ACTION POTENTIAL ELICITED CALCIUM SIGNALING IN MAMMALIAN PERIPHERAL NERVE AXONS: TOWARD AN OPTOGENETIC NEURAL INTERFACE by ARJUN KENNETH FONTAINE B.S. University of Colorado, Boulder, 2009 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Bioengineering Program 201 7
ii This thesis for the Doctor of Philosophy degree by Arjun Kenneth Fontaine has been approved for the Bioengineering Program by Richard Weir, Advisor Richard Benninger, Chair John Caldwell Emily Gibson Diego Restrepo M ay 13 201 7
iii Fontaine, Arju n (PhD, Bioengineering Program) Action Potential Elicited Calcium Signaling in Mammalian Peripheral Nerve Axons: Toward an Optogenetic Neural Interface Thesis directed by Research A ssociate Professor Richard Weir ABSTRACT Prosthetic limb control is considerably limited by the lack of an adequate neural interface. A system does not yet exist that provides high specificity neural communication in a non invasive manner. The recently emergent tools of optogenetics may offer an entirely new approach to neural in terfacing, with exciting and broad reaching implications. The primary goal of this thesis research was to identify and characterize a physiological signal that may be conducive to optogenetic read out of neural activity in the peripheral nervous system. In this aim I have investigated activity dependent calcium signaling in mammalian peripheral nerve axons, the pres ence of which was unknown prior to this research. An experimental paradigm was designed by which intra axonal loading of a synthetic dextran conjugated calcium indicator was achieved via diffusional and/or axoplasmic transport in an in vitro mouse nerve model Action potential elicited calcium transients were demons trated at axon nodes of Ranvier, and s patial and temporal dynamics of this sign al are presented in this thesis. Of particular significance to neuroprosthetic application is the characterization of a linear frequency amplitude modulation. The underlying physiological mechanisms and pathways were also investigated within this work. Pharmacological experiments indicate that the calcium transient arises by transmembrane influx, with insignificant contribution from intracellular organelle release. T type Ca V channels and
iv plasm a lemmal sodium calcium exchanger (NCX) are determined to be predominant mechanisms of calcium entry, likely functioning separately in heterogeneous axon types. In addition, the utility of the activity dependent calcium fluorescence as a prosthetic control signal is shown in proof of concept experiment s The acti on potential associated calcium signal is streamed in real time to control a prosthetic hand finger (Bebionic, RSL Steeper), and graded frequency modulated signals were used post hoc to demonstrate the correlative variable motor output that can be commande d with these signals. The form and content of this abstract are approved. I recommend its publication. Approved: Richard F. Weir
v TABLE OF CONTENTS CHAPTER I: INTRODUCTION AND LITERATURE REVIEW ................................ ................................ ..................... 1 II: ACTIVITY DEPENDENT CALCIUM SIGNALING AND DYNAMICS ................................ .......... 1 6 III: SOURCES AND MECHANISMS OF ACTIVITY DEPENDENT CALCIUM .............................. 4 3 IV: PROSTHETIC HAND ACTUATION WITH OPTICAL CALCIUM SIGNALS ............................ 5 7 V: DEEP TISSUE TWO PHOTON IMAGING IN BRAIN AND PERIPHERAL NERVE WITH A COMPACT HIGH PULSE ENERGY (YTTERBIUM) FIBER LASER ................................ 6 4 VI: DISCUSSION ................................ ................................ ................................ ................................ ................... 7 5 REFERENCES ................................ ................................ ................................ ................................ ........................ 8 4
1 CHAPTER I INTRODUCTION AND LITERATURE REVIEW Background & Significance Limb amputation is a prevalent issue arising from vascular complications, trauma, cancer a nd congenital related incidence In the United States there are nearly 2 million people living with limb loss. The National Center for Health Statistics estimates there are approximately 185 ,000 new amputations every year and about 30% of these are upper limb amputations 1,2 Individuals with limb deficiencies face substantial challenges in perfor ming everyday tasks and activities, but with proper limb replacement, the quality of life for those with limb loss could be greatly improved. Figure 1 .1 : Recent advanced prosthetic hands: TouchBionics i LIMB (left), Bebionic Hand (center), and the United States Defense Advanced Research Projects Agency (DARPA) sponsored prosthetic hand (right) Upper limb prosthetic devices have come a long way in restoring some basic functional capabilities and helping to improve the self image of amputees, but the full potential of prosthetic devices remains unfulfilled. In the past two decades, there has been great development of anthropomorphic prosthetic hands. Lead by groups such as Touch Bionics, BeBionic, organizations under the DA RPA Revolutionizing Prosthetics
2 initiative, and others, these novel devices have the mechanical capability to drive many degrees of freedom of a biological hand, opening the possibility for dexterous manipulation. See 3,4 for a review of current prosthetic hand design and performance. Biomimetic tactile sensors have also been developed for use in prosthetic hands, such as the SynTouch BioTac 5 7 which contains the sensory modalities of the human fingertip; force, vibration and temperature. Today, the largest barrier preventing true limb replacement is th e lack of an adequate control interface that successfully enables intuitive and natural control of such a device. Without this communicati on to the nervous system, artificial limbs and their control remain primitive, capable of only crude movements and la cking sensory fe edback. It has been documented that sensory feedback is necessar y for dexterous manipulation 8 A usefu l neural interface to exploit advanced devices must selectively and specificall y communicat e with neural pathways in order to enable fine (single digit) motor control and localized sensory percepts. Such an interface would go a long way in connecting patients with their assistive aids, allowing prostheses to be naturally and fully articulated, and as a result, far more helpful in assisting with their disability. State of the art electromyographic (EMG) control utilizes skin surface electrodes to harness motor commands from electrical signals of active muscle tissue. While this technique is relatively simplistic and non invasive, it has significant limitations: it lacks the ability to communicate with small or deeply embedded muscles for single digit co mmand or fine motor control, offe rs no means of sensory feedback, and is dependent on the presence of residual muscles.
3 The peripheral nerve offers a logical location to interface to the body because at this le vel of the nervous system, simplified neural outputs (axons) of complex CNS circuitry are contained in a spatially concentrated bundle. In a myelinated nerve axon the action potential is propagated by saltatory conduction between small intermittent gaps in myelin known as nodes of Ranvier. This saltatory conduction takes place by depolarization and repolarization mediated primarily by voltage gated sodium and potassium channels respectively. The sodium channel Na V 1.6 is highly concentrated at the node 9 while potassium channels K V 1.1 and K V 1.2 are located under the myelin sheath in the juxtaparanodal region of axon 10 12 (Figure 1. 2 ). Figure 1.2: Voltage gated sodium and potassium channel distribution at the peripheral axon node of Ranvier. Immunofluorescence labeling of Na V channels (green) concentrated at the node, myelin axon junction protein CAM Caspr (blue) located paranodally, and K V 1 potassium channels (red) located in the juxtaparanode. Scale bar 10 m. (Image: Schafer and Rasband, 2006 12 ) A peripheral nerve interface to adequately control prosthetic hand technology referenced above would require minimally invasive axon communication with high selectivity and specificity in order to achieve precise signals for motor control (individual digit command) and localized tactile feedback. Existing peripheral nerve
4 interfaces fail to meet these criteria. Cuff electrodes, such as the flat interface nerve electrode (FINE) employ an electrical communication to the nerve from the periphery, while flatt ening the cross section of the nerve to improve access to axons within the bundle. The FINE has made strides on the stimulation (read in) front, by demonstrating the ability to stimulate groups of axons independently to individual muscles in the human low er extremity 13,14 and sensory pe rcepts in the human upper limb 15 H owever, the device is not suited to read out neural activity due to the non fe asibility of deciphering minute electrical signals within a nerve, from peripheral (extra epineurial) electrodes. Other designs include arrays of needle electrodes such as the Utah Slant Electrode Array (USEA) 16,17 which impale s the nerve to place electrodes across the cross section of the nerve. While suffering from tissue damage and biocompatibility issues, such an array still does not achieve desirable selectivity/specificity. Singular needle like electrodes have been implanted within nerves to record and stimulate neural activity, such as the longitudinal intrafascicular electrode (LIFE) 18,19 and the transverse implanted multichannel electrode (TIME) 20 While these tools may be useful in some experimental situations, they only provide communication with small sub regions of nerve, and are not suitable for longer term interface applications. The recently emergent field of optogenetics offers a unique set of tools which may offer an imp roved solution to the peripheral nerve interface. This interdisciplinary field of advanced optical techniques and genetic engineering involves the monitoring or stimulation of light sensitive proteins expressed within specific cell types. In contrast to electrode based techniques of neural recording/stimulation which cannot decipher signals from single neural pathways (rather they provide crude aggregate information
5 from many units), optical interrogation can specifically query single neurons or neuron pr ocesses because the spatial resolution of modern optical systems is well within the size of these features. Neural activity can be monitored by optically detecting fluorescence transients that are coupled to action potentials. There are a number of prob es that could potentially provide this coupling, the mo st promising candidate being a genetically encoded calcium indicator (GECI) such as GCaMP 21 23 which increases its fluorescence in response to elevated intracellular calcium concentration associated with action potentials. GCaMP consists of circularly permuted green fluorescent protein (cpGFP), with the calcium binding protein calmodulin (CaM) and the CaM interacting M13 peptide inserted into the GFP sequence. In the calcium unbound state, an opening in the GFP beta barrel results in solvent quenching of the chromophore. When CaM binds calcium, a conformational change of the CaM M13 complex restricts solvent access to the chromo phore and thus increases its brightness. To date, GCaMP sensors have been widely used to study neural cir cuitry in in vivo animal models, and have been used in conjunction with two photon excitation for neuronal activity imaging 800 m deep in the mouse brain 24 The latest generation, GCa MP6 25,26 achieves reliable single action potential detection, with signal brightness exceeding that of synthetic calcium indicator dyes such as Oregon Green BAPTA 1. There has been effort to create red shifted GCaMP a nalogues using red fluorescent proteins, which are excited at wavelengths that can penetrate deeper into tissue due to reduced scattering and absorption, with a consequent reduction in phototoxicity. Additionally, a red shifted probe could increase spectr al separation with co expressed optogenetic proteins such as ChannelRhodopsin2 (ChR2, below) for dual wavelength systems. The latest
6 generation of red protein calcium indicators such as the mRuby based RCaMP, and the mApple based R GECO, are beginning to rival GCaMP in performance 27,28 and have peak two photon absorption spectra at ~1040n m which overlaps well with cost effective and powerful fiber lasers. These latest generation red GECIs generate response amplitudes (dF/F o ) comparable to GCaMP6 probes, but are still inferior in their overall dynamic range and absorption cross sections A probe which monitors electrical activity by directly measuring membrane potential rather than a secondary signal such as calcium is naturally highly desirable. A direct voltage read out can provide sub threshold potential monitoring that secondary as measured with direct voltage read out. Genetically encoded voltage indicators (GEVIs) have developed greatly in recent years 29 31 but still face major limitations which hamper their use in a peripheral nerve interface. Sensitivity, dynamic range, and signal to noise ra tio of GEVIs are significantly inferior to that of GECIs, and detecting measurable in vivo signals is still a challenge. Furthermore, detecting and using action potential trains for neural control with a GEVI may be unfavorable in some respects in compari son to GECI read out for the following related reasons: (1) The GEVI signal, if that does not increase in amplitude (brightness) with the sustainment of action potent ial firing, whereas a GECI mediated signal is analogue and compounds to form a neural activity (pulse train frequency) with a GEVI, a time window to integrate spikes would be required, whereas the GECI signal is inherent ly an
7 amplitude proportional to numbers of action potentials Data collected with GCaMP sensors have illustrated the proportional relationship of signal amplitude to number of burst action potentials 25,32 while another study has demonstrated the linear summation in calcium amplitude with action potential frequency 33 In addition to optogenetic rep orters, there exists a class of optogenetic effector proteins, which can actuate molecular processes within cells in a light driven fashion. ChannelRhodopsin2 (ChR2) is a widely used tool in neuroscience today, for optical stimulation of neurons. This no n specific cation channel opens under blue light activation, depolarizing the membrane and causing action pot ential firing, given sufficient light stimulation. While ChR2 has largely been used for studies in the CNS 34 38 its ability to stimulate peripheral nerve axons has also been demonstrated, under activation via an optical cuff 39,40 The opt ical cuffs in these studies deliver ed blue light indiscriminately to sciatic nerve fibers by two distinct designs. In the first study, 16 small LEDs were embedded in a glass cuff to circumferentially illuminate the nerve from all directions. In the other, an optical fiber delivered light to the nerve cuff which encased the nerve in a reflecti ve aluminum sheet to uniformly illuminate the nerve. Efforts are being made to create new ChannelRhodopsin variants with improved light sensitivity and channel kinetics (for increased firing frequencies) 41 as well as red shifted proteins 42,43 Such probes enable light driven stimulation of neural activity, and could potential ly be incorporated in an optogenetic neural interface to deliver sensory information. In an optogenetic peripheral nerve interface, optical penetration will be a limiting factor determining the portion of axons within a nerve bundle that can be
8 queried. A dvances in multi photon microscopy have improved the tissue depths at which it is possible to record. Denk et. al 44 demonstrated two photon fluorescence to a depth of over 1000um in mouse neocortex u sing a Ti:Al 2 O 3 regenerative amplifier while deep tissue calcium recordings from GCaMP3 have been resolved using multi photon microscopy 22,24 A recent study 45 which employed three photon microscopy in vivo using a fiber based excitation source was able to obtain high contrast images of fluorescently labeled vasculature up to 1,300um deep in the intact mouse brain. See Chapter 5 for a more comprehensive review of multi photon studies. In order to use ca lcium detection for activity read out in a peripheral nerve interface, an action potential coupled change in axonal calcium concentration must be present. It is well established that activity dependent calcium transients occur in the soma of a neuron, as well as the dendrites and axon terminals (F igure 1. 3 ) These transients in calcium signal important functions such as gene transcription, signal integration and neurotransmitter release. However, there is less literature demonstrating activity dependent intracellular ca lcium rise along the nerve ( axon) itself. Chiu et. al 46 has shown action potential induced calcium influx in rodent optic nerve axons, occurring in a uniform f ashion over the axon. The same lab demonstrated an activity dependent c alcium signal occurring at the n ode of Ranvier in frog sciatic nerv e axons 47 In 2015, Grundemann and Clark 48 reported action potential dependent calcium influx in CNS Pu rkinje cell axons, also occurring locally at nodes of Ranvier. How ever, until very recently there had been no published data showing activity dependent calcium transients in myelinated axons of the mammalian peripheral nervous system In work done in parallel with the research presented in this thesis,
9 Zhang and David 49 also report stimulation induced calcium influx in r odent peripheral axon nodes of Ranvier, however the two bodies of work differ significantly in their experimental approaches and focuses, with novel characterization of neuroprosthetic related signal dynamics being a central aim of the present work Zhang and David 49 characterized calcium responses to prolonged stimuli and were limited to single large diameter axons. The present s tudy considers a wide range of short er duration signals which may be more physiologically relevant to neural interfacing; signals extend ing from single action potentials to steady state trains while also monitoring a bulk distribution of variable size axons within the same nerve preparation Figure 1. 3 : Schematic of a motor unit. Activity dependent calcium signals are known to occur at the motor unit cell body and axon terminals (red circles), but these locations are impractical for a prosthetic limb neural interface. Calcium dependent neural detection is desired at the axon (green circle). Image schematic credit: Sensory Motor Performance Program, Rehabilitation Institute of Chicago.
