrophages around the electrode tips and neuronal processes outside of the normal fascicular structure. (C) Cross-section from a short-term animal of the nerve proximal to the array but still beneath the silicone cuff (a not-to-scale outline of the cuff is shown in green). The epineurial area adjacent to the silicone (outer edge of the nerve) shows a large degree of hypercellularity, which includes numerous foreign body giant cells (Fig. 3). (D) Same nerve as in (A), stained with MAC387, which confirms that a large number of cells adjacent to the silicone are phagocytic cells. Scale bars: A = 500 lm; B = 100 lm; C, D = 1 mm.
with contralateral unimplanted nerves. A difference was also observed in fiber density when we compared implanted distal sections and contralateral nerves across all animals (Table 2). For fiber count, distal sections had a lower total number of fibers than
proximal sections only in the short-term group. When evaluating data across all animals, both distal and proximal sections had significantly lower fiber packing percentages than contralateral nerves. Where sufficient statistical power was available, we saw no significant
4655
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 5. (A) Longitudinal section taken at the implant site from the short-term group, labeled for MAC387 (green) and NF200 (red). This section was taken near the surface of the nerve, at the base of the array. Numerous MAC387 positive cells surround each electrode and fill the epineurial space. Axons are also disrupted around electrodes. (B) Image from the chronic group, showing that MAC387 (green) positive cells are still present at long time points and continue down the electrodes into the endoneurial space, although they are not as concentrated as in the epineurial space. Cells positive for MAC387 are also found in the endoneurial space away from the electrodes. (C) Section adjacent to that shown in (B), labeled for S-100b (green), NF200 (red) and DAPI (blue). Axons and their associated myelin sheaths are diverted around the electrodes, greatly increasing the width of the fascicles under the implant. (D) Inset from (C). Severed or dead-end regenerating axons are seen adjacent to an electrode. Scale bars: A–C = 1 mm; D = 200 lm.
differences in mean fascicular areas, fiber counts or g-ratio values at any location in the nerve when evaluating data across all animals. Average fiber diameter distribution for contralateral nerves showed a trimodal distribution, with peaks around 4.5–5.5, 10.5– 11.5 and 13.5–14 lm. For statistical evaluation, small (
Contents lists available at ScienceDirect
Acta Biomaterialia journal homepage: www.elsevier.com/locate/actabiomat
The foreign body response to the Utah Slant Electrode Array in the cat sciatic nerve M.B. Christensen, S.M. Pearce, N.M. Ledbetter, D.J. Warren, G.A. Clark, P.A. Tresco ⇑ Department of Bioengineering, College of Engineering, University of Utah, 36 South Wasatch Drive, Rm 3100, Salt Lake City, UT 84112, USA
a r t i c l e
i n f o
Article history: Received 19 November 2013 Received in revised form 17 June 2014 Accepted 12 July 2014 Available online 17 July 2014 Keywords: Peripheral nerve Electrode Neural prosthesis Foreign body response Histomorphometry
a b s t r a c t As the field of neuroprosthetic research continues to grow, studies describing the foreign body reaction surrounding chronic indwelling electrodes or microelectrode arrays will be critical for assessing biocompatibility. Of particular importance is the reaction surrounding penetrating microelectrodes that are used to stimulate and record from peripheral nerves used for prosthetic control, where such studies on axially penetrating electrodes are limited. Using the Utah Slant Electrode Array and a variety of histological methods, we investigated the foreign body response to the implanted array and its surrounding silicone cuff over long indwelling periods in the cat sciatic nerve. We observed that implanted nerves were associated with increased numbers of activated macrophages at the implant site, as well as distal to the implant, at all time points examined, with the longest observation being 350 days after implantation. We found that implanted cat sciatic nerves undergo a compensatory regenerative response after the initial injury that is accompanied by shifts in nerve fiber composition toward nerve fibers of smaller diameter and evidence of axons growing around microelectrode shafts. Nerve fibers located in fascicles that were not penetrated by the array or were located more than a few hundred microns from the implant appeared normal when examined over the course of a year-long indwelling period. Ó 2014 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
