University of Southern California Dissertations and Theses

Physiology of the inner ear: the role of the biophysical properties of spiral ganglion neurons in encoding sound intensity information at the auditory nerve

(USC Thesis Other)

Physiology of the inner ear: the role of the biophysical properties of spiral ganglion neurons in encoding sound intensity information at the auditory nerve

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
Physiology of the Inner Ear:
The role of the biophysical properties of spiral ganglion neurons in encoding sound
intensity information at the auditory nerve
by
Alexander L. Markowitz

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
Neuroscience

May 2020

Alexander L. Markowitz
ii
Acknowledgments
I would like to express my deepest gratitude to all the people who have helped me with
the completion of this dissertation. First, I would like to thank Dr. Radha Kalluri for giving
me the opportunity to learn about cellular biophysics and experimental design in her
laboratory. Radha has always encouraged me and my fellow lab members to think
creatively and to have fun working in the lab. With these goals in mind, we created a great
work environment together. You have been an excellent mentor and I have become a
better thinker, listener, and presenter under your leadership.

I would also like to thank my dedicated committee members Drs. Jeannie Chen, Neil
Segil, and Carolina Abdala. I appreciate you all for contributing to my training in the many
grant writing workshops, department talks, and conferences throughout my time at USC.

Thank you to the smart, hard-working members of the Kalluri laboratory: Dr. Chris
Ventura, Maya Monges-Hernadez, Megana Iyer, and Dr. Daniel Bronson. Behind the
scenes of any great production, is a team all relying on each other to do their small role
the best they can. The Kalluri lab is no different. Thank you Maya for keeping spirits high
in the early days when we were trying to get the experiment up and running. Thank you
Chris and Dan for every thoughtful critique and recommendation of how to best present
my research. Thank you Megana for your continuous enthusiasm for learning and
scientific exploration.

I would like to thank the USC Neuroscience Graduate Program for giving me the
opportunity to move to Los Angeles to study neuroscience in their interdisciplinary
program. Getting accepted into this program was one of the happiest moments of my life
and it has been a pleasure being a member of the community. I wish the best of luck to
all my fellow NGP students on their doctoral training and their future endeavors.

I would like to acknowledge the institutions that provided funding for my academic
research. Thank you to the USC Neuroscience Graduate Program, the Hearing and
Communication Training Program, American Otological Society, and the National
Institute of Deafness and other Communication Disorders of the National Institutes of
Health.

Finally, I would like to express my gratitude for the love and support of my friends and
family. John, you’ve been my solid foundation through it all and have made me stronger,
wiser, and more resilient. Matthew, thank you for proofreading the many pieces of writing
I’ve sent you throughout my training. Aaron, thank you for being there for me since day
one of the cross-country road trip we took to move to LA from Maryland. Dad and Mom,
thank you for supporting me and encouraging me to be resilient. Thank you to all the
people in my life who have supported me to get to this milestone including Sarah, Ethan,
Michael, Sharon, Sheila, Jen, Angela, Steven, Brandon, Jon, Louise, Robin, Tim, Jessica,
Lauren, and Lana.
iii
Table of Contents
Acknowledgements………………...……………………………………………………………ii
List of Figures………..……………..…………………………………………………………..vii
Abstract……………………..………………….............…………………………………...…viii
Introduction ....................................................................................................................... 1
Chapter 1. ......................................................................................................................... 7
Literature review on how sound intensity coding at the auditory nerve ................................. 7
Overview of auditory transduction ................................................................................. 7
Neuroanatomy and physiology of the peripheral auditory system ..................................... 9
Primary auditory neurons encode for sound frequency and intensity .............................. 10
Ribbon synapse properties are organized along the basal pole of the inner hair cell ........ 13
Ion channel properties of auditory neurons are highly diverse ........................................ 16
The role of biophysical properties of auditory neurons in underlying differences in
thresholds and spontaneous rate remains poorly understood. ....................................... 20
A better understanding of how auditory neurons work may provide insight into future
therapies for deafness ................................................................................................ 22
Chapter 2. ....................................................................................................................... 26
Functional diversity in the auditory nerve emerges from a maturational gradient in biophysics
and morphology ............................................................................................................ 26
Introduction ............................................................................................................... 26
Methods .................................................................................................................... 28
Results ..................................................................................................................... 36
iv
Developing spiral ganglion neurons can be classified based on their pattern of
connectivity with hair cells. ...................................................................................... 36
Biophysical properties of developing spiral ganglion neurons change with normalized
basal position ......................................................................................................... 41
Current- and voltage-clamp features together predict normalized basal position ........... 52
Spatial gradients in biophysical properties may be maturational gradients. .................. 57
Discussion ............................................................................................................... 64
Biophysical properties of SGN vary as spatio-temporal gradients along the base of the
inner hair cell. ........................................................................................................ 64
Comparison between Acute and Cultured Preparations ............................................. 67
The biophysical properties of type II SGN compared to type I SGN ............................ 70
The relevance of somatic recordings ........................................................................ 71
Chapter 3. ....................................................................................................................... 73
SGN somatic recordings in the acute semi-intact preparations of cochleae display isopotential
behavior despite their bipolar morphology ....................................................................... 73
Introduction ............................................................................................................... 73
Results ..................................................................................................................... 74
Discussion ................................................................................................................ 81
Chapter 4. ....................................................................................................................... 85
Classifying neurons based on simply measurable biophysical properties ........................... 85
Introduction ............................................................................................................... 85
Methods .................................................................................................................... 86
Results ..................................................................................................................... 88
v
Univariate Data Analysis to predict SGN subtypes using Logistic Regression .............. 88
Multivariant analysis boosts the performance of the model to predict SGN subtypes .... 92
Discussion ................................................................................................................ 95
Chapter 5. ....................................................................................................................... 98
Generation of Inner Ear Hair Cells by Direct Lineage Conversion of Primary Somatic Cells . 98
Introduction ............................................................................................................... 98
Methods .................................................................................................................. 100
Somatic cells transformed into induced hair cells via direct reprogramming ............... 100
Results ................................................................................................................... 101
Passive membrane properties of induced hair cells in each experimental condition ... 101
Voltage-clamp responses of co-cultured induced hair cells are similar to primary hair
cells. ................................................................................................................... 104
Discussion .............................................................................................................. 106
Chapter 6. ..................................................................................................................... 108
Discussion .................................................................................................................. 108
How does diversity in post-synaptic biophysical properties of spiral ganglion neurons
contribute to sound intensity coding? ......................................................................... 108
Limitations of the experimental design and conclusions .............................................. 113
Future Experiments .................................................................................................. 115
Exploring the role of glutamatergic input on SGN subtypes ...................................... 115
Determine whether pre-synaptic ribbon active zone morphology correlates with post-
synaptic ion channel properties of cochlear afferents. ............................................. 118
Exploring the role of efferent input on SGN subtypes .............................................. 120
vi
References ................................................................................................................... 122