10 There are a number of mechanisms that could potentially cause an axonal calcium ris e. It has been suggested that intra axonal calcium is coupl ed with sodium. A study in mouse optic nerve axons 50 presented data on this coupling, showing sodium dependent calcium elevation not in a manner consistent with voltage gated calcium channel infl ux, but rather calcium influx via the reverse mode plasma membrane Na/Ca ex changer. Many studies have investigated the role of axonal calcium in neurodegenerative diseases, particularly, the calcium dysregulation that occurs in response to energy failure and/or injury. Calcium overload can engage a positive feedback cycle of fu rther calcium release, and activate calcium sensitive degradative enzymes. Numerous mechanisms have been implicated in the calcium overload of disease state axons 51 56 A study 57 which investigated intra axonal calcium release during ischemia, report s that this pathological calcium elevation is largely dependent on release from intracellular stores: via the mitochondrial Na/Ca exchanger, and endoplasmic reticulum via ryanodine receptor mediated release, and IP 3 signaling by phospholipase C. It is sugge sted that these three pathways may act in concert to modulate the release of calcium in the ischemic state. Furthermore, it has been shown that intra axonal calcium release by endoplasmic reticulum, mediated by ryanodine receptors and IP 3 receptors, contr ibutes to the degeneration of axons in white matter (myelinated axon) injury 58 While this research of axonal calcium as it pertains to disease and injury is not the objective of this thesis research, these studies point out the existence of axonal calcium release/flux mechanisms which might be relevant in act ivity dependent calcium signaling. Intracellular organelles that contribute to cytosolic calcium elevation
11 in disease states cou ld potentially give rise to activity dependent transients. Calcium induced calcium release from ryanodine receptors on ER coul d feasibly occur in response to the action potential if a n accompanying calcium influx is present It is also possible that membrane depolarization during the action potential could initiate a signaling cascade in which IP 3 is activated and causes calcium release from IP 3 receptors on ER This would require the activation of phos pholipase C and the presence/hydrolysis of the membrane associated phospholipid PIP 2 to produce IP 3. In addition, it could be speculated that action potential asso ciated sodium increase within the cytosol could initiate calcium extrusion from the mitochondrial Na + /Ca 2+ exchanger, a mechanism which is broadly suggested to be a pathway of mitochondrial calcium efflux in many tissues 59 Indeed, endoplasmic reticulum release and mitochondrial Na + /Ca 2+ exchange are concluded to be contributors to the activity de pendent calcium in frog sciatic nerve 47 while other studies in rodent axons point to T type voltage gated calcium channels, and plasma membrane Na + /Ca 2+ exchanger (NCX) as prominent influx pathways 48,49 Transmembrane influx pathways such as voltage gated calcium channels and the NCX are the simplest potential pathways fo r an activity dep endent calcium transient, as depolarization of the membrane leads to direct influx without intermediate signaling steps. Studies discussed in this chapter have experimentally and pharmacologically suggested the presence of certain calcium handling mechanisms, but there is minimal labeling (i.e. immu nofluorescence) data showing such mechanisms. The Todorovic lab has discovered and confirmed by immunofluorescence that one isoform of the T type calcium channel (Ca V 3.2) is functionally expressed in nociceptive axons, largely in
12 unmyelinated C type fibers 60 In 1997, Steffense n et. al. showed immunoflurescent reactivity of type 1 Na/Ca exchanger (NCX1) in peripheral axons 61 However, there is an overall absence of literature on the p resence of calcium channels in myelinated axons. Research Aims My research aims to develop a novel neural interface tech nique that overcomes the limitations of existing technologies a nd enables high resolution neural detection using an optogenetic approach. The hypothesis that drives this research is that an optogenetic peripheral nerve interface can prov ide a minimally invasive bi directional (motor and sensory) communication system with high (single axon) selectivity and specificity to control advanced prostheses. While prosthetic limb control has served as the primary motivation for this research, this development could have far reaching impact in other clinical areas such as clos ed loop control/study of organ function (i.e pancreas, bladder), visual neuroprosthesis in retinal blindness, and functional stimulation in ne uromuscular rehabilitation. The work discussed in this document will largely address the motor command (read out) side of this biological i nterface. In an in vitro rodent nerve preparation, I have investigated the po ssibility of using fluorescence based calcium detection to read out action potential activity in peripheral nerves. Indeed, our group has been among the first to show an activity dependent calcium signal in mammalian peripheral nerve axons. The aims of my thesis strive to establish calcium detection in peripheral nerve axons as a worthwhile approach to selective/ specific motor command read out, and
13 inves tigate the underlying physiological mechanisms and pathways of this calcium signaling. Aim 1: Investigation of activity dependent calcium signaling in peripheral nerve axons This was the cardinal question; at the outset of my research in 2011 on the possibility of calcium signaling phenomenon was yet known to exist in the mammalian peripheral nervous system. Aim 1 was to investigate whether there was a detectable calcium event conducive to fluorescent (optogenetic) monit oring of neural activity. Aim 2: Characterization of calcium signal dynamics The goal of Aim 2 was to characterize the dynamics of the biological calcium signal in order to understand and evaluate its utility in a neuroprosthesis. I investigated the response of the axon loaded fluorescent calcium indicator to frequency and duration of stimulus (action potential) trains. Specifically, I sought to characterize the gradation of signal amplitude as a function of the stimulus frequency, characterize the spatial behavior of th e calcium relative to the node, and test/quantify the calcium response to small stimuli (down to the si ngle action potential). Aim 3: Investigation of calcium sources/mechanisms The g oal of this phase of research was to identify the sources and mechanisms of the action potential induced calci um response characterized in aims 1 and 2. Involvement/existence of the following pathways were considered: 1.) Calcium influx from extracellular space (mechanisms: L,T type Ca V Na/Ca exchanger (NCX))
14 2.) Intracellular c alcium release from mitochondria (mitochondrial Na/Ca exchanger) 3.) Intracellular c alcium release fro m endoplasmic reticulum (ER) (possible mechanisms: ryanodine receptors (RyRs) and inositol triphosphate (IP 3 ) receptors) Figure 1. 4 : Schematic of an axon node of Ranvier showing the general components of calcium response investigation: (1) Calcium influx, (2) Mitochondrial Na/Ca exchanger, (3a) ER ryanodine receptors, and (3b) ER IP 3 receptors Aim 4: Multi photon imaging depth study Relatively superficial optical depth of imaging in peripheral nerve tissue due to highly scattering myelin is a current limitation for axon accessibility. In this study, I aim to demonstrate significantly improved imaging depth in peripheral nerve tissue using a high pulse energy, near IR laser for two photon excitation. The impact of this l aser technology on our goal of maximizing imaging depth lies in its capability of achieving short pulse widths ~150fs to produce very high pulse energies, with low pulse
15 repetition rate to keep average power manageable, at a highly penetrating near IR oper ational wavelength. I collected deep tissue images/me asurements and also quantified the differing scattering length constants of nerve (white matter) and brain (grey matter).
16 CHAPTER II ACTIVITY DEPENDENT CALCIUM SIGNALING AND DYNAMICS Introduction The following chapter will discuss the work of aim 1 and aim 2, as stated in Chapter 1. The motivation behind these aims was to establish the presence (or lack thereof) of an activity dependent calcium signal in mammali an peripheral nerve axons, and characterize the dynamics of such a signal within the context of neuroprosthetic control. The experimental design that was ultimately developed to detect and characterize calcium signals within nerve axons was the product an d culmination of much trial/error over a period of time. The experimental techniques required were either minimally established or unestablished, and thus many facets of the experiment were refined and developed over the course of this research. Processe s that were iteratively improved or developed throughout the research include: loading a synthetic calcium indicator into myelinated nerve axons, maintaining nerves in optimal health (conducting action potentials) in vitro for 4 7 hours, and stimulating an d verifying action potential propagation in the in vitro nerve preparation. Various approaches and experiments that were employed over the first couple years of thesis research in route to a successful approach will be discussed sparingly. And the experi mental design and methods that eventually enabled the routine elicit ation and record ing of calcium signals will be described in detail under Methods. Load ing a calcium indicator into nerve axons was the first challenge in investigating calci um signaling. Often, the simplest way to load cells with a fluorescent
17 dye is by using an acetoxymethyl (AM) conjugated molecule, which is cell permeable until hydrolyzed by intracellular esterases which activate the fluorescent molecule and retain it wit hin the cell by rendering it non membrane permeable. The simplicity of a bath applied, trans membrane approach was highly desirable and this method was strongly attempted. However, the axon enveloping myelin sheath proved to be too great a barrier for me mbrane permeant loading. Molecules such as Fluo 4 AM and Oregon Green 488 BAPTA 1 AM (In vitrogen) were tested in this ve in. When it became apparent that this loading approach was not feasible, the idea of loading non membrane permeable dye molecules via diffusion into the nerve axon cylinders from a cut end was considered. This required a tight fitting custom made glass suction pipette to deliver the dye into axons by drawing a cut end of nerve into the dye containing pipette. Initially, experiments were performed with peripheral nerves (sciatic and tibial), as well as the optic nerve because at that time there was published data showing calcium signaling in this nerve using a similar approach 46 In the first phase of this strategy this was attempted with dyes such as the membrane impermeant Oregon Green BAPTA 1, Rhod 2, flu orescein sodium salt tracer, as well as the Fluo 4 AM Axon labeling was achieved with varying degrees of success, ranging from no labeling to fairly robust labeling. Numerous methods were deployed to test whether a calcium change (induced either non physiologically or physiologically) could be detected wi th dyes loaded in this manner. To first test for intra axonal calcium dye functionality, I attempted to elevate intracellular calcium concentration by high K membrane depolarization (high concentration KCl) and application of the calcium ionophore, ionomy cin. Clear, unambiguous changes in calcium fluorescence were never
18 to these reagents, and/or the intracellular dye conditions were not suitable to report a calcium chan ge. I also attempted to elicit responses with electrical stimulation of the nerve. The dye loading suction pipette was outfitted with electrical leads and served as the stimulating electrode. To have a confirmation of action potential generation and p ropagation in the nerve, I included an innervated muscle to the sample preparation when testi ng peripheral nerve tissue (F igure 2.1). I excised the sciatic nerve along with its tibial branch and flexor digitorum brevis (FDB) muscle. The tendons of the mu scle were pinned to the chamber dish, and muscle twitch in response to sciatic nerve stimulation was monitored. Figure 2.1: Early phase nerve loading/stimulation/imaging schematic involving muscle for twitch verification of action potentials As with the high K and ionomycin experiments, no robust calcium fluorescent signals could be unambiguously detected with these dyes. In a few instances, small elevations in signal could be observed in correlation with the electrical stimulus, but such signals wer e not routinely repeatable.
19 Figure 2.2: Early phase axon labeling with non dextran conjugated calcium indicators. Experiments with longitudinally (pipette) loaded axons with Fluo 4AM (left) and cell impermeant Oregon Green BAPTA 1 (right) did no t label axons as well as dextran conjugated dyes. Scale bars : 10 Figure 2 3 : Early phase experimental data which failed to yield a calcium associated fluorescence transient. A nerve/muscle was prepared as shown in ( Figure 2.1 ), co loaded with the red spectrum calcium indicator Rhod 2 (A) and green fluorescein tracer (B). Fluorescein appeared to label the axon strongly while the Rhod 2 labeling was weak. (C) Tran s verse intensity profiles across the node of Ranvier are pi ctured in A & B. The dotted line c561 2 indicates the intensity profile at peak electrical stimulus. (D) The signal change along the transverse profile for the calcium sensitive Rhod 2 channel does not yield a signal change in the intra axonal pixels. Apparent signal change on the edge of this profile is due to a motion artifact caused b y muscle twitch in the chamber.
20 It was when I used a dextran conjugated dye molecule that I was able to achieve robust calcium signals in response to electrical stimulation. The 10,000 MW dextran is a large hy drophilic and relatively inert molecule which is conjugated to fluorescent dyes largely for the purpose of cellular retention. The dextran conjugated calcium dye (Calcium Green 1 Dextran, Thermo Fisher) often yielded bright, visually apparent fluorescence changes in response to electrical stimulation. It is not certain why the previous experiments using non dextran conjugated dyes were unable to produce a strong calcium fluorescence, given the fact that some of these dyes had labeled the axon, and have similar optical and calcium sensing properties to the Calcium Green 1. However, a s may be observed in Figure 2.2 and 2.3 the axon loading with non dextran conjugate d dyes was weak in comparison to labeling achieved with the dext ran conjugated indicator ( Figure 2. 8 ) The evident improvement in cellular labeling and retention observed with the dextran conjugated molecule, possibly aided by i ncreased axoplasmic transport, may likely have contributed to its success. In hindsight, substantially higher concentrations of the non dextran molecules could also have been at tempted, however solubility/ DMSO restraints severely limited the feasible AM dye concentration. The following section describes in detail the general experimental protocol that was used to detect and study the axonal calcium signaling: Methods Nerve Preparation The sciatic nerve and its tibial nerve branch is excised from adult wild type mice, and loaded from the tibial end with a synthetic calcium indicator (2 mM Calcium Green
21 1 Dextran, ex/em = 506/531 nm) and a calcium insensitive dye (1 mM Rhodamine B Dextran, ex/em = 570/590 nm) dissolved in a bu ffer containing 130 mM KCl and 30 mM MOPS, pH 7.2. (The Rhodamine B is co applied to provide bright axon labeling, as the baseline cytosolic Calcium Green 1 signal is relatively low). The tibial end is freshly cut in a zero calcium buffer (Mouse Saline w ith Ca 2+ replaced with Mg 2+ and 1mM EGTA added) to ensure open axon cylinders before being suctioned into a tight fit electrode with the dye buffer to facilitate longitudinal axonal dye loading via diffusion and/or axoplasmic transport. The suction elect rode on the tibial nerve also serves to record electrical activity within the nerve. The sciatic end of the nerve is drawn into a suction electrode for stimulation of compound action potentials (CAPs). The nerve sample is stored intermittently in a mouse saline solution (in mM: 126 NaCl, 5 KCl, 1.8 CaCl 2 1 MgCl 2 10 MOPS Buffer pH 7.2, 30 glucose) during experimental preparation and is 4 20 NaHCO 3 1.2 NaH 2 PO 4 2 CaCl 2 30 glucose ) bubbled with 95%0 2 /5%CO 2 for the duration of the two hour dye loading period and subsequent imaging/electrophysiology. The relatively high glucose concentration of 30mM used in these solutions was the upper bound concentration in a group of studies which tested ambient glucose on glycogen and axon function 62,63 Their data show that this concentration is not likely to have deleterious effects, and in fact supports sustained action potential propagation ( CAP amplitude) by the concomitant buildup of Schwann cell glycogen reserves.