1. Introduction Microelectrode arrays offer the potential to record from and stimulate nerve tissue, lending themselves to clinical and basic research applications. Such devices have found clinical use as implants to aid the deaf [1], as sacral nerve stimulators for the treatment of bladder incontinence [2,3], as peroneal nerve electrodes for the treatment of foot-drop [4,5] and as deep brain stimulators for the treatment of Parkinson’s disease [6,7]. More advanced prosthetic devices, such as robotic arms for amputees or nerve stimulation for patients with spinal cord injury, are in the investigational stages of development. To date, extraneural or cuff-style electrodes have been the most widely used and best studied type of microelectrode for peripheral nerve applications [8]. However, other types of electrodes that penetrate the nerve fascicle or completely transect the nerve and allow it to regenerate through the electrode also are being developed for neuroprosthetic control [9–11]. The available evidence indicates that both the amount of injury and the foreign body response (FBR) that accompanies the use of these different types
⇑ Corresponding author. Tel.: +1 801 587 9263; fax: +1 801 585 5361. E-mail address: [email protected] (P.A. Tresco). http://dx.doi.org/10.1016/j.actbio.2014.07.010 1742-7061/Ó 2014 Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
of electrodes, as well as their containment systems, vary considerably and likely reflect differences in design, materials used, surgical approach and assessment methods. For example, a number of studies have reported no difference in morphometric parameters following implantation of cuff electrodes [12–16], indicating that the implantation of such devices, which do little physical damage to the nerve, can be relatively benign with regard to changes elicited from the FBR with regard to nerve fiber composition. Other studies using extraneural cuff-style electrodes have reported alterations in morphometric parameters, including changes in fiber density [17–19] and fiber diameter distributions [19–21], which are likely due to differences in electrode design and methodological approach. In contrast, the histological changes associated with the FBR to sieve, regenerative and intrafascicular electrodes – electrode types that induce more physical damage upon insertion or application – tend to be greater and include changes in fiber count [22–26], fiber density [26], fiber packing [22] and fiber diameter and/or g-ratio (ratio of axon diameter to the outer diameter of the myelinated fiber) distributions [25–28] as they involve a variable amount of nerve axotomy, depending on the size and design of the electrode used. In general, qualitatively speaking, the FBR to all types of electrodes is associated with enhanced connective tissue deposition surrounding the implant [16,26,28–39] and the presence of macrophages at the biotic–abiotic interface [38–41]. The
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
major issue is one of safety and biocompatibility, as reports exist where the reaction to such implants can negatively affect nerve function [25]. A few studies have examined the tissue response to Utah electrode arrays that penetrate peripheral nerves and reside over long indwelling periods in large animals. Of these, results indicate that such penetrating arrays can compress individual nerve fibers under the electrode tip after acute implantation, where the compression zone can be over 100 microns below the electrode tip [42]. Whether such sequela transforms into prolonged nerve injury is not known. In a longer term study, where a number of different containment systems and different sized arrays were explored with both functioning and non-functional implants, it was shown that connective tissue is deposited beneath the array, suggesting a potential mechanism for pulling the electrode away from the implanted nerve. In that study it was also reported that nerve fiber diameter distribution was shifted toward smaller diameter fibers; however, the authors only compared data from single animals [43]. To build upon these studies, the foreign body response to the Utah Slant Electrode Array (USEA) encapsulated in a silicone cuff was studied in the sciatic nerve of cats over long indwelling periods. Unlike previous studies, all of the cats were purpose bred and received the same type of recording electrode and containment system placed in one nerve in the same location. Contralateral, unimplanted nerves served as controls. The FBR was assessed at the implantation site and 1 cm above and below it. Using immunohistochemical methods, we examined the degree of activated macrophages in both the epineurial and endoneurial spaces at or just proximal to the implant site. We also examined fascicle area, fiber count, fiber density, fiber packing (the percentage of fascicle area occupied by fibers) and mean g-ratio values, as well as fiber diameter distributions, in the implanted and contralateral nerves using high-power light level microscopy. We found that all implanted nerves were associated with an FBR accompanied by persistent inflammation at all implant surfaces, including the silicon containment system and wire bundles. We also observed that the initial injury of implantation is accompanied by a compensatory response that includes connective tissue deposition and a shift in nerve fiber diameters toward smaller diameter fibers, which appear to restore the total number of nerve fibers and their associated degree of myelination toward normal. 