Silicone-based Intracortical Implants with Brain-like Stiffness Reduces the Brain Foreign Body Response
Edward N. Zhang Department of Biomedical Engineering McGill University, Montreal December 2019
A thesis submitted to McGill University in partial fulfillment of the requirements of the degree Master of Engineering
© Edward N. Zhang 2019
Page | 1
Table of Contents 1. Abstracts ...... 4 1.1. English Abstract ...... 4 1.2. Résumé Français ...... 5 2. Acknowledgements ...... 7 3. Contribution of Authors ...... 9 4. Project Description ...... 10 4.1. Motivation ...... 10 4.2. Project Goals ...... 10 4.3. Declaration of Novelty ...... 11 5. Introduction: Brain Implants ...... 12 5.1. Brain Implant’s in Clinical Practice ...... 14 5.1.1. Clinical Significance and Applications of Deep Brain Stimulation Implants ...... 14 5.1.2. Clinical Significance and Applications of Intracortical Brain Implants ...... 15 5.1.3. Clinical Need for Brain Implants ...... 16 5.2. Brain Implant Challenges in Long-term Reliability and High-Quality Recordings ...... 18 5.2.1. Brain Implant Designs ...... 18 5.2.2. Issues in Electrode Functionality and Signal Quality in Chronic Recordings ...... 21 5.2.3. Failure Modes of Brain Implants ...... 24 5.3. The Brain Foreign Body Response ...... 25 5.3.1. Roles of Reactive Astrocytes in the Brain FBR ...... 27 5.3.2. Roles of Activated Microglia in the Brain FBR ...... 29 5.3.3. Glial Scar in the Chronic Brain FBR ...... 30 5.4. Stiffness Mismatch Between Brain Implants and Brain Tissue Exacerbates Brain FBR ...... 33 5.4.1. Neuron and Glial Response to Stiff and Compliant Substrates ...... 35 5.4.2. Strain Caused by Brain Implants on Brain Tissue ...... 36 5.4.3. In Vivo Observation of Exacerbated brain FBR due to Young’s Modulus Mismatch ...... 39 5.5. Current Polymer Materials for Compliant Implant Fabrication and Challenges ...... 43 5.5.1. Polyimide ...... 45 5.5.2. Parylene C ...... 47 5.5.3. SU-8 ...... 48 5.5.4. Other Polymers ...... 49 5.6. Current Insertion Methods of Compliant Intracortical Implants and Challenges ...... 52 5.6.1. Dissolvable Coatings ...... 52 5.6.2. Insertion Shuttles ...... 54 5.7 Summary ...... 55
Page | 2
6. Silicone Implants with Brain-like Stiffness Delivered Using Micro-Molded Dissolvable Sugar Shuttles Reduce the Brain Foreign Body Response ...... 56 6.1. Abstract ...... 56 6.2 Introduction ...... 57 6.3. Results and Discussion ...... 59 6.3.1 Fabrication of Sacrificial Sugar Molds ...... 59 6.3.2. Fabrication of Ecoflex Implants ...... 62 6.3.3. Characterization of Ecoflex and PDMS Young’s Modulus ...... 64 6.3.4. Encasing of Implants in Dissolvable Sugar Shuttles and Implantation into Rats ...... 67 6.3.5. In Vivo Assessment of Brain Foreign Body Response ...... 70 6.4. Conclusion ...... 75 6.5. Experimental Section ...... 77 6.5.1. Ecoflex Implant Fabrication ...... 77 6.5.2. Characterization of Ecoflex and PDMS Young’s Modulus ...... 78 6.5.3. Insertion of Implants ...... 79 6.5.4. Analysis of Brain Foreign Body Response ...... 82 6.6 Supplementary Data ...... 84 6.6.1. SU-8 Mold Fabrication ...... 84 6.6.2. PDMS Implant Fabrication ...... 87 6.6.3. Silicon Implant Fabrication ...... 89 7. Discussion ...... 94 7.1 Fabrication of Ecoflex and PDMS Implants ...... 94 7.1.1 Fabrication of Sacrificial Sugar Molds ...... 94 7.1.2 Vacuum Assisted Molding (VAM) of Soft Implants ...... 96 7.2 VAM of Dissolvable Sugar Shuttles for Delivery of Soft Implants into the Brain ...... 98 7.3 In Vivo Analysis of the Brain FBR to Ecoflex, PDMS, and Silicon Implants ...... 102 8. Conclusion ...... 103 8.1 Future Work ...... 104 9. References ...... 106
Page | 3
1. Abstracts
1.1. English Abstract
Brain implants have significant implications in treating neurological disorders and diseases and are a key enabler of brain machine interface (BMI) technology. However, brain implants’ widespread adoption in clinical settings is undermined by challenges in obtaining high-quality recordings and issues of long-term reliability caused by the elicited brain foreign body response
(FBR). The brain FBR is exacerbated by the substantial mismatch in Young’s modulus (E) between current implants made from materials like silicon (150 - 200 GPa) and brain tissue (0.4 - 15 kPa).
Recent research directed towards improving the mechanical compatibility of brain implants has given rise to more compliant implants made from materials like polyimide (4 - 8 GPa) that are flexible but are still orders of magnitude stiffer than the brain.
Here, we present novel fabrication and implantation methods for the softest sub- millimetre intracortical implant to date made from Ecoflex, a silicone elastomer with E = 20 kPa.
Non-functional Ecoflex implants (300 µm wide x 200 µm thick x 3 mm long) were fabricated using a sacrificial sugar mold replicated via soft lithography that was coupled with vacuum-assisted molding (VAM). To address the challenge of inserting soft implants into the brain, the implants were encased inside micro-molded dissolvable sugar shuttles (700 µm wide x 450 µm thick x 8 mm long) as temporary structural support for the implants to reliably penetrate brain tissue and to be delivered to the target location accurately. The brain FBR of Ecoflex implants was compared to those elicited by PDMS and silicon implants 3-week post implantation in rats.
Immunohistochemistry results show lower brain FBR elicited by Ecoflex implants compared to both silicon and PDMS implants suggesting that a functional soft implant made from materials
Page | 4 closer to brain stiffness such as Ecoflex could potentially demonstrate superior high-quality recordings and long-term reliability by reducing the brain FBR.
1.2. Résumé Français
Les implants cérébraux jouent un rôle important dans le traitement des troubles et maladies neurologiques et constituent un élément clé de l’interface cerveau-machine (IMC). Cependant, l’adoption généralisée des implants cérébraux en milieu clinique est limitée par la difficulté à obtenir des enregistrements de haute qualité et les problèmes de fiabilité à long terme causés par la réaction à un corps étranger (RCE) dans le cerveau. Cette RCE est exacerbée par l’importante différence entre le module de Young (E) du cerveau (0,4-15 kPa) et celui des implants cérébraux existants, généralement composés de silicium (100-200 GPa). Des recherches récentes portant sur l’amélioration de la compatibilité mécanique des implants cérébraux ont engendré des implants plus souples, faits par exemple de polyimide (4-8 GPa), qui font preuve d’une plus grande flexibilité mais demeurent plus rigides que le cerveau de plusieurs ordres de magnitude.
Nous présentons ici de nouvelles méthodes pour la fabrication et l’implantation de l’implant cérébral le plus souple à ce jour, fait d’Ecoflex, un élastomère de silicone d’une rigidité
E=20 kPa. Les implants non fonctionnels en Ecoflex (300 µm de largeur x 200 µm d’épaisseur x 3 mm de longueur) ont été fabriqués à l’aide d’un moule sacrificiel en sucre reproduit par lithographie douce combiné au moulage assisté par le vide (MAV). Pour relever le défi de l’insertion des implants souples dans le cerveau, ces derniers ont été encapsulés dans des navettes de sucre soluble (700 µm de largeur x 450 µm d'épaisseur x 8 mm de longueur) produites par micromoulage. Les navettes agissent à titre de support structural temporaire
Page | 5 pour que les implants puissent pénétrer de manière fiable dans le tissu cérébral et être livrés à l’emplacement cible avec précision. La RCE cérébrale des implants en Ecoflex a été comparée à celles générées par des implants faits de PDMS et de silicium trois semaines après leur implantation chez des rats. Les résultats obtenus en immunohistochimie montrent qu’en général, la RCE induite dans le cerveau par les implants en Ecoflex est moindre comparativement à celles induites par les implants de silicium et de PDMS. Cela suggère qu’un implant souple fonctionnel fait de matériaux d’une rigidité plus proche de celle du cerveau, tel qu’Ecoflex, pourrait potentiellement mener à des enregistrements de qualité supérieure et à une meilleure fiabilité à long terme en réduisant la RCE dans le cerveau.
Page | 6
2. Acknowledgements
I would like to express my sincere thanks to my supervisors Prof. David Juncker and Prof. Timothy
Kennedy for taking the risk of recruiting me as a master’s student. I came into my master’s with a desire to venture into biomedical engineering but from an electrical engineering background.
Thank you both for the opportunity that you have offered me to pursue my academic endeavors as well as providing me the learning experiences to develop as a scientist. Without your support throughout the course of my master’s, the work presented in this thesis would not have been possible.
To the colleagues that I have befriended in both the Juncker lab and the Kennedy lab, I am incredibly grateful for your continued support and friendship during my master’s. This experience would not have been the same without the moments of struggles and laughter that were shared together both inside and outside of the lab. Particularly, I would like to thank Alia
Alameri and Jean-Pierre Clément for both of your selfless support and assistance in my project.
I would also like to thank Dr. Alex Hernandez and Auxtine Micalet for their contributions to the vacuum assisted molding portion of the project. To Dr. Andy Ng, thank you for your mentorship throughout my master’s that helped me to stay on track and to remain encouraged despite the initial setbacks.
I would also like to thank Dr. Zhao Du, Alireza Mesgar, Jun Li, Mcolisi Dlamini, and
Christophe Clément for the microfabrication trainings and assistance. Guangyu Bao and Zhenwei
Ma, thank you both for the training and assistance in atomic force microscopy and tensile testing.
I have learned and developed a variety of technical skills because of all of you. Also, a special
Page | 7 thanks to my colleagues and friends from the Biomedical and Biological Engineering Student
Society for a productive and enjoyable council experience.
During my master’s I have received generous financial support from the Biomedical
Engineering department in the form of a recruitment award as well as an excellence award—I cannot be more grateful for the department’s willingness to support my studies. I am also thankful for the support from the CHIR-NSERC Collaborative Health Research Program Grant
(NSERC 493633-16, CIHR 357055) which funded my stipend. Lastly, thank you Mom and Dad for everything that the both of you have done for me and for your continued support in my academic pursuits. All of this would not have been possible without your efforts and sacrifices made in my interest.
Page | 8
3. Contribution of Authors
This thesis was written by Edward Zhang with revision help from Prof. David Juncker and Prof.
Timothy Kennedy. The French version of the thesis abstract was edited by Rosalie Martel. The manuscript was written by Edward with revision help from Prof. Juncker, Prof. Kennedy, and Dr.
Andy Ng. All experiments were completed by Edward except for rat surgeries and tissue cryosectioning which was done by Jean-Pierre Clément and confocal microscopy of the tissues which was done by Alia Alameri. VAM prototyping was done with Auxtine Micalet with input from
Dr. Alex Hernandez. MRI was a paid service done by Marius Tuznik. All data analysis and figure preparation were done by Edward.
Page | 9
4. Project Description
4.1. Motivation
Brain implants are devices that interface directly with the central nervous system (CNS) and serve as a bridge between the CNS and the outside world. These devices have the potential to treat neurological disorders and diseases that currently have no other recourse. However, current brain implants have difficulties in obtaining high-quality recordings and have issues in long-term reliability caused by the elicited brain FBR. This limits brain implants’ efficacy as medical treatments and prevents their widespread adoption for clinical use. Hence, the motivation for our work is to reduce the elicited brain FBR so that brain implants could become viable medical treatments. Current brain implants made from materials such as silicon induce significant tissue damage and exacerbate the brain FBR due to the significant mechanical mismatch between brain implants and brain tissue. The field has acknowledged the need to develop more compliant implants and has since developed flexible implants made from materials such as polyimide to reduce the brain FBR. However, these so-called compliant implants are still orders of magnitude stiffer than brain tissue hence we seek to develop an implant that has a stiffness closer to that of the brain to further reduce the brain FBR.
4.2. Project Goals
This project has three mains goals:
• fabricate a biocompatible brain implant (non functional) with a Young’s modulus similar to
that of the brain
• successfully deliver the implant into rat subjects without buckling or deflection
• assess the brain FBR evoked by the soft implant compared to PDMS and silicon implants
Page | 10
4.3. Declaration of Novelty
To the best of our knowledge, this work presents the softest fabricated sub-millimetre intracortical implant to date and is the first time Ecoflex has been used as the fabrication material for brain implants. Furthermore, our work is the first to couple sacrificial sugar molds and VAM to fabricate silicone elastomer-based brain implants to address the issues of residual membrane formation on device features and damage to soft micro-scale devices during the release process in conventional molding methods. Our work also presents a novel encasing method of the soft implants in sugar using VAM for the reliable delivery of the implants into neural tissue. This method allows for unprecedented control and reproducibility of the sugar encasing dimensions that has been unachievable with previous methods of coating. We also conduct a quantitative study of the brain FBR study providing data of the elicited brain FBR in vivo to implants of 20 kPa,
1.6 MPa, and 180 GPa offering a comparison range not explored in previous investigations.
Page | 11
5. Introduction: Brain Implants
Brain implants are devices that interface directly with the central nervous system (CNS) and serve as a bridge between the CNS and the outside world. The beginnings of brain implants were rooted in academic studies of neural circuitry through the recording and modulation of neuronal activity19,20, but they quickly became a technology of interest for clinical applications due to their potential for the treatment of neurological disorders and diseases as well as their potential restorative capabilities for individuals with motor dysfunction or limb loss.
Intracortical Implant Generally, brain DBS Implant
Cortex (Grey implants can be Matter) Thalamus categorized into two
different types, deep White Matter brain stimulation (DBS)
implants, and
intracortical implants.
These two types of
Figure 1 | Cross section of brain showing locations of implantation for DBS implants operate at and intracortical implants. Thalamus, subthalamic nucleus, globus pallidus (typical regions of DBS stimulation) shown in blue. Cortical region for different locations of intracortical stimulation shown in dark pink. the CNS as well as have different intended purposes. DBS implants are primarily used to electrically stimulate subcortical regions of the CNS such as the thalamus, subthalamus, and globus pallidus to supress or mitigate the manifested symptoms of neurological disorders. Intracortical implants are used to record from and stimulate cortical regions of the CNS such as the motor cortex and visual cortex that is
Page | 12 anywhere between 1 to 4.5 mm in thickness in the human brain21. Figure 1 illustrates areas of operation of DBS and intracortical implants in the CNS. The neuronal activities recorded or modulated by intracortical implants are utilized by brain machine interfaces (BMI) in a close feedback loop to enable volitional control of prosthetics or communication devices for individuals who have compromised motor or sensory pathways. This type of functionality requires high temporal and spatial resolution that is enabled by the high density of electrodes found on intracortical implants which can achieve a level of resolution down to the level of single neurons22. Further details of brain implant designs are discussed in Section 5.2.1.
The crux of BMIs is
the bidirectional design of
intracortical implants to
both record from and
stimulate cortical
circuits24. This allows BMIs
to provide an alternate
mechanism for
Figure 2 | Schematic of BMI for controlling a robotic prosthetic arm. Multiple movement, implanted microelectrode arrays (MEA) record from large populations of cortical neurons and the recorded activity is mathematically transformed into 3D communication, or arm-trajectory signals used to control the robotic prosthetic arm. Reproduced from ref.23 with permission from Springer Nature, copyright 2001. sensations aside from the normal pathways of an individual that may have been compromised by injury or disease. Figure
2 illustrates the role of brain implants in a BMI to provide the volitional control of a robotic prosthetic arm.
Page | 13
5.1. Brain Implant’s in Clinical Practice
5.1.1. Clinical Significance and Applications of Deep Brain Stimulation Implants
The clinical significance of brain implants for chronic stimulation as a therapeutic method began with the works of Alim Benabid who observed improved symptoms of Parkinson’s disease (PD) after electrically stimulating the VIM thalamic nucleus of patients in 198725. The pioneer work of
Benabid encouraged others from the scientific and medical communities to explore the use of
DBS for clinical applications leading to the development and subsequent approval of Medtronic’s unilateral DBS system for the treatment of essential tremor (ET) and severe cases of PD in 1997 by the Food and Drug Administration (FDA)26. Although DBS demonstrated compelling results in the treatment of ET and PD, it was nonetheless an invasive treatment with unknown adverse long-term effects. As a result, DBS underwent an additional five years of clinical trials to ensure the safety of the technology before the FDA approved it for the treatment of general PD in 200226.
The gradual acceptance of DBS ensued accordingly as it continued to demonstrate unparalleled efficacy in treating ET and PD since its FDA approval. Multiple clinical investigations report DBS significantly reduced tremor in ET patients27,28 and addressed many motor-related symptoms manifested in PD such as involuntary movements29,30 and dyskinesias31,32. The clinical success of DBS prompted it to become the preferred treatment for late-stage PD over alternatives such as thalamotomy or pallidotomy33 and has motivated investigations into the use of DBS for a variety of neurological disorders and diseases such as dystonia34,35, Alzheimer’s disease (AD)36,37, Tourette syndrome38,39, Huntington’s disease40,41, obsessive-compulsive disorder42,43, depression44,45, pain46,47, epilepsy48,49, and even obesity50.
Page | 14
5.1.2. Clinical Significance and Applications of Intracortical Brain Implants
Originally used as research devices to record and to modulate cortical circuits in non-human subjects, the clinical significance of intracortical implants took an upturn in the late 1980s coinciding with critical developments in brain implants enabled by the advancements in microfabrication technologies. The convergence of the advancements in our understanding of neural circuity and the continued progress in the development of intracortical implants attracted a growing amount of clinical interest in recent years expanding the use of intracortical implants as mere research tools to potential clinical treatments.
The first FDA approved clinical trial of BrainGate’s neuromotor prostheses systems led by
Leigh Hochberg in 2004 demonstrated the extraordinary potential of intracortical implants17. The subject of the clinical trial, a 25-year-old man who sustained a knife wound resulting in complete tetraplegia, had a microelectrode array (MEA) surgically implanted into his primary motor cortex that translated his imagined limb motions into volitional controls of a prosthetic arm17. This was one of the many demonstrations of clinical application of intracortical implants with others including a tetraplegic subject who controlled a prosthetic arm to drink coffee from a bottle51, a subject with spinal cord injury (SCI) having restored certain tactile sensations in their hand52, and an amyotrophic lateral sclerosis subject who controlled a computer cursor to type sentences53.
Although the functionality offered to these individuals by brain implant technology still lacked in comparison to what is normal, the fact that some abilities that were lost due to their conditions were recovered is nevertheless very encouraging for those living with neurological disorders and diseases that currently have no other recourse. However, the persistent challenge of obtaining high-quality recordings from brain implants and issues with long-term reliability prevent the
Page | 15 widespread adoption of the technology in clinical settings. Further advancements of the technology to overcome these challenges is necessary.
5.1.3. Clinical Need for Brain Implants
The clinical need for brain implants is evident by the number of neurological disorders and diseases that currently have either unsatisfactory or no treatment options that could be addressed with brain implants. The demographic shift of the world’s population towards a growing proportion of elderly people is directly correlated with the increase in the prevalence
14 13.20 of age-related 12 neurological 10 8 5.83 diseases such as 6 5.10
Disease Disease (Million) 4 3.00 2.53 PD and AD12,20. 1.88
2 1.40 1.13 Indivviduals with Indivviduals with Disorder or These diseases Tourette Syndrome Essential Tremor Parkinson's Disease Alzheimer's Disease greatly decrease 2010 2050 the quality of life Figure 3 | United States population affected by neurological disorders of the individuals and diseases in 2010 and the corresponding projection in 203012,13. affected and places a heavy burden on the health care system. In the United States, for PD alone, it is projected that the occurrence of the disease will increase by more 220% between 2010 and
2050 from 1.13 million to 2.54 million individuals (Figure 3)13. Similarly, AD is projected to experience more than a 250% increase in the occurrence of the disease between 2010 and 2050 from 5.1 million to 13.2 million individuals12. In addition to neurological diseases, individuals suffering from paralysis could also benefit from brain implants by allowing for volitional control of prosthetics (Figure 2). The Christopher & Dana Reeve Foundation released statistics for 2013
Page | 16 indicating the prevalence of paralysis in the United States, reporting that approximately 5.4 million individuals are living with paralysis54. Out of the 5.4 million
individuals, more than 85% of the occurrence is related to a neurological condition (Figure 4)1.
The promising
.45M .65M clinical results and the
apparent need for 1.48M 1.61M brain implants have
attracted significant 1.00M support from Other Cause Stroke Multiple Sclerosis Spinal Cord Injury Cerebral Palsy governmental
Figure 4 | 2013 population of individuals with paralysis in the United States research initiatives to represented by the different causes of paralysis1.
further develop the technology in these past few years. For example, the United States launched the Brain Research through Advancing Innovative Neurotechnologies (BRAIN) initiative in 2013 led by the National
Institute of Health (NIH) funding multiple investigations in the study of DBS and intracortical implants for clinical use55. The United State’s Defence Advanced Research Project Agency
(DARPA) also provided substantial support for developing brain implant technologies by funding programs such as Revolutionizing Prosthetics in 2006, Reorganization and Plasticity to Accelerate
Injury Recovery (REPAIR) in 2010, Restorative Encoding Memory Integration Neural Device
(REMIND) in 2009, Reliable Neural Interface Technology (RE-NET) in 2010, and Neural
Engineering System
Page | 17
Design (NESD) in 201656. This suggests that the implications of brain implants for use outside of a clinical setting could be possible given that the intentions of many of the DARPA supported programs aim to restore neural and behavioral function as well as to improve human training and performance using BMIs56. Outside of the United States there exists the Human Brain
Project launched by the European Union in 2013 and the China Brain Project launched by the
Chinese Academy of Sciences in 201657. The need for brain implant technologies has elicited substantial support and opportunities from multiple governmental research initiatives. The realization of brain implants for the treatment of neurological disorders and diseases now is subject to the advancement of the technology itself.
5.2. Brain Implant Challenges in Long-term Reliability and High-Quality Recordings
5.2.1. Brain Implant Designs
As discussed in the beginning of Chapter 5, DBS implants are primarily used to unilaterally or bilaterally stimulate subcortical regions of the brain to treat neurological disorders58. The design of DBS implants has changed very little since inception. Because the DBS implants operate in deep regions of the CNS and stimulate large populations of neurons at a time, they are much longer and larger in comparison to intracortical implants. DBS implants resemble a flexible wire with a diameter of around 1.5 mm and 40 – 60 mm in length59. They are often fabricated with 4
- 8 platinum/iridium contacts that are 1.5 mm in height with different spacings between the contacts. Figure 5 A shows examples of Medtronic 3387 and 3389 DBS implants made from platinum/iridium3.
Intracortical implants on the other hand, were originally developed as research tools to record neuronal activity, for the study of mechanisms that regulate attention, movement, and
Page | 18 behavior19. There were subsequently used to modulate target neurons for the use in BMIs. The designs of intracortical implants and materials used are also much more varied compared to that of DBS implants. The key enabler of intracortical implants is the advancement of microfabrication technologies in the 1980s leading to the development of semiconductor based implants such as
Michigan probes in 1986 (Figure 5 B)60 and Utah arrays in 1991 (Figure 5 C)61. Subsequent intracortical implants have been typically based upon the Michigan probe’s planar shank design, the 3D MEA design of the Utah array, or the microwire array design.
A B C
Figure 5 | (A) Image of Medtronic 3387 (left) and 3389 (right) DBS implants with diameters of 1.27 mm and contact heights of 1.5mm3. (B) Image of a 16-channel Michigan probe14. (C) Scanning electron micrograph of the MEA based on the Utah array used by Hochberg17. Figures 5A and 5C reproduced from ref.3 and ref.17 with permission from Springer Nature, copyright 2009 and copyright 2006 respectively. Figure 5B reproduced from ref.14. Copyright © 2004, IEEE.
Intracortical implants with planar designs vary largely in dimension and have been as thin as 1 µm with widths of less than 4 µm62 to, for example, being 125 µm in thickness and 300 µm in width63. This large range in dimensions can be due to reasons such as the intended function of the implant (i.e. single neuron recording or local field potential recording), the materials used for fabrication, and the necessity of coatings for either aiding implantation or bioactive functions.