22 Electrophysiology CAPs are generated and recorded throughout the experiment (MultiClamp 700B Amplifier, Axon Instruments, PCLAMP 10 Software) using 50 s square pulses to confirm and monitor good nerve viability. The stimulation voltage threshold for maximum CAP amplitude is determined, and a modestly supra threshold voltage is subsequently used to elicit calcium signals. Partial CAPs can be elicited with as low as 1 volt or less, and experimental voltages did not exceed 10 volts. Maximum CAP amplitudes varied between 4 8 mV with variability likely attributed to the degree of intactness of the nerve as well configuration within electrodes T he recorded amplitude would substantial ly diminish when the recording electrode contained the high potassium dye loading buffer. Figure 2. 4 : (Top) Experimental configuration of nerve loading, electrophysiology, and confocal imaging setup. (Bottom left) Custom fabricated suction microelectrode for nerve loading and stimulation/recording. (Bottom right) Image of nerve preparation depicted in top schematic.
23 Ele ctrode and Chamber Fabrication E lectrodes were custom fabricated; capillary glass was heat extruded and the annulus custom sized to the nerve diameter with a diamond scribe and fire polishing. The elec trical leads were made from T ef lon coated silver wire, the ends of which software (Figure 2.5) and 3d printed. No. 1.5 optical glass was fixed to the c hamber with a silicone adhesive, and ferritic steel plates were epoxied to the top surface of the chamber for magnetic fixation of the electrodes. Figure 2.5: SolidWorks rendering of chamber used for nerve electrop hysiology and optical recording, shown without optical glass and steel plates. Optical Imaging/Recording Dye labeled axons were imaged in a region of nerve near the tibial electrode. The nerve was gently pressed down to the imaging on an inverted microscope. The harp device was constructed as a U shaped steel wire with silk string s spanning the arms of the Testing of CAP amplitude with and without the harp resting on the nerve indicated that the device did not damage the nerve or inhibit action potential propagation. Fluorescence imaging was performed on a spinning disk confocal microscope (Intelligent Imaging Innovations, Marianas)(Figure 2. 6 ). A 515nm laser line was used
24 to excite the Calcium Green 1 and a 561nm line was used for the Rhodamine B tracer. Pixels were binned (2x2) to improve the frame read out time for fast imaging. To record calcium transients, timelapse images were acquired at 12 20Hz, during which the nerve was stimulated by an electrical stimulator triggered via TTL pulse from the microscope. Fluorescence was imaged onto an EMCCD camera (Photometrics Evolve) through a 525/50nm emi ssion filter for the Calcium Green 1 channel and 617/73nm for the RhodamineB channel. Images were collected with a 63X, 1.4NA oil immersion objective lens. Image data was extracted from SlideBook6.0 software and analyzed and pr ocessed using cus tom MatLab script Figure 2. 6 : Spinning disk confocal microscope (Marianas, 3i) along with PCLAMP, amplifier and electrophysiological equipment for action potential generation and recording.
25 Signal Detrending and Quantification Optical signal recordings inherently contained decay due to photobleaching. This photobleaching signal component was removed for more accurate quantification of signal amplitudes, and presentation (example, Figure 2. 7 ). The raw signal is normalized t o the pre stimulus level. S ignal with the stimulus induced transient portion removed (i.e. pr e stimulus data points and post stimulus /recovery data points ) is fit with a double exponential function which is then extrapolated over the whole signal. The exponential fit is subtracted from the signal to yield the detrended signal. Signal amplitudes are calculated from the subtracted signal as the mean amplitude of peak signal minus mean of ten data points pre stimulus baseline. Figure 2 7 : E xample of photobleaching decay removal. The raw optical recording containing photobleaching is normalized to pre stimulus level and fit with a double exponential function. The exponential fit is subtracted to remove the underlying photobleaching signal c omponent
26 Results Axons were robustly labeled by this dye loading technique. Additionally, the dextran conjugated dye appeared to retain well within the axon cylinders and little, if any, dye leakage occurred during these experiments, as observed by the fact that the baseline axonal fluorescence in both the RhodamineB and Calcium Green 1 channels did not diminish with time throughout experiments. Rather the brightness in both channels would increase throughout imaging sessions as the dye containin g electrode remained in place, and dye continued to load by diffusion and/or axoplasmic transport for the duration of an experiment. Because the dye was loaded from the nerve end, the intra axonal dye concentration and thus baseline brightness was strongl y correlated to the distance from the electrode pipette. This gradient in dye concentration had to be considered when searching for a region of nerve in which to image a calcium response. When the Calcium Green 1 concentration was too high, any calcium t ransient would be baseline state fluorescence, precluding detection of any calcium change. Conversely, a small dye concentration would not enable detection of a calcium e vent above background noise. Figure 2. 8 shows an example of well loaded tibial nerve axons in both the RhodamineB and Calcium Green 1 channels, in which at least six nodes of Ranvier are present in the field.
27 Figure 2. 8 : Mouse tibial nerve axons loaded via pipette diffusion / axoplasmic transport, rhodamineB channel (left) and Calcium Green 1 channel (right). Activity dependent calcium elevation was observed at axon nodes of Ranvier. Figure 2. 9 illustrates an example of an action potential elicited calcium response, and the associated fluorescent signal. The axon propagates 50 action potentials, delivered at 100 Hz by electrical stimulation (designated by lower black bar), and the resulting inc rease in calcium associated fluorescence is visually apparent at the node illustrating a sharp increase in fluorescence at the onset of stimulus, and abrupt decay upon cessation of the stimulus. This relatively modest signal amplitude shows an approximat ely 11% change in fluorescence in response to the short burst of action change in this experiment is influenced by the dye diffusion/transport aspect of the experiment which dynamically changes the dye/calcium buffering dynamics throughout the experiment, and thus it is possible to obtain quite different signal amplitudes depending on the dye conditions.
28 Figure 2. 9 : Illustration of an action potential elicited calcium fluorescence res ponse to 50 action potentials (100Hz). Top panel images show an individual node of Ranvier before the action potential stimulus (A), during (B), and after (C). Bottom panel is the quantitative optical signal produced by the node ; the signal is extracted from a region of interest (ROI) directly at the node ( approximately 3 x 2 ) Thick black bar indicates acti on potential stimulus. It was typical to observe multiple nodes of Ranvier responding to action potential stimulation in the same focal plane. Figure 2. 10 is an example of this case: six nodes yield calcium fluorescence changes in response to a 1 s train of action potentials. Here, signal amplitudes among the six nodes range from 11 24%.
29 Figure 2. 10 : Field of tibial nerve axons with at least six nodes of Ranvier yieldi ng a calcium coupled fluorescence change in response to a 1 s train of action potentials (100Hz). Thick black bar denotes action potential stimulus. Inset scale bars: 1 s and 5% signal change. The calcium signal originates directly at the node epicenter and propagates distally and proximally, a short distance into the internode. The maximum signal amplitude diminishes as a function of distance from the node whil e latency of signal onset increases ( Figure 2. 11 ). The signal amplitude follows a quasi exponential decrease with distance, as data from five axons was fit better on average with an exponential fit than with a linear fit. Fluorescence data was used to calculate mean calcium wave propagation velocity ( 33.5 6.1 m/s ) and spatial length constant ( 11.8 1.0 m ) f or a typical half second long action potential burst applied in these experiments (50 APs, 100H z, n=5). The velocity calculated here is in reasonable agreement with
30 measurements of other studies in the cytoplasm of retinal pigmented epithelial cells (15.8 1.7 m/s ) 64 and cilia (14.9 2.1 m/s and 22 3 m/s ) 65,66 Figure 2. 11 : Longitudinal calcium signal propagation. (A) Axon segment with node of Ranvier (scale bar 10m) (B) Activity dependent calcium fluorescence in response to a 50 action potential burst (100Hz) recorded at incremental distances away from the node epicenter (signal trace colors correspond to ROIs in top image). Plotted signals illustrate the latency in signal onset as well as reduced signal amplitude with distance from the node. (C) Peak normalized signal amplitude versus distance from node from five axon nodes. Yellow trace corresponds to nodal data of A and B. Changes in calcium fluorescence could be detected in response to a single action potential (Figure 2.12) While the signal to noise ratio of such signals is low, averaging revealed a signal change of 2.8% .5, n=3 nodes)
31 Table 2.1: Calcium signal propagation velocity and amplitude data from five nodes of Ranvier. Figure 2. 12 : Fluorescence response to a single action potential in two nodes of Ranvier. Five recordings were acquired. The largest amplitude response, the mean signal of the five sweeps, and the signal decay exponential fit with decay constants are shown. As the number of action potentials in a stimulus is increased, the fluorescence amplitude increases accordingly. H owever, given constant action potential frequency as the number of action potentials is increased, the signal amplitude will begin to plateau as it reaches or nearly reaches a steady state Figure 2.1 3 illustrates this property; the signal am plitude increases steeply as the number of action potentials is
32 Figure 2.13 : Data from three nodes in a nerve illustrate the fluorescence amplitude modulation with increasing number of action potentials, from 1 80 action potentials, at a constant frequency of 100Hz. The signal amplitude begins place. Panel (right) shows the amplitude measurements are well fit by a double exponential function (mean R 2 0 .993). init ially increased, but plateaus as the steady state amplitude for that frequency is reached. The figure shows the amplitude approach a plateau for three different nodes as the 100 Hz pulse train is extended to include more action potentials. Additionally, I s place in this process, and as such, collected amplitude data as the number of action potentials was ramped up, and down.
33 As the data show, no significant hysteresis is apparent as the amplitudes of the ramp eviate. In addition to this correlation between signal amplitude and number of action potentials, it was important to characterize the dependence of the calcium signal amplitude on stimulus frequency As the force/velocity of a motor command is neurally dictated by the action potential frequency, it is a useful feature of a motor control signal prosthetic act uator in proportion to frequency modulated calcium could enable intuitive prosthesis manipulation for a user. Nodal calcium responses to action potential trains of varying frequencies were tested for relatively long pulse trains of 2 s down to shorter b ursts of .5 s. Frequencies 1 4 ) and the data showed a linear relationship bet ween amplitude and frequency This linear relation of calcium fluorescence with action potential frequency is consistent with similar studies in pyramidal neuron dendrites, in which calcium accumulation summated linearly with action potential frequency 33,67 With a 2 s stimulus, signals were recorded for frequencies between 25 70Hz as this would be a typical range of firing for a relatively sustained motor ou tput. However, for short bursts, motor neurons are capable of firing significantly faster, up to approximately 150Hz. As such, t he responses to shorter stimuli were recorded with frequencies ranging from 25 150Hz. The activity dependent calcium signals were again acquired by ramping up the frequency and subsequently ramping down the frequency to test for hysteresis. No significant hysteresis was observed
34 Calcium signals rapidly reach a steady state plateau, where a fluctuation between a lower and upper level takes place. Sakmann et. al 33 provided one model of the summation of calcium transients in trains of action potentials in the following mathematical derivation The model assumes linear superposition of transients and constant a ction potential frequency. W ith single action potential transient amplitude ap the model predicts mean [Ca 2+ ] i summation above the unit amplitude for ap T he [Ca 2+ ] i above baseline at the beginning of the (n+1) th action potential is given by: (eq. 1) The steady state is reached when the decay of [Ca 2+ ] i is just compensated by the upper steady state [Ca 2+ ] i amplitudes are calculated from eq.1 as n (eq. 2) In the steady state, the mean [Ca 2+ ] above baseline is given by the time dependent integral during a stimulus interval: (eq. 3)
35 The work by Sakmann et. al. provides a crude model of the system, but does not comprehensively describe certain asp ects of the calcium phenomenon and could be built upon. Their equations do not fully describe the transition between the rising phase of [Ca 2+ ] i and the steady state. Furthermore, there may be more complex dynamics occurring during the final decay of a summated calcium build up, as the decay constants ( dec ) observed in the present work (i.e. Figure 2.1 4 Figure 2.1 7 and Figure 2.18 ) are significantly greater than the unit action potential decay (Figure 2. 12 ). However, t he linearity of calcium fluorescence with stimulus frequency which we observe is in accordance with the linear characterization derived in eq. 3. (Theoretical slop e is plotted with experimental slopes in Figure 2.1 6 ). Taken with the assumption that the activity dependent calcium increase comes from influx of action potential associated calcium quanta and not calcium induced calcium intracellular release, as our data indicates (Chapter 3), these observations would support the claim that calcium has been monitored in the linear buffering range of the indicator, rather than in its likely concentration (linear) regime of dye indicator buffering, as the intracellular baseline calcium concentration (~100nM) is well below the indicator Kd (~380nM) 68,69 and fluorescence transients from that baseline have occupied a small portion of the available dynamic range of the indicator (< ~ .3 and 14 respectively). Another piece of evid ence supporting that these experiments were performed in the linear indicator buffering range is the data of linear frequency modulation for non steady state stimuli. In addition to the linear frequency dependent data presented in Figure 2.1 4 in which a mplitudes were taken from sustained, steady state signals, the frequency amplitude
36 relationship for shorter, non steady state stimuli of .5 s was also observed to be linear (Figure 2. 1 5 ). Frequency modulated data collected in different points of the sat uration curve shown in Figure 2.1 3 would be not be expected to both be linear, if the dye was in a non linear (saturating) regime of calcium buffering. The u nit action potential amplitude A and decay quantities described in the previous model are inherently dependent on the calcium buf fering environment. This will be impa cted by both endogenous buffers and the calcium indicator itself. While indicator buffering could change by small factors over the course of experiments due to the dynamic dye loading, the timescale of this transient is relatively large and would not be expected to significantly alter A or This fact is supported by the lack of hysteresis like e ffec ts in time dependent amplitude data (Figure 2.1 3 Figure 2.1 4D ) Although single action potential amplitude and decay measurements were not regularly obtained for nodes used to collect frequency modulation data (Figure 2.14), approximations of mean sample in Figure 2.13 are reasonably close to observed mean amplitudes for given frequencies (
37 Figure 2.1 4 : Activity dependent calcium signal amplitude is modulated by action potential frequency. (A) Graded calcium signals in response to a 2 s stimulus, and (B) to a shorter .5 s stimulus. (C) Frequency modulated calcium fluorescence traces with bars indicating mean stimulus period amplitude. (D) Stimuli/recordings were performed whi le stepping up action potential frequency as well as stepping down the frequency, showing no significant hysteresis effect. (E) Signal amplitude versus frequency for 3 nodes (3 nerve samples) as recorded in panel C, and their mean (F) illustrate linear de pendence with a slope of .07% fluorescence/Hz.