2. Methods 2.1. Animals All studies were conducted with the approval of the Institutional Animal Care and Use Committee at the University of Utah. Fourteen purpose-bred cats, with a mean weight of 3.13 ± 0.14 kg, were implanted with USEAs, with indwelling times ranging from 13 to 350 days. Animals were kept in a common housing area and were allowed to roam freely after recovery from surgical implantation. For analysis over the indwelling period, animals were grouped into four categories: acute (2 weeks; n = 2), short-term (5–6 weeks; n = 4), long-term (10–15 weeks; n = 3) and chronic (22–26 weeks; n = 4). 2.2. Microelectrode array insertion Prior to surgery, animals were anesthetized with an intravenous Telazol injection (10 mg kg 1) that was maintained by isoflurane inhalation (1–2.5%). A single USEA (Fig. 1A) was implanted into the sciatic nerve as previously described, approximately 2 cm proximal to the bifurcation of the tibial and peroneal nerves [43,44]. Briefly, the surgical site was shaved and treated with
4651
betadyne, and an incision was made from the hip to the knee. The biceps femoris and vastus lateralis muscles were separated to expose the sciatic nerve. The array was then positioned on the nerve and inserted using a pneumatically actuated impulse inserter (Cyberkinetics, Salt Lake City, UT). Silicone (kwik-cast, WPI, Sarasota, FL) was applied in two stages. First, the nerve was gently lifted and the two-part mixed silicone was applied under the nerve, after which the nerve was replaced so that the silicone could conform to it. Second, the silicone was poured over the array to form a conforming cuff. Wires exited subcutaneously from the array to a connector that was secured to either the femur or the iliac crest (see Clark et al. [44] for more details on connector securement). Sufficient slack was left in the wire bundle to prevent tethering forces on the array. Fig. 1B shows a picture of a representative implanted system, including the electrode, silicone containment system, wire bundle and connector. Contralateral or unimplanted nerves served as controls.
2.3. Euthanasia and tissue preparation Animals were terminally anesthetized using a 10:1 ketamine/ xylazine cocktail administered intravenously at 0.2 ml kg 1. Following anesthesia, animals were transcardially perfused with pH 7.4 phosphate-buffered saline (PBS) followed by 4% paraformaldehyde in PBS. The sciatic nerve and its two main branches, the tibial and common peroneal nerves, were exposed from the latero-posterior thigh by blunt dissection of the biceps femoris and photographed in situ. The main sciatic trunk from each leg was dissected free, post-fixed in the same fixative overnight and stored in PBS with 0.01% sodium azide at 4 °C. Implanted arrays and the associated silicone cuffs were removed from the nerve following the post-fixation period. Following array and cuff removal, nerves were cut in crosssection approximately 1 cm proximal and distal to the implantation site and at 3 cm proximal to the bifurcation on the contralateral nerve to produce nerve segments approximately 2 mm in length (Fig. 1C). The nerve segments were then further post-fixed in 4% paraformaldehyde/2% glutaraldehyde in PBS for 24 h followed by equilibration in 2% osmium tetroxide in 0.1 M sodium cacodylate buffer for 24 h. Sections were dehydrated through a graded series of ethanols, cleared with propylene oxide and infused with Embed 812 (Electron Microscopy Sciences, Hatfield, PA). Sections were cut at 0.7 lm using an Ultracut EMUC6 (Leica, Bannockburn, IL), placed on Superfrost Plus glass slides, and stained with thionin and acridine orange (Sigma–Aldrich, St. Louis, MO). Sections were dried and coverslipped using Cytoseal XYL (Richard-Allan Scientific, Kalamazoo, MI). For immunohistochemistry, nerve sections at the implant site or immediately proximal to the implant were equilibrated in a 30% sucrose solution, embedded in Tissue Freezing Medium (Triangle Biomedical Sciences, Durham, NC) and sectioned using a Leica CM3050 S cryostat at 40–50 lm. Free-floating sections were placed in a blocking solution consisting of 4% (v/v) goat serum (Invitrogen, Carlsbad, CA), 0.05% (v/v) Triton-X 100 and 0.01% (w/v) sodium azide overnight at 4 °C. The sections were then treated with primary antibodies (Table 1) in blocking solution for 48 h at 4 °C, followed by three room-temperature rinses in PBS (2 h per rinse). The appropriate secondary antibodies in blocking solution were added for 48 h at 4 °C and followed by three additional rinses. All sections were counterstained with 10 lM DAPI to visualize cell nuclei. Sections were then mounted on slides and coverslipped using Fluoromount-G (Southern Biotech, Birmingham, AL). Retrieved arrays and the silicone containment system were processed similarly for immunohistochemical analysis using antisera against macrophages.