Similarly, intracortical implants with the array designs can also vary but often have dimensions similar to what is seen in the original Utah array with shank thicknesses
Page | 19
Table 1 | Specifications of various intracortical implants belonging to the three main designs (microelectrode arrays, planar shanks, microwire arrays)
Microelectrode Arrays Planar shanks Microwire Arrays Substrate • Silicon61,64,65 • Epoxy66 -- Materials • Polyimide65 • Parylene C67 • Polyimide68 Conductor • Polysilicon, titanium, • Gold, iridium oxide66 • Stainless Steel70 Materials iridium69 • Tungsten, titanium67 • Platinum, iridium71 • Gold, platinum61 • Titanium, iridium, • Tungsten72 • Titanium, gold68 aluminum65 Insulation • Silicone rubber, • Epoxy66 • Teflon70 Materials silicon dioxide, • Parylene C67 • Polyimide71,72 silicon nitride64 • Polyimide68 • Polyimide61,65 Dimensions • 10 – 15 µm thick, 40 • 130 µm wide, 127 • 400 – 500 µm spacing, µm wide, 2.5 mm µm thick, 25 µm 50 µm diameter, 8 long shanks with 200 electrode diameter, mm long70 µm spacing between 150 µm electrode • 500 µm spacing, 50 shanks64 spacing66 µm diameter71 • 90 µm diameter • Recording disk of • 250 – 375 µm spacing, base, 1.5 mm long, 100 µm diameter, 50 µm diameter, 5 400 µm pitch61 and shaft thickness mm long72 • 25 µm thick, 160 µm of 20 µm, 30 µm wide, 1.2 mm long wide, and 5 mm shanks with 200 µm long67 spacing65 • 48 µm thick, 280 µm wide, 4 mm long68
Fabrication • Micro-machining, • Micro-assembly, • Handmade Methods micro-assembly, micro-molding, assembly70,71 microfabrication61,69 microfabrication66 • Various deposition • Various deposition methods and deep methods, reactive ion etching microfabrication, (DRIE)65 reactive ion etching (RIE)67,68 Material • 190 GPa Young’s • 3 GPa storage • 200 GPa Young’s Stiffness modulus61,64,73 modulus66 modulus70,74 • 36.8 GPa Young’s • 200 GPa Young’s • 160 GPa Young’s modulus65 modulus67 modulus71,75 • 2 GPa68 • 380 GPa Young’s modulus72,76 Comments • 128 channels69 • 32 channels66 • 16 channels70-72 • 100 channels61 • 3 channels67 • 18 channels65 • 8 channels68
Page | 20
of close to 90 µm and pitch distance of 400 µm61. The materials used to fabricate intracortical
implants are also diverse with commonly used materials being silicon, polyimide, and Parylene
C for the implant substrate, and tungsten, gold, and titanium for the conductors. The choice of materials used is important as the material properties ultimately affect implant performance in vivo from both an electrical and biological standpoint. A summary of intracortical implant designs can be seen in Table 1. Further discussion of implant materials and fabrication methods will be presented in Section 5.5.
One particular material property that is important to consider in fabricating brain implants in the Young’s modulus. A characterization of stiffness, the Young’s modulus is largely dependent on the materials used to fabricate the implant. It defines the relationship between the stress and strain of a material and can be acquired experimentally via tensile testing. The mismatch in Young’s modulus between current brain implants and that of the brain tissue has been acknowledged as a major contributor to the brain FBR which will be further discussed in
Section 5.3, 5.4, and 5.5.
5.2.2. Issues in Electrode Functionality and Signal Quality in Chronic Recordings
Challenges in long-term reliability and high-quality recordings are ubiquitous in clinical and academic investigations of brain implants preventing their widespread implementation as effective medical treatments aside from DBS. With respect to intracortical implants, there exists an abundance of literature reporting failures of devices in chronic settings. For example, in the
BrainGate clinical trial led by Hochberg, 54 out of the 96 electrodes on the MEA experienced considerable decrease in recorded activity after 6 months of implantation17. Similarly, a study
Page | 21 done by Rousche et al. (1998) with MEAs implanted into the cerebral cortex of cats reported that only ~60% of the electrodes recorded some type of neuronal activity 6 months after implantation77. Issues of long-term reliability are not isolated to implant type, region of implantation, or implant material. Nicolelis et al. (2003) implanted high-density microwire arrays made from either Teflon-coated stainless steel or Teflon-coated tungsten into different cortical areas of rhesus monkeys and reported on average only 54% of the microwires yielded signals from neurons with the lowest being 26% (25 out of 96 microwires) 30 days after implantation78.
Another study done by Williams et al. (1999) implanted polyimide-coated tungsten microwire arrays into the cortex of guinea pigs and reported on average 80% of the electrodes were active
3 weeks after implantation, but interestingly observed an increase in the average percentage of active electrodes to 84% 9 weeks and onwards after implantation79. However, the average percentage of active electrodes at both 3 weeks and 9 weeks onward was calculated based on the number of animals surviving at the time effectively decreasing the sample population from 8 microwire arrays to either 1 or 2 microwire arrays. Consequently, the increase observed in the percentage of active electrodes does not necessarily reflect non-active electrodes reverting to an active state but rather reflects the average number of active electrodes under the most favorable experimental conditions.
In addition to issues of long-term reliability, there also exists many studies reporting issues with high-quality recordings of intracortical implants in a chronic setting. In a study conducted by Chestek et al. (2011), implanted MEAs in the motor cortices of rhesus monkeys exhibited an average decrease of potential amplitude of 2.4% per month over the course of 9,
10, and 31 months (Figure 6 A)7. Other types of issues with recording reliability include abrupt
Page | 22 disappearances of neuronal activity that were observed by Jackson et al. (2007) when recording a well-isolated neuron with stable signals that abruptly ended on the 6th day (Figure 6 B)16. It is presumed that the abrupt disappearance of the signal was due to the movement of the electrode that may have injured the specific neuron being recorded underscoring the delicate and challenging nature of recording neuronal activity. The translation of signal instability to BMI performance was reported by Perge et al. (2013) whereby clinical trials of the BrainGate system that exhibited significant changes in apparent firing rate in 84% of the recorded neurons yielded a directional bias in 56% of all performance assessments in participant cursor control, resulting in suboptimal
A C
B
Figure 6 | Examples of different types of signal instabilities. (A) Peak-to-peak voltage amplitude change from largest recorded neurons on each channel of 4 MEAs in rhesus monkeys7. (B) Superimposed action potential spike waveforms from a well-isolated neuron recorded using Teflon-coated tungsten microwire arrays in macaque monkeys16. (C) Mean firing rate change of 26 neuronal units (bottom) correlated with control instability of cursor (top) in human subjects with MEAs. Bottom insets represent cursory trajectories with the arrow indicating bias direction. Bottom: Unit 3 (thick black line) rate change strongly correlated with decreased performance. The 3 thicker lines represent the 3 units with significant rate changes18. Figures 6A and 6C reproduced from ref.7 and ref.18 with permission from IOP Publishing, copyright 2011 and 2013 respectively. Figure 6B reproduced from ref.16 with permission from The American Physiological Society, copyright 2007.
Page | 23 performance (Figure 6 C)18. Compromised signal quality can also affect spike detection with a given required threshold contributing to inaccuracies and errors in the processing of neuronal signals for use in other BMI applications.
5.2.3. Failure Modes of Brain Implants
The failure modes of brain implants responsible for issues of long-term reliability and high-quality recordings can be classified into those arising from an engineering origin and those arising from a biological origin. Examples of the failure modes arising from engineering origins include mechanical failure of interconnects80, deterioration or corrosion of electrodes72,81, and degradation of insulating and passivating layers72,80,82,83. Examples of the possible failure modes arising from biological origins include increased tissue impedance from gliosis72,77,84-92, loss of recording signals due to neuronal loss63,87-89,93-97, and compromised implant functionality due to a chronic blood-brain barrier (BBB) breach72,98. Often failure modes arise from a combination of both engineering and biological origins. For example, a chronic breach in the BBB by the brain implant leads to an influx of red blood cells consequently leading to an increase in hemoglobin from red blood cell breakdown in the brain. This in turn increases the amount of reactive oxygen and nitrogen species present in the brain99,100 which are known to be major drivers of corrosion in implants101-103. The underlying phenomenon that is responsible for the biological responses from the central nervous system (CNS) to brain implants is known as the brain foreign body response (FBR) and many studies suggest that the greatest challenge to obtaining long-term brain implant functionality and reliable recordings is due to this inflammatory response that the CNS mounts against the brain implant.
Page | 24
5.3. The Brain Foreign Body Response
The brain FBR refers to the consequent inflammatory and wound healing process that takes place after the insertion of a foreign material into the CNS104. The cascade of cellular responses that occurs following a CNS injury is complex in nature and involves many different populations of cells. However, the consensus in the field is that glial cells, which are non-neuronal cells in the
CNS, specifically astrocytes and microglia play significant roles in the brain FBR104-108. As a result, the assessment of biocompatibility of brain implants can be and is most commonly characterized by the temporal changes of the brain FBR with a focus on the responses made by these glial cells and the assessment of neuronal density surrounding the implant.
The insertion of a brain implant into the CNS will inevitability damage vasculature, the extracellular matrix (ECM), and both glial and neuronal processes regardless of the insertion technique or the size of the implant. Distinct cellular responses to this insult can be categorized into different phases based on the stage of the inflammatory and the wound healing process.
The haemorrhagic phase spans the time point immediately after the implantation to around 3 days post implantation108. During this phase, the immediate breech of the BBB causes an influx of haematogenous cells such as neutrophils, monocytes, macrophages, and leukocytes to the injury site108 releasing a blend of cytokines including tumor necrosis factors (TNF)69, interleukins
(ILs)109,110, and fibroblast growth factors (FGFs)111 inducing migration of activated microglia and reactive astrocytes, the pathological counterparts of normal microglia and astrocytes, to the injury site. The ensuing subacute phase is characterized by a decrease in the haematogenous cell population and an increase in the population of activated microglia and reactive astrocytes108.
The activated microglia phagocytize unwanted debris around the injury site and the reactive
Page | 25 astrocytes begin the scarring process. The subacute phase takes place between day 3 to day 8 post implantation and is followed by the consolidation phase that takes place between day 8 to day 20 post implantation where the contraction of the scar tissue takes place108. Although there is not a clear consensus from literature on the exact phases of the brain FBR, it can be proposed that these 3 phases can be considered to constitute the acute brain FBR with the subsequent brain FBR occurring after these phases known as the chronic brain FBR. A summary of the cascade of immune events following the insertion of a brain implant can be seen in Figure 7.
Figure 7 | Timeline of brain FBR events following the insertion of an intracortical implant.
Page | 26
5.3.1. Roles of Reactive Astrocytes in the Brain FBR
Normal astrocytes serve many different functions in the CNS such as the regulation of cerebral blood flow112, brain metabolic activity107,113, and extracellular pH and potassium homeostatasis107,114. The normal morphology of astrocytes are represented by a star-like appearance105 and when reacting to injury, take on a different morphology characterized by hypertrophy and the upregulation of glial fibrillary acidic protein (GFAP), vimentin, and nestin104,107,115,116 (Figure 8).
The responses of astrocytes to
an injury begins when pro-
inflammatory cytokines such as IL-
1β117, TNFα118, and IL-6119 evoke a
change from resting astrocytes into
reactive astrocytes107,116. Following Figure 8 | GFAP immunostaining of astrocytes illustrate the morphology of normal (left) and reactive (right) astrocytes showing visible hypertrophy2. Scale bar: 25 µm. the transformation, the early Figure 6B reproduced from ref.2 with permission from The American Physiological Society, copyright 2014. responses of reactive astrocytes
include repairing the BBB115, remodeling the ECM, clearing the injury site of debris by protease secretion120, releasing cytokines such as TGF-β and TGF-α108,120, and secreting neurotrophins to enhance neuron survival116. In order to better understand the subsequent roles that reactive astrocytes play during the brain FBR, it is important to realize that the population of reactive astrocytes present during an injury is heterogenous105,115 exhibiting different responses according to the nature and severity of the injury as well as to the different chemical cues present121. Some of the later roles
Page | 27 that reactive astrocytes play include regulation of inflammation, isolation of lesions, regulation of the BBB, reorganization of the neural circuit, and formation of the glial scar115.
In some cases, the activities of reactive astrocytes have seemingly antagonistic effects.
For example, when regulating inflammation, reactive astrocytes can produce either pro or anti- inflammatory molecules. Additionally, reactive astrocytes may produce chondroitin-6-sulfate proteoglycans122,123 and tenascin124, proteins that inhibit axon regeneration. Consequently, reactive astrocytes forming the glial scar have been widely considered to be a barrier to axon regeneration and that by attenuating or preventing the formation of the glial scar, axon regeneration at the injury site can take place. However, a recent study conducted by Anderson et al. (2016) reported that in genetically modified mice that either prevented, attenuated, or deleted the presence of glial scarring at lesion sites caused by induced SCI not only failed to demonstrate regeneration of transected axons but also significantly increased axonal dieback125.
The diverse and occasionally contradictory nature of reactive astrocytes during the brain FBR is complex and not completely understood, but it can be surmised that the varied roles of reactive astrocytes represent the necessary responses of these cells to the multifaced challenges of the brain FBR.
Page | 28
5.3.2. Roles of Activated Microglia in the Brain FBR
Microglia are the resident immune cells of the CNS. They primarily act as phagocytes secreting proteolytic enzymes that degrade cellular debris during regular cell turnover104 and shaping neuronal synapses by phagocytosing inactive dendritic spines126. Upon injury, microglia become
activated, proliferating and taking on a more
compact globular morphology by retracting
their processes108 (Figure 9) and exhibiting
an increased expression of cytokines such as
TNF-α127, IL-1128, and IL-6108,110 as well as Figure 9 | Overlay of Iba-1, Ox6, and Hoechst staining of microglia to illustrate the morphology of resting (left) and activated (right) microglia showing other proteins such as CD68 and CD11b104. enlarged somata, thickened proximal processes, and retracted distal processes8. Reproduced from ref.8 under CC BY 2.0 At the site of the injury, the primary role of activated microglia is to act as cytotoxic cells that kill pathogenic organisms or as phagocytes that degrade cellular debris and damaged ECM104. They are also known to actively strip inhibitory synapses from postsynaptic sites which is believed to be a neuroprotective role126,129,130. Similar to reactive astrocytes, activated microglia release cytotoxins such as proteases, TNF-α, but also neurotrophins that promote neuron survival and axonal regeneration104,108. Likewise, the seemingly contradictory nature of activated microglia suggests that they too can take up a variety of roles to respond to the diverse challenges manifested during the brain FBR.
Studies have shown that the presence of insoluble foreign material in the brain can lead to the surrounding activated microglia’s continuous release of neurotoxic substances131. Multi-
Page | 29 nucleated giant microglia cells were observed by Edell et al. (1992) directly surrounding silicon- based implants in rabbit cortices as well as by Gällentoft et al. (2016) surrounding implanted nanowires in rats resembling the behaviour of macrophages outside of the CNS when they encounter a foreign object that cannot be degraded96,132,133. These activated microglia are key culprits in producing pro-inflammatory and cytotoxic soluble factors that cause neurodegeneration with cytokines such as TNF-α and monocyte chemoattractant protein-1
(MCP-1) having direct toxic effects on the surrounding neurons134. The formation of a kill zone, a region that has significantly lower or non-existent neuronal density but with a significant population of activated microglia can be as large as 60 μm87,96. Furthermore, the sustained pro- inflammatory responses of activated microglia along with the altered expression of matrix metalloproteinases causes a further breakdown of the BBB, which exacerbates the brain FBR and has been cited as a contributing factor that reduces the stability of reliable recordings and long- term functionality of brain implants98,135. Therefore, it is of interest to reduce the pro- inflammatory response of activated microglia to prevent exacerbation of the brain FBR.
5.3.3. Glial Scar in the Chronic Brain FBR
The glial scar is the most distinctive manifestation of the chronic brain FBR. Both reactive astrocytes and activated microglia take part in the formation of the scar that surrounds the injury site and it is presumed that the scar isolates the deleterious inflammatory effects of the brain
FBR from the rest of the CNS107,115,121,136. A schematic of scar formation is presented in Figure 10.
Page | 30
Figure 10 | Chronic brain FBR elicited by a typical brain implant showing a glial scar composed of reactive astrocytes and activated microglia encapsulating the implant and isolating it from the surrounding neurons.
Although this is a positive development in the brain FBR from the perspective of protecting the rest of the CNS from the injury, the insulative properties of a glial scar presents a challenge for brain implants by separating the implant from the neurons of interest hindering both ion diffusion as well as increasing the impedance between the electrodes and the target neurons87,89. Theoretical models predict that the action potential of a neuron will not be reliably detected beyond a distance of ~130 μm from a recording site137 with in vivo experiments corroborating the theoretical claim reporting the maximum distance between a recording site and the target neuron to be between 50 and 140 μm138.
Page | 31
Studies conducted by Turner et al. (1999) observed the formation of a glial scar take place at around 2 weeks post implantation87. Furthermore, a later study by the same group observed that glial scars begin to form a compact sheath at ~6 weeks post implantation with a constant thickness until the termination of the study at 12 weeks, at which point the glial scars measured between 50 – 100 μm in thickness around the implant85. The increase in tissue impedance resulting from the glial scar has been widely considered a hindrance to the recording capabilities of brain implants. However, the impact of the impedance itself on the recording capabilities of brain implants is not well defined. For example, Prasad et al. (2012) reported an inverse correlation between the recording quality and impedance measurements of chronically implanted tungsten microwire arrays in rats139. Similarly, Johnson et al. (2005) reported that a decrease in electrode impedances was paralleled with an improvement in the signal-to-noise- ratio (SNR) of the MEA recordings in their study92. However, other findings such as the one by
Malaga et al. (2015) reported that the glial scars that formed around MEAs implanted into monkeys did not affect the signal quality of neuronal recordings despite an observed increase in impedance140. Furthermore, the action potential decrease with respect to time observed by
Chestek et al. was (2011) presumed by the study to be the effect of material and engineering failures rather than the effect of the glial scar formation, because the decline of the action potential continued consistently beyond 12 weeks at which point the glial scar should have formed and consolidated7. These divergent findings reflect that further investigation is warranted of the complexities of the relationship between the increase in impedance cause by glial scars and the recording quality of brain implants.
Page | 32
Although the barrier formed by the glial scars creates an inhibitory environment for axonal regeneration which repels any neural processes away from the recording site, prevention of glial scar formation following an injury is reported to further increase neuronal loss and demyelination suggesting that the glial scar has a vital role in neuronal survival and maintaining tissue integrity125,141-143. It is known that glia scars undergo extensive remodeling and over time become progressively more permissive to axon growth. The findings regarding the diverse and occasionally antagonistic roles of both activated microglia and reactive astrocytes in the brain
FBR require careful consideration of how to best approach the issue of the chronic brain FBR. We must be careful in modulating the inflammatory response such that the necessary protective and supportive roles of glial cells at the site of the injury are not diminished in favor of abating the overall brain FBR.
5.4. Stiffness Mismatch Between Brain Implants and Brain Tissue Exacerbates Brain FBR
Many strategies have been explored in recent years to reduce the brain FBR with varying degrees of success, although nothing has been considered a definitive solution to date. Surface modifications of brain implants have been investigated to reduce inflammation by the controlled release of anti-inflammatory drugs144 or nanoparticles145. Coatings on the surface of brain implants, such as the use of hydrogels, have also been explored to create biomimetic surfaces to improve the biocompatibility of brain implants146,147. Insertion speed148,149 and tip geometries 150 of implants have been studied to address the adverse effects of tissue dimpling during implantation. Amongst all these possible solutions to mitigating the brain FBR, the most successful approach should be a combination of some of these solutions given the multifaceted nature of the brain FBR.
Page | 33
With this in mind, the mechanical characteristics of brain implants have been amongst the most studied topics with regards to addressing the brain FBR. Many different mechanical parameters of brain implants have been investigated in the context of the inflammatory response or tissue damage including the surface area151, size152, tip geometry78,96, density153, and texture of the brain implant154. Although all these various mechanical parameters do affect the brain FBR to certain degrees, a study done by Szarowski et al. (2003) downplayed most of their roles with respected to the chronic FBR. In this study, micro-machined silicon implants with different mechanical parameters were implanted into the cortex of rats and the brain FBR was studied over the course of 1 day and 1, 2, 4, 6, and 12 weeks post implantation. Although the acute brain
FBR (2 weeks and less) varied with the various mechanical parameters, Szarowski reports that the sustained brain FBR (beyond 4 weeks) was the same regardless of the different mechanical parameters of the implants as they all ultimately resulted in a compact glial sheath indicating that the chronic brain FBR is independent of device shape, geometry, and surface characteristics85.
As mentioned, however, in the beginning of Chapter 5, one mechanical parameter of brain implants that does significantly affect the chronic brain FBR is the drastic mismatch between the Young’s modulus of the brain and that of the materials used in current brain implants. Cells from the CNS form among the softest tissue in the human body and have been reported to have a Young’s modulus between 40 Pa to 20 kPa depending on the measurement methods, frequency and strain magnitudes of the measurements, preparation of the sample, and the population of CNS tissue measured155. In contrast, the Young’s modulus of both silicon and tungsten are orders of magnitudes higher at 130 – 190 GPa73 and 390 – 410 GPa156 respectively.
Given that brain implants are typically affixed to the skull, it becomes apparent that any relative
Page | 34 displacement of the soft brain tissue with respect to the brain implant would cause tissue damage which would perpetually exacerbate the chronic brain FBR.
5.4.1. Neuron and Glial Response to Stiff and Compliant Substrates
The notion that cells are sensitive to the mechanical properties of their surroundings is well established157,158. The ECM of the brain is significantly softer than the ECM that forms in other soft tissues due to the ECM in the CNS being largely devoid of the fibrous matrix proteins such as collagens and fibronectins155,159,160. Given that the environment surrounding neurons and glial cells are relatively soft in nature, it is reasonable to hypothesize that these cells would require implants of similar mechanical properties to the ECM. Neuronal cells from a mouse spinal cord, for example, have been demonstrated by Flanagan et al. (2002) to prefer polyacrylamide (PAA) gels of 500 dyne/cm compared to stiffer PAA gels of 5500 dyne/cm (50 – 550 Pa) with neurons cultured on the softer gels exhibiting better growth and forming more than 3 times as many neurite branches than neurons cultured on the stiffer gels161. Similarly, Kostic et al. (2007) cultured mouse hippocampal cells on PAA substrates between 500 Pa to 7.5 kPa and reported the inhibition of neurite extensions and less formation of primary dendrites of neurons cultured on the stiffer substrates of 7.5 kPa compared to those cultured on the softer substrates162. They further concluded that the stiff substrates trigger force-dependent activation of receptor-like protein tyrosine phosphate alpha (RPTPα) which inhibits neurite extensions offering an explanation as to why softer substrates are more favorable for neurons.
Page | 35
With respect to glial cells, Moshayedi et al. (2010) cultured astrocytes from neonatal rat cortices on PAA gels with storage modulus ranging from 100 Pa to 10 kPa and observed that the astrocytes cultured on the stiff gels exhibited significantly more processes and adopted a more spread out morphology compared to those cultured on softer gels (Figure 12)163. In a subsequent
Figure 12 | Fluorescence images of microglia cultured on compliant and stiff PAA substrates stained for TLR4 and PPARγ (left). Scale bar: 50 µm. Fluorescence images of astrocytes cultured on compliant and stiff PAA substrates stained for caspase-1 and IL-1β (middle). Scale bar: 50 µm. Astrocyte and microglia morphology on compliant and stiff PAA substrates (right). Scale bar 30 µm11. Reproduced from ref.11 under CC BY 3.0.
study conducted by Moshayedi, it was reported that that astrocytes and microglia cultured on the stiff substrates displayed characteristics of reactive astrocytes and activated microglia typical of the late chronic brain FBR, exhibiting significant upregulation of IL-1β and caspase-1 for the astrocytes and toll-like receptor 4 (TLR4), receptors mediating microglia activation, and peroxisome proliferator-activated receptor gamma (PPARγ), markers of microglia activation, for the microglia (Figure 10)11. This suggests that even in the absence of injury, the interplay between stiff materials and neurons and glia cells can evoke unfavorable responses.