38 Figure 2 .1 5: Fluorescence amplitude versus frequency for .5 second stimuli is linear. (A) Four independent nodes, and (B) Mean. The frequency modulation slope of a given node could not be correlated with its size, based on analysis from six nodes (from six nerve samples) of nodal diameter, paranodal diameter, and paranodal/nodal ratio (Figure 2. 1 6 ). Varying amplitude frequency slopes may be attributed to differing unit influx quantity and/or variable dye conditio ns.
39 Figure 2.1 6 : (A) Frequency modulation slopes from six independent sample nodes and the Sakmann et. al. theoretical slope, calculated with single unit parameters of (Figure 2. 13 ). Slopes versus nodal diameter, paranodal diameter, and paranodal/nodal ratio, do not reveal correlation between nodal size and slope.
40 Figure 2. 1 7 : (Left) Mean decay for small signals ( shown in Figure2.1 4 E and Nodes 1 3 in Figure 2.1 6A ) demonstrate little correlation between decay time and stimulus frequency. (Right) Some l arger amplitude data sets show dependence on decay time and stimulus frequency. Blue, red and green lines correspond to nodes 4,2 and 1 respectively in (Figure 2.1 6A ). Additionally I considered how the signal decay constant ( decay ) is affected by stimulus frequency. When signal amplitudes are modest, the data showed insignificant correlation between these two parameters However, some datasets that had large signal amplitudes did demonstrate some degree of correlation between decay time and signal size (Figure 2.1 7 ) The reasoni ng for this inconsistency is not entirely clear, however one possibility is that the axon calcium extrusion capability is overloaded by high calcium accumulation in some cases, possibly due to non optimal and var iable in vitro nutrie nt support (i.e. ATP deficit) Another important property of the calcium signal is its modulation by pulse train length While this property was expected, it was important to confirm. In the context of neuroprosthetic control it will be highly useful for the driving calciu m signal to subsist for the full duration of a stimulus (action potential command) and to decay to baseline when stimulus is not present. This further establishes this type of signal command as a direct and intuitive neural read out and control method. F igure 2. 1 8 shows calcium fluorescence responses to constant frequency action potential trains of graded duration
41 from .5 to 2 seconds, illustrating the fluorescence persisting for the duration of the stimulus, with amplitude holding relatively constant. Mean decay constants were not significantly different among any of the four pulse durations (ANOVA, p=.5161). Figure 2. 1 8 : (A) Modulation of calcium fluorescence by action potential train duration. Constant frequency (125Hz) action potential trains are applied at .5, 1, 1.5 and 2 s (colored bars, bottom), and the calcium response duration closely follows the stimulus while preserving steady amplitude. (B) Mean decay constants from three independent sampl e nodes at each pulse duration. Discussion This chapter has presented work in developing an experiment to elicit and test activity dependent calcium signals and the dynamical characterization of such signals.
42 By longitudinally loading mouse peripheral nerve axons with dextran conjugated fluorescent molecules (Rhodamine B and Calcium Green 1) via suction electrode, we were able to elicit and study axonal calcium transients induced by action potentials. These activity dependent calcium elevations originate in axons at nodes of R anvier and propagate from the epicenter at an average of 33 .5 m/s (n=5 ) as the calcium amplitude decays. Calcium signals were detected from single action potentials, and the signal amplitude was characterized as the number of action potentials in a stimulus train is increased. The data show a consequent increase in calcium wit h action potential number and eventual plateauing as the ampl itude begins to saturate. We have also demonstrated the frequency modulated nature of these signals, and shown a linear frequency amplitude relationship in the physiological firing range. The signal persists for the duration of an action potential train and decays upon stimulus cessation; thus the temporal bounds of the activ ity is translated well by the calcium signal. By demonstrating that neural activity in a peripheral nerve can be transduced to an optically emitting signal, we take a step forward in establishing feasibility of optogenetic read out as a means to interfa ce to the peripheral nervous system. Via calcium, we have an action potential coupled optical signal which is proportional in amplitude to the neural activity and which turns on during activity and turns off when the activity ends, with a short decay. Wh ether or not this mechanism can be exploited for useful neuroprosthetic control will depend on optical tissue penetration constraints and the development of a cuff mounted device that can optically scan the neural tissue appropriately.
43 CHAPTER III SOURCES AND MECHA NSISMS OF ACTIVITY DEPENDENT CALCIUM Introduction Within this aim, I investigated possible pathways and mechanisms of the activity dependent calcium signals which are presented in detail in Chapter 2. Previous literature (see review in Chapter 1) has shown that across various scenarios, a number of axonal calcium signaling pathways exist. Some studies suggest trans membrane influx via exchang ers and voltage gated channels, and others have shown that intracellular calcium release can take place via endoplasmic reticulum and mitochondria. Figure 3. 1 : Schematic of an axon node of Ranvier showing the general components of calcium response invest igation: (1) Calcium influx, (2) Mitochondrial Na/Ca exchanger, (3a) ER ryanodine receptors, and (3b) ER IP 3 receptors
44 While axonal calcium signaling is often complex and interrelated, I sought to determine at least the primary mode(s) of calcium elevatio n (F igure 3.1: influx, intracellular organelle release). And furthermore, I hoped to point to molecular mechanisms underlying these modes. Methods The general approach to investigating potential pathways was to apply (or remove) various molecular agents to the nerve wh ich are known to inhibit certain mechanisms, and examine the degree of reduction (if any) of the activity dependent calcium amplitude in this drug state The nerve preparation and calcium indicator loading was performed as described in Chapter 2 Methods In addition to this preparation, the epineurium was cut open in attempt to facilitate drug diffusion from the bath solution into the nerve bundle. Since signals were monitored and quantified for a prolonged period of time in these pharmacological studi es, further attention was paid to the stability and health of the in vitro nerve. Viability of the nerve as a whole was assessed with the CAP amplitude. CAP amplitudes were recorded repeatedly between the time of nerve excision/preparation in the electro de chamber and the end of the incubation/dye loading period. Perfusion of 95% O 2 /5% CO 2 was applied during the loading period as it was in Chapter 2 studies If the CAP amplitude degraded over this time period, then long term signal monitoring was not obtained. Experimental Protocol Activity dependent calcium signals were elicited by short bursts of action potentials; 50 action potentials or less at 100Hz. Signals were generated by the stimulus at inter vals of 2 5 minutes, and this interval was held constant within a given
45 experiment. Regions within the nerve with low to modest dye labeling were selected for time dependent recordings; as mentioned in Chapter 2 the axonal dye loading, particularly near t he loading electrode, would gradually increase throughout the course of experiments, and thus a region with high initial dye content risked saturation and false signal abolition. The axonal fluorescence level in a given region, based on constant laser pow er and detector gain settings, was used to gauge the extent of dye loading, and the necessary levels to keep axons non dye saturated for over an hour became clear. Once a stable baseline signal was established at one or more nodes, the drug was applied by bath exchange with drug containing Mouse Saline The activity dependent calcium amplitudes were monitored for 35 70 minutes after drug addition. As with the CAP amplitude, if a baseline calcium signal amplitude could not be established without deterioration it was not monitored for pharmacological effect. Results To investigate dependence on trans membrane calcium influx, bath calcium was removed. (Ca 2+ was replaced with Mg 2+ and 1mM EGTA added for the calcium free buffer). The action potential elicited calcium signals were largely abolished within 15 minutes in zero calcium environment, and returned to near base line with calcium replenis hed (F igure 3.2). Statistical testing of the data indicate unequal means between Kramer test for multiple comparisons). This da ta strongly suggests an influx component to the signal.
46 Figure 3.2: Extracellular calcium removal causes near abolition of axona l calcium fluorescence transient. (A) Nodal calcium response to 20 action potential bursts (100Hz) in normal calcium, zero calcium, and replete calcium. (B) Summary of n=3 trials L type voltage gated calcium channels are ubiquitous in the nervous system and their involvement in the nodal calcium influx was tested. Nifedipine (10 ), the well known blocker of this class of calcium channels was applied to inhibit L type current 70 Nerve incubation in Nifedipine did not diminish the calcium signal amplitude suggesting other influx pathways (paired sample t test with 95% confidence, p=.5092 ) (F igure 3. 5E ). The T type Ca V channel blocker Mibefradil (IC 50 = 2.7 ) 71 was also applied (25 Interestingly, this blockade caused the near abolishment of some nodal signals while others appeared unblocked. This observation was substantiated by instances where, within the same region in a n erve sample, nodal calcium signals were blocked while other nodes in close proximity remained unblocked (Figure 3.3 A )
47 Figure 3.3: (A) Signal responses from two nodes in the same region of a nerve sample exposed to Mibefradil treatment. The signal from Node1 is abolished over a 70 minute period of incubation, while the Node2 signal remains unblocked. (Stimuli: 50 action potential bursts at 100Hz). (B) Signal responses from two nodes in the same imaging field of a nerve sample exposed to KBR7943 tr eatment. The signal from Node1 is abolished over a 50 minute period of incubation, while the Node2 signal remains largely unblocked. (Stimuli: 50 action potential bursts at 100Hz). Horizontal/vertical scale bars are 1s/3%dF/F o A number of nodes were monitored under Mibefradil application (n=10), and the data appear to show two distinct groups of responses (F igure 3.4 A), one of which is unaffected by the drug, and the other of which is significantly diminished by T type Ca V inhibition. Among the 10 no des tes ted, the groups are split into six which displayed blockage, and 4 which did not, i.e. the data was split roughly into equal groups among the data collected. A statistical two sample t test indicate separate group means (95% confidence, p=9e 10)
48 T he plasma membrane sodium calcium exchanger (NCX) was also investigated as a source of calcium influx. At resting and hyperpolarized membrane potential s the NCX operates in the forward mode which imports sodium into the cell down it s concentration gradient while pumping calcium out of the cell at a stoichiometry of three to one respectively However, during membrane depolarization and intracellular sodium elevation (i.e. during the action potential) the NCX can operate in revers e mode, pumping sodium out and calcium in. A n approximate reversal potential for the NCX can be calculated using typical neuronal ionic concentrations: 1 0 mM and 100nM for sodium and calcium respectively, and extracellular concentrations of 125mM and 2mM respectively. Using the Nernst equation, the equilibrium potentials for the two ion species are E Na = + 63 mV and E Ca = + 124mV. The driving force for three sodium ions is 3(E Na V m ), where V m is the membrane potential, and the driving force for o ne calciu m ion (valence = 2) is 2(E Ca V m ). There will be no exchange when these driving forces are equal, i.e. when 3(E Na V m ) = 2(E Ca V m ) Rearranging this expression gives the reversal potential for the exchanger: Given a resting membrane potential of approximately 70mV, t his valu e indicates reverse mode sodium calcium exchange occurs during substantial depola rization from the resting state; in fact, the reversal potential will become more negative as sodium enters the axon during the action potential. A primary example of reverse mode NCX operation is the calcium transient which occurs during the cardiac action potential in which the activity dependent calcium is dependent on sodium current and the presence of NCX 72
49 Two specific blockers of the NCX were employed to test its contribution. KBR7943 73 and ORM10103 74 are potent inhibitors of the reverse mode NCX (IC 50 =.3 and 1 The results for each drug were similar; again two groups of responses were present in the data. Some nodes were either mildly affected or unaffected, while o thers displayed a significant degree of blockage (Figure3.4 B,C ). As with the Mibefradil treated samples, nodal responses were statistically separated into two groupings, as indicated with a two sample t test at 95% confidence (p=4.3e 4, KBR7943 and p= .0012, ORM10103). From this data, a natu rally arising hypothesis became that one set of axons exhibits a primarily T type Ca V mediated calcium signal while another set has NCX dominated calcium influx. In this scenario, co application of both blocke rs would inhibit all, or at least more, nodes than with either drug alone. Mibefradil (T type Ca V inhibitor) and KBR7943 (NCX inhibitor) were applied in conjunction. However, the result s of this experiment (Figure 3.4 D) did not support this hypothesis, and require alternate explanation. It was observed that no fewer nodal signals were diminished with drug co application. While less clear than single drug applications alone, th e distribution exhibited bimodality, as tested with the As 75 : Collectively our data implicates the involvement of Ca V,T and NCX in peripheral nerve activity dependent calcium transients, but clear roles of these mechanisms and their allocation among axon type warrants additional future work.
50 Size parameters of nodes were compared between groups of nodes which had significantly different amplitude response s to a pharmacologic blocker (Figure 3.5). The data shows little significant size difference between nodes, based on nodal diamet er, paranodal diameter, and paranodal/nodal ratio. In addition to plasma membrane associated mechanisms, intracellular release from ER and mitochondria was tested. The drug Thapsigargin 76 is a commonly used molecule to empty ER of its calcium. It does so by selectively blocking endoplasmic reticulum Ca 2+ ATPase, an ATP driven p ump which sequesters calcium into the ER. Emptying ER of its calcium in theory would block activity dependent calcium signal due to ER release. In numerous trials with Thapsigargin, we did not observe a statistically significant reduction in calcium sign al amplitude (paired sample t test with 95% confidence, p=.4226) (Figure 3. 6 ) To look at the mitochondria, the drug FCCP was first applied. This drug blocks the electron transport chain and thus effectively shuts down mitochondria. Unfortunately, the drug caused fast death of the action potential, and so it could not be used to study the effect mitochondrial inhibition on activity dependent calcium. Instead, the mitochondrial sodium calcium exchanger (mNCX) inhibitor CGP37157 77 was employed. As with Thapsigargin, application of this blocker did not result in signi ficant reduction of activity dependent calcium amplitude (paired sample t test with 95% confidence, p=.431 ) (Figure 3. 6 ).
51 Figure 3. 4 : T type Ca V channel and NCX blockers show inhibition on activity dependent calcium response. (A) T type Ca V channel blocker Mibefradil (25 nearly abolishes the calcium signal in some nodes while other nodes are unblocked. (B) and (C) NCX inhibitors ORM10103 (15 KBR7943 (15 respectively also show differential blockage: a set of nodal signals are significantly diminished while others show little or no blockage. (D) Co application of Mibefradil and KBR7943 surprisingly show similar results/distribution as seen for single drug application. (E) Activity dependent calcium signals were not blocked or diminished by 10uM Nifedipine.