4652
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 1. (A) Representative scanning electron micrograph of a USEA consisting of 10 rows of 10 electrodes each, with each row ranging in length from 0.6 to 1.5 mm, with 0.1 mm steps between rows. Each electrode is 80 lm wide at the base and tapers to a point, spaced at 400 lm between centers, and project out of a 0.2 mm thick silicon substrate measuring 4.2 4.2 mm. (B) Representative photograph of complete implant system, which includes an array (not visible) with encapsulating green cuff, lead wires and connector. (C) Schematic (not to scale) showing positioning of the array in the cat sciatic nerve and the location of nerve segments processed for quantitative histological analysis located 1 cm proximal and distal from the center of the implant site.
Table 1 List of primary and secondary antibodies used. Antibody
Antigen
Cell Type(s)
Isotype
Concentration (lg ml
NF-200 S-100b b-III Tubulin MAC387
Neuronal 200 kD intermediate filament b-chain of S-100 dimer protein Isotype 3 of b-tubulin L1 or calprotectin protein
Neurons Glial/Schwann cells Neurons Monocytes/macrophages
Rabbit IgG Mouse IgG1 Mouse IgG2b Mouse IgG1
8.3 14.6 7.6 0.5
2.4. Quantitative analysis Images were obtained using ImagePro Plus 4.0 (MediaCybernetics, Bethesda, MD) and a color CCD camera (Photometrics, Tucson, AZ) attached to a Nikon E600 microscope. For immunohistochemistry, sections were imaged at 100 final magnification. For thin sections, whole nerve images were obtained at 200 final magnification, while fascicle images were obtained at 1000 final magnification. For both whole nerve and fascicle imaging, serial overlapping images were taken over the entire nerve or fascicle. Using the photomerge command in Adobe Photoshop CS, serial images were reconstructed to form a mosaic image. The reconstructed images were then processed through a series of contrast enhancement steps in Photoshop to further differentiate myelin from background. Fascicle areas were calculated by manually tracing the inner edge of the perineurium and total myelinated fiber counts were performed manually. All other morphometric measurements were obtained automatically using ImagePro Plus. 2.5. Statistics Comparisons were made between nerve sections taken proximal and distal to the implant, as well as from the contralateral
1
)
Vendor Sigma–Aldrich Sigma–Aldrich Sigma–Aldrich Millipore
nerve, for nerve cross-sectional area, fiber count, fascicle area, fiber density, fiber packing and mean g-ratio values across each animal group using paired t-tests with a Bonferroni correction factor. A comparison was also performed across all groups at all locations using a one-way analysis of variance with a Tukey’s post hoc test. The fiber diameter distributions did not exhibit normal distributions, so they were evaluated using a Wilcoxon signed-rank test by ranking and comparing all bins less than 10 lm (to compare the amount of small diameter fibers) and greater than or equal to 10 lm (to compare the amount of large diameter fibers). p-values below 0.05 were considered significant. All data are represented as average ± SEM. 3. Results 3.1. Macroscopic analysis At the time of dissection, contralateral nerves and their derivative tibial and peroneal branches had a white appearance and were surrounded by a clear epineurial layer (Fig. 2A). In general, the main trunk of the sciatic nerve was elliptical in shape. Implanted nerves were covered with a greater amount of connective tissue, which was primarily present around and within the silicon cuff
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
and around the lead wires (Fig. 2B). The lead wires exiting the encapsulation cuff and continuing subcutaneously to the connector were covered in a sheath of connective tissue whose thickness varied from animal to animal, as well as along the wire bundle within a given animal. The connective tissue was non-adherent to adjacent muscles. In acute cases (2 weeks), the connective tissue exhibited a higher degree of vascularization that was not cleared by perfusion compared with contralateral nerves (Fig. 2C). At all time points, blood products that were not cleared by perfusion were observed under the explanted USEA. Granulation tissue beneath the array was present in longer term and chronic animals (>10 weeks), and increased as the indwelling time increased (Fig. 2D–F). No functional impairment in freely moving activity was observed during the course of the study. 