vii
List of Figures
Figure 1: ..................................................................................................................................................... 40
Figure 2: ..................................................................................................................................................... 45
Figure 3: ..................................................................................................................................................... 47
Figure 4: ..................................................................................................................................................... 52
Figure 5: ..................................................................................................................................................... 53
Figure 6: ..................................................................................................................................................... 60
Figure 7: ..................................................................................................................................................... 63
Figure 8: ..................................................................................................................................................... 80
Figure 9: ..................................................................................................................................................... 89
Figure 10: ................................................................................................................................................... 93
Figure 11: ................................................................................................................................................. 102

viii
Abstract
In this dissertation, my objective is to test whether the ion channel properties play
a role in shaping the diverse physiology of SGN. I hypothesize that the ion channel
properties of SGN are diverse and can be systematically linked to their function to encode
different sound intensities. By understanding how the ion channel properties play a role
in shaping the function of SGN, we can better understand how sound intensity is encoded
at the auditory nerve. Furthermore, a better understanding of how SGN work may provide
insight into future therapies of deafness, such as hidden-hearing loss, which is selectively
degenerates low-SR SGN.
In chapter 1, I provide a literature review of the concepts and previous experiments
that have led me to develop my hypotheses. I provide an in-depth summary of the
neuroanatomy and physiology of the cochlea, organ of Corti, and the inner hair cell active
zones. In addition, I provide a summary of how studying this question can aid in the
shared public health interest to learn about specific types of deafness. My goal in this
chapter is to show my conceptual understanding of the literature and to provide insight
for the next person who may pick up where I left off when studying this question.
Chapter 2 is the focal point of the dissertation and presents my most significant
contributions to the field. In this chapter, I present the results from a series of experiments
where I perform simultaneous patch-clamp recordings and single cell labelling on acute
semi-intact preparations of cochleae. These results link the cellular biophysics of spiral
ganglion neurons to their putative intensity-coding subgroup, thus, filling our gap in
knowledge in how the biophysical properties of SGN play a role in sound intensity
encoding. Also, we provide an inference statistical model that predicts SGN subtypes
ix
solely on the biophysical properties of the neuron. Not only does this model provide a tool
to predict SGN subtypes, the model concisely defines the relationship of how the
biophysical properties are shaped among the SGN subtypes.
Chapter 3 is a deeper analysis of the electrophysiological measurements
presented throughout the dissertation. The bipolar morphology of SGN in the semi-intact
preparation raises the possibility of inadequate space-clamp and nonisopotential
behavior. It is important to carefully evaluate this possibility since non-isopotential
behavior would mean that the results presented in chapter 2 could stem from unknown
combination of variations in the spatial extent of voltage clamp, somatic ion channel
densities as well as the size of the currents shunting through the axial conductance and
adjacent dendritic compartments. The analysis shows that the SGN recorded in the semi-
intact preparation largely have as isopotential compartments, and therefore, lead us to
not have significant concern for our biophysical assessment.
Chapter 4 expands upon the prediction models introduced in chapter 2. Here, I
reduced the demand of the model to predict just one of two categories of SGN (putative
high-SR or low-SR SGN) based on their biophysical properties. As a result of these loosen
model constraints, I was able to use methods to validates the model’s accuracy that were
not available in the initial model fitting process. The bimodal classifier prediction model
presented here would benefit the field’s progress in easily separating and studying these
two subgroups of SGN.
Chapter 5 details my contribution of the collaborative project performed by
members the Segil, Ichida, and Kalluri laboratories. The collaboration focused on
providing an experimental protocol to directly reprogrammed somatic cells into cochlear
x
hair cell-like cells. Here, I show how the electrophysiological properties of these hair cell-
like cells are similar to primary hair cells. The results produced in this chapter add to a
larger body of work showing that the molecular, morphological, and physiological
properties of the direct reprogrammed cells match those of normal hair cells. With this
study, we contribute to the field of restorative and regenerative hearing science by
providing a method to produce large quantities of hair cell-like cells to test the lethality of
potential ototoxins and preventative treatments.
Finally, in chapter 6, I discuss the overall conclusions of the dissertation including
how the biophysical properties of spiral ganglion neurons play a role in sound intensity
coding. I provide a summary of the limitations and setbacks I encountered during the
course of performing these experiments. Lastly, I detail future experiments I believe would
be beneficial to the field that directly stem for the ideas and results provided in this
dissertation.