52 Figure 3 .5: Axon size data for the pharmacologically tested nodes in (Figure 3.4). Group1 data are nod es that had significant reduction in calcium amplitude due to their respective blocker, and Group 2 data are nodes which did not show a substantial decrease in calcium amplitude due to the drug. Measurements of nodal diameter, paranodal diameter, and paranodal/nodal ratio were compared between the two groups. Two sample t tests (re sults below each scatter plot) did not show statistically different size parameters between groups at the .05 significance level, with the exception of one category: nodal diameter in the group of Mibefradil and KBR7943 co application
53 Figure 3. 6 : Application of the ER Ca 2+ ATPase blocker Thapsigargin (6 and the mitochondrial Na/Ca exchanger blocker CGP37157 (50 significantly reduce the activity dependent nodal calcium amplitude. As a control for pharmacological experiments involving signal quantification for extended periods of time, signal amplitudes in normal M. Saline bath were monitored for 100 minutes in four independent samples, and amplitudes subsisted without degradation o ver the time period. In addition to this control data, countless nodes exhibited sustained robust signal responses 2 5 hours into the imaging session, including much of the dynamical data presented in Chapter 2.
54 Figure 3.7: Control signal amplitude da ta over a 100 minute period of monitoring from four independent nerve samples does not show degradation in signal over time. (Stimuli: 5 action potential bursts at 100Hz). Discussion In this chapter I have investigated mechanisms involved in activity d ependent calcium transients at mammalian peripheral nerve axon nodes of Ranvier. I loaded nerve axons with a dextran conjugated calcium sensor and studied nodal calcium elevations in response to electrode elicited action potentials while applying pharmaco logic agents to study possible molecular pathways for this signal. Virtually all axons in the nerve are loaded by our experimental technique, and thus we have studied the bulk distribution of axons and have not distinguished between axon types (i.e motor versus sensory). Associated with this experiment were a number of inherent difficulties, and while potential confounding factors were identified and addressed, these are worth mentioning. The dynamic dye loading throughout the experiment makes possible the changing of calcium signal amplitude with time, if the calcium indicator is operating at either low range or very high range concentrations. As mentioned in Methods, this issue is controlled for by selecting axon regions with initial dye content condu cive to long term monitoring, and analyzing data collected in this reasonable dye range. A signal abolition due to dye masking could be easily distinguished from molecular/pharmacologic signal diminishment by prior
55 familiarization of intensity values at w hich dye masking occurs. However, due to relatively long time intervals required to monitor the nerve, small underlying changes in calcium signal amplitude due to dye dynamics are possible, even in intermediate dye ranges. Consequently, precise quantific ation of signal changes over relatively long periods of time is difficult, but significant changes in physiological signal can be deduced. A related factor in the experiment is the diffusion time for successful pharmacologic action at the target (the axon of recording). The epineurium was opened to some degree in these experiments, however the nerve is not fully de sheathed and thus the diffusion pathway for a pharmacologic agent is still substantial and variable between nerves/drugs. So by nature, thi s experiment is limited in its precision, but is adequate for determining major molecular contributors. Removal of extracellular calcium essentially abolished the signal, which returned to near baseline upon reintroduction of bath calcium. This stro ngly suggests there is involvement of calcium influx from the extracellular space, however L type Ca V appear un involved, as Nifedipine application was without effect. We did not observe direct contribution of ER or mitochondrial release as tested with the SERCA blocker Thapsigargin and the mNCX inhibitor CGP37157. Rather, our data suggests that both T type Ca V and plasma membrane NCX are involved in peripheral axon calcium signals. Interestingly, individual inhibition of both these mechanisms resulted in a range of remaining unblocked. These data could suggest multiple axon types within the data, and it would be convenient to speculate that there could be a mechanistic diff erence in
56 anatomically separate. Results of co application of T type Ca V and NCX blockers were counterintuitive however, as they did not show enhanced inhibition from single bl complex, and additional experimentation with larger sample sizes may help to clarify distribution and involvement of these mechanisms.
57 CHAPTER IV PROSTHETIC HAND ACTUATION WITH ACTIVITY DEPENDENT OPTICAL CALCIUM SIGNALS Introduction In the following chapter, I demonstrate the concept of using activity dependent calcium signals presented thus far in this thesis, as a mode of control for p rosthetic device actuation. I have shown that fluorescent calcium transients can provide information on the presence and intensity of neural activity. As such, monitoring these signals in motor axons may allow a command intended for the absent biological actuator, and drive the prosthetic device in of the finger of an electrically powered prosthetic hand is commanded in r eal time by axonal activity, via the optical calcium associated signal. In addition to this real time actuation, graded frequency modulated signals are streamed post hoc to the prosthetic hand to demonstrate varying degrees of motor flexion in response to graded neural command intensity. Methods Nerve P reparation Mouse sciatic/tibial nerves were excised and prepared with a fluorescent calcium sensor (CalciumGreen1 Dextran), as described in Chapter 2 Methods. Neural activity is induced with sciatic ne rve electrode stimulation and a single axon node of Ranvier at the distal (tibial) end is monitored with a spinning disk confocal fluorescence microscope. Images from the microscope/ SlideBook6.0 software (Intelligent Imaging
58 Innovations) were collected an d streamed live at 7 frames/s to custom MATLAB (Mathworks Inc. Natick, MA) script. Control Interface and Signal C ommand A setup function established the serial communication protocol between the laptop computer and the prosthetic hand and reset the hand into the open position. A second function received the time lapse captures from Slidebook and translated image data into an optical signal by averaging nodal ROI pixel intensities at each frame. Photobleaching of the signal was kept minimal by the reduct ion of laser power and exposure, and any mild decay due to photobleaching was not removed. The signal commanded a velocity control paradigm whereby the amplitude above a threshold of two percent signal change was mapped to flexion velocity of the prosthetic digit B elow this threshold the digit reversed direction and ext ended to its resting state Velocity gains were adjusted in order to ensure f ull range of motion A motor command was sent to the prosthetic hand using a USB serial command interface from the MATLAB script. Position encoder values from the prosthetic finger motor were recorded simultaneously and converted to joint angle measurements post hoc. The serial commands from the laptop interface were sent to an Arduino (SparkFun Electronics, Boulder, CO) which converted the serial commands into I2C communication for the custom motor controller systems installed in the prosthetic hand. The electronics in the original Bebionic hand were replaced with a custom motor controller syste m (Sigenics Inc., Chicago, IL) and included a central controller board and
59 Each penny board was connected by a four wire I2C bus with each board associated with an individual actuator and finger. The Bebionic hand (RSL Steeper Inc., United Kingdom) is a commercially available five DoA, multi functional prosthesis with a manually positioned thu mb abduction joint (Figure 4.1) The speed of the digits remained the same as the original commercial device (1 second to open/close or 15 revolutions per minute). Results The selected axon yielded a visible fluorescence response to the emulated motor command. The calcium response, consistent with many observed nodal responses, originated at the node center and propagated distally and proximally into the internodal region of the axon. The nodal region, which was taken as the motor control signal, show ed an approximately 12% change in fluorescence intensity. Under the velocity control paradigm employed here, the digit drove in flexion for the duration of the supra threshold optical signal, and extended back to its resting state as the si gnal diminished below threshold (Figure 4.2). Previously recorded signals that were collected over a range of action potential frequencies were also used to drive prosthetic finger actuation. As characterized in earlier work (Chapter 2), average fluorescence amplitu des of sustained stimulus are linearly modulated by the frequency of action potentials. Such graded signals can thus encode intensity of the motor command. The prosthetic finger actuated to varying degrees of flexion, depending on its action potential fr equency modulated control signal ranging from 25 125Hz (Figure 4. 3 ).
60 Figure 4.1: Commercially available Bebionic hand used for finger actuation experiments
61 Figure 4. 2 : (A) Confocal images of a CalciumGreen 1 Dextran loaded axon node of Ranvier used to control finger actuation, shown before, during and after the activity induced fluorescent signal (scale bar 10 calcium fluorescence signal in response to the 1s, 100Hz train of action potentials (black bar). (C) Pr of the optical signal from panel B. (D) Corresponding finger joint angle.
62 Figure 4.3 : (A) Graded calcium fluorescence transients in an axon node of Ranvier in response to a range of action potential frequencies. (B) Resulting finger joint angles of the prosthetic finger as driven with controls signals from panel A.
63 Discussion In the present work we have demonstrated prosthesis actuation under real time control of neural activity in a peripheral nerve fiber. The neural activity is transduced to an optically emitting signal via a fluorescent calcium sensor, which was loaded into the a xonal cylinder via diffusion and or axoplasmic transport. The presence of activity dependent calcium signaling in mammalian peripheral nerve axons was presented and characterized in Chapters 2 & 3. Calcium elevation at nodes of Ranvier appear to be the r esult of transmembrane influx via predominantly T type voltage gated calcium channels and/or the plasmalemmal sodium calcium exchanger (NCX). By interfacing microscope software with custom MATLAB script, fluorescence intensity data could be streamed into a real time signal of neural activity and command prosthetic finger motor control. The configuration was used to drive the prosthetic finger in a velocity proportional manner, to achieve basic flexion and extension of the digit. Furthermore, via pre reco rded signals, the prosthesis was driven to varying flexion joint angles by neural signals of differing action potential frequency, illustrating the potential for neural intensity modulated motor output.
64 CHAPTER V DEEP TISSUE TWO PHOTON IMAGING IN BRAIN AND PERIPHERAL NERVE WITH A COMPACT HIGH PULSE ENERGY (YTTERBIUM) FIBER LASER Introduction Optical imaging is a powerful tool for studying and probing structure and function of neural circuits in the brain. Due to the highly scattering nature of neural tissue, optical imaging in non fixed, intact tissue was limited in the past to very superficial depths. The advent of two photon microscopy has greatly improved upon limitations of single photon imaging by (1) increasing penet ration depth in tissue with longer wavelength light, (2) ability to achieve higher collection efficienc ies through scattering media 78,79 (3) reducing photo damage due to more localized photo excitation and (4) reduced scattering signal from out of focus planes due to the localized excitation The non linear optical process of two photon excitation requires the simultaneous absorption of two photons, w hich necessitates a pulsed laser source that can achieve high pulse energies, while maintaining non damaging average power. Standard commercial two photon microscopes using Ti:Sapphire lasers operating at 80 MHz repetition rates and several nJ pulse ener gies and ~100 200 fs can provide penetration depths approximately 3 fold that of single photon confocal systems 80 However, these systems are not sufficient for many applications in which still greater optical penetration is desired. Recent groups have presented non standard two photon laser systems to further extend this depth limit. Many of these techniques aim to increase the peak
65 power of the laser pulse while maintaining t he same average power. This is achieved with a reduction of the laser pulse duty cycle, either by decreasing the pulse repetition rate or shortening the pulse width. This results in enhanced signal generation while avoiding potential tissue photo damage associated with higher average power. Theer et. al 44 imaged fluorescein labeled vasculature and green fluorescent protein (GFP) labeled neurons up to a millimeter below the surface of mouse cortex us ing a TiAl 2 O 3 regenerative amplifier which operated at center = 925 nm, 600 mW maximum average power, 200 KHz pulse repetition rate (3 with 150 fs minimum pulse width. More recently, Kobat et. al 81 us ed an optical parametric oscillator (OPO) operating at an IR shifted center wavelength of 1280 nm to obtain structural images 1.6 mm deep in a live mouse brain. The OPO in th e Korbat et al. study delivered pulses at 80 MHz with 120 mW average power (1.5 n J pulse energy), 140 fs pulse width, and imaged vasculature labeled by retro orbital injection of Alexa680 Dextran. Neuronal activity has been recorded 800 m below the pia in somatosensory cortex from the genetically enco ded calcium indicator GCaMP3 24 using a regenerative amplification system similar to that in [1 ]. N euronal activity can also be activated at similar depths via two photon excitation of ChR 2 36 The use of three photon micro scopy has also been employed 45 in a study which imaged red fluorescent protein (RFP) labeled neurons and Texas Red labeled vasculature in vivo up to 1.4 mm deep in mouse brain. Imaging was perf ormed with 1675 nm excitation light (67 nJ pulses, 65 fs duration, 1 MHz repetition rate and 22 mW maximum power ) with a novel high pulse energy source using soliton self frequency shift (SSFS) in a photonic crystal rod.
66 Unfortunately, solid stat e laser systems employed in these deep imaging studies tend to be quite expensive, bulky, and generally require regular maintenance and alignment. Fiber lasers address these disadvantages by virtue of their compact design and portability, cost, and mainte nance free operation, but have traditionally been inferior to solid state lasers for pulsed operation due to the accumulation of nonlinear phase shifts which can degenerate the pulse. Over the course of the last decade however, mode locked fiber laser des ign has evolved rapid ly. In 2005, Buckley et. al 82 presented a ytterbium (Yb) doped femtosecond fiber laser with which they achieved pulse energies up to 14 nJ, dechirped to ~100 fs puls e duration. Messerly et. al 83 obtained 25 nJ pulses with < 200 fs duration in a Yb doped fiber laser w ith increased pump power and spectral filtering. Fiber lasers have been used in two photon studies with stable excitation for fluorescence co rrelation spectroscopy (FCS) 84 and wavelength tunable excitation from 850 nm to 1100 nm via coupling with a photonic crystal fiber (PCF) 85 Wise et. al have im proved upon the pulse generation of these studies by excluding anomalous group velocity dispersion (GVD) in the laser cavity and operating in an all normal dispersion (ANDi) regime, ultimately increasing the pulse energy potential. The initial demonstrati on yielded pulses of 3 n J 86 The group has sinc e achieved 26 nJ pulses with 325 mW average power, at a 12.5 MHz repetition rate with 165 fs pulse duration 87 In a more recent study, 31 nJ pulse energy (2.2 W average power, 70 MHz repetition rate, 80 fs pulse duration) was demonstrated with higher overall power using a double clad (DC) gain segment 88 In the present study, we investigated whether we could achieve two photon optical depth measurements similar to recent studies, using a simple turnkey operated
67 fiber laser. Additionally, we sought to characterize depth measurements in the peripheral nerve (white matter) since probing nerves, myelinated spinal tracts, and white matter in the brain is also of substantial importance and interest in both neuroscientific and neuroprosthetic applications. Methods Images were acquired using the Yb doped laser ( Y Fi, KM Labs), based on the ANDi (All Normal Dispersion) ultrafast fiber technology 88 with high pulse energy capability and short pulse operation. The laser operated at a center wavelength of 1035 nm, with repetition rate of 5 MHz and pulse width of 140 fs at the laser aperture. This system was coupled with custom optics to an Olympus FluoView 1000 laser scanning microscope (Figure 5.1) and images were collected with a long working distance 40x objective lens (Olympus LUMPlanFl/IR, .8 NA, W ater WD 3.3 mm) for the brain slice and phantom samples, and a 100x (Olympus LUMPl anFl, 1.00 NA, W ater WD 1.5 mm) for the sciatic nerve sample Figure 5.1: Light path schematic with custom optical elements.