3.2. Microscopic analysis All fascicles were examined at a final magnification of 1000 to examine individual myelinated axons and surrounding cellularity. Contralateral nerves showed no evidence of hypercellularity in the epineurial space or of infiltrating macrophages in the endoneurial space, and showed little or no evidence of axonal turnover. Fiber distribution and myelin thickness appeared uniformly distributed in all contralateral nerves. In contrast, implanted nerves showed varying degrees of cytoarchitectural changes. For proximal sections, animals in the acute and short-term groups showed some evidence of degenerative processes, whereas animals in the long-term and chronic groups showed little evidence of ongoing inflammation (Fig. 3A). Distal sections of implanted nerves displayed infiltration by presumptive macrophages in the endoneurial space, particularly in smaller fascicles. These macrophages appeared to be involved in active phagocytosis of myelin and in clearing cellular debris at all time points (Fig. 3B). Implanted nerves showed signs of Wallerian degeneration characterized by axon degradation, myelin breakdown and the presence of myelin-laden phagocytic cells. Signs of axonal regeneration were also evident as evidenced by an apparent shift in fiber distributions toward small diameter fibers without a change in the total number of myelinated fibers.
4653
Nerve sections that were taken just proximal to the implant site under the silicone cuff showed an inflammatory reaction around the periphery of the nerve which included connective tissue deposition, presumptive macrophages and foreign body giant cells (Figs. 3C–E and 4A). To further analyze this reaction, tissue from animals in the short-term, long-term and chronic groups was examined using immunohistochemical techniques. Using cellspecific markers for macrophages, we found a pronounced inflammatory reaction at the periphery of the nerve which correlated to the reaction seen in thin sections (Fig. 4C and (D). We also examined the presence of macrophages at the implant site. In cross-sections, we observed evidence of electrode penetration deep within fascicular structures (Fig. 4A). Associated with fascicular penetration, we also found evidence of axonal disruption, including axonal sprouting near the electrode tracks (Fig. 4B). Longitudinal sections revealed an inflammatory reaction near the base of the array (Fig. 5A) that followed the electrode tracks down to the tips. The macrophage concentration around the tips located in the endoneurial space always appeared to be reduced compared with the tips located in the epineurial space (Fig. 5B). There was also positive staining for macrophages in the endoneurial space away from the electrode tips (Fig. 5B). Longitudinal sections also showed disruption of axons and their associated myelin due to the presence of electrodes. Several axons appear to have been severed, whereas many appeared to curve around the electrodes, substantially increasing the width of the fascicles (Fig. 5C and (D). Nerve fibers just outside the reactive zone appeared normal. Retrieved arrays were examined grossly and using immunohistochemical techniques. No adherent tissue was observed on explanted arrays. Fluorescent imaging revealed adherence of cells on the device surface, both at the base and along the electrode shafts, with many of these cells being MAC387 positive. We evaluated several morphometric parameters, including fiber counts, fascicle areas, fiber densities and packing, mean g-ratio values and fiber diameter distributions. Several differences were found, including the fascicle areas in the acute group, which showed an increase in area in distal sections compared with proximal sections (Table 2). Fiber density also showed a significant difference in the long-term group when comparing distal sections
Fig. 2. Representative images of the appearance of the implanted nerve. (A) Contralateral perfused nerve at similar point of implantation (arrow). (B) Representative implantation site from the short-term group. Connective tissue build-up has occurred around the silicone and wires (arrows). (C) Nerve from the acute group. Even after complete exsanguination, the connective tissue surrounding the nerve shows a high degree of uncleared blood vessels, signifying an ongoing angiogenic process of unconnected blood vessels (arrow). (D) In the short-term group, the tissue beneath the array appears firm, clearly showing each electrode penetration profile. (E) In the longterm group, the tissue beneath the array exhibited evidence of blood products, particularly around the edges of the array footprint. (F) At the longest time point of 350 days post-implantation, the tissue under the array showed signs of peri-prosthetic granulation tissue and appeared mushy (arrow).