5.4.2. Strain Caused by Brain Implants on Brain Tissue
The displacement of the brain tissue with respect to the brain implant can be attributed to the forces that result from rotational acceleration of the head164 or from respiratory or cardiac
Page | 36 pulsations known as micromotions165,166. These displacements of brain tissue cause interfacial strain at the brain-implant interface and motivated multiple finite element modeling (FEM) studies to explore the effects of the Young’s modulus mismatch between the brain tissue and the brain implant on the tissue strain has been postulated to be a contributing factor to the exacerbated chronic brain FBR15,167. In one FEM study, Subbaroyan et al. (2005) simulated the strain arising from a silicon, polyimide, and hypothetical implant with greater compliance with
Young’s modulus of 200 GPa, 3 GPa, and 6 MPa respectively15. Subbaroyan reports that the strain profile of silicon implants arising from a radial tethering force displacing the implant by just 1 µm resulted in symmetrical strain areas in the brain tissue along the sidewalls of the implant that resulted from the shearing of tissue extending up to 100 µm from the implant interface15. When comparing the strain profiles arising from the radial tethering force of the silicon implant to the polyimide and hypothetical implant, Subbaroyan reports a parallel correlation between the decrease in strain profile with the decrease in implant Young’s modulus at all locations of the implant which can be seen in Figure 1215.
Page | 37
A B
Figure 12 | (A) FEM strain profile in brain tissue resulting from a 1 µm displacement arising from the radial tethering force of a silicon implant. (B) Normalized strain values of a silicon, polyimide, and soft implant at the implant surface, mid-point, and tip arising from a radial tethering force15. Reproduced from ref.15 with permission from IOP Publishing, copyright 2005.
In an in vivo study conducted by Sridharan et al. (2015), the mechanical interactions between the rat brain and silicon (stiff) and nanocomposite (compliant) shanks were investigated by using a 10g load cell for dynamic force measurements and the estimated stress was acquired from the experimental force measurements. It was validated in this study that the compliant implants greatly minimized the mechanical stress at the brain-implant interface arising from brain micromotion. The average micromotion induced stress from the silicon shanks were around
170 Pa compared to the 50 Pa induced by the compliant implant. However, given the relatively low stresses induced by the micromotion of the implants, it is more likely that the exacerbated brain FBR is due to the strains arising from tethering.
Page | 38
5.4.3. In Vivo Observation of Exacerbated brain FBR due to Young’s Modulus Mismatch
In vivo studies that investigated the effects of brain implant stiffness on the brain FBR indicate that in general, there is a reduction of the observed brain FBR with a more compliant implant.
Some of these studies directly compared the induced brain FBR between brain implants of varying stiffness, and all of them reported a significant decrease in the chronic brain FBR elicited by implants with a lower Young’s modulus. In a study done by Harris et al. (2011), the brain FBR was compared between a traditional tungsten microwire implant (411 GPa) and a mechanically adaptive polymer implant made of poly(vinyl acetate) (PVAc) and tunicate whiskers that had a tensile storage modulus of 5 GPa, but decreased to 12 MPa when exposed to physiological conditions. The study reported significantly higher neuronal density within 100 µm of the compliant implant compared to that of the tungsten implant 4 weeks post implantation as well as a less exacerbated glial scar 8 weeks post implantation168. Du et al. (2017) reported similar findings with a soft microwire (974 kPa) fabricated from the conductive polymer poly(3,4- ethylenedioxythiophene) and polyethylene glycol (PEDOT-PEG) with polydimethylsiloxane
(PDMS). Du compared the brain FBR elicited by the soft implant to that of a tungsten implant
(400 GPa) and reported significantly higher expressions of GFAP and decreased neuronal density around the stiff implant compared to that of the soft implant at 8 weeks post implantation169.
Given the influx of newly developed brain implants with various Young’s modulus less than that of the traditional silicon and tungsten-based brain implants, Lee et al. (2017) conducted a study to determine if a critical Young’s modulus exists, beyond which the favorable effects in mitigating the chronic brain FBR plateaus. In this study, direct comparisons were made in mice of the brain FBR elicited by tethered implants made from silicon (150 GPa), polyimide (1.5 GPa),
Page | 39 off-stoichiometry thiol-enes-epoxy (OSTE+) hard (300 MPa), and OSTE+ soft (6 MPa). A significant decrease in the fluorescence intensity of biomarkers for activated microglia, and BBB leakiness was observed at 8 weeks post implantation between the polyimide implant and the silicon implant. However, neuronal density comparisons between the implants did not show a specific correlation with the Young’s modulus of the implants with the highest neuronal density expressed around the polyimide implant, then the silicon implant, then the OSTE+ soft implant, and lastly the OSTE+ hard implant 8 weeks post implantation. Similarly, the expression of reactive astrocytes also demonstrated unexpected results with the only conclusive correlation between a lower Young’s modulus and decreased GFAP expression made between the OSTE+ soft implant and the silicon implant170. Given these findings, Lee suggested that further reducing the Young’s modulus below that of the polyimide implant does not result in additional benefits in reducing the brain FBR.
However, Moshayedi et al. (2014) reported a significant difference in the brain FBR elicited by stiff PAA gels (30 kPa) and compliant gels (100 Pa) that were inserted into rats but untethered to the skull11. In this study it was reported that compared to the soft gels, the stiff gels expressed significantly higher amounts of GFAP, IL-1β, and Cd11b indicating an exacerbated brain FBR elicited by the stiffer gels at both 1 and 3 weeks post implantation (Figure 13)11. The results in Lee’s study could have been attributed to the varying thickness of the PEG used to coat the implants. Given that the implants of the study were ~20 µm thick and the PEG coating was
~50 µm thick with measurement standard deviation of more than 30%, the variation of the PEG coating thickness could contribute to
Page | 40
variability and bias the results and could A be a source of error in the reported
observations of elicited brain FBR.
Furthermore, the initiators responsible
B for cross-linking between the thiol and
the epoxy resin of OSTE+ have been
reported to be cytotoxic, which could
explain the lower observed neuronal
densities surrounding the OSTE+ Figure 13 | (A) Immunohistochemistry of brain FBR in rats of 100 Pa PAA gel staining for Hoechst showing neuronal nuclei (blue), CD11b showing activated implants compared to that of the silicon microglia (red), and GFAP showing reactive astrocytes (green) (left). Immunohistochemistry of brain FBR of 30 and polyimide implants171. kPa Pa PAA gel (right). Scale bar: 50 µm. (B) Quantitative analysis of the fluorescence intensity GFAP, IL-1β, and Cd11b of stiff (30 kPa) and compliant (100 Pa) PAA gels11. Another justification of the need Reproduced from ref.11 under CC BY 3.0. for soft implants is with respect to brain
Figure 14 | Comparison of Ecoflex and PDMS stress-strain curves from tensile tests. Ecoflex stress-strain curves with (n = 6) stiffness is shown on the left and PDMS stress-strain curves with (n = 6) is shown on the right.
Page | 41 tissue displacement and an implant’s capacity for axial strain. Voluntary movements of the head in people produces anywhere between 1 to 3 mm relative brain displacement172,173. Normal processes such as learning new skills or even exercising can temporarily increase portions of the human brain, such as the motor cortex, by 2-3 percent in volume174,175. With respect to both tethered and untethered implants, implants made from materials with a lower modulus have better axial strain capacity and readily stretch and compress according to the displacement or swelling of the brain. Figure 14 compares the stress-strain curves of PDMS and Ecoflex that we acquired experimentally. These are two silicone elastomers with a Young’s modulus of 1.6 MPa and 20 kPa respectively. It can be seen from the graph that in order to achieve a 10% relative strain, PDMS required a stress force more than 700% greater than Ecoflex. This indicates that softer materials such as Ecoflex will be superior in coupling with brain tissue during tissue displacement or swelling, forming better electrical connections as well has inducing less tissue damage.
Lastly, it is important to note that all the studies done with regard to brain FBR were conducted in animal models. Although it would be optimal to have data from human subjects, the study of brain FBR response to implants cannot be systematically conducted in people.
Additionally, alternative models such as brain organoids derived from human stem cells currently do not reflect their in vivo counterparts as they lack vasculature, have discontinuous neuroepithelial structures, and most importantly have an absence of immune cells176. This is not to say that animal models are limited in their capacity to represent a human brain FBR, but rather they are very informative. Although animal models have in some fields had a poor track record of success in translating from research to clinical trials in drug development, in terms of studying
Page | 42 developmental biology and fundamental biological processes such as the brain FBR, it is likely that animal models will be highly informative. The key differences lie in successfully modeling the pathology. For example, in the case of AD, a major cause of the poor translation from successful research in animal models to clinical studies are the limitations in modeling of the Alzheimer’s pathology itself. Most Alzheimer’s research is conducted on transgenic mice that produce amyloid plaques and the resultant research aims to mitigate the production of amyloid plaques177. The translation of successful studies of animal models unfortunately fall short during clinical trials because the animal models only mirror a limited aspect of Alzheimer’s multifaceted nature, which we do not completely understand. In the case of the brain FBR, this biological process occurs as a natural response to sustained injuries without the need for a transgenic model as wild type animal models can inherently develop a brain FBR to an injury. From the study of brain development to pathways crucial to animal physiology, these fundamental cellular responses are highly conserved amongst mammals, suggesting that they will be valuable animal models for the study of the human brain FBR178,179.
5.5. Current Polymer Materials for Compliant Implant Fabrication and Challenges
In recent years, polymers have been found to be a promising class of materials for the fabrication of brain implants due to their biocompatibility, material stability, mechanical compliance, and insulative capability. The development and advancement of polymer-based technologies for
MEMS (microelectromechanical systems) was a key enabler for the innovation of a variety of intracortical implants that are significantly more compliant than standard silicon-based implants.
Polyimide, Parylene-C, and SU-8, well documented materials used for MEMs devices, have also emerged as the predominant materials used to
Page | 43
Figure 15 | Graph illustrating the different Young’s modulus of various materials used for brain implants compared to brain tissue. fabricate compliant brain implants180. Implants created from these materials are flexible allowing them to reduce the strain induced upon brain tissue arising from micromotion, tethering, and other relative displacements of brain tissue to the implant. Nevertheless, these polymer materials are still orders of magnitude stiffer than brain tissue (Figure 15). Hydrogel-based implants have been developed in recent years, but designs have been limited to thin-films consequently limiting their applications to, for example, the stimulation of the sciatic nerve and not as brain implants181,182.
Due to their flexibility and a reduction of the elicited brain FBR, brain implants with a stiffness below 10 GPa have widely been accepted as compliant134,183. Consequently, compliant implants cover a large range of stiffness that requires further categorization to reflect their relative stiffness compared to brain tissue. For this purpose, we propose to assign the term soft implant to any compliant implant with a Young’s modulus below 1 MPa. It is noteworthy to point out that to this date there has yet to be an intracortical implant that can be categorized as soft, with the exception of a silicone microwire fabricated by Du et al. (2017)169 at 974 kPa and a hydrogel-elastomer coaxial cable introduced by Sheng et al. (2019)184 at 16.56 kPa. Although
Page | 44 technically impressive, these implants act as single recording elements and are limited in their application as intracortical implants due to their large size and lack of spatial resolution.
5.5.1. Polyimide
Polyimide was amongst the first polymers used specifically for the fabrication of compliant brain implants154. Originally, it was commonly used as an insulation material in electronics, but was later used to fabricate brain implants when it was recognized for its
potential biocompatibility, high mechanical
strength, and chemical stability185,186. The first
chronically implantable intracortical electrode
array based on polyimide was introduced by
Rousche et al. (2001) and was fabricated using
standard photolithography techniques. The
intended use of polyimide was to provide strain
relief against the forces of micromotion between
the tissue and implanted device. In vivo data from Figure 16 | Process flow of a polyimide implant: (a) Physical vapor deposition of rats with the implant located in the barrel cortex sacrificial aluminum (Al) layer on silicon substrate (b) Spincoating of polyimide onto 154 of Al and baking at 300 °C (c) Spincoating of achieved a good signal to noise ratio of 5:1 . photoresist onto polyimide for patterning of electrode traces (d) Deposition of gold Other studies with regards to tissue compatibility conductive layer (e) Spincoating polyimide to insulate conductive layer and baking at 350 °C (f,g) Deposition and patterning of Al onto of polyimide reported that compliant polyimide polyimide as mask for final etching (h) Etching of unwanted polyimide via reactive implants induced minimal tissue reaction187 and ion etching (i,j) Al mask and sacrificial Al layer removed via etching10. Reproduced from ref.10 with permission from elicited significantly lower activation of microglia IOP Publishing, copyright 2014. and reactive astrocytes in vivo compared to silicon
Page | 45 implants of similar dimensions and shape170. It has been reported, however, that polyimide is prone to swelling and swells by 4-6% due to moisture uptake upon implantation134. The swelling of polyimide is linked to a decrease in electrode performance stemming from possible interference with the electrical circuit or electrode insulation.
Fabrication of polyimide-based implants
typically involves standard photolithographic
procedures beginning with the spincoating of
polyimide onto a substrate such as silicon. After
spincoating, the polyimide undergoes a soft bake
process followed by UV exposure10,188. The final
structure of the polyimide substrate is realized
after development and curing at temperatures
between 300 – 500 °C. Due to the high
temperatures required to cure polyimide, the Figure 17 | Process flow of a Parylene C implant: (a) SiO2 growth (b) SiO2 patterning (c) Metal evaporation and patterning (d) curing process is a limiting factor to the selection DRIE to form trench/mold (e) Deposition of Parylene C (f) Opening Parylene C window of materials used to fabricate a polyimide implant. (g) XeF2 silicon isotropic etching (h) Parylene C deposition (i) Exposure of electrodes and contact pads (j) DRIE to Electrode traces can be patterned onto the release Parylene C implant from silicon wafer6. Reproduced from ref.6 with permission from polyimide substrate with various deposition IOP Publishing, copyright 2017. techniques such as electron beam physical vapor deposition10. In order to obtain freestanding and flexible substrates, a sacrificial layer is usually deposited onto the silicon substrate before the spincoating of polyimide where the structures
Page | 46 can be released via processes such as conventional wet etching or anodic metal dissolution.
Polyimide has been noted to have poor adhesion to certain materials and has been reported to delaminate from metals such as platinum, aluminum, and copper which can undermine the functionality and reliability of implants in a chronic setting189.
5.5.2. Parylene C
Parylene C is a specific type of poly(p-xylylene), and is another polymer commonly used to fabricate compliant implants. It is known for its biological inertness190 and met rigorous guidelines of USP Class VI and ISO 10993 for biocompatibility190. Although it has been reported to be more fragile and less robust in comparison to polyimide, its ability to be processed using deposition techniques at room temperature is more advantageous when considering connection and assembling processes of brain implants180. Additionally, Parylene C exhibits little optical scattering and high transmittance in the visible spectrum allowing it to be used to fabricate implants with a need for optical transparency190. Examples of compliant Parylene C intracortical implants include one based on a tube structure that allows for optogenetic stimulation6 and a sheath structure probe allowing for the ingrowth of neural processes for improved tissue integration191. With respect to silicon implants, in one study, Parylene C implants resulted in a
12-17% neuronal loss surrounding an inserted implant192 compared to a loss of 40% resulting from silicon implants indicating better mechanical compatibility88.
Fabrication of Parylene C based implants has primarily entailed chemical vapor deposition of Parylene C onto a substrate or into a mold6,191,193. Electrical interconnects can then be patterned onto the Parylene C by methods such as lift off. Typically, Parylene C can be released from substrates and molds with manual peeling191 but can also utilize sacrificial release layers
Page | 47 when necessary. However, Parylene C’s relatively low glass transition temperature between 35 to 90 °C194 limits compatible processes in fabricating Parylene C based brain implants due to an induced bulk change in material properties when the glass transition temperature is surpassed.
5.5.3. SU-8
SU-8, a commonly used negative photoresist in microfabrication, has also been used to create compliant intracortical implants in recent years. It is known for its chemical, thermal, and mechanical stability180. Similar to Parylene C, SU-8 is also highly transparent for wavelengths greater than 400 nm180. However, the biocompatibility of SU-8 is questionable with some studies claiming good biocompatibility195 while others such as Verkear et al. (2009) reporting that <10% of primary neurons survived when cultured with untreated SU-8. Although cell viability improved to 86% with subsequent heat treatment of SU-8 and isopropanol sonication, XPS analysis revealed the presence of fluorine and antimony which are potentially neurotoxic elements196.
Despite the inconclusive biocompatibility of SU-8, it has been used to fabricate brain implants due to its flexibility and available processing techniques. Nevertheless, Huang et al. (2014) fabricated a compliant SU-8 implant and reported that the SU-8 implant induced less glial aggregation compared to rigid silicon implants and was capable of achieving a SNR greater than
7197. This suggests that with appropriate treatment of SU-8, possible toxic leachates can be removed. Additionally, Yang et al. (2019) took advantage of the processing techniques available for SU-8 and fabricated sub-micron thick SU-8 probes with dimensions comparable to neurites.
The mechanical stiffness of the implant scaled with the thickness the SU-8 implants allowing it to form intimate coupling with the surrounding neurons and was reported to illicit a negligible immune response compared to other implants62.
Page | 48
Typical fabrication processes of SU-8
implants entail conventional
photolithographic processes. It is generally
spin coated onto a substrate such as silicon or
glass, followed by a baking step and then a UV
exposure process9,197. Electrical traces can
then be deposited onto the developed SU-8
Figure 18 | Process flow of an SU-8 implant: substrate via techniques such as sputtering or (i) Sacrificial Al deposition onto silicon via sputtering (ii) Spin coating and baking of SU-8 thermal evaporation197. Similar to polyimide, (iii) UV exposure of SU-8 and sputtering of chromium and gold conductive layers (iv) Spin coating and pattering of positive photoresist to sacrificial layers using aluminum or titanium is define electrodes (v) Wet etching to pattern metals (vi) Removal of positive photoresist are required between the SU-8 layer and the (vii) Deposition of SU-8 passivation layer (viii) Release of device from silicon substrate9. Reproduced from ref.9 with permission from underlying substrate for the release of the IOP Publishing, copyright 2010. completed device. A fabrication challenge of
SU-8 is the induced stress between SU-8 and the substrate during fabrication that can lead to shape distortions and fractures. Stresses in SU-8 films can be reduced through optimization of the process steps but are difficult to completely resolve which can ultimately compromise the performance of the implant.
5.5.4. Other Polymers
Other polymers used to fabricate compliant implants include liquid crystal polymers (LCP)198, polyvinyl acetate (PVAc)199, and OSTE170. Although implants made of these materials all have shown markedly decreased Young’s moduli compared to that of silicon, they are still orders of magnitude stiffer than brain tissue. PDMS and other silicone variants200,201 and various
Page | 49 hydrogels181,202 that are much softer than the aforementioned polymers are more similar to brain tissue stiffness thus can be classified as soft materials. However, devices made from these soft materials are largely restricted to surface electrode arrays and peripheral nerve cuffs and not intracortical implants. The only known functional implant fabricated from silicone is from Du et al. (2017) where a syringe was used to extrude a PEDOT-PEG and PDMS conductive microwire that was subsequently dipped into fluorosilicone and had a measured Young’s modulus of 974 kPa169. However, the fabrication technique can only produce wires with a single recording site limiting the functional use of the device.
Given that polymers such as silicones are more closely matched with brain tissue in terms of stiffness and are known for their biocompatibility and various superior physical properties, it begs the question why has there yet to be an intracortical implant made from soft materials with similar stiffness to brain203. Many indications in the literature point to the difficulty in the fabrication and implantation of such intracortical implants (implantation challenge further discussed in Section 5.6). Fabrication methods for soft materials on the micro-scale face several challenges and are not standardized. Methods such as laser micro-machining or laser ablation of soft materials require expensive laser systems, are difficult to scale-up, and lack the high-fidelity features achievable with replica molding204-206. More economical methods such as 3D-printing soft materials lacks the ability to achieve high resolution micro features with current resolutions limited to ~100 µm207-209. Hence replica molding is still the most popular method for soft material fabrication as it is capable of achieving high resolution features but limited to thicker devices due to difficulties in releasing the molds from softer materials with thinner features210. Further difficulties arise when molding with micro-scale features where capillary forces dominate fluid,
Page | 50 inertial, and viscous forces resulting in increased difficulty generating high-fidelity replications.
Additionally, at the micro-scale, replica molding suffers from the formation of a residual membrane at the top surface preventing the realization of independent features when cured.
This is caused by the excess polymer that overflows the mold during the pouring process and is difficult to remove. Although many have proposed solutions such as cure inhibition211, wet- etching212-214, dry-etching99,215, or physical clamping of the mold to squeeze out the residual polymer216-218, these techniques often lead to sticky surfaces, entail the use of toxic chemicals, and are unreliable.
On the electrical side, the integration of electrodes and conductive traces with soft substrates is a non-trivial consideration. In order to have a successful soft brain implant, the conductive traces and electrodes must be able to accommodate the stretching of soft substrates without compromising the electrical integrity. This requires creative approaches such as designing traditional conductive materials like gold into fractal or serpentine shapes to achieve flexible interconnects for soft electrical devices20,219,220. In recent years, research in soft electronics has gained momentum and other works include using organic conductors such as
PEDOT:PSS221 that have a naturally lower Young’s modulus compared to traditional metal conductors or liquid metals222 for making the electrical components of soft electronics. However, the research and fabrication methods for soft electronics is still in its infancy. Hence, there is much to be investigated in choosing the optimal materials and fabrication methods to obtain good levels of performance, reliability, and integration of the different materials in soft electronics.
Page | 51
5.6. Current Insertion Methods for Compliant Intracortical Implants and Challenges
With the decrease in Young’s modulus, it becomes inevitable that compliant implants will face challenges in being able to penetrate the brain tissue during insertion. When the implant stiffness is too low, it fails to sustain the necessary penetration force to enter the brain tissue and buckles during insertion or deflects when inserted into the brain causing additional tissue damage and missing the intended target location4,193,223. Consequently, there has been a collective effort in the field to develop solutions to aid the insertion of compliant implants. Methods of assisting the insertion of compliant implants fall under two predominant categories—using dissolvable coatings that temporarily increases the implant’s stiffness or using rigid insertion shuttles as a temporary backbone.
5.6.1. Dissolvable Coatings
Dissolvable coatings that have been used to aid the insertion of compliant implants include silk5,193, tyrosine-derived polymers224, sugar10,225,226, and non-cytoxic carboxymethylcellulose
(CMC)63. A successful dissolvable coating provides enough structural integrity for the implant to penetrate the brain and dissolves within a reasonable amount of time once inserted into the brain. Furthermore, the dissolved coating should not elicit an immune response from the surrounding tissue and should be a smooth and uniform coating to avoid further damage to the tissue during insertion.
In this respect, most of these materials meet the criteria of a successful coating. For example, silk has a tunable dissolution rate and dissolves via protease degradation whereby the by-products do not elicit an immune response227. Furthermore, silk is easily sterilized due to it being unaffected by autoclave temperatures. Similarly, tyrosine-derived polymers and sugars
Page | 52 also dissolve quickly after implantation into the brain and are metabolized. CMC, however becomes a gel once it enters the brain and does not completely degrade restricting the proximity of the implant to the neurons of interest63,228 making it a less advantageous material compared to silk, sugar, and tyrosine-derived polymers.
Although all these materials have demonstrated sufficient structural support to aid in a successful insertion of compliant implants, they have yet to be proven for practical use with soft implants. Furthermore, a common weakness shared by many of these materials is the method that is used to coat them onto an implant—dip-coating. Dip coating is difficult to scale and lacks reproducibility due to the need to control for concentration of the dipping solution and withdrawal speeds10. Furthermore, the geometry of the coatings is limited and softer materials are difficult to dip coat at the micro-scale due to the relative viscous nature of the coating solutions. Other methods such as layer-by-layer casting in a mold (Figure
19)5 are labor intensive requiring manual application of each silk layer onto the implant with a wait time of 15 min between each layer. Figure 19 | Flexible polyimide implant coated with silk using layer-by- layer casting to aid in the insertion process5 Additionally, this method Reproduced from ref.5 with permission from John Wiley and Sons, copyright 2013. cannot be used with
Page | 53 materials like sugar for coating which requires ambient temperatures well above 100 °C during the coating process. Hence, it is of interest to innovate a new method of coating that offers control over coating dimensions, is easily scalable, and can be extended to a wide variety of materials for aiding the implantation of soft materials.
5.6.2. Insertion Shuttles
Rigid insertion shuttles are another commonly used method to deliver compliant implants into the brain4,65,67,229,230. Often shuttles are attached to the backsides of a compliant implant. The greatest challenge in using shuttles is successfully coupling the implants with the shuttles and decoupling the two after implantation. Due to the need to decouple the shuttles after implantation, mechanical coupling at the micro-scale is difficult and unpractical. Hence most methods of coupling use biodissolvable adhesives such as PEG169,229 or electrostatic interactions4 between the shuttle and the implant. However, there exists limitations with both methods as biodissolvable adhesives do not interact in the same way with all materials restricting the material selection for the fabrication of the compliant implant. On the other hand, coupling based on electrostatic interaction can be extremely unreliable. Having a successful electrostatic coupling requires fine tuning the Figure 20 | Flexible polyimide implant coupled with a surface modified silicon shuttle using self-assembled monolayer coating of a hydrophilic compound. Reproduced from ref.4 with permission from Elsevier, copyright 2009. surface characteristics of both the implant and shuttle complicating the fabrication process.