68 The maximum power measured at the objective lens was 185 mW. Fluorescence was epi collected and filtered with a bandpass filter BP 500 570 nm (535AF70, Omega ) onto a non descanned photomuliplier tube detector. Internal dispersion compensation capability of the laser was adjusted for maximal fluorescence excitation. Yellow fluorescent protein (YFP) labeled samples yielded substantial fluorescence at this wavelength, and thus measurements in brain were acquired with YFP labeled neurons in piriform cortex of mouse br ain slices Axons in mouse sciatic nerve were labeled in vitro by longitudinal, diffusion dye loading of Rhodamine B Dextran. For this dye loading technique, the proximal end of an excised nerve is drawn into a tightly fitting suction pipette containing dye loading buffer (2 mM Rhodamine B Dextran, 130 mM KCL, 30 mM MOPS buffer, pH 7.2) Samples were placed on the optical glass of an imaging chamber and submerged with a mouse saline solution (in mM: 126 NaCl, 5 KCl 1.8 CaCl 2 1 MgCl 2 10 MOPS Buffer pH 7.2, 30 glucose) In addition to brain and nerve samples, imaging was performed in a brain tissue like phantom consisting of 1% agarose gel containing 1.1 m mean diameter non fluorescent scattering beads (Sigma) at a concentration of 5.4 x 10 9 beads/ml t o emulate the scattering len gth constant of brain tissue 44 and as tracers, latex amine modified, polystyrene fluorescent beads (ex/em = 520/540 nm, Sigma) with a 2 m mean diameter at a concentration of 2.8 x 10 7 beads/ml. For deep tissue z stack image sets, the average power was minimized at the sample surface and scaled exponentially with depth, in order to avoid tissue damage at superficial z locations. Image analysis was performed with ImageJ Sof tware. S ignal (S) was calculated as an average intensity of the labeled features of interest (i.e neuron cell
69 bodies or axons) while the background (B) is an average of non feature pixel intensities selected with ROIs of non feature background across an i mage. Results YFP labeled neurons were visible up to approximately 600 m deep in the piriform cortex, and axons were discernable at depths up to 75 m in nerve. Beads from the phantom were imaged up to ~950 m. The extent of additional depth achieved in the fluorescent bead phantom relative to actual brain tissue was somewhat unexpected. However, this could be partially attributed to the fact that the YFP excitation spectrum may be less optimal than that of the phantom fluorescent beads for this laser wavelength (1035nm) ; YFP has a two photon excitation peak of 960nm 89 and has a blue shifted single photon excitation peak relative to the fluorescent beads (512nm and 520nm) respectively 90 Figure 5.2 shows f ive equally spaced optical sections in brain and nerve. The contrast of neuron components in the images degrades with depth due to decreased signal generation caused by scattering of excitation light, and increased out of focus fluorescence generation. The degradation in this signal to background ratio with depth is signific antly more pronounced in the nerve compared to brain. In both tissues, the maximum available laser power had not been expended at the maximum depth at which neuronal bodies / axons were visible. In other words, the reduction of signal to background ratio was dominated by out of focus fluorescence rather tha n available laser power.
70 Figure 5. 2 : Signal to background ratio of cell processes (cell bodies or axons) decreases over a shorter depth in nerve than brain cortex or fluorescent bead phantom. Images of YFP labeled neurons in mouse piriform cortex (top) and Rhodamine B labeled axons of mouse sciatic nerve (middle) and phantom (bottom) at five equally spaced z planes. (Contrast was adjusted in each image for presentation) Figure 5. 3 : X Z cross sectional view of YFP labeled neurons in mouse cortex
71 Figure 5. 4 : Three dimensional renderings of cortex (A), nerve (B) and phantom (C) z stack data. Image sets were collected while exponentially scaling the laser power with depth. To quantify the large difference in optical penetration depth between brain and nerve, constant power z stack image sets were collected in each tissue type. Their respective attenuation curves (Fi gure 5. 5 ) show an almost 20 fold greater attenuation in t he myelinated nerve relative to non myelinated cortex. The signal (S) was calculated as an average intensity of the labeled features of interest (i.e neuron cell
72 bodies or axons), while the background (B) is an average of non feature pixel intensities. Figure 5. 5 : Attenuation curves collected in piriform cortex (A) and sciatic nerve (B) using constant laser power show a length constant (l) in cortex nearly 20 fold greater than that in myelinated nerve. Discussion Using an ultrafast fiber laser system, we were able to image deeper into tissue than standard two photon microscopes, but did not surpass depths achieved in recent studies employing regenerative amplification or OPOs. In addition to quantifying imaging de pth in brain, we sought to establish the general depth at which it is possible
73 to optically reach axons in white matter tracts, and to demonstrate the significantly limited penetration relative to brain. The attenuation length constant for nerve measured here is in reasonable agreement with the white matter attenuation length constant reported in a previous study (~20 30 m in YFP labeled spinal colu mn with 1043 nm excitation) 91 when considering experimental variations between studies such as shorter excitation wavelength and higher volume fraction labelling in the present study. T he length constant of 256 m measured in brain is comparable with that measured in a previous study 44 While this preliminary study has demonstrated utility and efficacy of fiber lasers for deep tiss ue two photon microscopy, the quantitative characterization of imaging depth into these tissue types can be improved upon. Firstly, the greater depth achieved in the brain like phantom which contained red shifted florescent beads compared with t he YFP expressing brain samples suggests that the cortex imaging depth may significantly improve with a higher wavelength excitable fluorophore. The laser pulse width is another parameter which influences the optical penetration depth. However, reducing the pulse width only modestly will result in little enhancement of the imaging depth limit. A 25% increase in imaging depth, for example, would require a reduction in pulse width from 140fs (present study) to approximately 70fs, with more substantial dept h improvement occurring w ith pulse widths below 50fs 80 Furthermore, while the tissue specific scattering degrades the signal generation with depth, the experimental technique of fluorescent tissue labelling can strongly influence the depth measurement, and therefore the tissue characterization. The optical penetration in white matter is particularly limited due to the high degree of scattering
74 from myelin, which creates repeated interfaces of differing refractive indices (i.e water and lipid). This scattering reduces the photo excitation at the focus, thus decreasing the signal of interest. However, as the ability to resolve features of interest in deep tissue studies i s dependent on the signal to background ratio, background fluorescence (non focal signal generation) is a large factor intrinsic to the depth measurement. Detailed quantitative theory on this imaging depth limit with respect to the fluoropho re distributio n is given in 80 (eq. 22). Given this significance, it is important to note that this background fluorescence is not predominantly a tissue property, but rather it is largely depende nt on the fill factor of fluorescent labeling; with sparser labeling of axons or soma resulting in lower background, and vice versa 92 This is particularly pertinent in the nerve, where the diffusion loading tech nique resulted in the labeling of virtually all axons. Because of this high fill factor nerve labeling, the nerve depth measurement we present here is a lower bound approximation of the imaging depth that is possible with this laser. Therefore, the relat ive difference in depth we report for brain and nerve is also an approximation, as the labelling fill factor and thus background fluorescence generation are quite different between the two tissues. An accurate measure of this difference will necessitate e qual fill factor labeling in each tissue, requiring novel techniques.
75 CHAPT ER VI DISCUSSION & FUTURE DIRECTIONS Discussion The primary underlying goal driving this work has been to develop a novel approach to neural interfacing, specifically a peripheral nerve interface for control of advanced prosthesis. As discussed in Chapter 1, there are major limitations in current neural interface strategies, and advanced prosthesis integration is still largely precluded by the lack of a su fficient mode of communication to the nervous system. Perhaps the most paramount attribute of a neural interface, which remains inadequate useful if not necessary for fine neural recording/stimulation; even pattern recognition algorithms used to scour aggregate neural recordings have not proven capable of reliably deciphering neural pathways needed for individual digit command and dexterity. Such a vastly improved neural interface would not only be valuable in the realm of prosthetic device control, but also in the neuromodulation o f any neurally innervated tissue, whether it be functional substitution, therapeutic, or for basic study and understanding. Within this goal of an improved neural interface technology, this work has striven to establish a mechanism that will enable optogeneti c read out of neural activity. Activity Dependent Calcium and Dynamics The presence of an activity dependent calcium transient was investigated as a potential means for neural read out. Calcium signaling in response to membrane
76 depolarization is ubiquitous throughout the nervous system, occurring in neuron soma, dendrites, axon terminals, and all muscle types This fac t provided good reason to speculate that such signaling also occurs along the process of the axon as well. Additionally, at the outset of this research, there had been studies which began to investigate calc ium signaling along axons 46,47 yet there was no data on activity dependent calcium signaling in mammalian peripheral nerve axons. Here, action potential elicited calcium signaling has been demonstrated and characterized in the mammalian peripheral axon, using an in vitro rodent nerve model with axon loaded synthetic ca lcium indicator. As was the case in a prior s tudy in the frog periphery 47 a nd rodent purkinje neurons 48 we observed activity dependent calcium at the axon nodes of Ranvier. It was not surprising to find this signaling at the node, as it is a channel dense (primarily voltage gated sodium channels) and metabolically active region that facilitates saltatory conduction. It has also been suggested that important glial axon signaling occurs at the node of Ranvier 12,51,52,93 The calcium signal was observed to arise directly at the node center, and propagate bi directionally a short distance into the internode, and t ransients in calcium could be detected in response to single action potentials A number of dynamical analyses were investigated (Chapter 2). Calcium wave propagation velocity was calculated (33 .5 6.1 m/s n=5) and spatial decay of amplitude was determined to fall off quasi exponentially with a mean spatial length constant of (11.8 1.0 m n=5) The calcium fluorescence amplitude in response to a train of action potentials rose sharply at low numbers of actio n potentials before reaching a steady state plateau where influx reaches a balance
77 characterized the shape of the signal. A pertinent characterization to neuroprosthetic application was that of frequency mo dulation. As expected, the amplitude of calcium fluorescence summated with the number of action potentials in the stimulus (pulse train frequency). In agreement with studies of dendritic calcium accumulation 33,67 the amplitude frequency relationship was linear. This finding fits with mathematical modeling that shows a li near dependence of mean amplitude on action potential frequency, given a consistent unit calcium quantity added per action potential. The frequency amplitude slopes were variable with a mean of .07% fluorescence change / Hz. The frequency versus amplitu de dependence of a node did not appear to be correlated with its size, and varying slopes may be attributed to differing calcium influx quantities an d/or variable dye conditions. Calcium Mechanisms Pharmacological experiments were performed to inves tigate the mechanisms and pathways of the calcium elevation. This work suggests that the calcium signal arises from transmembrane influx rather than intracellular release from ER or mitochondria, as blockers of calcium handling mechanisms on these organel les did not si gnificantly decrease the signal amplitude. Specifically, data presented in this thesis implicates the involvement of T type Ca V and NCX (Na Ca exchanger). These findings are consistent with the study in CNS purkinje axons 48 in which T type Ca V are found to be responsible for nodal calcium influx. A recent study that, along with this work, demonstrates nodal calcium signaling in the mouse PNS 49 also points to T type Ca V and NCX, and suggests that these two proteins are co localized at the node with each contributing to the increase in calcium. This finding of co contribution within the same
78 axon is not necessarily clear in the present study which may suggest larger pro portional contribution from each mechanism. As such, it is worth pointing out experimental differences. The study by Zhang et al. employed microelectrode dye injection and coul d only monitor a single axon in each experiment; they studied large diameter (>5 axons of the phrenic nerve. Furthermore, their concentration of the T type Ca V blocker M ibefradil was likely below the IC 50 for these channels 94,95 and thus, the dat a are likely to underestimate contribution of T type Ca V. In the present study, nodal responses with inhibitors of both these pathways fell into two groupings: signals from some nodes were significantly diminished or blocked by drug application, w hile others were largely unblocked. Co application of T type Ca V and NCX blockers did not inhibit all signals as I hypothesized would occur, but rather yielded a distribution of response amplitudes ranging from unblocked to blocked. This distribution of responses did appear to have grouping similar to the single drug data, and indeed was determined to be bi bi modality. Taken together, this data would suggest that the distribution of peripheral nerve axo ns studied do not have a single mode of calcium handling, nor is it likely that there are simply two functional classes of calcium signaling (i.e. T type Ca V dominated and NCX dominated). Rather, the distribution of calcium influx mechanisms may be even m ore heterogeneous. A potential explanation for the observation that some nodal signals remained unaffected with drug co application is that there is a third predominant mode of calcium elevation. Additional work will be required to achieve a more complet e understanding of the seemingly heterogeneous axon distribution as it relates to calcium signaling mechanisms. P o ssible pathways that could contribute to a
79 third mode are other types of voltage gated calcium channels such as N, P/Q, and R type channels, as well as ionotropic and metabotropic glutamate receptors activated by Schwann cell coupling. These mechanisms could be tested pharmacologically in future experiments. experimental design employed here was not optimally suitable for precise signal change quantification over a drug incubation period. A miniscule effect of a blockade would not be detect ed; however, substantial pharmacological signal effects could be deduced. Calcium Signal Actuation of Prosthesis To demonstrate the ability of calcium signals to function as motor commands in a neuroprosthesis, action potential elicited calcium fluoresc ence signals from the in vitro nerve were used in real time to drive a prosthetic finger. Graded, action potential frequency modul ated signals were also routed to the prosthesis to illustrate the corresponding gradation in motor output. This was a rudime ntary experiment to associated optical signals could be used for direct prosthetic motor control and could drive motor output in proportion with the neural intensity of the Ca signal. Achieving this in vivo and furthermore in a freely moving animal, is a significantly more challenging task; one that requires the development of additional aspects of the neural interface system. Sufficient viral transduction of a GECI to axons of interest and an optical fibe r bundle coupled endoscopic device with adequate tissue scanning capability will be integral components in transitioning to in vivo models.