4654
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 3. (A) Representative 1000 image of the endoneurial space in the long-term and chronic groups proximal to implant. Macrophages are not typically present in the proximal space after the initial degradation of downstream axons due to Wallerian degeneration. This section shows little evidence of fiber diameter or g-ratio distribution changes, although quantitative data from this group suggests that there is a shift in fiber diameter distributions towards smaller diameter fibers compared with contralateral sections (see Fig. 6). (B) Representative 1000 image of the endoneurial space in all groups distal to the implant. Numerous foamy macrophages can be seen actively degrading axons and their associated myelin sheaths (arrows). There appears to be preferential degradation of large diameter fibers. (C) Epineurial space showing hypercellularity in tissue adjacent to silicone cuff. (D) Higher magnification view from (C). Several different cells types are present, although individual cell boundaries are often not distinguishable. Cells are occasionally visible in the vasculature, even after perfusion. (E) Foreign body giant cells are present in the hypercellular zone encapsulating presumptive myelin debris (arrow). Scale bars: A, B, E = 50 lm; C, D = 100 lm.
Fig. 4. (A) Cross-section of nerve from the chronic group at the implant site, labeled with DAPI (blue) and NF200 (green), showing penetration of the electrodes deep within the fascicular structure. An outline of a USEA, modified from a photograph, is shown for reference. From the USEA outline, electrode penetration is estimated to be 1.1 mm in this figure (or the sixth row of the array). (B) Cross-section of a nerve from the long-term group at the electrode nerve interface, labeled with MAC387 (green) and b-III tubulin (red), showing macrophages around the electrode tips and neuronal processes outside of the normal fascicular structure. (C) Cross-section from a short-term animal of the nerve proximal to the array but still beneath the silicone cuff (a not-to-scale outline of the cuff is shown in green). The epineurial area adjacent to the silicone (outer edge of the nerve) shows a large degree of hypercellularity, which includes numerous foreign body giant cells (Fig. 3). (D) Same nerve as in (A), stained with MAC387, which confirms that a large number of cells adjacent to the silicone are phagocytic cells. Scale bars: A = 500 lm; B = 100 lm; C, D = 1 mm.
with contralateral unimplanted nerves. A difference was also observed in fiber density when we compared implanted distal sections and contralateral nerves across all animals (Table 2). For fiber count, distal sections had a lower total number of fibers than
proximal sections only in the short-term group. When evaluating data across all animals, both distal and proximal sections had significantly lower fiber packing percentages than contralateral nerves. Where sufficient statistical power was available, we saw no significant
4655
M.B. Christensen et al. / Acta Biomaterialia 10 (2014) 4650–4660
Fig. 5. (A) Longitudinal section taken at the implant site from the short-term group, labeled for MAC387 (green) and NF200 (red). This section was taken near the surface of the nerve, at the base of the array. Numerous MAC387 positive cells surround each electrode and fill the epineurial space. Axons are also disrupted around electrodes. (B) Image from the chronic group, showing that MAC387 (green) positive cells are still present at long time points and continue down the electrodes into the endoneurial space, although they are not as concentrated as in the epineurial space. Cells positive for MAC387 are also found in the endoneurial space away from the electrodes. (C) Section adjacent to that shown in (B), labeled for S-100b (green), NF200 (red) and DAPI (blue). Axons and their associated myelin sheaths are diverted around the electrodes, greatly increasing the width of the fascicles under the implant. (D) Inset from (C). Severed or dead-end regenerating axons are seen adjacent to an electrode. Scale bars: A–C = 1 mm; D = 200 lm.
differences in mean fascicular areas, fiber counts or g-ratio values at any location in the nerve when evaluating data across all animals. Average fiber diameter distribution for contralateral nerves showed a trimodal distribution, with peaks around 4.5–5.5, 10.5– 11.5 and 13.5–14 lm. For statistical evaluation, small (