Page | 54
Furthermore, there is a need to ensure that the surface of the shuttle and implant retains their surface characteristics at the time of implantation to achieve the desired coupling and uncoupli6ng mechanism. Sterilization of electrostatic coupled devices becomes limited and complex. Additionally, soft materials such as silicone tend to stick to the surfaces of shuttles due to hydrophobic interactions4 and can be carried out of the brain along with the shuttle when it is removed despite surface modifications.
Persistent challenges in inserting compliant implants successfully into the brain and in the fabrication of soft materials require additional effort and novel techniques to overcome shortcomings of existing methods. Without advances in these areas, the implementation of soft implants cannot come to fruition, reflecting the current state of the field, with an absence of soft implants despite an evident need.
5.7 Summary
Given the evident medical benefits and clinical need for brain implants, there still exists obstacles that prevent the widespread adoption of brain implants in clinical settings. Issues of long-term reliability and the challenge of obtaining high-quality recordings from brain implants has been linked to the elicited brain FBR that is exacerbated by the substantial mismatch in stiffness between current implants and brain tissue. Our work is motivated by existing challenges in current fabrication and implantation methods that prevent the realization of a truly soft brain implant. We aim to create the first soft brain implant with a stiffness similar to mammalian brain tissue. Our approach to creating the softest sub-millimetre intracortical implant to date, and our investigation of the brain response to the soft implant, is presented in the following chapter.
Page | 55
6. Silicone Implants with Brain-like Stiffness Delivered Using Micro-Molded Dissolvable Sugar Shuttles Reduce the Brain Foreign Body Response
Edward N. Zhang1,2, Jean-Pierre Clément3*, Alia Alameri1,2*, Andy Ng1,2, Timothy E. Kennedy3,
David Juncker1,2,3
1Department of Biomedical Engineering, McGill University, Montreal, Quebec, Canada
2McGill University and Genome Quebec Innovation Centre, Montreal, Quebec, Canada
3Department of Neurology and Neurosurgery, McGill University, Montreal, Quebec, Canada
*These authors contributed equally to this work
6.1. Abstract
Brain implants have significant implications for the treatment of neurological disorders and diseases. The clinical application of brain implants, however, is limited by the challenge to obtain high-quality recordings limitations to long-term reliability caused by the brain foreign body response (FBR). Critically, the FBR is exacerbated by the Young’s modulus (E) mismatch between current implants (150 - 200 GPa) and brain tissue (.4 - 15 kPa). Current research to improve the mechanical compatibility of brain implants has given rise to compliant implants made from materials like polyimide (4 - 8 GPa) that are flexible but are still orders of magnitude stiffer than brain. Here, we present novel fabrication and implantation methods that enable the realization of the softest sub-millimetre intracortical implants to date, composed of Ecoflex, a silicone elastomer with E = 20 kPa. Implants were fabricated using sacrificial sugar molds coupled with vacuum-assisted molding (VAM). Ecoflex implants were then encased inside micro-molded dissolvable sugar shuttles as temporary structural supports for reliable and accurate implantation. We then compared the brain FBR evoked by Ecoflex implants to PDMS and silicon
Page | 56 implants 3-week post implantation in rats. Immunohistochemical analysis revealed lower brain
FBR elicited by Ecoflex implants compared to silicon and PDMS. These findings support the conclusions that a functional soft implant made from materials closer to brain stiffness such as
Ecoflex could potentially provide superior high-quality recordings and long-term reliability by reducing brain FBR.
6.2 Introduction
Brain implants have significant implications for the treatment of neurological disorders and diseases. Challenges, however, in long-term reliability and in the maintenance of high-quality recordings of brain implants are well documented in literature17,80,231 and have prevented the widespread adoption of brain implants as viable medical treatments. The brain foreign body response (FBR) elicited by current brain implants is a primary culprit in the decline of implant functionality due to the formation of dense glial scars that isolate implants from the neurons of interest85,90,92 and promote neuronal death surrounding the implant63,94,97. The substantial mismatch in Young’s modulus (E) between current brain implants made from materials such as silicon (150 - 200 GPa) and brain tissue (0.4 – 15 kPa) results in chronic tissue damage and an exacerbated brain FBR owing to tissue strain at the tissue-implant interface15,99 and the intrinsic interplay between implant stiffness and the response of glial and neuronal cell232-234. Research directed towards improving the mechanical compatibility of brain implants has focused on increasing implant flexibility to reduce tissue strain by fabricating implants from more compliant materials like Parylene C191,193,229 and polyimide10,154. With multiple studies indicating that more compliant implants reduce the brain FBR, it is noteworthy that there has yet to be a brain implant developed with a Young’s modulus similar to that of the brain. Possible explanations for this
Page | 57 include difficulty in manipulating soft materials during fabrication and major challenges inserting soft implants into the brain.
Fabrication of soft materials on the micro-scale faces several challenges. Methods such as laser micro-machining204,205 and 3D-printing207,208 require expensive infrastructure, are difficult to scale, and lack the high-fidelity features achievable with conventional replica molding.
Standard replica molding methods, however, are limited to stiffer materials and thicker constructs as thinner and softer materials are susceptible to damage from the release process210.
At the micro-scale, replica molding also suffers from the formation of residual membranes arising from small amounts of excess polymer that overflows the mold during the mold filling process.
These membranes interfere with the generation of independent features and overcoming this challenge has inspired solutions such as cure inhibition of the excess polymer211, physically clamping the mold to squeeze out excess polymer216,218, and wet213,214 or dry-etching 99,215 of the residual membranes. Shortcomings of these solutions, however, include rough or sticky device surfaces, the use of toxic chemicals, and unreliable results.
Furthermore, current compliant implants either buckle before penetrating the brain or deflect once inside the brain4,193,223. This poses an even greater challenge for softer implants close to the brain stiffness. Methods for inserting compliant implants into the brain fall under two predominant categories—using dissolvable coatings as structural aids or rigid insertion shuttles as temporary backbones. Dissolvable coatings that have been used include tyrosine-derived polymers224, sugar10, and carboxymethylcellulose63. However, there is a lack of reproducibility and control in coating dimensions and the coating methods are often laborious and difficult to
Page | 58 scale. With insertion shuttles, the main challenge is the reliable coupling and decoupling of the implant from the shuttle during the implantation process4.
In order to overcome these challenges, we present a robust and reliable method of fabrication and implantation of the softest sub-millimetre intracortical implant to date. The implant is fabricated out of Ecoflex, a biocompatible platinum-cured silicone elastomer235 with E
= 20 kPa. The brain FBR of Ecoflex implants were compared to those elicited by PDMS (1.6 MPa) and silicon implants (180 GPa) to study the effects of stiffness on the brain FBR. The Ecoflex and
PDMS implants were fabricated using sacrificial sugar molds coupled with vacuum-assisted molding (VAM). Standard DRIE was used to fabricate the silicon implants. Ecoflex, PDMS, and silicon implants that were 300 µm in width, 200 µm in thickness, and 2.5 mm long and were delivered into the cortex of rats by encasing them in dissolvable sugar shuttles (720 µm in width,
430 µm in thickness, and 8 mm long) that dissolve within minutes after insertion.
Immunohistochemical analysis of brain slices of rats sacrificed at 3-week time post-implantation was used to assess the brain FBR elicited by the different implants.
6.3. Results and Discussion
6.3.1 Fabrication of Sacrificial Sugar Molds
The realization of Ecoflex implants that can be inserted into the brain required three distinct fabrication processes: fabrication of the sacrificial sugar molds, VAM of the Ecoflex implants in the sugar molds, and encasing the Ecoflex implants in micro-molded dissolvable sugar shuttles using VAM. Sacrificial sugar molds were used to fabricate the Ecoflex implants to avoid damaging the soft implants during the release process caused by conventional molds. Figure 21a illustrates the fabrication process for the sugar molds. For the process to fabricate the SU-8 master mold
Page | 59 see supplementary Figure S1. A SU-8 master mold was created using conventional photolithography to obtain high resolution implant features 200 μm in thickness and 300 μm in width, see Figure 21b. MD700, an oligo-urethane methacrylate containing a bifunctional perfluoropolyether chain, was used to create a negative mold from the SU-8 master mold (Figure S4). MD700 was chosen for its low surface energy of 12-15 mN/m236,237 that allowed for easy release from the SU-8 master mold without the need for surface treatments of the master mold. Additionally, the low surface energy of the MD700 mold allowed the release of the
Figure 21. Fabrication process of sacrificial sugar molds. a-i) UV curing of MD700 prepolymer on top of positive SU-8 master mold to create negative MD700 mold. a-ii) Heat curing of PDMS prepolymer on top of MD700 mold to create positive PDMS mold. a-iii) Removal of PDMS mold from solidified negative sugar mold. b) Microscope image of positive SU-8 master mold and negative sacrificial sugar mold. Scale bars: 2 mm, 300 μm (insets).
subsequent positive PDMS mold without silanization. The PDMS mold was then used to create the negative sugar mold. PDMS was chosen because of its capacity to withstand high process temperatures238 when fabricating the sacrificial sugar molds and because it does not stick to sugar owning to its low surface energy of 20 – 30 mN/m239.
Page | 60
To fabricate the sugar molds, a 1:2:3 mixture of light corn syrup to water to table sugar was microwaved to 180°C resulting in a caramelized sugar mixture with a dark brown color210.
Sugar mixtures typically reach the hard crack stage, the point where once cooled becomes hard and glass-like, between 146 - 154°C240. As a result of curing the Ecoflex prepolymer inside the molds at 60°C to fabricate Ecoflex implants, the sacrificial sugar molds required a glass transition temperature (Tg) higher than 60°C. Raising the Tg of the sugar mold beyond 60°C was done by cooking the sugar mixture to a higher temperature, 180°C. By extending the caramelization process, more monosaccharide molecules such as glucose in the sugar mixture become
241 polymerized, raising the resultant Tg of the sugar mold . The molten sugar was then poured onto the PDMS mold and degassed in a vacuum oven at 240°C corresponding to a sugar temperature of 150°C, to remove bubbles from the sugar mixture that would compromise the features of the sugar mold. After removing the bubbles, the sugar mixture was left to cool at room temperature (RT). The PDMS mold was released from the sugar mold after the sugar solidified, as illustrated in Figure 21b. A comparison between the SU-8 master mold and the sacrificial sugar mold reveals high-fidelity replication of the SU-8 mold features. We found that the sugar molds left at RT with a relative humidity of 40% began to degrade within an hour due to the hydroscopic nature of sugar. To preserve the sugar molds, we stored them in a desiccator with CaCl2. The sugar molds can be well preserved for over a month.
Sugar is highly advantageous compared to other materials in fabricating sacrificial molds.
Its relatively high Young’s modulus of 100 MPa allow for replication of high fidelity features210.
Additionally, sugar is cheap, highly soluble in water242, and non-toxic. Although we released
Ecoflex implants with a thickness of 200 μm from the sacrificial sugar molds, in principal, the
Page | 61 utility of sugar molds can be extended to materials softer than Ecoflex, at thinner dimensions, and with more complex designs.
6.3.2. Fabrication of Ecoflex Implants
VAM, a process adapted from our lab’s previous work in fabricating freestanding polymer microfilter membranes243, was used to infill both Ecoflex and PDMS prepolymers into the sacrificial sugar molds to overcome the residual membranes that may occur in conventional replica molding, shown in supplementary Figure S5. Figure 22a illustrates the VAM process for fabricating Ecoflex implants. For PDMS implants see supplementary Figure S3.
Figure 22. Fabrication process of Ecoflex implants. a-i) Ecoflex prepolymer pipetted onto sugar mold inlet with mold sealed using heat release tape (HRT) for vacuum assisted molding (VAM). a-ii) Venting desiccator after degassing infills Ecoflex prepolymer into mold where prepolymer is cured. a-iii) Removing HRT and dissolving sugar mold releases Ecoflex implants. All Ecoflex, implants were 300 µm in width, 200 µm in thickness and 2.5 mm long cut to size using a razor blade b) Degassing (left) of mold with air leaving through Ecoflex prepolymer indicated by bubbles forming in prepolymer. Inset shows magnified view of bubbles. Venting of desiccator begins after no more bubbles form in prepolymer resulting in infilling within seconds (right). c) Microscope image of Ecoflex implant (right) with SEM image of surface roughness (left) showing a relatively pristine surface. Scale bars: 300 μm, 25 μm (inset).
To fabricate Ecoflex implants using VAM, heat release tape (HRT) was used to seal the sugar mold leaving only the inlet of the mold exposed. After the sugar mold was sealed with HRT
Page | 62 and the Ecoflex prepolymer was loaded into the PDMS ring surrounding the mold inlet, the whole mold was degassed in a vacuum desiccator. Given that there is no outlet, air removed from the negative mold features exits through the prepolymer indicated by the formation of bubbles shown in Figure 22b. Consequently, the absence of bubble formation in the prepolymer indicates that the desiccator can be vented for infilling. When the desiccator is vented, the ambient pressure inside the mold returns to the atmospheric pressure causing the Ecoflex prepolymer to infill the mold. An excess amount of prepolymer was loaded into the PDMS ring to ensure that no air enters the mold during infilling which can cause defects in the molded Ecoflex implants.
Degassing the mold takes approximately 20 min and infilling finishes in less than 30 seconds as seen in supplementary Video 1. Immersing the sugar molds in water released the cured Ecoflex within minutes. The Ecoflex implants float in the water bath and can be picked up using a microscope slide avoiding distortion and damage to the implant. The resulting implants were 5 mm long, 300 µm wide, and 200 µm thick and were manually cut into two 2.5 mm long sections using a razer blade (Figure 22a-iii) for implantation into the cortex of rats. With VAM, we have also successfully fabricated 300 µm wide, 200 µm thick, and 25 mm long implants demonstrating the robustness of the method for fabricating soft implants at longer dimensions.
Because Ecoflex has a pot life of 30 min, whereby its viscosity doubles, we initially observed difficulties completely infilling the molds given the amount of time required for degassing. To help keep the viscosity of the prepolymer low, Ecoflex was mixed with a 10% silicone thinner and cooled in an ice container after mixing before VAM. Figure 22c shows the high-fidelity replication of the Ecoflex implant from the SU-8 master mold. The SEM image
(VEGA3 TESCAN, 10 kV, 10.17 WD) of the Ecoflex implant indicates that the only roughness on
Page | 63 the implant surface arises from wrinkles formed by the Ecoflex during curing. These wrinkles arise as a means for the Ecoflex to minimize the system’s potential energy to account for the compressive stress induced during the in-plane compression resulting from the Ecoflex and sugar mold strain mismatch244-246. Hence, VAM coupled with the sacrificial sugar mold is introduced as an effective and robust method of molding soft materials. VAM also enables the fabrication of more complex designs at smaller dimensions, although the dimensions and design must account for the viscosity and pot life of the polymer.
6.3.3. Characterization of Ecoflex and PDMS Young’s Modulus
Table 2. Young’s modulus (E) values of Ecoflex, PDMS, silicon, and brain tissue.
Material E No Additional Heat Treatment E with Additional Heat Treatment Ecoflex 00-20 20.36 ± .67 kPa 20.21 ± .38 kPa PDMS (Sylgard184) 1.07 ± .02 MPa 1.60 ± .05 MPa Silicon 179 GPa247 -- Brain Tissue 1 – 10 kPa248,249 --
The stiffness of a brain implants is primarily represented by its Young’s modulus. To characterize the Young’ modulus of both Ecoflex and PDMS implants, we performed tensile testing on specimens of both materials fabricated according to ASTM D412 Type C specifications for vulcanized rubber and thermoplastic elastomers (Figure S10). Since the implants were autoclaved for surgeries and were embedded in dissolvable sugar shuttles at high temperatures, we investigated the effects of these subsequent high temperature processes on the Young’s modulus of the implants. Figures 23a and 23b show the stress-strain curves acquired from the tensile tests of Ecoflex and PDMS specimens with and without the subsequent processes. Both
Ecoflex and PDMS exhibit typical stress-strain curves of hyperelastic materials with three distinct regions depicted as A, B, and C where at strain rates within region A, the silicone elastomers are
Page | 64 relatively stiff compared to strain rates in region B, followed by an increase in stiffness at strain rates in region C. Region A, the initial linear elastic segment of the stress-strain curve at low strains where Hooke’s law is valid250 was used to calculate the Young’s modulus of both PDMS and Ecoflex using Equation 1 where E is the Young’s modulus, σ is the applied stress and ε is the resultant strain.
Equation 1.
휎 퐸 = 휀
The Young’s modulus measured at region A representing 35% strain and 10% strain for
Ecoflex and PDMS respectively is shown in Table 1 which indicates that Ecoflex has a larger strain range for elastic deformation. Enlarged stress-strain curves of region A for PDMS and Ecoflex can be seen in supplementary Figure S11. Since the implants are untethered to the skull after insertion into the brain, the only source of strain that is experienced by the implants is from brain tissue micromotions that results in tissue displacements of 2 – 30 μm in rats251. Given this relatively small displacement compared to the dimensions of our implants, the Young’s modulus obtained from region A of the stress-strain curves serves as appropriate characterizations of the stiffness of our implants. Ecoflex implants after the subsequent processes yielded a Young’s modulus of 20.21 ± .38 kPa which is not significantly different to the Young’s modulus from the specimens that did not undergo subsequent processes of 20.36 ± .67 kPa (p-value = .672).
Page | 65
Figure 23. Comparison of Ecoflex and PDMS stress-strain curves from tensile tests. a) Ecoflex stress-strain curves with (blue, n = 6) and without (red, n = 6) additional heat treatment show no significant changes in Ecoflex stiffness (p-value = .672). A, B, and C identifies regions of the stress-strain curves that show a distinct change in material stiffness with respect to strain. Linear elastic segments in region A were used to calculate the Young’s modulus of Ecoflex with (dark blue, r2 = .9661) and without (dark red, r2 = .9519) additional heat treatment. b) PDMS stress-strain curves with (blue, n = 6) and without (red, n = 6) additional heat treatment show significant changes in PDMS stiffness (p-value <.001). Respective linear elastic segments used to calculate Young’s modulus shown in dark blue (r2 = .9963) and dark red (r2 =.9961).
We initially considered conventional PDMS (SYLGARD 184) for fabricating the brain implants due to the biocompatibility, ease of fabrication, and thermal and chemical stability of silicones203. However, our tensile test measurements of PDMS yielded a Young’s modulus of 1.07
± .02 MPa that increased by over 160% after subsequent processes to 1.60 ± .05 MPa, a value that is orders of magnitude higher than the Young’s modulus of brain tissue, which is ~1 - 10 kPa
(Table 2). We next considered a softer PDMS variant (SYLGARD 527) with a reported Young’s modulus of 5 kPa252 (Table 2). However, the tacky nature and excess of silicone oil present in the soft PDMS were undesirable for fabrication. Ecoflex 00-20, a platinum-catalyzed and biocompatible silicone235, was chosen as the final material for the fabrication of the soft brain implant. As an addition-cure silicone, Ecoflex can be molded without any shrinkage203. Compared to PDMS, Ecoflex has a significantly larger fracture strain (supplementary Table S1) of over 950%,
Page | 66 compared to PDMS at ~200%, and is less susceptible to aging effects experienced by PDMS because of its better water resistivity253. These reasons suggested that Ecoflex would be superior to PDMS for the fabrication of brain implants to be used in a chronic setting.
6.3.4. Encasing of Implants in Dissolvable Sugar Shuttles and Implantation into Rats
As mentioned before, one of the biggest challenges for compliant implants is the ability to reliably deliver them into the brain given their inability to penetrate brain tissue. To overcome this challenge, we used VAM to encase Ecoflex, PDMS, and silicon implants into dissolvable sugar shuttles with the encasing process shown in Figure 24a. PDMS molds were used for the VAM of Time = 10 s Time = 1 min 30 s the dissolvable sugar shuttles and were replicated from a positive 3D-printed master mold with a thickness of 450 μm and a width 700 μm. The dimensions of the sugar shuttles were optimized such that they had sufficient structural integrity to reliably delivery the soft implants into the brain without breaking during insertion. The implants were then embedded into the PDMS mold and sealed with a 1:20 PDMS slab as seen in Figure 24b. The lower stiffness PDMS slab allowed for superior conformal contact with the stiffer PDMS mold which was necessary for a good seal during VAM. A thin layer of powdered sugar was coated onto the PDMS mold to help prevent
Ecoflex implants from sticking onto the PDMS mold. We observed that without the powdered sugar layer, Ecoflex implants stuck onto the PDMS mold when attempting to release the sugar shuttles from the mold.
The same mixture ratio of sugar used to make the sacrificial sugar molds was used to make the dissolvable sugar shuttles. However, in this case the sugar mixture was microwaved to a
Page | 67
Figure 24. Encapsulation process of implants in dissolvable sugar shuttles. a-i) Replication of PDMS negative mold from 3D printed positive master mold. a-ii) Insertion of implants into PDMS mold and sealing mold with PDMS slab for VAM. a-iii) Infilling of molten sugar to encase implants using VAM. b) VAM setup with two Ecoflex, PDMS, and silicon implants embedded into the PDMS mold sealed with a PDMS sealing slab. c) Microscope image of dissolvable sugar shuttle with Ecoflex implant encased. Scale bar: 1 mm. d) Insertion of sugar shuttle with Ecoflex implant encased into rat brain showing insertion and dissolving of the sugar shank.
lower temperature of 140 °C. The lower temperature was used to avoid burning the sugar given that the sugar mixture used to fabricate the dissolvable sugar shuttles was placed inside the vacuum oven for VAM at 240 °C for 20 min, which was considerably longer than the time the sugar mixture stayed in the vacuum oven for the fabrication of the sacrificial sugar molds. The vacuum oven temperature of 240 °C continued to heat the sugar at around 150 °C keeping the sugar mixture molten allowing for the air inside the PDMS mold to exit through the sugar.
Once sugar filled the mold and solidified, the sacrificial sugar shuttles with the implants embedded can be easily released from the PDMS mold using a tweezer. We observed that the
Page | 68
PDMS mold must be sufficiently thin so that it can be allowed to flex (below 3 mm thick). The flexibility of the PDMS mold allowed for a more reliable release of the sugar shuttles. If the PDMS mold is too stiff, then releasing the sugar shuttles with tweezers induces stress on the shuttles causing them to break, especially at the location where the implants are encased due to less sugar reinforcement. Figure 24c shows an Ecoflex implant encased in the sugar shuttle. Long term cultures of the sugar shuttles with encased implants in culture media without antibiotics showed no signs of contamination confirming the sterility of the fabrication process. Like the sugar mold, the dissolvable sugar shuttles can be kept in a vacuum desiccator with CaCl2 desiccant to prevent the sugar shuttle melting. The shuttles were stable when stored this way for over a month.
Figure 24d shows the insertion of
an Ecoflex implant encased in sugar into a
rat brain where the sugar dissolves quickly,
taking a little more than 1 min, and the
tissue displaced initially by the sugar
shuttle can be seen to refill the displaced
area as the sugar shuttle dissolves. Sugar
shuttles were designed to have an opening
angle of 17° to minimalize tissue
dimpling96. Although the 3D printer
resolution tip resulted in a flat shuttle tip Figure 25. MRI of rat brain showing successful delivery of Ecoflex implant without implant deflection of distortion. Frontal view (top) and sagittal view of around 200 μm in width, dimpling of the (bottom).
tissue during insertion was minimal, as
Page | 69 seen in the supplementary Video 2, given the small opening angle. Figure 25 reveals the successful delivery of an Ecoflex implant into the rat brain, visualized using MRI, indicating no deflection or distortion of the implant after insertion. Our method of encasing in sugar exhibits good reproducibility and control of coating dimensions, previously not achieved by coating the implant with dissolvable materials to aid insertion. Due to the high temperature required to keep the sugar mixture liquid, materials for the fabrication of soft implants will be limited to those that withstand the high temperatures. However, the VAM method can be used to coat soft implants using other dissolvable coatings, which potentially circumvents the material limitation for the fabrication of soft implants.
6.3.5. In Vivo Assessment of Brain Foreign Body Response
To directly compare the brain FBR elicited by Ecoflex, PDMS, and silicon implants, immunohistochemical studies were done on brain slices from rats sacrificed 3-weeks post implantation. Specifically, assessment of the brain FBR was based on the relative levels of expression of Iba-1, CD-68, GFAP, and IgG. Iba-1 is a cytoplasmic protein that is distributed relatively uniformly in the cytoplasm and processes of microglia. Detection of Iba-1 protein allowed us to assess the density and distribution of microglia. Detection of CD-68, a transmembrane protein with increased expression in activated microglia, helped provide a distinction between activated and ramified microglia254. GFAP is a major intermediate filament protein expressed by astrocytes that is upregulated in reactive astrocytes. It allows us to assess the distribution and density of reactive astrocytes255. Detecting the presence of IgG in brain tissue, the most abundant antibody isotype circulating in blood serum, indicates a breach in the
Page | 70 blood-brain-barrier (BBB). Assessment of the level of IgG in brain parenchymal helped us evaluate the extent of the BBB breach caused by the three types of implants.