80 Future Directions Continued advancement in a number of fields will be required for the implementation of an optogen etic neural interface, particularly in the peripheral nervous system. Viral Vector Design Optogenetic proteins, namely a GECI and ChR2 variant, will need to be expressed at suitable levels in axons for an optical nerve interface. Viral transduction of o ptical proteins in peripheral axons has already been achieved 39 but much work can be done to improve the targeting and efficiency of viral vectors. Increased cell specific psuedotyping for targeting and transduction efficiency, as well as immunosuppression techniques for improved transduction are active areas of research, driven in part by the prospect of lucrative gene therapies. See 96 99 for review of current state of v iral gene transfer methods. Laser Development Lasers for multi photon interrogation will likely need to be improved to achieve maximal tissue penetration, while ultimately becoming more compact and portable. Substantial progress in laser development has taken place in the past decade (see Chapter 5) and laser e ngineering efforts are likely to continue the enhancement of tissue penetration depth. Maximizing laser pulse energy while decreasing the repetition rate (duty cycle) of pulsing will aid tissue penetration while minimizing damaging energy deposition in th e tissue. Increasing power output, particularly at far infrared wavelengths, and in more compact laser configurations such as time lens
81 lasers 100 will greatly benefit the capabilities of optical interrogation in the brain and peripheral nerve. Optical Fiber Coupled Microscopy Beyond the laser development, an optical system must interface with the nerve to deliver/detect light between the laser and neural targets. Light deliverance would take place through a flexible optical fiber bundle, and the device must be capable of rapid three dimensiona l scanning at the tissue end of this optical conduit. Electro wetting lens technology 101 has been demonstrated in the Gopinath and Bright labs show ing promise for rapid z dimension laser scanning which has traditionally been the far slo wer scanning axis in three dimensional microscopy. The Gibson lab has demonstrated the use of an electro wetting lens incorporated in a fiber coupled microscope for in vivo confoc al imaging in cortex over three dimensional volumes 102 This device would b e integrated in the intra co r poreal end of the optical fiber, as part of a novel cuff assembly that abuts the nerve and fix es the optical fiber/lens assembly in a direct interface with the nerve tissue Two dimensional (x y) scanning would be achieved by piezoelectric actuators, microscanners or other device, situated at the extra corporeal end of the fiber. Such a system could be used to access Ca 2+ signals and stimulate in specific axons. With adequate scanning speed a nd range within the nerve a set of nodes would be serially scanned to moni tor their activity. Detectable nodes within the scan range would first be tested to identify their physiological role by running through a protocol of visualized/imagined movements with the subject. Nodal signals would be pre calibrated to map signal amp litude to appropriate prosthetic motor output, by deriving
82 the frequency amplitude slope. Ideally, at least one nodal response could be identified and pre calibrated for each digit. By virtue of the analog nature of a calcium response to trains of action potentials, pertinent nodes could be scanned in a serial fashion with reasonable sampling rate per node, given sufficient scan speed. PNS Neuromodulation for Disease Treatment and Study Another promising application of this work is that of therapeutic neuromodulation. The concept of treating, diagnosing and studying visceral diseases with the modulation of peripheral nerves has recently received significant attention. As 103 the advance ment of neural interfacing technology is essential to opening up such strategies. An the context of neuroprosthetic control, has the potential to be a tremendous boon to this field, namely due to the cell/process specificity and non invasiveness which it can provide. Of course the deployment of this approach in bioelectronic medicine will also largely depend on the requisite advancements that have been described in thi s discussion. In many cases, a closed loop, bi directional interface which can read (record) neural activity and modulate (stimulate/silence) activity within the same system is highly desirable. The activity dependent calcium signaling presented in this work may serve as a read out, while a ChR2 or similar opsin would provide stimulation. In addition to the stimulation afforded by ChR2, light driven silencing could also be achieved with the yellow light activated chloride pump halorhosopsin. Optogenetic actuation is already being applied to numerous functions withi n the peripheral nervous
83 system 98 The range of disorders which could poten tially be studied and treated with such a system is exhaustive, but may include diabetes, cardiac complications, pain rheumatoid arthritis. Vagus nerve stimulation, for example, has been shown to reduce seizure frequency in intractable primary generalized epilepsy 104 and attenuate inflammation in human subjects with rheumatoid arthritis 105 Neuromodulation could feasibly encompass three broad modes of function: (1) afferent stimulation/silencing (i.e. somatosensation/pain/seizure modulation), (2) hormone release/signaling perturbation (i.e. pancreatic insulin), and (3) muscle contraction (i.e. car diac pacing, bladder control). Summary This work show s the feasibility of using optical signals as a readout of electrical activity in peripheral nerve axons. From single axons we can detect activity down to the single action potential, and use these signals to drive the movement of individual digits of a p rosthetic hand. Because motoneurons and their axons are the final common pa th of motor output 106 these calcium signals are the result of all the normal processing of the brain and spinal cord and provide the capability of precise and rapid movements of a prosthetic device. Thus, the work in this thesis demonstrates the feasibility of potential therapies using optogenetics.
84 REFERENCES 1. Ziegler Graham, K., MacKenzie, E. J., Ephraim, P. L., Travison, T. G. & Brookmeyer, R. Estimating the Prevalence of Limb Loss in the United States: 2005 to 2050. Arch. Phys. Med. Rehabil. 89, 422 429 (2008). 2. Amputee Coalition. Limb Loss Statistics. (2016). doi:10.1017/CBO9781107415324.004 3. Belter, J. T., Segil, J. L., Dollar, A. M. & Weir, R. F. Mechanical design and performance specifications of anthropomorphic prosthetic hands: a review. J. Rehabil. Res. Dev. 50, 599 618 (2013). 4. Saikia, A. et al. Recent advancements in prosthetic hand technology. J. Med. Eng. Tech nol. 1902, 1 10 (2016). 5. Wo/2009/023, 334, W. P. & 2009. Biomimetic Tactile Sensor for Control of Grip. Wipo.Int c, (2009). 6. Wettels, N., Popovic, D., Santos, V. J., Johansson, R. S. & Loeb, G. E. Biomimetic tactile sensor for control of grip. in 200 7 IEEE 10th International Conference on 923 932 (2007). doi:10.1109/ICORR.2007.4428534 7. Su, Z., Fishel, J. A., Yamamoto, T. & Loeb, G. E. Use of tactile feedback to control exploratory movements to characterize object c ompliance. Front. Neurorobot. 6, 1 9 (2012). 8. Johansson, R. S. Sensory Control of Dextrous Manipulation in Humans.pdf. Hand and Brain 381 414 (1996). 9. Caldwell, J. H., Schaller, K. L., Lasher, R. S., Peles, E. & Levinson, S. R. Sodium channel Na(v)1.6 is localized at nodes of ranvier, dendrites, and synapses. Proc. Natl. Acad. Sci. U. S. A. 97, 5616 20 (2000). 10. Rasband, M. N. et al. Potassium channel distribution, clustering, and function in remyelinating rat axons. J. Neurosci. 18, 36 47 (1998). 11. Rasband, M. N. & Trimmer, J. S. Developmental clustering of ion channels at and near the node of Ranvier. Dev. Biol. 236, 5 16 (2001). 12. Schafer, D. P. & Rasband, M. N. Glial regulation of the axonal membrane at nodes of Ranvier. Curr. Opin. Neurobiol. 16, 508 514 (2006). 13. Schiefer, M. A. et al. Selective activation of the human tibial and common peroneal nerves with a flat interf ace nerve electrode. J. Neural Eng. 10, 056006 (2013).
85 14. Schiefer, M. A., Polasek, K. H., Triolo, R. J., Pinault, G. C. J. & Tyler, D. J. Selective stimulation of the human femoral nerve with a flat interface nerve electrode. J. Neural Eng. 7, 26006 (201 0). 15. Tan, D., Schiefer, M., Keith, M. W., Anderson, R. & Tyler, D. J. Stability and selectivity of a chronic, multi contact cuff electrode for sensory stimulation in a human amputee. Int. IEEE/EMBS Conf. Neural Eng. NER 12, 859 862 (2013). 16. Sharma, A. et al. Long term in vitro functional stability and recording longevity of fully integrated wireless neural interfaces based on the Utah Slant Electrode Array. J. Neural Eng. 8, 45004 (2011). 17. Normann, R. A., McDonnall, D., Clark, G. A., Stein, R. B. & Branner, A. Physiological activation of the hind limb muscles of the anesthetized cat using the utah slanted electrode array. Proc. Int. Jt. Conf. Neural Networks 5, 3103 3108 (2005). 18. Micera, S. et al. Decoding of grasping information from neural signals recorded using peripheral intrafascicular interfaces. J. Neuroeng. Rehabil. 8, 53 (2011). 19. Dhillon, G. S. & Horch, K. W. Direct Neural Sensory Feedback and Control of a Prothetic Arm. 13, 468 472 (2005). 20. Boretius, T. et al. A transverse intrafascicular multichannel electrode (TIME) to interface with the peripheral nerve. Biosens. Bioelectron. 26, 62 69 (2010). 21. Akerboom, J. et al. Optimization of a GCaMP Calcium Indicator for Neura l Activity Imaging. J. Neurosci. 32, 13819 13840 (2012). 22. Tian, L. et al. Imaging neural activity in worms, flies and mice with improved GCaMP calcium indicators. Nat. Methods 6, 875 881 (2009). 23. Akerboom, J. et al. Genetically encoded calcium indi cators for multi color neural activity imaging and combination with optogenetics. Front. Mol. Neurosci. 6, 2 (2013). 24. Mittmann, W. et al. Two photon calcium imaging of evoked activity from L5 somatosensory neurons in vivo. Nat. Neurosci. 14, 1089 1093 (2011). 25. Badura, A., Sun, X. R., Giovannucci, A., Lynch, L. A. & Wang, S. S. H. Fast calcium sensor proteins for monitoring neural activity. Neurophotonics 1, 025008 (2014). 26. Emiliani, V., Cohen, A. E., Deisseroth, K. & Husser, M. All Optical Interrogation of Neural Circuits. J. Neurosci. 35, 13917 26 (2015). 27. Dana, H. et al. Sensitive red protein calcium indicators for imaging neural activity. 2, 1 24 (2016).
86 28. Oheim, M. et al. New red fluorescent calcium indicat ors for optogenetics, photoactivation and multi color imaging. Biochim. Biophys. Acta Mol. Cell Res. 1843, 2284 2306 (2014). 29. Kn o pfel, T., Gallero Salas, Y. & Song, C. Genetically encoded voltage indicators for large scale cortical imaging come of age. Curr. Opin. Chem. Biol. 27, 75 83 (2015). 30. St Pierre, F. et al. High fidelity optical reporting of neuronal electrical activity with an ultrafast fluorescent voltage sensor. Nat. Neurosci. 17, 884 9 (2014). 31. Vogt, N. Voltage sensors: challenging, but with potential. Nat. Methods 12, 921 924 (2015). 32. Chen, T. W. et al. SUPPLEMENTAL Ultrasensitive fluorescent proteins for imaging neuronal activity. Nature 499, 295 300 (2013). 33. Helmchen, F. & Sakmann, B Ca2 + Buffering and Action Potential Evoked Ca2 + Signaling in Dendrites of Pyramidal Neurons. Biophys. J. 70, (1996). 34. Wang, H. et al. High speed mapping of synaptic connectivity using photostimulation in Channelrhodopsin 2 transgenic mice. Proc. Na tl. Acad. Sci. U. S. A. 104, 8143 8 (2007). 35. Aravanis, A. M. et al. An optical neural interface: in vivo control of rodent motor cortex with integrated fiberoptic and optogenetic technology. J. Neural Eng. 4, S143 S156 (2007). 36. Andrasfalvy, B. K., Zemelman, B. V., Tang, J. & Vaziri, A. Two photon single cell optogenetic control of neuronal activity by sculpted light. Proc. Natl. Acad. Sci. U. S. A. 107, 11981 6 (2010). 37. Zhao, S. et al. Cell type specific channelrhodopsin 2 transgenic mice for optogenetic dissection of neural circuitry function. Nat. Methods 8, 745 752 (2011). 38. Britt, J. P., McDevitt, R. A. & Bonci, A. Use of channelrhodopsin for activation of CNS neurons. Curr. Protoc. Neurosci. 1, (2012). 39. Towne, C., Montgomery, K. L., Iyer, S. M., Deisseroth, K. & Delp, S. L. Optogenetic Control of Targeted Peripheral Axons in Freely Moving Animals. PLoS One 8, (2013). 40. Llewellyn, M. E., Thompson, K. R., Deisseroth, K. & Delp, S. L. Orderly recruitment of motor units under optical control in vivo. Nat. Med. 16, 1161 1165 (2010).
87 41. Berndt, A. et al. High efficiency channelrhodopsins for fast neuronal stimulation at low light levels. Proc. Natl. Acad. Sci. U. S. A. 108 7595 600 (2011). 42. Inagaki, H. K. et al. Optogenetic control of Drosophila using a red shifted channelrhodopsin reveals experience dependent influences on courtship. Nat. Methods 11, 325 32 (2014). 43. Packer, A. M. et al. Two photon optogenetics of dendritic spines and neural circuits. Nat. Methods 9, 1202 1205 (2012). 44. Theer, P., Hasan, M. T. & Denk, W. Two photon imaging to a depth of 1000 microm in living brains by use of a Ti:Al2O3 regenerative amplifier. Opt. Lett. 28, 1022 1024 (2003). 45. Horton, N. G., Wang, K., Wang, C. C. & Xu, C. In vivo three photon imaging of subcortical structures of an intact mouse brain using quantum dots. 2013 Conf. Lasers Electro Optics Eur. Int. Quantum Electron. Conf. CLEO/Europe IQEC 2013 7, 205 209 (2013). 46. Zhang, C. L. Action Potentials Induce Uniform Calcium Influx in Mammalian Myelinated Optic Nerves. J. Neurophysiol. 96, 695 709 (2006). 47. Zhang, C. L., Ho, P. L., Kintner, D. B., Sun, D. & Chiu, S. Y. Activity dependent regulation of mitochondrial m otility by calcium and Na/K ATPase at nodes of Ranvier of myelinated nerves. J. Neurosci. 30, 3555 3566 (2010). 48. Gr u ndemann, J. & Clark, B. A. Calcium Activated Potassium Channels at Nodes of Ranvier Secure Axonal Spike Propagation. Cell Rep. 12, 171 5 1722 (2015). 49. Zhang, Z. & David, G. Stimulation induced Ca 2+ influx at nodes of Ranvier in mouse peripheral motor axons. J. Physiol. 00, n/a n/a (2015). 50. Verbny, Y., Zhang, C. L. & Chiu, S. Y. Coupling of calcium homeostasis to axonal sodium in axons of mouse optic nerve. J. Neurophysiol. 88, 802 816 (2002). 51. Tsutsui, S. & Stys, P. K. Metabolic injury to axons and myelin. Exp. Neurol. 246, 26 34 (2013). 52. Fern, R. F., Matute, C. & Stys, P. K. White matter injury: Ischemic and nonischemic. Glia 62, 1780 1789 (2014). 53. Stirling, D. P. & Stys, P. K. Mechanisms of axonal injury: internodal nanocomplexes and calcium deregulation. Trends Mol. Med. 16, 160 170 (2010).