Most of the implant holes left by Ecoflex, PDMS, and silicon implants clearly reflected the rectangular cross-section and dimensions of the inserted implants (Figure 26). However, some tissues had irregularities in the implant holes and ones with significant irregularities or tissue folds were not used for data analysis, see supplementary Figure S12. The irregularities of the implant holes manifested either as a deviation from the normal rectangular shape or in a significant discrepancy of the expected dimensions. In total, one Ecoflex, two PDMS, and one silicon samples were removed from data analysis for CD-68. Two Ecoflex and one silicon samples were removed from data analysis for IgG. These irregularities in the implant holes could have resulted from processes such as tissue mounting after staining, folds and tears introduced during cryosectioning, and explant of implants from the tissue sample.
Figure 26 shows a representative panel of the morphology of the implant tracks and expression levels of CD68, Iba-1, and GFAP from Ecoflex, PDMS, and silicon implants at 3-weeks post implantation. These images suggest significantly lower brain FBR elicited by either Ecoflex or PDMS implants compared to silicon implants. The amount of IgG, however, was not significantly different between the three different implants (Figure S14). It is also interesting to note that for sugar shams, implantation sites showed almost no inflammation 3-weeks post implantation. In fact, the implant hole initially caused by the relatively large dissolvable ~430 µm thick and ~720 µm wide sugar shuttles appear to have been completely repopulated by the initially displaced tissue. This finding is congruent with previous reports256, further supporting the conclusion that immediate insults to the brain, or stab wounds, are not the major driving force
Page | 71 for the development of the chronic brain FBR. Hence, despite the relatively large footprint of the dissolvable sugar shuttles, they do not appear to have a major impact on the brain FBR elicited by the Ecoflex, PDMS, and silicon implants.
Figure 26. CD-68, Iba-1, and GFAP expressions elicited by Ecoflex, PDMS, and silicon implants as well as sugar shams 3-weeks post implantation. (animal n = 5). Scale bar: 200 µm.
Page | 72
Figure 27 shows the normalized intensity curves of the levels of CD-68, Iba-1, GFAP, and
IgG immunoreactivity elicited by Ecoflex, PDMS, and silicon implants. For CD-68, there was a significantly lower relative intensity expression level elicited by both Ecoflex and PDMS implants compared to silicon implants (p-value < .05). However, between Ecoflex and PDMS implants, there was no statistically significant difference in the relative intensities of CD-68 detected.
Nevertheless, Ecoflex implants showed a 7 % decrease in the mean relative intensity levels of CD-
68 compared to PDMS implants for binned intensities 50 µm away from the implant site.
Similarly, although there were no significant differences detected in the relative intensity of GFAP between PDMS and Ecoflex implants, Ecoflex implants showed a 20% decrease in the relative intensity levels compared to PDMS implants at the same binning distance. Furthermore, there was no significant difference in the relative intensities of GFAP between PDMS and silicon implants but there was a significant difference between Ecoflex and silicon implants. The same observations were made for levels of Iba-1 where there were no significant differences between the relative intensities of Ecoflex and PDMS implants, but a significant difference was detected between Ecoflex and silicon implants and no significant difference between PDMS and silicon implants. Additionally, Ecoflex implants showed a 10% decrease in the relative intensity levels of
Iba-1 compared to PDMS implants for binned intensities 50 µm away from the implant site. With respect to IgG, there was no significant difference in the relative intensities elicited by all three implant types. Given that the implants were untethered to the skull, we didn’t expect there to be a persistent BBB breach stemming from the shear stress caused by the small displacements of brain tissue relative to the implants due to brain micromotion. This is supported by the relatively low normalized intensity levels of IgG expression for all implant types.
Page | 73
Although there was no statistical difference in the relative intensity levels between PDMS and Ecoflex implants for GFAP and Iba-1 immunoreactivities, Ecoflex consistently demonstrated lower average relative intensity levels compared to PDMS at binned intensities 50 µm away from the implant which was the region that had the most brain FBR activity. Furthermore, Ecoflex showed significantly lower levels of Iba-1 and GFAP compared to silicon
Figure 27. Immunoreactivity for CD-68, Iba-1, GFAP, and IgG presented as normalized intensity curves elicited by Ecoflex, PDMS, and silicon implants 3-weeks post implantation (animal n = 5). Lower relative intensity values signify less brain FBR. At least two tissue sections were imaged for each target expression per rat. Trendlines added for better visualization. Error bars represented the standard error. implants when PDMS implants failed to do so. Because the implants were untethered, the elicited brain FBR of all three implant types most likely stems from the inherent interplay between the glial cells and the implant stiffness as opposed to the induced tissue strain that causes tissue damage in tethered studies. The findings provide evidence in the absence of significant tissue
Page | 74 strain arising from the displacement of brain tissue relative to the implant, stiff implants still induced a significantly more intense brain FBR.
All rats in the study developed superficial infections around the suture site around one- week post implantation and were successfully treated with antibiotics with the infections subsiding within one to two days. The superficial infections did not affect the FBR elicited as none of the rats developed meningitis and implant sites were isolated from the superficial infection by healthy tissue. A major contributor to the statistical non-significance of the relative intensity levels PDMS and Ecoflex implants is the relatively large variability of the levels of the immunoreactivity detected amongst different tissue samples, indicated by the large standard error bars in Figure 27. Variability can arise from a multitude of factors such as whether there was a major blood vessel around the implantation site, the staining process, and variations in the depth of cortical tissue analyzed. Mediating this issue will require a larger sample size to acquire a more robust dataset less susceptible to the influence of these variabilities.
6.4. Conclusion
In this study, we successfully fabricated the softest sub-millimetre intracortical implant to date at 20 kPa made from Ecoflex, a platinum-catalyzed silicone elastomer. In order to overcome current fabrication challenges in the field that prevented the realization of soft implants with a
Young’s modulus closer to that of brain tissue, we introduced a novel method of fabrication coupling sacrificial sugar molds with VAM. This allowed the replication of soft implants with micro-scale dimensions without damaging the implant features during the release process.
Furthermore, with VAM, we were able to avoid the issue of residual membranes that prevent the
Page | 75 creation of independent implant features. The technique of coupling VAM with sacrificial sugar molds can be extended to other soft materials as well as other designs.
We also introduce a new method of coating compliant implants with materials to temporarily provide the structural integrity necessary for the implants to be inserted into the brain. In this case, sugar was used to coat the compliant implants using VAM, with Ecoflex, PDMS, and silicon implants embedded in dissolvable sugar shuttles. Our method of coating is simple, scalable, and produced reproducible results that enabled superior control of the dimensions and shapes of the coatings not possible with previous methods. It was shown in our study that that the dissolvable sugar shuttles, despite having a relatively large footprint, elicited a very small brain FBR 3-weeks post implantation. The initial implant hole caused during the surgery was repopulated by the displaced brain tissue and the implant tracks of sugar shams were almost identical to the surrounding normal tissue. Our findings demonstrate that the soft Ecoflex implants reduced levels of activated microglia and reactive astrocytes compared to PDMS and silicon implants suggesting that compliant implants closer to the brain’s stiffness reduces the brain FBR. This provides a novel contribution to the literature, supporting the conclusion that soft implants below the stiffness of current compliant (4 – 8 GPa) implants does in fact further decrease the brain FBR. With the introduced methods of fabrication and implantation of the soft
Ecoflex implants, we provide initial steps towards realizing a truly soft and functional brain implant with the potent to improve the limited reliability and functionality that plagues current brain implants.
Page | 76
6.5. Experimental Section
6.5.1. Ecoflex Implant Fabrication
Mold Fabrication for Ecoflex Implants
For PDMS and silicon implant fabrication processes, see Figure S3 and S9 respectively. Ecoflex implants were fabricated using vacuum assisted molding (VAM) coupled with use of a sacrificial sugar mold. All molds used in this paper were replicated from a positive SU-8 master mold (Figure
S1). A MD700 (Fluorolink MD700, Solvay) negative mold was replicated from the SU-8 master mold by pouring a mixture of 2% by weight of a photo initiator (Darocur 1173, Sigma) with
MD700. The MD700 prepolymer was degassed on top of the SU-8 mold for 1 hr to remove any bubbles on the mold features and then cured using an UV flood curing system (IntelliRay 600,
UViTron International) for 2 mins at 180 mW/cm2. After curing, the MD700 mold was released from the SU-8 mold and sonicated in isopropanol (IPA) for 15 min to remove any uncured oligomers. To fabricate the sugar molds, PDMS (SYLGARD 184, DOW) 1:10 crosslinker to base was mixed using a planetary centrifugal mixer (AR-100, THINKY) at 2000 rpm for 90 s and degassed before pouring onto the MD700 molds. After pouring, the PDMS was cured for 24 hrs at 60°C and subsequently released from the MD700 mold and placed onto the bottom of a silicone container with the features facing upwards. A sugar mixture (1:2:3 corn syrup to water to table sugar) was poured onto the PDMS replica and heated in a microwave until the sugar mixture reached 180°C.
Afterwards, the container was quickly transferred to a vacuum oven (Lindberg/Blue M, Thermo
Scientific) and degassed at 240°C at 25 inHg for 5 min to remove bubbles that can compromise the sugar mold features. After venting the vacuum oven, the silicone container was removed from the oven and left to cool at room temperature (RT) until the caramelized sugar solidified.
Page | 77
Once solidified, the sugar was removed from the silicone container and the PDMS mold was released from the sugar block resulting in a negative sugar mold.
Fabrication of Ecoflex Implants
For VAM, the sugar mold was sealed with a heat release tape (HRT) (319Y-4LS, Nitto) leaving only the inlet of the mold exposed. A PDMS O-ring (2 mm thick, 9/16 in inner diameter) was punched from a cured PDMS (1:10) slab using a hollow punch and placed around the inlet of the mold to form a reservoir. 300 µl of chilled Ecoflex (Ecoflex 00-20, Smooth-On) (1:1) mixed at 2000 rpm for 90 s with 10% by weight of silicone thinner (Silicone Thinner, Smooth-On) and .001% green silicone pigment (Silc Pig, Smooth-On) was pipetted into the reservoir using a viscous pipette and the mold was placed into a vacuum desiccator and degassed for 20 min. Afterwards, the mold was removed from the desiccator to allow the infilling of the Ecoflex prepolymer. After the
Ecoflex cured (60 °C for 1 hr), the HRT was removed from the sugar mold and the Ecoflex implants were released from the sugar mold by dissolving the mold in a water bath.
6.5.2. Characterization of Ecoflex and PDMS Young’s Modulus
Ecoflex and PDMS Young’s modulus was acquired via tensile testing (5965 Series Universal
Testing System, Instron). All Ecoflex in this paper was mixed 1:1 of part A to part B with an added silicone thinner that was 10% of the total weight of the Ecoflex prepolymer using a planetary centrifugal mixer at 2000 rpm for 90 s for both mixing and degassing spins. All PDMS in this paper was mixed 1:10 of curing agent to base unless otherwise noted. PDMS was mixed with the same parameters as Ecoflex. The Ecoflex or PDMS prepolymer was poured onto an aluminum mold fabricated according to the American Society for Testing of Materials (ASTM) D412 Type C specifications (Figure S10) to make the test specimens. Bubbles formed from pouring were
Page | 78 removed with a pipette and the Ecoflex prepolymer was cured at 60°C for 1 hr and PDMS prepolymer was cured at 60°C for 24 hr. An additional experimental condition in which the cured
Ecoflex or PDMS specimen underwent a dry autoclave cycle for 45 min at 121 °C and a subsequent heat treatment in a vacuum oven for 20 min at 240 °C was conducted to investigate the effects of the sterilization and sugar encapsulation processes of the Ecoflex or PDMS implants on their
Young’s modulus. The thickness of each test specimen was measured using a micrometer (P/N
TL268, Proster) before being mounted onto the tensile machine. A 10 N load cell was used for
Ecoflex and a 1 kN load cell was used for PDMS. The crosshead velocity was set at 3 mm/s. Test specimens were extended until failure or until the tensile machine reached its extension limit.
Test specimens that failed at locations other than the test section were discarded and remade for testing. In total 6 test specimens were used to acquire the Young’s modulus of each experimental condition. All test data was obtained using Bluehill Universal.
6.5.3. Insertion of Implants
Sugar Encasing of Ecoflex, PDMS, Silicon Implants
All implants were encased in hard dissolvable sugar shuttles for insertion into the brain. To encase the implants in sugar, VAM was used. A positive 3D-printed master mold was fabricated using a DLP printer (Miicraft 100, Miicraft) and a photocurable resin (BV-002A, Miicraft). Features of the mold were 450 μm thick, 700 μm wide, and 8 mm long. 50 μm layer thickness was used at an exposure of 0.8 s per layer. The 3D printed mold was washed with IPA in a sonicator for 10 min and then dried with a nitrogen gun and post cured in a UV flood curing system for 2 mins at
180 mW/cm2. The molds were left in the oven at 60°C for 24 hr to evaporate any uncured resin that may inhibit the curing of PDMS. PDMS prepolymer (1:10) was poured onto the post-treated
Page | 79
3D-printed molds and then cured at 60°C for 24 hr and subsequently released from the 3D printed mold. Ecoflex and PDMS implants were cut into 2.5 mm long pieces using a razor blade. A layer of powdered sugar was deposited onto the PDMS mold to prevent Ecoflex implants from sticking to the PDMS mold after encasing. All implants were placed into the PDMS mold using tweezers and positioned close to the tip sections of the sugar shuttles. The PDMS mold was sealed with a
1:20 PDMS sealing slab (2mm thick) that had inlet cut out using a hole puncher (3/8 in inner diameter). The mold with the embedded implants was then dry autoclaved for 45 min at 121°C.
A 1:2:3 mixture of corn syrup to water to table sugar was heated in a silicone container using a microwave until the sugar temperature reached 140°C. The autoclaved PDMS mold was placed into the molten sugar with the inlet facing downwards and quickly placed into a vacuum oven and degassed at 240°C at 25 inHg for 20 min. Once the mold was taken out of the vacuum oven, sugar immediately infilled the mold. The mold was quickly removed from the molten sugar mixture and left to solidify at RT in a biosafety cabinet. Once solidified, the implants encased in sugar were released from the PDMS mold under sterile conditions using a tweezer and placed in a vacuum desiccator with CaCl2 until use.
Surgical Implantation of Implants
All surgical procedures were done in accordance with the Canadian Council on Animal Care guidelines for the use of animals in research and were approved by the Montreal Neurological
Institute Animal Care Committee. Stereotaxic surgery was performed on 6 female rats (Sprague
Dawley Rat, Charles River) between 10 and 12 weeks of age weighing between 275 and 300 g to evaluate the in vivo brain FBR elicited by the Ecoflex, PDMS, and silicon implants. Animals were anesthetized with 3% isoflurane in 0.8 L/min oxygen prior to surgery and maintained during the
Page | 80 procedure at 2 to 2.5% isoflurane via a facemask. Ophthalmic ointment was applied to both eyes to prevent dryness and damage to the rat cornea. Afterwards, 10 mg/kg body weight of carprofen
(RIMADYL) and 0.5 mL/10 g body weight of 0.9% sterile saline was administered subcutaneously.
A pedal reflex test was performed before the beginning of each surgery and anesthesia level was monitored closely during the surgery by observing the breathing rate of the rats. Rats were placed on a stereotaxic (Model 940, KOPF) and the hair on top of the heads shaved. Rats were then covered with sterile surgical drapes and the craniums exposed with a scalpel blade. Four 1.5 mm holes were drilled into the rat skulls at coordinates obtained from a rat brain atlas where the implants were to be delivered (Figure S13). The holes were drilled using a handheld drill
(Foredom Micromotor 1070, Foredom) until a thin layer of skull was left. The remaining skull was removed manually using a 25G syringe that was bent into a hook shape at the tip to avoid accidently drilling into the brain tissue. The coordinates were chosen to avoid major ventricular systems. Each rat had an Ecoflex, PDMS, silicon, and sugar sham implant sufficiently spread out such that there was no overlap in the evoked brain FBR. All implants were mounted on an electrode holder (Model 1770, KOPF) and inserted until the encased implant can be seen to have entered the brain tissue at which point the implant was left until the sugar shuttles dissolve
(supplementary Video 2). Sterile PBS was used to promote the remaining sugar to dissolve and the incision wound of the rats were sutured after surgery. Six rats were used to evaluate the brain
FBR at the 3-week time period. Rats with superficial wound infections were treated with subcutaneous injections (10 mg/kg body weight) of enrofloxacin (Baytril) and wounds were cleaned daily for five days from the onset of the infection.
Page | 81
6.5.4. Analysis of Brain Foreign Body Response
Histology
Rats sacrificed at the designated timepoints were anesthetized with intraperitoneal injection of
Avertin (240 mg/kg body weight) and then transcardially perfused with ice cold PBS followed by
4% (w/v) paraformaldehyde (PFA) in PBS. Silicon implants were carefully removed from the tissues and PDMS and Ecoflex implants were left in the brain tissue during processing and sectioning. Brains were removed from rats and cryoprotected in 30% sucrose prior to freezing.
Sections were cut using a cryostat at 20 µm thicknesses and placed on microscope slides. For immunohistochemistry, two groups of co-staining were performed, and the respective concentrations and groupings can be seen in Table 2. At least four tissue sections from each rat at locations 400 - 700 µm deep were used for each co-stain group and at least two regions per rat were analyzed. Tissues sections were hydrated in 1X PBS and subsequently incubated in blocking solution (0.3% Triton X-100, 10% fetal bovine serum (FBS), 3% bovine serum albumin
(BSA) in 1X PBS) for 2 hrs at RT. Afterwards, blocking solution was removed and primary antibodies in new blocking solution added to the tissue sections and incubated at 4 °C overnight.
After primary antibody incubation, the tissue sections were washed with 1X PBS three times for
10 mins each. Secondary antibodies were diluted in new blocking solution and added to the tissue and incubated for 2 hrs at RT. After secondary incubation, tissues were washed with 1X PBS for
15 mins followed by staining with Hoechst dye in 1X PBS for 15 min. Two additional 15 min washes in 1X PBS were performed before the tissues were air dried and mounted with mounting medium (Fluoro-Gel, Electron Microscopy Sciences).
Page | 82
Table 3. Table of antibodies used for immunohistochemistry brain FBR analysis.
Antibody Product Code/Company Target Concentration/Speci Secondary Antibody es Iba-11 (019-19741/Wako) Microglia 1:500/Rabbit Alexa Fluor 488/A21206 GFAP1 (AB5541/Millipore) Astrocytes 1:500/Chicken Alexa Fluor 647/A21449 CD681 (MCA341GA/Bio-Rad) Activated Microglia 1:50/Mouse Alexa Fluor 555/A31570 NeuN2 (MAB377/Millipore) Neuronal Nuclei 1:200/Mouse Alexa Fluor 488/A21202 IgG2 -- IgG protein -- Alexa Fluor 546/A11081 * All co-stains were done with Hoechst (1:500) and all secondary antibody concentrations were 1:500
Tissue Analysis
Confocal fluorescent microscopy was used to asses the brain FBR elicited by Ecoflex, PDMS, and silicon implants. For each antibody target, images were acquired using the same laser power and detector settings to reduce variability. Laser power was adjusted such that there was less than
10% oversaturation in the pixels surrounding the implantation sites. Images were centered on the implant site and multi-channel images acquired simultaneously with a 20X objective. Each image stitched was composed of a 4 x 4 tile so that it was possible incorporate tissue up to 1 mm away from the implant site. Analysis was done using the MATLAB script I.N.T.E.N.S.I.T.Y v2.063.
Tissue 700 – 900 µm from the implantation site were used as control for the analysis of the tissues and used to calculate the background noise intensity threshold. To prevent tissue holes from affecting the mean fluoresce intensity of the tissue area used to calculate the background noise intensity threshold, the mean intensity of the control area was first acquired and any pixel one standard deviation below the mean intensity value was considered a hole and removed from the calculation for the background noise intensity threshold. The mean intensity of the remaining pixels was acquired, and the background noise intensity threshold set to one standard deviation below the mean intensity. The MATLAB script was used to determine the center of an implant track and a rectangular region of interest with dimensions outlining the implant track was
Page | 83 determined from measurements done in ImageJ. Concentric rectangular bins for every 10 µm away from the edge of the implant track for 400 µm (Figure S15) were created and the mean pixel intensity of pixels above the noise threshold of every bin was normalized against the background generating the intensity profiles as a function of distance for each antibody target.
The normalized intensity values were calculated using Equation 2 where AvgI>T is the mean intensity of all pixels above the noise threshold in each bin and AvgN is the mean noise floor intensity. Hence, it is expected that the intensity curves do not asymptote to 1 unless there is no staining signal in the corresponding data. Data was averaged for each implant type and reported as mean ± standard error. Statistics performed to determine significance of the differences in elicited brain FBR expression were evaluated for a region 50 µm from the edge of the implantation site using single factor ANOVA test.
퐴푣푔퐼 푁표푟푚푎푙푖푧푒푑 퐼푛푡푒푛푠푖푡푦 = >푇 퐴푣푔푁 Equation 3.
6.6 Supplementary Data
6.6.1. SU-8 Mold Fabrication
For process flow see Figure S1. A 4 in prime grade silicon wafer was thoroughly rinsed with acetone, isopropyl alcohol (IPA), and distilled water to remove organic residues. Following the rising step, the wafer was dried using a nitrogen gun and placed on a hotplate at 150°C to dehydrate for 10 min. After dehydration the wafer is centered onto a spin coater (WS-400, Laurell
Technologies) and 5 ml of SU-8 (SU-8 2050, MicroChem) was poured onto the wafer in a spiral formation carefully to avoid bubbles. The SU-8 was spun at 500 rpm for 5 s and then 1630 rpm for 30 s to achieve a thickness of 100 μm. Edge bead removal (EBR) was performed manually
Page | 84 using a cotton swab dipped in SU-8 developer (MicroChem) and the wafer was soft baked at 65°C for 5 min and 95°C for 10 min and then cooled to room temperature (RT) on the hotplate. A second 100 μm layer of SU-8 was coated with the same spinning parameters but with longer soft bake times to account for the insulation of the first layer (65°C for 10 min and 95°C for 20 min).
A 20k DPI transparency photomask (CAD/Art Services Inc.), Figure S2, was placed in proximity contact with the wafer at a separation distance of 100 μm and the wafer was exposed using a mask aligner (EVG620, EV Group) at 400 mJ/cm2 to crosslink. After exposure, the wafer was left to rest for 10 min before post exposure baking at 65°C for 10 min and 95°C for 26 min and then cooled to RT on the hotplate. The wafer was then placed into SU-8 developer to dissolve the unpolymerized SU-8 for 7 min or until no more residual unexposed resist can be seen. The fabricated SU-8 master mold was rinsed with fresh SU-8 developer followed by IPA and then dried with a nitrogen gun.
Page | 85
Figure S1. Fabrication process of SU-8 master mold.
Page | 86
Figure S2. Photomask design for SU-8 master mold with feature dimensions. Four implant variants were made.
6.6.2. PDMS Implant Fabrication
Negative MD700 molds were used for VAM of PDMS implants. The MD700 molds were prepared for VAM similar to the sacrificial sugar molds used for fabricating Ecoflex implants with the exception of the HRT used (REVALPHA 3198LS, Nitto). This tape was better at releasing the PDMS implants from the MD700 molds whereas the other HRT variant did not have the necessary
Page | 87 adhesion force to remove the PDMS implant from the MD700 mold reliably. The VAM process for PDMS implants was the same as VAM process of the Ecoflex implants. After the PDMS infilled into the mold, it was cured, and the HRT was removed the from mold with the PDMS implants attached. PDMS implants were released from the HRT by heating the HRT to 90 °C on a hotplate.
See Figure S3 for the fabrication process.
Figure S3. Fabrication process of PDMS implants.
MD-700 Mold
Figure S4. Microscope image of negative MD700 mold (left, contrast enhanced for better visualization) and MD700 mold setup for VAM (right). Scale bars: 2 mm, 300 μm (inset). Fabrication process of PDMS implants.
Page | 88
Ecoflex implant Residual membrane
Figure S5. Ecoflex implants with residual membrane issue (left) and PDMS implant without residual membrane (right). Scale bars: 300 μm.
6.6. 3. Silicon Implant Fabrication
Bosch etching was used to fabricate the silicon implants. A protective mask was created using photolithography. To create the mask, a 2 in prime grade silicon wafer with 200 µm thickness was cleaned with acetone and IPA followed by distilled water to remove organic residues. The wafer was dried using a nitrogen gun and baked on a hotplate at 150°C to dehydrate for 10 min.
Afterwards the wafer was spin-coated with SU-8 negative photoresist (SU-8 2015, MicroChem) at 500 rpm for 5 s and then 2000 rpm for 30 s using a spin coater (WS-400, Laurell Technologies) to achieve a thickness of 20 µm. EBR was performed using a Q-tip dipped in SU-8 developer
(MicroChem). Afterwards, the wafer was soft baked at 65°C for 2 min and 95°C for 4 min and then cooled to RT on the hotplate. A 10k DPI transparency photomask, Figure S6, (CAD/Art
Services Inc.) was placed in proximity contact with the photoresist and was exposed using a mask aligner (EVG620, EV Group) at 200 mJ/cm2 to crosslink. Afterwards, the wafer was left to rest for
10 min at RT before a post bake at 65°C for 2 min and 95°C for 5 min and then cooled to RT on the hotplate. The wafer was then placed in SU-8 developer to dissolve the unpolymerized SU-8 for 2 min. The wafer was rinsed with fresh SU-8 developer followed by IPA and distilled water and dried with a nitrogen gun.
Page | 89
Figure S6. Photomask design for SU-8 mask for silicon implant fabrication. All features were 300µm wide ranging from 1 to 4 mm in length in .5 mm increments.
After the mask was successfully created, the wafer was mounted onto a 6 in silicon carrier wafer with a mounting adhesive (Crystal Bond 555 HMP, Agar Scientific). Anisotropic Bosch etching was used with an 8:20 passivation to etching time. The polymer deposition step was as followed: 8 seconds, 65 sccm C4F8, 1 sccm SF6, ICP power 450 W, platen power (CCP) 10 W. The etching step was as followed: 20 seconds, 65 sccm SF6, 15 sccm C4F8, ICP power 450 W, platen power (CCP) 25 W. An etch rate of 700 nm/step (~1.4 µm/min) was achieved. After successful etching, the silicon carrier wafer was heated to 70°C causing the mounting adhesive to flow and releasing the etched silicon implants. Lastly the silicon implants were sonicated in IPA for 20 min to remove any adhesive residue. See Figure S7 for process flow.
Page | 90
Figure S9. Fabrication process of silicon implants.
Figure S10. Aluminum mold with ASTM D412 Type C specifications used to make Ecoflex (bottom left) and PDMS (bottom right) test specimen (left). Dimensions of test specimen (right).
Page | 91
Figure S11. Comparison of Ecoflex and PDMS stress-strain curves from tensile tests. left) Ecoflex stress-strain curves with (blue, n = 6) and without (red, n = 6) additional heat treatment show no significant changes in Ecoflex stiffness. right) PDMS stress-strain curves with (blue, n = 6) and without (red, n = 6) additional heat treatment show significant changes in PDMS stiffness.
Table S1. Fracture strains of PDMS and Ecoflex. Ecoflex fracture strain limit was beyond 950% but was limited by the range of the tensile machine.
Material Fracture Strain No Additional Heat Fracture Strain with Additional Heat Treatment Treatment Ecoflex 00-20 Tensile Limit (950%) Tensile Limit (950%) PDMS 233% 183%
Figure S12. Example of implant hole Figure S13. Coordinates of implantation irregularity. Samples such as these were not sites relative to bregma. 2.5 mm anterior to used for data analysis. Scale bar: 200 µm. bregma, 3 mm left and right. -.5 mm anterior to bregma 4 mm left and right.
Page | 92
Figure S14. IgG expressions elicited by Ecoflex (left), PDMS (middle), and silicon (right) implants 3-weeks post implantation. (animal n = 5). Scale bar: 200 µm.
Figure S15. 10 µm concentric rectangular bins of regions of interest around implant site for 400 µm.
Page | 93
7. Discussion
This section presents an extended discussion of the fabrication results from Chapter 6, which was a major novel component of this work, as well as the in vivo brain FBR results. The silicone implant fabrication process will be further discussed, followed by the insertion of the implants into the brain, and finally the in vivo results.
7.1 Fabrication of Ecoflex and PDMS Implants
7.1.1 Fabrication of Sacrificial Sugar Molds
The key advantage of sacrificial sugar molds is that they readily dissolve in water circumventing the need for manual release and result in a device without damage and distortions (Figure 22 C).
Although some soft materials such as PDMS can be released manually from molds made from low surface energy materials such as MD700 (Figure S4 and S5), softer materials like Ecoflex in the kPa range and below or thinner devices would require a release process without manual manipulation. We observed that the release of the Ecoflex implants happened within minutes once the sugar mold was immersed in water and long before the complete sugar mold dissolved.
Once a thin layer of the sugar mold dissolved, the disintegrated mold features were no longer in contact with the soft implants resulting in their release.
As mentioned previously in Section 6.3.2, sugar molds are highly advantageous compared to sacrificial molds made from other materials. They are cheap, highly soluble in water, non-toxic, and have a relatively high Young’s modulus of 100 MPa allowing for high fidelity replication.
Furthermore, sugar molds can be released easily from silicones such as PDMS making it amenable to soft lithography. In order to fabricate sugar molds, however, sugar mixtures must be caramelized at 180°C. The caramelization process creates bubbles due to the evaporation of
Page | 94 water content in the sugar mixture that results in defects in the sugar mold. Consequently, the caramelized sugar required degassing at high temperatures in a vacuum oven at 240°C that corresponded to a sugar temperature of 150°C to remove these bubbles. Another processing step that requires attention is the extent of caramelization of the sugar mixture which affects the Tg of the resultant sugar mold. Sugar molds that were not well caramelized and contained too much unpolymerized monosaccharide molecules such as glucose had low Tg. We observed that sugar mixtures initially cooked to 150°C, the temperature at which caramelization takes place, resulted in sugar molds that had features disintegrating when placed in a 60°C oven for 1 hr to cure the
Ecoflex implants. This was solved by cooking sugar mixtures at 180°C to create sugar molds with a higher Tg. Although the Tg of the sugar molds can be raised to a certain extent by changing the composition of the sugar mixture and cooking the sugar mixture at higher temperatures, molded materials that require high curing temperatures beyond the maximum achievable Tg of hard sugar would not be compatible with sacrificial sugar molds.
Another limitation of sugar molds is their hydroscopic nature. We observed that sugar molds left at RT with a relative humidity of 40% began to degrade within an hour. Although the integrity of the sugar molds can be preserved by placing the mold in a vacuum desiccator, the hydroscopic nature of the sugar molds required that the processing steps for the fabrication of implants take place in low-humidity environments. Hence, we opted for implant features that are 300 µm wide and 200 µm thick, larger compared to conventional planar implant designs of
150 µm wide and 50 µm thick, to ensure high fidelity replications from the sacrificial sugar molds.
Although we have not investigated what minimum feature size can be achieved with sugar molds, it can be assumed that smaller features would degrade faster. Theoretically, there would exist a
Page | 95 feature size where the degradation of the mold features would take place before the fabrication process completed even in a low-humidity environment. This also means that soft materials with water content would not be compatible with sacrificial sugar molds.
7.1.2 Vacuum Assisted Molding (VAM) of Soft Implants
VAM coupled with sacrificial sugar molds is a simple and robust method for fabricating soft micro- scale devices not possible with other methods of molding due to challenges in releasing the soft constructs from non-sacrificial molds discussed earlier and the formation of residual membranes in conventional molding techniques. The residual membrane issue arises when overpouring into molds result in excess prepolymer that cannot be completely removed from the mold surface and when polymerized creates final constructs that have a polymerized film connecting all the free-standing features (Figure S5). Overpouring prepolymer into micro-molds is unavoidable and dispensing viscous prepolymer into micro-molds using pipettes at the microliter range is inaccurate. Consequently, we present the method of VAM to circumvent the issue of having excess prepolymer on the surface of the molds.
Our sacrificial sugar molds are set up for VAM by using a heat release tape (HRT) seal covering the top of the mold (except for an inlet) which prevents the overflow of excess polymer.
We observed that mold surfaces must be clean and absent from surface imperfections such as bumps or dirt in order to generate a good seal between the HRT and the sugar mold which is important for preventing the overflow of excess prepolymer as well as the complete infilling of the prepolymer into the mold. It is important to note that we dispensed approximately 300 µl of
Ecoflex prepolymer to cover the mold inlet contained by a PDMS ring (Figure 22) despite the total volume of the mold only requiring 20 µl to fill. The extra amount of prepolymer was required to
Page | 96 ensure that the prepolymer contained within the PDMS ring continued to cover the inlet during the infilling process such that air does not enter the mold which would cause incomplete filling.
The dead-end design of our VAM system where there is only one inlet and no outlet hold several advantages to economical injection mold designs sealed with tape or materials like PDMS that require active injection and have both inlets and outlets to ensure the complete infilling of the prepolymer into the molds. For active injection, in designs where there are many individual features like dead end channels, in order to completely fill the mold features an outlet must be created for every dead-end channel which could be impossible for small and dense channel designs due to limitations in space between features and the amount of labor required. VAM can be particularly advantageous for applications where the mold designs terminate with an array of features that are tens of microns in size such as electrode array designs. In these cases, it is extremely difficult to fill the features using active injection due to the high injection pressure required which can delaminate the seals. VAM is also much more cost-effective compared to traditional injection molding. In addition to the cost-prohibitive nature, traditional injection molding also requires air-tight fitting of two mold halves that is incompatible with sacrificial sugar molds that break when introduced to a high amount of compressive stress.
The total VAM process for Ecoflex was relatively fast. On average it took around 20 min for the sealed sugar molds to be degassed in a desiccator before the bubbling of the prepolymer contained within the PDMS ring stopped indicating that no more air is exiting the mold through the prepolymer. At this point, the vacuum desiccator can be vented to begin the infilling process which completes within a minute. Once the infilling process is competed, the sealed mold is placed into an oven at 60°C for one hour to cure the Ecoflex implants which is subsequently
Page | 97 released by dissolving the sacrificial sugar molds. The total fabrication time from VAM to a completed Ecoflex implant took no more than 2 hrs. VAM, however, is not without its limitations.
For example, in order to completely fill a mold, all air molecules inside the mold features must be thoroughly degassed which can take upwards of 20 – 30 mins depending on the viscosity of prepolymer being infilled, the pump power being used, and to a certain extent the air permeability of the mold material. However, for materials such as Ecoflex that have a short pot life, 20 – 30 mins could result in the prepolymer becoming too viscous to infill. Possible solutions to this could be extending the pot life of the prepolymer through chilling or using thinning compounds and by using vacuum systems with a lower baseline pressure to reduce degassing times. Furthermore, softer and less viscous variants of silicones can be used with VAM to fabricate the implants as well as other materials such as hydrogels. Our fabrication method of soft implants is compatible with other material systems given that these materials exist in a liquid phase during the molding process and is not soluble in water during the release process.
7.2 VAM of Dissolvable Sugar Shuttles for Delivery of Soft Implants into the Brain
As previously mentioned, one of the biggest challenges for compliant implants is the successful delivery of the implants into the brain. This challenge was magnified in our work given that our
Ecoflex implants are much softer and harder to handle than current compliant implants.
Inspired by the success of VAM for the fabrication of Ecoflex and PDMS implants, we adapted the method to create micro-molded dissolvable sugar shuttles with the soft implants embedded inside. To the best of our knowledge, we are the first to adapt the VAM process to mold caramelized sugar.
Page | 98
Given the dimensions of our soft implants which were 300 µm wide 200 µm thick and
2.5 mm long, the dissolvable sugar shuttles needed to be larger than the embedded implants in order to maintain enough structural integrity to reliably deliver the implants into the brain.
Consequently, we chose sugar shuttle feature sizes that were a little over twice the size of the soft implants and achieved features that were 430 µm thick, 720 µm wide, and 8 mm long, which were replicated into a PDMS mold. The 8 mm length of the sugar shuttles allowed for easier handling during surgery and was the length necessary to be successfully mounted onto the implant holder of the stereotaxic surgical device with sufficient sugar shuttle length to reach 3 mm deep into the brain of rat subjects. The large surface area of the sugar shuttle not only ensured the structural strength to reliably penetrate the brain tissue but also ensured that the dissolution of the sugar shuttle was slow enough to allow for the implant to reach the target location. Currently, it takes around 2 min for our sugar shuttles to completely dissolve and the dissolution time can be modified by changing the size of the sugar shuttles.
Thin PDMS molds of 2 mm in thickness were used for the VAM of the sugar shuttles. It was important to have a thin PDMS mold that was flexible to improve the yield rate of the sugar shuttles. We observed that when releasing sugar shuttles from PDMS molds that were 4 mm thick, the yield rate was around 25% compared to around 85% for PDMS molds that were 2 mm thick. If the PDMS mold was too stiff, then releasing the sugar shuttles with tweezers induces stress on the shuttles causing them to break especially at the location where the implants are encased due to less sugar reinforcement. However, flexing of the PDMS mold allowed for the sugar shuttles to “pop” out without excessive use of tweezers. We also observed that it was necessary to precoat the PDMS molds with a layer of powered sugar. This
Page | 99 prevented the Ecoflex implants from sticking onto the PDMS mold. Without the initial layer of powdered sugar, embedded Ecoflex implants stuck to the PDMS mold after the sugar shuttles were released.
Unlike VAM of Ecoflex implants, VAM of the sugar shuttles occurred in a vacuum oven at
240°C which corresponds to a sugar temperature of 150°C. The sugar mixture used was the same as what was used to fabricate the sacrificial sugar molds and was microwaved to 150°C before being put into the vacuum oven with the PDMS mold. Typically, degassing the molds finished around 15 min but was generally degassed for 20 min for more reliable infilling as the vacuum oven pressure varied between 15 to 25 inHg. We observed that infilling the caramelized sugar completed faster than the infilling of Ecoflex, possibly due to the lower viscosity of the molten sugar compared to the Ecoflex prepolymer. However, because VAM of sugar shuttles require high temperatures, this limits the material selection for the fabrication of the compliant implant. Additionally, the use of sugar shuttles results in an increased footprint of the overall implant construct used during the implantation process. Although our histological study (Figure 26 D-P) showed that the increased footprint due to the sugar shuttles does not contribute significantly to the brain FBR, it is still beneficial to have a construct with a smaller footprint to prevent excessive tissue damage and bleeding during surgery. However, our developed method of embedding implants into shuttles is not limited to sugar as VAM can be used to embed implants in other materials such as silk which could circumvent the issues of a large shuttle footprint and limitations of the materials used to fabricate the compliant implant.
It is also important to note that our current fabrication process requires the manual insertion of the soft implants into the PDMS mold. This results in slightly varied embedded
Page | 100 locations of the implants within the sugar shuttles that affects the accuracy of the delivery of the implants into the brain. This could be addressed by designing reference structures within the PDMS mold guiding and ensuring more accurate placement of the soft implants.
Compared to other methods of encasing compliant implants in sugar constructs such as dip-coating, our method offers considerable control of shuttle dimensions and superior reproducibility. Furthermore, dip-coating soft implants is laborious and difficult to scale, and the viscosity of molten sugar prevents reliable dipping of the soft implants at the micron-scale into the sugar. Compared to insertion aids such as surface modified shanks, our sugar shuttle shows far superior reliability in successful insertion of compliant implants as it does not depend on complex coupling and decoupling mechanisms between the implants and the insertion aids.
It is important to consider the subsequent steps in fabricating functional implants and how connectors and tethers can be integrated with our method of embedding the soft implants into dissolvable sugar shuttles. There are two approaches to integrating functional implants with our sugar encasing method. We can embed the whole functional implant with the connecters and tethers into the mold and encase it inside the sugar shuttle and dissolve the sugar encasing completely after the successful implantation of the penetrating portion of the implant while releasing the connectors and tethers. With this method, as with the material limitation of the soft implants, connector and tether materials must be able to withstand the high temperatures during the VAM process. The other approach is to create a more modular implant where the penetrating portion of the implant is encased inside the sugar shuttle. Once successfully delivered into the brain and the sugar shuttle dissolves, the connectors and tethers can be attached to the backend of the soft implant that is protruding from the brain.
Page | 101
7.3 In Vivo Analysis of the Brain FBR to Ecoflex, PDMS, and Silicon Implants
Our findings support the notion that soft implants made from PDMS and Ecoflex both elicit a reduced brain FBR compared to stiff silicon implants. Given that the implants were untethered, our results indicate that even in the absence of tissue strain arising from tethering forces, softer materials demonstrate a reduced brain FBR. However, the difference in the elicited brain FBR between PDMS and Ecoflex implants was statistically insignificant. This is within our expectations as the difference between the Young’s modulus of silicon implants to PDMS and
Ecoflex implants was over 5 and 6 orders of magnitude respectively whereas the difference in
Young’s modulus between PDMS and Ecoflex implants is an ~80-fold increase. Nevertheless, it is important to point out for both GFAP and Iba-1 expression levels, Ecoflex demonstrated a statistically significant reduction of the relative intensity levels compared to silicon implants whereas PDMS did not as shown in Figure 27 A and C implying better compliancy of Ecoflex implants. Furthermore, Ecoflex implants consistently elicited a lower average relative binned intensity 50 µm away from the implantation site where the brain FBR was most intense compared to PDMS implants. This suggests that with a larger sample size to offset the uncertainty caused by the large variabilities, such as the presence or absence of vasculature near the implantation site, observed in our dataset, there could be a statistically significant trend between Ecoflex and PDMS implants with respect to a reduced brain FBR. It is also noteworthy to mention that despite an increased insertion footprint caused by the sugar shuttle, the acute injury did not significantly affect the tissue response at 3-week post implantation where the initial displaced tissues quickly repopulated the area of the sugar
Page | 102 shuttle after it dissolved (Figure 26 D-F) which is congruent with previous findings in literature that stab wounds from implants do not contribute to the chronic brain FBR.
8. Conclusion
In this work, we present a novel fabrication method that utilize sacrificial sugar molds coupled with VAM to overcome challenges in the current methods used to fabricate soft devices and successfully fabricated the softest sub-millimetre intracortical implant to date. The implants were
300 µm wide, 200 µm thick, and 2.5 mm long and were made from the silicone elastomer Ecoflex with E = 20 kPa which is over 5 orders of magnitude reduction in stiffness compared to current compliant implants made from materials like Polyimide. Additionally, we present the first instance where such a soft implant was successfully delivered into the brain by introducing a novel method of embedding the soft implants into micro-molded dissolvable sugar shuttles that were 430 µm thick, 720 µm wide, and 8 mm long fabricated using VAM. The use of the dissolvable sugar shuttles to deliver the implants demonstrated better reliability and success rates compared to other reported methods, such as the use of surface modified insertion shank, in delivering soft polymer implants.
Immunohistochemistry results of the brain FBR 3-week post-implantation in rat subjects support the notion that softer implants indeed elicited a reduced brain FBR compared to stiff implants, and in general Ecoflex implants (20 kPa) elicited a reduced brain FBR compared to
PDMS implants (1.6 MPa). This indicates that softer implants with stiffness similar to brain tissue could potentially improve the chronic functionality of brain implants in clinical setting by reducing the brain FBR.
Page | 103
8.1 Future Work
In this work we fabricated sacrificial sugar molds with conservative dimensions that ensured the successful and high-fidelity replication of our Ecoflex implants. As mentioned before, we have not investigated the minimum feature size achievable using sacrificial sugar molds given their hydroscopic nature. Hence, sugar molds with smaller dimensions should be designed and assessed for their use as sacrificial sugar molds. A study where the feature integrity of sugar molds with design features ranging from 10 µm to 300 µm in width and 10 µm to 200 µm in thickness with 10 µm increments should be assessed using a microscope, with the molds subjected to different humidity levels as well as temperatures for different durations of times.
This will provide a comprehensive understanding and assessment of the feature size limitations of sacrificial sugar molds.
In lieu of exploring the minimum implant size that can be achieved using sacrificial sugar molds coupled with VAM, it would be of interest to investigate the minimum dimensions of the sugar shuttles required to reliably deliver the soft implants into the brain. This investigation should be done after achieving the smallest implants from the sacrificial molds. Our current sugar shuttles are relatively large and were chosen to achieve high yield rates during the release from
PDMS molds. However, with smaller embedded implants, the sugar shuttles should also decrease in footprint. In this study, the dimensions of the sugar shuttle should first be anchored on the dissolution rate of sugar. This can be done by immersing sugar shuttles of various dimensions in water heated to body temperature and quantifying the reduction of the dimensions of the sugar shuttles at specific durations of time. This can tell us the minimum size of the sugar shuttles required to successfully deliver implants into the brain given different times required to insert
Page | 104 the implants. Afterwards, it would be necessary to investigate the minimum sugar shuttle dimensions achievable with an embedded soft implant such that the yield rate from releasing the shuttles from the PDMS mold is sufficiently high. The minimum dimensions from these studies can serve as the starting point in testing the structural strength of sugar shuttles with embedded soft implants for reliably penetrating brain tissue. A study of the buckling force versus the sugar shuttle dimensions would give us the optimized sugar shuttle dimensions for delivering the soft implants.
Our ultimate goal is to fabricate a functional soft implant with a stiffness approaching to that of brain tissue. Hence, it would be important to work towards modifying the current methods of fabricating non-functional soft implants to accommodate processes required to realize a functional soft implant. There are a variety of options that could be used for the electrically active components of the soft implant, from conductive polymers such as PEDOT:PSS to liquid metals such as eutectic gallium-indium. Since what we have created in this work is essentially the soft substrate and passivation layer of a soft intracortical implant, conductive traces could be added to the substrate via a multitude of methods from spin-coating to various deposition methods. An example of a feasible approach could be to redesig the SU-8 master mold to incorporate channels in the Ecoflex substrate. Liquid PEDOT:PSS could then be infilled into the substrate using VAM. Once PEDOT:PSS infills the channels of the Ecoflex substrate, the solvent could be evaporated leaving the conductive polymer traces in the channels of the Ecoflex substrate. A mask could then be aligned to the Ecoflex substrate and a thin insulation layer of
Ecoflex spin coated and cured onto the substrate leaving electrode contact points uncovered for interfacing with neurons. Similarly, conductive hydrogel traces can be patterned in a similar
Page | 105 manner to achieve a lower stiffness device. The development of additional methods would be required to adapt our current fabrication techniques to seamlessly integrate with the fabrication methods of the electrical components of a functional implant.
On the biological side, future work will aim to extend the scope of the brain FBR assessment to neuronal responses. For example, we can examine tenascin expression levels to assess the degree of axon fiber regeneration due to the different implants stiffness as material stiffness has been known to affect axonal growth. We can also investigate the effects of the implant stiffnesses on neuronal density surrounding the different implants by staining for NeuN.
Other aspects of neuronal behavior that would be interesting to study are the inhibitory verse excitatory responses of neurons surrounding implants by studying the expression levels of markers such as synaptophysin and gephyrin. Furthermore, in vivo studies should be conducted beyond the 3-week timepoint. Timepoints such as 9-weeks reflect the chronic brain FBR and would provide a better assessment of the chronic brain FBR compared to 3-weeks which reflect the acute brain FBR due to trauma and would be more indicative of the brain FBR experienced by chronic implants.
9. References 1 Foundation, C. D. R. Prevalence of paralysis in the United States. (2013). 2 Pekny, M. & Pekna, M. Astrocyte reactivity and reactive astrogliosis: costs and benefits. Physiological reviews 94, 1077-1098 (2014). 3 Kopell, B., Machado, A. & Butson, C. Stimulation technology in functional neurosurgery. Textbook of Stereotactic and Functional Neurosurgery, 1401-1425 (2009). 4 Kozai, T. D. Y. & Kipke, D. R. Insertion shuttle with carboxyl terminated self-assembled monolayer coatings for implanting flexible polymer neural probes in the brain. Journal of neuroscience methods 184, 199-205 (2009). 5 Tien, L. W. et al. Silk as a multifunctional biomaterial substrate for reduced glial scarring around brain‐penetrating electrodes. Advanced Functional Materials 23, 3185-3193 (2013). 6 Zhao, Z., Kim, E., Luo, H., Zhang, J. & Xu, Y. Flexible deep brain neural probes based on a parylene tube structure. Journal of Micromechanics and Microengineering 28, 015012 (2017).