88 54. Petrescu, N., Micu, I., Malek, S., Ouardouz, M. & Stys, P. K. Sources of axonal calcium loading during in vitro ischemia of rat dorsal roots. Muscle and Nerve 35, 451 457 (2007). 55. Mattson, M. R. Calcium and neurodegeneration. Aging Cell 6, 337 350 (2007). 56. Wojda, U., Salinska, E. & Kuznicki, J. Calcium ions in neuron al degeneration. IUBMB Life 60, 575 590 (2008). 57. Nikolaeva, M. A. Na+ Dependent Sources of Intra Axonal Ca2+ Release in Rat Optic Nerve during In Vitro Chemical Ischemia. J. Neurosci. 25, 9960 9967 (2005). 58. Stirling, D. P., Cummins, K., Wayne Chen, S. R. & Stys, P. Axoplasmic reticulum Ca2+ release causes secondary degeneration of spinal axons. Ann. Neurol. 75, 220 229 (2014). 59. Lin, G. G. & Scott, J. G. NCLX: The mitochondrial sodium calcium exchanger. 100, 130 134 (2012). 60. Rose, K. E. et al. Immunohistological demonstration of Ca3.2 T type voltage gated calcium channel expression in soma of dorsal root ganglion neurons and peripheral axons of rat and mouse. Neuroscience 250, 263 74 (2013). 61. Steffensen, I., Waxman, S. G., Mills, L. & St ys, P. K. Immunolocalization of the Na(+) Ca2+ exchanger in mammalian myelinated axons. Brain Res. 776, 1 9 (1997). 62. Brown, A. M., Evans, R. D., Black, J. & Ransom, B. R. Schwann cell glycogen selectively supports myelinated axon function. Ann. Neurol. 72, 406 418 (2012). 63. Evans, R. D., Brown, A. M. & Ransom, B. R. Glycogen function in adult central and peripheral nerves. J. Neurosci. Res. 91, 1044 1049 (2013). 64. Delling, M., DeCaen, P. G., Doerner, J. F., Febvay, S. & Clapham, D. E. Primary cilia are specialized calcium signalling organelles. Nature 504, 311 314 (2013). 65. Doerner, J. F., Delling, M. & Clapham, D. E. Ion channels and calcium signaling in motile cilia. Elife 4, 1 19 (2015). 66. Delling, M., DeCaen, P G., Doerner, J. F., Febvay, S. & Clapham, D. E. Primary cilia are specialized calcium signalling organelles. Nature 504, 311 314 (2013). 67. Maravall, M., Mainen, Z. F., Sabatini, B. L. & Svoboda, K. Estimating Intracellular Calcium Concentrations and B uffering without. Biophys. J. 78, (2000).
89 68. Simons, T. J. B. Calcium and neuronal function. Neurosurgical Review 11, 119 129 (1988). 69. Molecular Probes. Indicators for Ca2+, Mg2+, Zn2+ and Other Metal Ions. Mol. Probes Handb. (2010). 70. Helton, T. D. Neuronal L Type Calcium Channels Open Quickly and Are Inhibited Slowly. Journal of Neuroscience 25, 10247 10251 (2005). 71. Clozel, J. P., Ertel, E. A. & Ertel, S. I. Discovery and main pharmacological properties of mibefradil (Ro 40 5967), the first selective T type calcium channel blocker. J. Hypertens. Suppl. 15, S17 25 (1997). 72. Larbig, R., Torres, N., Bridge, J. H. B., Goldhaber, J. I. & Philipson, K. D. Activation of reverse Na+ Ca2+ exchange by the Na+ current augments the cardiac C a2+ transient: evidence from NCX knockout mice. J. Physiol. 588, 3267 76 (2010). 73. Watanabe, Y., Koide, Y. & Kimura, J. Topics on the Na+/Ca2+ exchanger: pharmacological characterization of Na+/Ca2+ exchanger inhibitors. J. Pharmacol. Sci. 102, 7 16 (20 06). 74. Jost, N. et al. ORM 10103, a novel specific inhibitor of the Na+/Ca2+ exchanger, decreases early and delayed afterdepolarizations in the canine heart. Br. J. Pharmacol. 170, 768 778 (2013). 75. Ashman, K. M., Bird, C. M. & Zepf, S. E. Detecting Bimodality in Astronomical Datasets. Astron. J. 108, 2348 2361 (1994). 76. Thastrup, O., Cullen, P. J., Drbak, B. K., Hanley, M. R. & Dawson, a P. Thapsigargin, a tumor promoter, discharges intracellular Ca2+ stores by specific inhibition of the endopla smic reticulum Ca2(+) ATPase. Proc. Natl. Acad. Sci. U. S. A. 87, 2466 2470 (1990). 77. Nicolau, S. M. et al. Mitochondrial Na+/Ca2+ exchanger blocker CGP37157 protects against chromaffin cell death elicited by veratridine. J. Pharmacol. Exp. Ther. 330, 844 854 (2009). 78. Crosignani, V. et al. Deep tissue fluorescence imaging and in vivo biological applications. J. Biomed. Opt. 17, 116023 (2012). 79. Schroeder, L., Riley, J., Kang, S. S., Lugar hammer, M. & Gandjbakhche, A. LIGHT COLLECTION FROM LIVIN G TISSUE BY NON CONTACT. 241, 153 161 (2012). 80. Theer, P. & Denk, W. On the fundamental imaging depth limit in two photon microscopy. J. Opt. Soc. Am. A 23, 3139 3149 (2006).
90 81. Kobat, D., Horton, N. G. & Xu, C. In vivo two photon microscopy to 1.6 mm depth in mouse cortex. J. Biomed. Opt. 16, 106014 (2011). 82. Buckley, J. R., Wise, F. W., Ilday, F. & Sosnowski, T. Femtosecond fiber lasers with pulse energies above 10 nJ. Opt. Lett. 30, 1888 1890 (2005). 83. Messerly, M. J., Dawson, J. W., Barty, C. P. J., An, J. & Kim, D. 25 nJ passively mode locked fiber laser at 1080 nm. Conf. Lasers Electro Optics 2006 Quantum Electron. Laser Sci. Conf. CLEO/QELS 2006 6 7 (2006). doi:10.1109/CLEO.2006.4627950 84. JR, U., ES, P., RG, M., R, H. & CK, J. Evaluat ion of a femtosecond fiber laser for two photon fluorescence correlation spectroscopy. Microsc Res Tech 69, (2006). 85. Unruh, J. R. et al. Two photon microscopy with wavelength switchable fiber laser excitation. Opt. Express 14, 9825 9831 (2006). 86. Chong, A., Buckley, J., Renninger, W. & Wise, F. All normal dispersion femtosecond fiber laser. Opt. Express 14, 10095 10100 (2006). 87. Chong, A., Renninger, W. H. & Wise, F. W. All normal dispersion femtosecond fiber laser with pulse energy above 20 nJ. Opt. Lett. 32, 2408 2410 (2007). 88. Kieu, K., Renninger, W. H., Chong, A. & Wise, F. W. Sub 100 fs pulses at watt level powers from a dissipative soliton fiber laser. Opt. Lett. 34, 593 595 (2009). 89. Two photon action cross sections. Developmental Re source for Biophysical Imaging Optoelectronics, Cornell University Available at: http://www.drbio.cornell.edu/cross_sections.html. 90. Scordato, A. & Schwartz, S. Yellow Fluorescent Protein (YFP) Excitation. MicroscopyU (2016). Available at: https://www. microscopyu.com/techniques/fluorescence/nikon fluorescence filter sets/yellow fluorescent protein yfp excitation. 91. Farrar, M. J., Wise, F. W., Fetcho, J. R. & Schaffer, C. B. In vivo imaging of myelin in the vertebrate central nervous system using third harmonic generation microscopy. Biophys. J. 100, 1362 1371 (2011). 92. Kao, Y. T., Zhu, X., Xu, F. & Min, W. Focal switching of photochromic fluorescent protei ns enables multiphoton microscopy with superior image contrast. Biomed. Opt. Express 3, 1955 (2012). 93. Lev Ram, V. & Ellisman, M. H. Axonal activation induced calcium transients in myelinating Schwann cells, sources, and mechanisms. J. Neurosci. 15, 262 8 37 (1995).
91 94. Mehrke, G., Zong, X., Flockerzi, V. & Hofmann, F. The Ca(++) channel blocker Ro 40 5967 blocks differently T type and L type Ca++ channels. J Pharmacol Exp Ther 271, (1994). 95. Leuranguer, V., Mangoni, M., Nargeot, J. & Richard, S. Inhi bition of T type and L type calcium channels by mibefradil: physiologic and pharmacologic bases of cardiovascular effects. J Cardiovasc. Pharmacol. 37, 649 661 (2001). 96. Grimm, D. & Zolotukhin, S. E Pluribus Unum: 50 Years of Research, Millions of Viruses, and One Goal Tailored Acceleration of AAV Evolution. Mol. Ther. 23, 1 13 (2015). 97. Kantor, B., Bailey, R. M., Wimberly, K., Kalburgi, S. N. & Gray, S. J. Chapter Three Methods for Gene Transfer to the Central Nervous System Adv Genet Volume 87, (2014). 98. Montgomery, K. L., Iyer, S. M., Christensen, A. J., Deisseroth, K. & Delp, S. L. Beyond 8, (2016). 99. Masamizu, Y. et al. Local and retrograde gene transfer into primate neuronal pathways via adeno associated virus serotype 8 and 9. Neuroscience 193, 249 258 (2011). 100. Niederriter, R. D., Ozbay, B. N., Futia, G. L., Gibson, E. A. & Gopinat h, J. T. Compact diode laser source for multiphoton biological imaging. Accept. Publ. Biomed. Opt. Express (2016). 101. Terrab, S., Watson, A. M., Roath, C., Gopinath, J. T. & Bright, V. M. Adaptive electrowetting lens prism element. Opt. Express 23, 2583 8 (2015). 102. Ozbay, B. N. et al. Miniaturized fiber coupled confocal fluorescence microscope with an electrowetting variable focus lens using no moving parts. Opt. Lett. 40, 2553 6 (2015). 103. Birmingham, K. et al. Bioelectronic medicines: A research roadmap. Nat. Rev. Drug Discov. 13, 399 400 (2014). 104. Gurbani, S. et al. Neuromodulation Therapy with Vagus Nerve Stimulation for Intractable Epilepsy: A 2 Year Efficacy Analysis Study in Patients under 12 Years of Age. Epilepsy Res. Treat. 2016, 1 5 (2016). 105. Koopman, F. A. et al. Vagus nerve stimulation inhibits cytokine production and attenuates disease severity in rheumatoid arthritis. Proc. Natl. Acad. Sci. U. S. A. 113, 8284 9 (2016).
92 106. grative action of the nervous system: A centenary appreciation. Brain 130, 887 894 (2007). Uncited References 1. Andresen, V. et al. Infrared multiphoton microscopy: subcellular resolved deep tissue imaging. Curr. Opin. Biotechnol. 20, 54 62 (2009). 2. Boyden, E. S. A history of optogenetics: the development of tools for controlling brain circuits with light. F1000 Biol. Rep. 3, 11 (2011). 3. Brill, N. et al. Nerve cuff stimulation and the effect of fascicular organization for hand grasp in nonhuman p rimates. Proc. 31st Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. Eng. Futur. Biomed. EMBC 2009 1557 1560 (2009). doi:10.1109/IEMBS.2009.5332395 4. Broichhagen, J. et al. Optical control of insulin release using a photoswitchable sulfonylurea. Nat. Commun. 5, 5116 (2014). 5. Brown, A. M., Evans, R. D., Black, J. & Ransom, B. R. Schwann cell glycogen selectively supports myelinated axon function. Ann. Neurol. 72, 406 418 (2012). 6. Bryson, J. B. et al. Optical control of muscle function by transplantation of st em cell derived motor neurons in mice. Science (80 ). 344, 94 97 (2014). 7. Buchen, L. Neuroscience: Illuminating the brain. Nature 465, 26 28 (2010). 8. Buffer, C. Calcium Calibration Buffer Kits. Imaging 1 10 (2011). 9. Cao, G. et al. Genetically targeted optical electrophysiology in intact neural circuits. Cell 154, 904 913 (2013). 10. Catterall, W. A. Voltage Gated Calcium Channels. Cold Spring Harb. Perspect. Biol. 3, (2011). 11. Cordella, F. et al. Literature Review on Needs of Upp er Limb Prosthesis Users. Front. Neurosci. 10, 1 14 (2016). 12. Dhillon, G. S., Lawrence, S. M., Hutchinson, D. T. & Horch, K. W. Residual function in peripheral nerve stumps of amputees: Implications for neural control of artificial limbs. J. Hand Surg. A m. 29, 605 615 (2004). 13. Evans, R. D., Brown, A. M. & Ransom, B. R. Glycogen function in adult central and peripheral nerves. J. Neurosci. Res. 91, 1044 1049 (2013).
93 14. Gonz a lez Fern a ndez, M. Development of Upper Limb Prostheses: Current Progress an d Areas for Growth. Archives of Physical Medicine and Rehabilitation 95, 1013 1014 (2014). 15. Harwin, W. S. Robots with a gentle touch: advances in assistive robotics and prosthetics. Technol Heal. Care 7, 411 417 (1999). 16. Helmchen, F. Topic Introduction Calibration of Fluorescent Calcium Indicators. Cold Spring Harb Protoc 923 930 (2011). doi:10.1101/pdb.top120 17. Jebari, K. Brain machine interface and human enhancement An ethical review. Neuroethics 6, 617 625 (2013). 1 8. Kawada, T. et al. A sieve electrode as a potential autonomic neural interface for bionic medicine. Conf. Proc. IEEE Eng. Med. Biol. Soc. 6, 4318 4321 (2004). 19. Kriegler, S. & Chiu, S. Y. Calcium signaling of glial cells along mammalian axons. J. Neuro sci. 13, 4229 4245 (1993). 20. Lebedev, M. A. & Nicolelis, M. A. L. Brain machine interfaces: past, present and future. Trends Neurosci. 29, 536 546 (2006). 21. Nghiem, B. T. et al. Providing a sense of touch to prosthetic hands. Plast. Reconstr. Surg. 135 1652 63 (2015). 22. Ortiz Catalan, M., Hakansson, B. & Branemark, R. An osseointegrated human machine gateway for long term sensory feedback and motor control of artificial limbs. Sci. Transl. Med. 6, 257re6 257re6 (2014). 23. Packer, A. M. et al. Two ph oton optogenetics of dendritic spines and neural circuits. Nat. Methods 9, 1202 1205 (2012). 24. Poliak, S. & Peles, E. The local differentiation of myelinated axons at nodes of Ranvier. Nat. Rev. Neurosci. 4, 968 980 (2003). 25. Sharma, A. et al. Long ter m in vitro functional stability and recording longevity of fully integrated wireless neural interfaces based on the Utah Slant Electrode Array. J. Neural Eng. 8, 45004 (2011). 26. Staal, J. A. et al. Initial calcium release from intracellular stores follow ed by calcium dysregulation is linked to secondary axotomy following transient axonal stretch injury. J. Neurochem. 112, 1147 1155 (2010). 27. Cytoplasmic Free Calcium Monit ored With a New Intracellularly Trapped Fluorescent Indicator. 325 334
94 28. Veltink, P. H. Sensory feedback in artificial control of human mobility. Technol. Health Care 7, 383 391 (1999). 29. Wang, H., Fu, Y., Zickmund, P., Shi, R. & Cheng, J. X. Coheren t anti stokes Raman scattering imaging of axonal myelin in live spinal tissues. Biophys. J. 89, 581 91 (2005). 30. Yizhar, O., Fenno, L. E., Davidson, T. J., Mogri, M. & Deisseroth, K. Optogenetics in Neural Systems. Neuron 71, 9 34 (2011). 31. Zhang, F. e t al. Optogenetic interrogation of neural circuits: technology for probing mammalian brain structures. Nat. Protoc. 5, 439 56 (2010).