Page | 106
7 Chestek, C. A. et al. Long-term stability of neural prosthetic control signals from silicon cortical arrays in rhesus macaque motor cortex. Journal of neural engineering 8, 045005 (2011). 8 VanGuilder, H. D. et al. Concurrent hippocampal induction of MHC II pathway components and glial activation with advanced aging is not correlated with cognitive impairment. Journal of neuroinflammation 8, 138 (2011). 9 Altuna, A. et al. SU-8-based microneedles for in vitro neural applications. Journal of Micromechanics and Microengineering 20, 064014 (2010). 10 Xiang, Z. et al. Ultra-thin flexible polyimide neural probe embedded in a dissolvable maltose- coated microneedle. Journal of Micromechanics and Microengineering 24, 065015 (2014). 11 Moshayedi, P. et al. The relationship between glial cell mechanosensitivity and foreign body reactions in the central nervous system. Biomaterials 35, 3919-3925, doi:https://doi.org/10.1016/j.biomaterials.2014.01.038 (2014). 12 Hebert, L. E., Scherr, P. A., Bienias, J. L., Bennett, D. A. & Evans, D. A. Alzheimer disease in the US population: prevalence estimates using the 2000 census. Archives of neurology 60, 1119-1122 (2003). 13 Bach, J. P., Ziegler, U., Deuschl, G., Dodel, R. & Doblhammer‐Reiter, G. Projected numbers of people with movement disorders in the years 2030 and 2050. Movement disorders 26, 2286- 2290 (2011). 14 Vetter, R. J., Williams, J. C., Hetke, J. F., Nunamaker, E. A. & Kipke, D. R. Chronic neural recording using silicon-substrate microelectrode arrays implanted in cerebral cortex. IEEE transactions on biomedical engineering 51, 896-904 (2004). 15 Subbaroyan, J., Martin, D. C. & Kipke, D. R. A finite-element model of the mechanical effects of implantable microelectrodes in the cerebral cortex. J Neural Eng 2, 103-113, doi:10.1088/1741- 2560/2/4/006 (2005). 16 Jackson, A. & Fetz, E. E. Compact movable microwire array for long-term chronic unit recording in cerebral cortex of primates. Journal of neurophysiology 98, 3109-3118 (2007). 17 Hochberg, L. R. et al. Neuronal ensemble control of prosthetic devices by a human with tetraplegia. Nature 442, 164 (2006). 18 Perge, J. A. et al. Intra-day signal instabilities affect decoding performance in an intracortical neural interface system. Journal of neural engineering 10, 036004 (2013). 19 Kozai, T. D. Y. The History and Horizons of Microscale Neural Interfaces. Micromachines (Basel) 9, 445, doi:10.3390/mi9090445 (2018). 20 Chen, R., Canales, A. & Anikeeva, P. Neural recording and modulation technologies. Nature Reviews Materials 2, 16093 (2017). 21 Fischl, B. & Dale, A. M. Measuring the thickness of the human cerebral cortex from magnetic resonance images. Proceedings of the National Academy of Sciences 97, 11050-11055 (2000). 22 Rossant, C. et al. Spike sorting for large, dense electrode arrays. Nature neuroscience 19, 634 (2016). 23 Nicolelis, M. A. Actions from thoughts. Nature 409, 403 (2001). 24 Rothschild, R. M. Neuroengineering tools/applications for bidirectional interfaces, brain– computer interfaces, and neuroprosthetic implants–a review of recent progress. Frontiers in neuroengineering 3, 112 (2010). 25 Sironi, V. A. Origin and evolution of deep brain stimulation. Front Integr Neurosci 5, 42-42, doi:10.3389/fnint.2011.00042 (2011). 26 Gardner, J. A history of deep brain stimulation: technological innovation and the role of clinical assessment tools. Social studies of science 43, 707-728 (2013).
Page | 107
27 Koller, W. C., Lyons, K. E., Wilkinson, S. B. & Pahwa, R. Efficacy of unilateral deep brain stimulation of the VIM nucleus of the thalamus for essential head tremor. Movement disorders: official journal of the Movement Disorder Society 14, 847-850 (1999). 28 Ondo, W., Jankovic, J., Schwartz, K., Almaguer, M. & Simpson, R. Unilateral thalamic deep brain stimulation for refractory essential tremor and Parkinson's disease tremor. Neurology 51, 1063- 1069 (1998). 29 Benabid, A. L., Chabardes, S., Mitrofanis, J. & Pollak, P. Deep brain stimulation of the subthalamic nucleus for the treatment of Parkinson's disease. The Lancet Neurology 8, 67-81 (2009). 30 Benabid, A. L. Deep brain stimulation for Parkinson’s disease. Current Opinion in Neurobiology 13, 696-706, doi:https://doi.org/10.1016/j.conb.2003.11.001 (2003). 31 Deuschl, G. et al. A randomized trial of deep-brain stimulation for Parkinson's disease. New England Journal of Medicine 355, 896-908 (2006). 32 Group, D.-B. S. f. P. s. D. S. Deep-brain stimulation of the subthalamic nucleus or the pars interna of the globus pallidus in Parkinson's disease. New England Journal of Medicine 345, 956-963 (2001). 33 Miocinovic, S., Somayajula, S., Chitnis, S. & Vitek, J. L. History, applications, and mechanisms of deep brain stimulation. JAMA neurology 70, 163-171, doi:10.1001/2013.jamaneurol.45 (2013). 34 Vidailhet, M. et al. Bilateral deep-brain stimulation of the globus pallidus in primary generalized dystonia. New England Journal of Medicine 352, 459-467 (2005). 35 Kupsch, A. et al. Pallidal deep-brain stimulation in primary generalized or segmental dystonia. New England Journal of Medicine 355, 1978-1990 (2006). 36 Lozano, A. M. et al. A phase II study of fornix deep brain stimulation in mild Alzheimer’s disease. Journal of Alzheimer's Disease 54, 777-787 (2016). 37 Laxton, A. W. et al. A phase I trial of deep brain stimulation of memory circuits in Alzheimer's disease. Annals of neurology 68, 521-534 (2010). 38 Houeto, J.-L. et al. Tourette’s syndrome and deep brain stimulation. Journal of Neurology, Neurosurgery & Psychiatry 76, 992-995 (2005). 39 Shahed, J., Poysky, J., Kenney, C., Simpson, R. & Jankovic, J. GPi deep brain stimulation for Tourette syndrome improves tics and psychiatric comorbidities. Neurology 68, 159-160 (2007). 40 Moro, E. et al. Bilateral globus pallidus stimulation for Huntington's disease. Annals of Neurology: Official Journal of the American Neurological Association and the Child Neurology Society 56, 290-294 (2004). 41 Kang, G. A., Heath, S., Rothlind, J. & Starr, P. A. Long-term follow-up of pallidal deep brain stimulation in two cases of Huntington's disease. Journal of Neurology, Neurosurgery & Psychiatry 82, 272-277 (2011). 42 Greenberg, B. D. et al. Three-year outcomes in deep brain stimulation for highly resistant obsessive–compulsive disorder. Neuropsychopharmacology 31, 2384 (2006). 43 Abelson, J. L. et al. Deep brain stimulation for refractory obsessive-compulsive disorder. Biological psychiatry 57, 510-516 (2005). 44 Mayberg, H. S. et al. Deep brain stimulation for treatment-resistant depression. Neuron 45, 651- 660 (2005). 45 Lozano, A. M. et al. Subcallosal cingulate gyrus deep brain stimulation for treatment-resistant depression. Biological psychiatry 64, 461-467 (2008). 46 Bittar, R. G. et al. Deep brain stimulation for pain relief: a meta-analysis. Journal of clinical neuroscience 12, 515-519 (2005). 47 Rasche, D., Rinaldi, P. C., Young, R. F. & Tronnier, V. M. Deep brain stimulation for the treatment of various chronic pain syndromes. Neurosurgical focus 21, 1-8 (2006).
Page | 108
48 Boon, P. et al. Deep brain stimulation in patients with refractory temporal lobe epilepsy. Epilepsia 48, 1551-1560 (2007). 49 Andrade, D. et al. Long-term follow-up of patients with thalamic deep brain stimulation for epilepsy. Neurology 66, 1571-1573 (2006). 50 Whiting, D. M. et al. Lateral hypothalamic area deep brain stimulation for refractory obesity: a pilot study with preliminary data on safety, body weight, and energy metabolism. Journal of neurosurgery 119, 56-63 (2013). 51 Hochberg, L. R. et al. Reach and grasp by people with tetraplegia using a neurally controlled robotic arm. Nature 485, 372 (2012). 52 Flesher, S. N. et al. Intracortical microstimulation of human somatosensory cortex. Science translational medicine 8, 361ra141-361ra141 (2016). 53 Gilja, V. et al. Clinical translation of a high-performance neural prosthesis. Nature medicine 21, 1142 (2015). 54 Blabe, C. H. et al. Assessment of brain–machine interfaces from the perspective of people with paralysis. Journal of neural engineering 12, 043002 (2015). 55 Jorgenson, L. A. et al. The BRAIN Initiative: developing technology to catalyse neuroscience discovery. Philosophical Transactions of the Royal Society B: Biological Sciences 370, 20140164 (2015). 56 Miranda, R. A. et al. DARPA-funded efforts in the development of novel brain–computer interface technologies. Journal of Neuroscience Methods 244, 52-67, doi:https://doi.org/10.1016/j.jneumeth.2014.07.019 (2015). 57 Hong, G. & Lieber, C. M. Novel electrode technologies for neural recordings. Nature Reviews Neuroscience, 1 (2019). 58 Amon, A. & Alesch, F. Systems for deep brain stimulation: review of technical features. Journal of Neural Transmission 124, 1083-1091 (2017). 59 Hickey, P. & Stacy, M. Deep brain stimulation: a paradigm shifting approach to treat Parkinson's disease. Frontiers in neuroscience 10, 173 (2016). 60 BeMent, S. L., Wise, K. D., Anderson, D. J., Najafi, K. & Drake, K. L. Solid-state electrodes for multichannel multiplexed intracortical neuronal recording. IEEE transactions on biomedical engineering, 230-241 (1986). 61 Campbell, P. K., Jones, K. E., Huber, R. J., Horch, K. W. & Normann, R. A. A silicon-based, three- dimensional neural interface: manufacturing processes for an intracortical electrode array. IEEE Transactions on Biomedical Engineering 38, 758-768 (1991). 62 Yang, X. et al. Bioinspired neuron-like electronics. Nature materials 18, 510 (2019). 63 Kozai, T. D. et al. Chronic tissue response to carboxymethyl cellulose based dissolvable insertion needle for ultra-small neural probes. Biomaterials 35, 9255-9268 (2014). 64 Bai, Q., Wise, K. D. & Anderson, D. J. A high-yield microassembly structure for three-dimensional microelectrode arrays. IEEE Transactions on Biomedical Engineering 47, 281-289 (2000). 65 Takeuchi, S., Suzuki, T., Mabuchi, K. & Fujita, H. 3D flexible multichannel neural probe array. Journal of Micromechanics and Microengineering 14, 104-107, doi:10.1088/0960- 1317/14/1/014 (2003). 66 Merlo, M. W., Snyder, R. L., Middlebrooks, J. C. & Bachman, M. Microelectrode arrays fabricated using a novel hybrid microfabrication method. Biomedical microdevices 14, 193-205 (2012). 67 Sohal, H. S. et al. The sinusoidal probe: a new approach to improve electrode longevity. Front Neuroeng 7, 10, doi:10.3389/fneng.2014.00010 (2014). 68 Wu, F., Im, M. & Yoon, E. in 2011 16th International Solid-State Sensors, Actuators and Microsystems Conference. 966-969 (IEEE).
Page | 109
69 Logan, A., Frautschy, S. A., Gonzalez, A. M., Sporn, M. B. & Baird, A. Enhanced expression of transforming growth factor beta 1 in the rat brain after a localized cerebral injury. Brain research 587, 216-225 (1992). 70 Tsai, M.-L. & Yen, C.-T. A simple method for fabricating horizontal and vertical microwire arrays. Journal of neuroscience methods 131, 107-110 (2003). 71 Zhang, X., Zhang, R. L., Zhang, Z. G. & Chopp, M. Measurement of neuronal activity of individual neurons after stroke in the rat using a microwire electrode array. Journal of neuroscience methods 162, 91-100 (2007). 72 Prasad, A. et al. Comprehensive characterization and failure modes of tungsten microwire arrays in chronic neural implants. J Neural Eng 9, 056015, doi:10.1088/1741-2560/9/5/056015 (2012). 73 Hopcroft, M. A., Nix, W. D. & Kenny, T. W. What is the Young's Modulus of Silicon? Journal of microelectromechanical systems 19, 229-238 (2010). 74 Chen, Z., Gandhi, U., Lee, J. & Wagoner, R. Variation and consistency of Young’s modulus in steel. Journal of Materials Processing Technology 227, 227-243 (2016). 75 Merker, J., Lupton, D., Töpfer, M. & Knake, H. High temperature mechanical properties of the platinum group metals. Platinum Metals Review 45, 74-82 (2001). 76 Škoro, G. et al. Dynamic Young’s moduli of tungsten and tantalum at high temperature and stress. Journal of Nuclear Materials 409, 40-46 (2011). 77 Rousche, P. J. & Normann, R. A. Chronic recording capability of the Utah Intracortical Electrode Array in cat sensory cortex. Journal of neuroscience methods 82, 1-15 (1998). 78 Nicolelis, M. A. et al. Chronic, multisite, multielectrode recordings in macaque monkeys. Proceedings of the National Academy of Sciences 100, 11041-11046 (2003). 79 Williams, J. C., Rennaker, R. L. & Kipke, D. R. Long-term neural recording characteristics of wire microelectrode arrays implanted in cerebral cortex. Brain Research Protocols 4, 303-313 (1999). 80 Barrese, J. C. et al. Failure mode analysis of silicon-based intracortical microelectrode arrays in non-human primates. Journal of neural engineering 10, 066014 (2013). 81 Patrick, E., Orazem, M. E., Sanchez, J. C. & Nishida, T. Corrosion of tungsten microelectrodes used in neural recording applications. J Neurosci Methods 198, 158-171, doi:10.1016/j.jneumeth.2011.03.012 (2011). 82 Wang, A. et al. Stability of and inflammatory response to silicon coated with a fluoroalkyl self- assembled monolayer in the central nervous system. Journal of biomedical materials research. Part A 81, 363-372, doi:10.1002/jbm.a.31034 (2007). 83 Hammerle, H. et al. Biostability of micro-photodiode arrays for subretinal implantation. Biomaterials 23, 797-804 (2002). 84 Sillay, K. A. et al. Long-Term Measurement of Impedance in Chronically Implanted Depth and Subdural Electrodes During Responsive Neurostimulation in Humans. Brain Stimulation 6, 718- 726, doi:https://doi.org/10.1016/j.brs.2013.02.001 (2013). 85 Szarowski, D. et al. Brain responses to micro-machined silicon devices. Brain research 983, 23-35 (2003). 86 Hoogerwerf, A. C. & Wise, K. D. A three-dimensional microelectrode array for chronic neural recording. IEEE Transactions on Biomedical Engineering 41, 1136-1146 (1994). 87 Turner, J. et al. Cerebral astrocyte response to micromachined silicon implants. Experimental neurology 156, 33-49 (1999). 88 Biran, R., Martin, D. C. & Tresco, P. A. Neuronal cell loss accompanies the brain tissue response to chronically implanted silicon microelectrode arrays. Experimental neurology 195, 115-126 (2005). 89 Liu, X. et al. Stability of the interface between neural tissue and chronically implanted intracortical microelectrodes. IEEE transactions on rehabilitation engineering 7, 315-326 (1999).
Page | 110
90 Williams, J. C., Hippensteel, J. A., Dilgen, J., Shain, W. & Kipke, D. R. Complex impedance spectroscopy for monitoring tissue responses to inserted neural implants. Journal of neural engineering 4, 410 (2007). 91 Kim, Y.-T., Hitchcock, R. W., Bridge, M. J. & Tresco, P. A. Chronic response of adult rat brain tissue to implants anchored to the skull. Biomaterials 25, 2229-2237 (2004). 92 Johnson, M. D., Otto, K. J. & Kipke, D. R. Repeated voltage biasing improves unit recordings by reducing resistive tissue impedances. IEEE Transactions on neural systems and rehabilitation engineering 13, 160-165 (2005). 93 Purcell, E., Seymour, J., Yandamuri, S. & Kipke, D. In vivo evaluation of a neural stem cell-seeded prosthesis. Journal of neural engineering 6, 026005 (2009). 94 McConnell, G. C. et al. Implanted neural electrodes cause chronic, local inflammation that is correlated with local neurodegeneration. Journal of neural engineering 6, 056003 (2009). 95 Giulian, D., Li, J., Li, X., George, J. & Rutecki, P. A. The impact of microglia-derived cytokines upon gliosis in the CNS. Developmental neuroscience 16, 128-136 (1994). 96 Edell, D. J., Toi, V., McNeil, V. M. & Clark, L. Factors influencing the biocompatibility of insertable silicon microshafts in cerebral cortex. IEEE Transactions on Biomedical Engineering 39, 635-643 (1992). 97 Saxena, T. et al. The impact of chronic blood–brain barrier breach on intracortical electrode function. Biomaterials 34, 4703-4713 (2013). 98 Saxena, T. et al. The impact of chronic blood-brain barrier breach on intracortical electrode function. Biomaterials 34, 4703-4713, doi:10.1016/j.biomaterials.2013.03.007 (2013). 99 Kozai, T. D., Jaquins-Gerstl, A. S., Vazquez, A. L., Michael, A. C. & Cui, X. T. Brain tissue responses to neural implants impact signal sensitivity and intervention strategies. ACS chemical neuroscience 6, 48-67 (2015). 100 Giulian, D. & Baker, T. J. Characterization of ameboid microglia isolated from developing mammalian brain. The Journal of neuroscience : the official journal of the Society for Neuroscience 6, 2163-2178 (1986). 101 Tengvall, P., Lundström, I., Sjöqvist, L., Elwing, H. & Bjursten, L. M. Titanium-hydrogen peroxide interaction: model studies of the influence of the inflammatory response on titanium implants. Biomaterials 10, 166-175 (1989). 102 Gilbert, J. L. et al. Direct in vivo inflammatory cell-induced corrosion of CoCrMo alloy orthopedic implant surfaces. Journal of biomedical materials research. Part A 103, 211-223, doi:10.1002/jbm.a.35165 (2015). 103 Potter-Baker, K. A. et al. Development of superoxide dismutase mimetic surfaces to reduce accumulation of reactive oxygen species for neural interfacing applications. Journal of Materials Chemistry B 2, 2248-2258 (2014). 104 Polikov, V. S., Tresco, P. A. & Reichert, W. M. Response of brain tissue to chronically implanted neural electrodes. Journal of neuroscience methods 148, 1-18 (2005). 105 Salatino, J. W., Ludwig, K. A., Kozai, T. D. & Purcell, E. K. Glial responses to implanted electrodes in the brain. Nature biomedical engineering 1, 862 (2017). 106 Chen, R., Canales, A. & Anikeeva, P. Neural recording and modulation technologies. Nature Reviews Materials 2, 16093, doi:10.1038/natrevmats.2016.93 (2017). 107 Landis, D. M. The early reactions of non-neuronal cells to brain injury. Annual review of neuroscience 17, 133-151 (1994). 108 Berry, M. CNS injuries: cellular responses and pharmacological strategies. (CRC Press, 2019). 109 BELL, M. J. et al. Interleukin-6 and interleukin-10 in cerebrospinal fluid after severe traumatic brain injury in children. Journal of neurotrauma 14, 451-457 (1997).
Page | 111
110 Woodroofe, M. N. et al. Detection of interleukin-1 and interleukin-6 in adult rat brain, following mechanical injury, by in vivo microdialysis: evidence of a role for microglia in cytokine production. Journal of Neuroimmunology 33, 227-236, doi:https://doi.org/10.1016/0165- 5728(91)90110-S (1991). 111 Logan, A., Frautschy, S. A., Gonzalez, A. M. & Baird, A. A time course for the focal elevation of synthesis of basic fibroblast growth factor and one of its high-affinity receptors (flg) following a localized cortical brain injury. The Journal of neuroscience : the official journal of the Society for Neuroscience 12, 3828-3837 (1992). 112 Koehler, R. C., Roman, R. J. & Harder, D. R. Astrocytes and the regulation of cerebral blood flow. Trends in neurosciences 32, 160-169 (2009). 113 Pellerin, L. et al. Activity‐dependent regulation of energy metabolism by astrocytes: an update. Glia 55, 1251-1262 (2007). 114 Dong, Y. & Benveniste, E. N. Immune function of astrocytes. Glia 36, 180-190 (2001). 115 Burda, J. E., Bernstein, A. M. & Sofroniew, M. V. Astrocyte roles in traumatic brain injury. Experimental Neurology 275, 305-315, doi:https://doi.org/10.1016/j.expneurol.2015.03.020 (2016). 116 Buffo, A., Rolando, C. & Ceruti, S. Astrocytes in the damaged brain: molecular and cellular insights into their reactive response and healing potential. Biochemical pharmacology 79, 77-89, doi:10.1016/j.bcp.2009.09.014 (2010). 117 Lin, H. W. et al. Astrogliosis is delayed in type 1 interleukin-1 receptor-null mice following a penetrating brain injury. Journal of neuroinflammation 3, 15, doi:10.1186/1742-2094-3-15 (2006). 118 Kordek, R. et al. Heightened expression of tumor necrosis factor alpha, interleukin 1 alpha, and glial fibrillary acidic protein in experimental Creutzfeldt-Jakob disease in mice. Proceedings of the National Academy of Sciences of the United States of America 93, 9754-9758, doi:10.1073/pnas.93.18.9754 (1996). 119 Swartz, K. R. et al. Interleukin-6 promotes post-traumatic healing in the central nervous system. Brain Res 896, 86-95 (2001). 120 Gorina, R., Font-Nieves, M., Marquez-Kisinousky, L., Santalucia, T. & Planas, A. M. Astrocyte TLR4 activation induces a proinflammatory environment through the interplay between MyD88- dependent NFkappaB signaling, MAPK, and Jak1/Stat1 pathways. Glia 59, 242-255, doi:10.1002/glia.21094 (2011). 121 Sofroniew, M. V. Molecular dissection of reactive astrogliosis and glial scar formation. Trends in Neurosciences 32, 638-647, doi:https://doi.org/10.1016/j.tins.2009.08.002 (2009). 122 McKeon, R. J., Schreiber, R. C., Rudge, J. S. & Silver, J. Reduction of neurite outgrowth in a model of glial scarring following CNS injury is correlated with the expression of inhibitory molecules on reactive astrocytes. The Journal of neuroscience : the official journal of the Society for Neuroscience 11, 3398-3411 (1991). 123 Silver, J. & Miller, J. H. Regeneration beyond the glial scar. Nature reviews neuroscience 5, 146 (2004). 124 Laywell, E. D. et al. Enhanced expression of the developmentally regulated extracellular matrix molecule tenascin following adult brain injury. Proceedings of the National Academy of Sciences of the United States of America 89, 2634-2638, doi:10.1073/pnas.89.7.2634 (1992). 125 Anderson, M. A. et al. Astrocyte scar formation aids central nervous system axon regeneration. Nature 532, 195-200, doi:10.1038/nature17623 (2016). 126 Kettenmann, H., Kirchhoff, F. & Verkhratsky, A. Microglia: new roles for the synaptic stripper. Neuron 77, 10-18, doi:10.1016/j.neuron.2012.12.023 (2013).
Page | 112
127 Sheng, W. S., Hu, S., Kravitz, F. H., Peterson, P. K. & Chao, C. C. Tumor necrosis factor alpha upregulates human microglial cell production of interleukin-10 in vitro. Clinical and diagnostic laboratory immunology 2, 604-608 (1995). 128 Giulian, D., Li, J., Leara, B. & Keenen, C. Phagocytic microglia release cytokines and cytotoxins that regulate the survival of astrocytes and neurons in culture. Neurochemistry international 25, 227-233 (1994). 129 Blinzinger, K. & Kreutzberg, G. Displacement of synaptic terminals from regenerating motoneurons by microglial cells. Zeitschrift fur Zellforschung und mikroskopische Anatomie (Vienna, Austria : 1948) 85, 145-157 (1968). 130 Chen, Z. et al. Microglial displacement of inhibitory synapses provides neuroprotection in the adult brain. Nature communications 5, 4486 (2014). 131 Weldon, D. T. et al. Fibrillar beta-amyloid induces microglial phagocytosis, expression of inducible nitric oxide synthase, and loss of a select population of neurons in the rat CNS in vivo. The Journal of neuroscience : the official journal of the Society for Neuroscience 18, 2161-2173 (1998). 132 Kozai, T. D. Y., Vazquez, A. L., Weaver, C. L., Kim, S.-G. & Cui, X. T. In vivo two-photon microscopy reveals immediate microglial reaction to implantation of microelectrode through extension of processes. Journal of neural engineering 9, 066001 (2012). 133 Gällentoft, L. et al. Impact of degradable nanowires on long-term brain tissue responses. Journal of Nanobiotechnology 14, 64, doi:10.1186/s12951-016-0216-7 (2016). 134 Jorfi, M., Skousen, J. L., Weder, C. & Capadona, J. R. Progress towards biocompatible intracortical microelectrodes for neural interfacing applications. Journal of neural engineering 12, 011001 (2014). 135 Nolta, N. F., Christensen, M. B., Crane, P. D., Skousen, J. L. & Tresco, P. A. BBB leakage, astrogliosis, and tissue loss correlate with silicon microelectrode array recording performance. Biomaterials 53, 753-762, doi:10.1016/j.biomaterials.2015.02.081 (2015). 136 Wanner, I. B. et al. Glial scar borders are formed by newly proliferated, elongated astrocytes that interact to corral inflammatory and fibrotic cells via STAT3-dependent mechanisms after spinal cord injury. The Journal of neuroscience : the official journal of the Society for Neuroscience 33, 12870-12886, doi:10.1523/jneurosci.2121-13.2013 (2013). 137 Eaton, K. P. & Henriquez, C. S. Confounded spikes generated by synchrony within neural tissue models. Neurocomputing 65, 851-857 (2005). 138 Henze, D. A. et al. Intracellular features predicted by extracellular recordings in the hippocampus in vivo. Journal of neurophysiology 84, 390-400, doi:10.1152/jn.2000.84.1.390 (2000). 139 Prasad, A. & Sanchez, J. C. Quantifying long-term microelectrode array functionality using chronic in vivo impedance testing. Journal of neural engineering 9, 026028 (2012). 140 Malaga, K. A. et al. Data-driven model comparing the effects of glial scarring and interface interactions on chronic neural recordings in non-human primates. Journal of neural engineering 13, 016010 (2015). 141 Bush, T. G. et al. Leukocyte infiltration, neuronal degeneration, and neurite outgrowth after ablation of scar-forming, reactive astrocytes in adult transgenic mice. Neuron 23, 297-308 (1999). 142 Faulkner, J. R. et al. Reactive astrocytes protect tissue and preserve function after spinal cord injury. The Journal of neuroscience : the official journal of the Society for Neuroscience 24, 2143- 2155, doi:10.1523/jneurosci.3547-03.2004 (2004). 143 Okada, S. et al. Conditional ablation of Stat3 or Socs3 discloses a dual role for reactive astrocytes after spinal cord injury. Nat Med 12, 829-834, doi:10.1038/nm1425 (2006).
Page | 113
144 Boehler, C. et al. Actively controlled release of Dexamethasone from neural microelectrodes in a chronic in vivo study. Biomaterials 129, 176-187, doi:10.1016/j.biomaterials.2017.03.019 (2017). 145 Mercanzini, A. et al. Controlled release nanoparticle-embedded coatings reduce the tissue reaction to neuroprostheses. Journal of Controlled Release 145, 196-202, doi:https://doi.org/10.1016/j.jconrel.2010.04.025 (2010). 146 Spencer, K. C. et al. Characterization of mechanically matched hydrogel coatings to improve the biocompatibility of neural implants. Scientific reports 7, 1952 (2017). 147 Hassler, C. et al. in 2015 7th International IEEE/EMBS Conference on Neural Engineering (NER). 561-564 (IEEE). 148 Casanova, F., Carney, P. R. & Sarntinoranont, M. In vivo evaluation of needle force and friction stress during insertion at varying insertion speed into the brain. Journal of neuroscience methods 237, 79-89 (2014). 149 Welkenhuysen, M., Andrei, A., Ameye, L., Eberle, W. & Nuttin, B. Effect of insertion speed on tissue response and insertion mechanics of a chronically implanted silicon-based neural probe. IEEE Transactions on Biomedical Engineering 58, 3250-3259 (2011). 150 Andrei, A., Welkenhuysen, M., Nuttin, B. & Eberle, W. A response surface model predicting the in vivo insertion behavior of micromachined neural implants. Journal of neural engineering 9, 016005 (2011). 151 Skousen, J. L. et al. in Progress in Brain Research Vol. 194 (eds Jens Schouenborg, Martin Garwicz, & Nils Danielsen) 167-180 (Elsevier, 2011). 152 Thelin, J. et al. Implant size and fixation mode strongly influence tissue reactions in the CNS. PloS one 6, e16267 (2011). 153 Lind, G., Linsmeier, C. E. & Schouenborg, J. The density difference between tissue and neural probes is a key factor for glial scarring. Scientific reports 3, 2942 (2013). 154 Rousche, P. J. et al. Flexible polyimide-based intracortical electrode arrays with bioactive capability. IEEE Transactions on biomedical engineering 48, 361-371 (2001). 155 Franze, K., Janmey, P. A. & Guck, J. Mechanics in neuronal development and repair. Annual review of biomedical engineering 15, 227-251 (2013). 156 Lassner, E. & Schubert, W.-D. Properties, chemistry, technology of the element, alloys, and chemical compounds. Vienna University of Technology, Vienna, Austria, Kluwer, 124-125 (1999). 157 Bershadsky, A. D., Balaban, N. Q. & Geiger, B. Adhesion-dependent cell mechanosensitivity. Annual review of cell and developmental biology 19, 677-695 (2003). 158 Discher, D. E., Janmey, P. & Wang, Y. L. Tissue cells feel and respond to the stiffness of their substrate. Science 310, 1139-1143, doi:10.1126/science.1116995 (2005). 159 Ruoslahti, E. Brain extracellular matrix. Glycobiology 6, 489-492 (1996). 160 Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Matrix elasticity directs stem cell lineage specification. Cell 126, 677-689, doi:10.1016/j.cell.2006.06.044 (2006). 161 Flanagan, L. A., Ju, Y.-E., Marg, B., Osterfield, M. & Janmey, P. A. Neurite branching on deformable substrates. Neuroreport 13, 2411-2415, doi:10.1097/01.wnr.0000048003.96487.97 (2002). 162 Kostic, A., Sap, J. & Sheetz, M. P. RPTPα is required for rigidity-dependent inhibition of extension and differentiation of hippocampal neurons. Journal of cell science 120, 3895-3904 (2007). 163 Moshayedi, P. et al. Mechanosensitivity of astrocytes on optimized polyacrylamide gels analyzed by quantitative morphometry. Journal of Physics: Condensed Matter 22, 194114 (2010). 164 Goldstein, S. R. & Salcman, M. Mechanical factors in the design of chronic recording intracortical microelectrodes. IEEE Transactions on Biomedical Engineering, 260-269 (1973).
Page | 114
165 Britt, R. H. & Rossi, G. T. Quantitative analysis of methods for reducing physiological brain pulsations. Journal of Neuroscience Methods 6, 219-229, doi:https://doi.org/10.1016/0165- 0270(82)90085-1 (1982). 166 Muthuswamy, J., Gilletti, A., Jain, T. & Okandan, M. in Proceedings of the 25th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (IEEE Cat. No. 03CH37439). 1941-1943 (IEEE). 167 Lee, H., Bellamkonda, R. V., Sun, W. & Levenston, M. E. Biomechanical analysis of silicon microelectrode-induced strain in the brain. Journal of neural engineering 2, 81 (2005). 168 Harris, J. et al. Mechanically adaptive intracortical implants improve the proximity of neuronal cell bodies. Journal of neural engineering 8, 066011 (2011). 169 Du, Z. J. et al. Ultrasoft microwire neural electrodes improve chronic tissue integration. Acta biomaterialia 53, 46-58 (2017). 170 Lee, H. C. et al. Histological evaluation of flexible neural implants; flexibility limit for reducing the tissue response? J Neural Eng 14, 036026, doi:10.1088/1741-2552/aa68f0 (2017). 171 Ejserholm, F. et al. Biocompatibility of a polymer based on Off-Stoichiometry Thiol-Enes+ Epoxy (OSTE+) for neural implants. Biomaterials research 19, 19 (2015). 172 Ji, S., Zhu, Q., Dougherty, L. & Margulies, S. S. In vivo measurements of human brain displacement. (SAE Technical Paper, 2004). 173 Zhu, Q., Dougherty, L. & Margulies, S. S. in Summer Bioenegineering Conference, Key Biscayne, Florida. 174 Wenger, E., Brozzoli, C., Lindenberger, U. & Lövdén, M. Expansion and renormalization of human brain structure during skill acquisition. Trends in cognitive sciences 21, 930-939 (2017). 175 Erickson, K. I. et al. Exercise training increases size of hippocampus and improves memory. Proceedings of the National Academy of Sciences 108, 3017-3022 (2011). 176 Qian, X., Song, H. & Ming, G.-l. Brain organoids: advances, applications and challenges. Development 146, dev166074 (2019). 177 Drummond, E. & Wisniewski, T. Alzheimer’s disease: experimental models and reality. Acta neuropathologica 133, 155-175 (2017). 178 Council, N. R. in Scientific Frontiers in Developmental Toxicology and Risk Assessment (National Academies Press (US), 2000). 179 Spear, L. P. Adolescent brain development and animal models. Annals of the New York Academy of Sciences 1021, 23-26 (2004). 180 Hassler, C., Boretius, T. & Stieglitz, T. Polymers for neural implants. Journal of Polymer Science Part B: Polymer Physics 49, 18-33 (2011). 181 Liu, Y. et al. Soft and elastic hydrogel-based microelectronics for localized low-voltage neuromodulation. Nature biomedical engineering 3, 58 (2019). 182 Huang, W. C. et al. Ultracompliant Hydrogel‐Based Neural Interfaces Fabricated by Aqueous‐ Phase Microtransfer Printing. Advanced Functional Materials 28, 1801059 (2018). 183 Weltman, A., Yoo, J. & Meng, E. Flexible, penetrating brain probes enabled by advances in polymer microfabrication. Micromachines (Basel) 7, 180 (2016). 184 Sheng, H. et al. Neural interfaces by hydrogels. Extreme Mechanics Letters, 100510 (2019). 185 Stieglitz, T. & Meyer, J. Implantable microsystems. Polyimide-based neuroprostheses for interfacing nerves. Medical device technology 10, 28-30 (1999). 186 Jiang, X. et al. In vitro and in vivo evaluation of a photosensitive polyimide thin-film microelectrode array suitable for epiretinal stimulation. Journal of neuroengineering and rehabilitation 10, 48 (2013).
Page | 115
187 Cheung, K. C., Renaud, P., Tanila, H. & Djupsund, K. Flexible polyimide microelectrode array for in vivo recordings and current source density analysis. Biosensors & bioelectronics 22, 1783- 1790, doi:10.1016/j.bios.2006.08.035 (2007). 188 Lee, K.-K. et al. Polyimide-based intracortical neural implant with improved structural stiffness. Journal of Micromechanics and Microengineering 14, 32 (2003). 189 Ordonez, J. S., Boehler, C., Schuettler, M. & Stieglitz, T. in 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 5134-5137 (IEEE). 190 Kim, B. J. & Meng, E. Review of polymer MEMS micromachining. Journal of Micromechanics and Microengineering 26, 013001 (2015). 191 Kuo, J. T. et al. Novel flexible Parylene neural probe with 3D sheath structure for enhancing tissue integration. Lab on a Chip 13, 554-561 (2013). 192 Seymour, J. P. & Kipke, D. R. Neural probe design for reduced tissue encapsulation in CNS. Biomaterials 28, 3594-3607 (2007). 193 Lecomte, A. et al. Silk and PEG as means to stiffen a parylene probe for insertion in the brain: toward a double time-scale tool for local drug delivery. Journal of Micromechanics and Microengineering 25, 125003 (2015). 194 Kahouli, A., Sylvestre, A., Jomni, F., Yangui, B. & Legrand, J. Ac-conductivity and dielectric relaxations above glass transition temperature for parylene-C thin films. Applied Physics A 106, 909-913 (2012). 195 Nemani, K. V., Moodie, K. L., Brennick, J. B., Su, A. & Gimi, B. In vitro and in vivo evaluation of SU-8 biocompatibility. Materials Science and Engineering: C 33, 4453-4459 (2013). 196 Vernekar, V. N. et al. SU‐8 2000 rendered cytocompatible for neuronal bioMEMS applications. Journal of Biomedical Materials Research Part A: An Official Journal of The Society for Biomaterials, The Japanese Society for Biomaterials, and The Australian Society for Biomaterials and the Korean Society for Biomaterials 89, 138-151 (2009). 197 Huang, S.-H., Lin, S.-P. & Chen, J.-J. J. In vitro and in vivo characterization of SU-8 flexible neuroprobe: From mechanical properties to electrophysiological recording. Sensors and Actuators A: Physical 216, 257-265 (2014). 198 Lee, S. E. et al. A flexible depth probe using liquid crystal polymer. IEEE Transactions on Biomedical Engineering 59, 2085-2094 (2012). 199 Nguyen, J. K. et al. Mechanically-compliant intracortical implants reduce the neuroinflammatory response. J Neural Eng 11, 056014, doi:10.1088/1741-2560/11/5/056014 (2014). 200 Minev, I. R. et al. Electronic dura mater for long-term multimodal neural interfaces. Science 347, 159-163 (2015). 201 Chou, N., Yoo, S. & Kim, S. A largely deformable surface type neural electrode array based on PDMS. IEEE Transactions on neural systems and rehabilitation engineering 21, 544-553 (2012). 202 Staples, N. A. et al. Conductive hydrogel electrodes for delivery of long-term high frequency pulses. Frontiers in neuroscience 11, 748 (2018). 203 Colas, A. & Curtis, J. Silicone biomaterials: history and chemistry. Biomaterials science: an introduction to materials in medicine 2, 80-85 (2004). 204 Kam, D., Shah, L. & Mazumder, J. Femtosecond laser machining of multi-depth microchannel networks onto silicon. Journal of Micromechanics and Microengineering 21, 045027 (2011). 205 Isiksacan, Z., Guler, M. T., Aydogdu, B., Bilican, I. & Elbuken, C. Rapid fabrication of microfluidic PDMS devices from reusable PDMS molds using laser ablation. Journal of Micromechanics and Microengineering 26, 035008 (2016). 206 Kim, T. N., Campbell, K., Groisman, A., Kleinfeld, D. & Schaffer, C. B. Femtosecond laser-drilled capillary integrated into a microfluidic device. Applied Physics Letters 86, 201106 (2005).
Page | 116
207 Ozbolat, V. et al. 3D Printing of PDMS improves its mechanical and cell adhesion properties. ACS Biomaterials Science & Engineering 4, 682-693 (2018). 208 Hinton, T. J., Hudson, A., Pusch, K., Lee, A. & Feinberg, A. W. 3D printing PDMS elastomer in a hydrophilic support bath via freeform reversible embedding. ACS biomaterials science & engineering 2, 1781-1786 (2016). 209 Femmer, T., Kuehne, A. J. & Wessling, M. Print your own membrane: direct rapid prototyping of polydimethylsiloxane. Lab Chip 14, 2610-2613, doi:10.1039/c4lc00320a (2014). 210 Moraes, C., Labuz, J. M., Shao, Y., Fu, J. & Takayama, S. Supersoft lithography: candy-based fabrication of soft silicone microstructures. Lab on a Chip 15, 3760-3765 (2015). 211 Carlborg, C. F., Haraldsson, T., Cornaglia, M., Stemme, G. & van der Wijngaart, W. A high-yield process for 3-D large-scale integrated microfluidic networks in PDMS. Journal of microelectromechanical systems 19, 1050-1057 (2010). 212 Dahiya, R. S. & Lorenzelli, L. in Proc. 38th Int. Conf. MNE 2012. 1-2. 213 Deng, Y.-L. & Juang, Y.-J. Polydimethyl siloxane wet etching for three dimensional fabrication of microneedle array and high-aspect-ratio micropillars. Biomicrofluidics 8, 026502 (2014). 214 Kleiman, M., Ryu, K. A. & Esser‐Kahn, A. P. Determination of Factors Influencing the Wet Etching of Polydimethylsiloxane Using Tetra‐n‐butylammonium Fluoride. Macromolecular Chemistry and Physics 217, 284-291 (2016). 215 Balakrisnan, B., Patil, S. & Smela, E. Patterning PDMS using a combination of wet and dry etching. Journal of Micromechanics and Microengineering 19, 047002 (2009). 216 Chiu, D. T. et al. Patterned deposition of cells and proteins onto surfaces by using three- dimensional microfluidic systems. Proceedings of the National Academy of Sciences of the United States of America 97, 2408-2413, doi:10.1073/pnas.040562297 (2000). 217 Juncker, D. et al. Soft and rigid two-level microfluidic networks for patterning surfaces. Journal of Micromechanics and Microengineering 11, 532 (2001). 218 Gregory, C. W., Sellgren, K. L., Gilchrist, K. H. & Grego, S. High yield fabrication of multilayer polydimethylsiloxane devices with freestanding micropillar arrays. Biomicrofluidics 7, 056503 (2013). 219 Rogers, J. A., Someya, T. & Huang, Y. Materials and mechanics for stretchable electronics. science 327, 1603-1607 (2010). 220 Lacour, S. P. et al. Flexible and stretchable micro-electrodes for in vitro and in vivo neural interfaces. Medical & biological engineering & computing 48, 945-954 (2010). 221 Lu, T., Finkenauer, L., Wissman, J. & Majidi, C. Rapid prototyping for soft‐matter electronics. Advanced Functional Materials 24, 3351-3356 (2014). 222 Dickey, M. D. Stretchable and soft electronics using liquid metals. Advanced Materials 29, 1606425 (2017). 223 Khilwani, R. et al. Ultra-miniature ultra-compliant neural probes with dissolvable delivery needles: design, fabrication and characterization. Biomedical microdevices 18, 97 (2016). 224 Lewitus, D., Smith, K. L., Shain, W. & Kohn, J. Ultrafast resorbing polymers for use as carriers for cortical neural probes. Acta biomaterialia 7, 2483-2491 (2011). 225 Singh, A., Zhu, H. & He, J. in The 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. 4298-4301 (IEEE). 226 Apollo, N. V. et al. Soft, flexible freestanding neural stimulation and recording electrodes fabricated from reduced graphene oxide. Advanced Functional Materials 25, 3551-3559 (2015). 227 Vepari, C. & Kaplan, D. L. Silk as a biomaterial. Progress in polymer science 32, 991-1007 (2007). 228 Lo, M.-c. et al. Coating flexible probes with an ultra fast degrading polymer to aid in tissue insertion. Biomedical microdevices 17, 34 (2015).
Page | 117
229 Kim, B. J. et al. 3D Parylene sheath neural probe for chronic recordings. J Neural Eng 10, 045002, doi:10.1088/1741-2560/10/4/045002 (2013). 230 Richter, A. et al. A simple implantation method for flexible, multisite microelectrodes into rat brains. Frontiers in Neuroengineering 6, doi:10.3389/fneng.2013.00006 (2013). 231 Simeral, J., Kim, S.-P., Black, M., Donoghue, J. & Hochberg, L. Neural control of cursor trajectory and click by a human with tetraplegia 1000 days after implant of an intracortical microelectrode array. Journal of neural engineering 8, 025027 (2011). 232 Cullen, D. K., Simon, C. M. & LaPlaca, M. C. Strain rate-dependent induction of reactive astrogliosis and cell death in three-dimensional neuronal–astrocytic co-cultures. Brain Research 1158, 103-115, doi:https://doi.org/10.1016/j.brainres.2007.04.070 (2007). 233 Georges, P. C., Miller, W. J., Meaney, D. F., Sawyer, E. S. & Janmey, P. A. Matrices with compliance comparable to that of brain tissue select neuronal over glial growth in mixed cortical cultures. Biophysical journal 90, 3012-3018, doi:10.1529/biophysj.105.073114 (2006). 234 LaPlaca, M. C., Cullen, D. K., McLoughlin, J. J. & Cargill, R. S., 2nd. High rate shear strain of three- dimensional neural cell cultures: a new in vitro traumatic brain injury model. Journal of biomechanics 38, 1093-1105, doi:10.1016/j.jbiomech.2004.05.032 (2005). 235 Park, G. et al. Immunologic and tissue biocompatibility of flexible/stretchable electronics and optoelectronics. Advanced healthcare materials 3, 515-525 (2014). 236 Finn, A. et al. High aspect ratio pattern collapse of polymeric UV-nano-imprint molds due to cleaning. Microelectronic Engineering 110, 112-118, doi:https://doi.org/10.1016/j.mee.2013.02.065 (2013). 237 Lin, C.-H. et al. Extreme UV diffraction grating fabricated by nanoimprint lithography. Microelectronic Engineering 98, 194-197 (2012). 238 Ren, K. et al. in 14th International Conference on Miniaturized Systems for Chemistry and Life Sciences, Groningen, The Netherlands. 554-556. 239 Fuard, D., Tzvetkova-Chevolleau, T., Decossas, S., Tracqui, P. & Schiavone, P. Optimization of poly-di-methyl-siloxane (PDMS) substrates for studying cellular adhesion and motility. Microelectronic Engineering 85, 1289-1293, doi:https://doi.org/10.1016/j.mee.2008.02.004 (2008). 240 Hartel, R. W., von Elbe, J. H. & Hofberger, R. in Confectionery Science and Technology (eds Richard W. Hartel, Joachim H. von Elbe, & Randy Hofberger) 211-244 (Springer International Publishing, 2018). 241 Jiang, B., Liu, Y., Bhandari, B. & Zhou, W. Impact of caramelization on the glass transition temperature of several caramelized sugars. Part I: chemical analyses. Journal of agricultural and food chemistry 56, 5138-5147 (2008). 242 Alves, L. A., Almeida e Silva, J. B. & Giulietti, M. Solubility of D-glucose in water and ethanol/water mixtures. Journal of Chemical & Engineering Data 52, 2166-2170 (2007). 243 Hernández-Castro, J. A., Li, K., Meunier, A., Juncker, D. & Veres, T. Fabrication of large-area polymer microfilter membranes and their application for particle and cell enrichment. Lab on a Chip 17, 1960-1969 (2017). 244 Bowden, N., Brittain, S., Evans, A. G., Hutchinson, J. W. & Whitesides, G. M. Spontaneous formation of ordered structures in thin films of metals supported on an elastomeric polymer. Nature 393, 146 (1998). 245 Genzer, J. & Groenewold, J. Soft matter with hard skin: From skin wrinkles to templating and material characterization. Soft Matter 2, 310-323 (2006). 246 Yin, J., Han, X., Cao, Y. & Lu, C. Surface wrinkling on polydimethylsiloxane microspheres via wet surface chemical oxidation. Scientific reports 4, 5710 (2014).
Page | 118
247 Bhushan, B. & Li, X. Micromechanical and tribological characterization of doped single-crystal silicon and polysilicon films for microelectromechanical systems devices. Journal of Materials Research 12, 54-63 (1997). 248 Budday, S. et al. Mechanical properties of gray and white matter brain tissue by indentation. Journal of the mechanical behavior of biomedical materials 46, 318-330 (2015). 249 Bouchonville, N. et al. AFM mapping of the elastic properties of brain tissue reveals kPa μm− 1 gradients of rigidity. Soft Matter 12, 6232-6239 (2016). 250 Schaefer, R. J. Mechanical properties of rubber. Harris' Shock and Vibration Handbook, Sixth edition, A. Piersol, T. Paez (Eds), McGraw-Hill Companies Inc, 33.31-33.18 (2010). 251 Gilletti, A. & Muthuswamy, J. Brain micromotion around implants in the rodent somatosensory cortex. J Neural Eng 3, 189-195, doi:10.1088/1741-2560/3/3/001 (2006). 252 Palchesko, R. N., Zhang, L., Sun, Y. & Feinberg, A. W. Development of polydimethylsiloxane substrates with tunable elastic modulus to study cell mechanobiology in muscle and nerve. PloS one 7, e51499 (2012). 253 Amjadi, M., Yoon, Y. J. & Park, I. Ultra-stretchable and skin-mountable strain sensors using carbon nanotubes-Ecoflex nanocomposites. Nanotechnology 26, 375501, doi:10.1088/0957- 4484/26/37/375501 (2015). 254 Korzhevskii, D. & Kirik, O. Brain microglia and microglial markers. Neuroscience and Behavioral Physiology 46, 284-290 (2016). 255 Eng, L. F. & Ghirnikar, R. S. GFAP and astrogliosis. Brain pathology 4, 229-237 (1994). 256 Potter, K. A., Buck, A. C., Self, W. K. & Capadona, J. R. Stab injury and device implantation within the brain results in inversely multiphasic neuroinflammatory and neurodegenerative responses. Journal of neural engineering 9, 046020 (2012).
Page | 119