1
Introduction
Sound is a pressure wave. These waves are fed through the tunnel of the external
auditory canal and transmitted through a series a membranes and bones in the middle
ear which transform the pressure wave into a mechanical force that strikes upon the
sensory organ of the peripheral auditory system, the cochlea. Sound has two
components, frequency and intensity, and the cochlea manages these two components
in separate ways. Frequency is encoded along the length of the cochlea, where high-
frequency sounds are encoded at the base of the cochlea while low-frequency are
encoded at the apex. The spatial separation of frequency coding at the cochlea is known
as its tonotopic map and is established by the structural properties of the cochlea.
At any frequency, the sound intensity can vary. Normal hearing allows us to hear
a vast dynamic range of sound intensities: from 0 dB to 120 dB. This range is equivalent
to a whisper all the way to the roaring of a jet engine. Intensity information is encoded by
the primary auditory neurons, the spiral ganglion neurons (SGN). The focus of this
dissertation is to understand the cellular biophysical mechanisms that enable sound
intensity encoding at the auditory nerve.
An overarching hypothesis in the field is that the full range of sound intensities to
be encoded in normal hearing is achieved using a population code across three broad
groups of SGN. In pioneering studies, in vivo sharp electrode recordings showed that the
SGN population consists of three distinct subgroups based on how these neurons
responded to increasing levels of sound intensity (Kiang et al., 1976; Liberman, 1978).
The first in vivo characteristic that defines each subgroup is observed when there was no
sound stimulus. When no sound stimulus is present, one subgroup of SGN that has a
2
relatively high spontaneous discharge rate (dubbed high-SR), one group that has a
relatively low spontaneous discharge rate (low-SR), and a third that is in the middle
(medium-SR). As the Kiang et al experimenters increased the stimuli’s sound intensity,
the high-SR SGN subgroup showed to be the most sensitive and increased its firing rate,
followed by the medium-SR, followed by the low-SR subgroup. As the sound level
continues to rise, each SGN subgroup monotonically increases their firing rates until they
reach a point of saturation. The high-SR/highly sensitive subgroups are the first to
saturate, followed by the medium-SR subgroup, and finally the low-SR subgroup. I
focused on the following question in this dissertation: what are the mechanisms that allow
SGN subgroups to be differentially sensitive to sound intensity ?
There are a wide range of hypotheses being considered. In summary, previous
studies have hypothesized that each inner hair cell conveys sound intensity information
into three SGN subgroups via differential activity of the pre-synaptic ribbon active zone
that makes a one-to-one pairing with a single SGN terminal. Ribbon active zones are
specialized to make precise, high-throughput synaptic transmission with its afferent
target. Within each hair cell, there are approximately 30 ribbon active zones with a large
variance in their structural properties. It is hypothesized that variability in ribbon synapse
morphology leads to the diversity in neuronal responses (Meyer et al. 2009). First, the
size of the ribbon varies with each ribbon synapse. Variations in ribbon size is a proxy for
the size of the active zone which would allow for different quantities of synaptic vesicles
to be primed and readily available for glutaminergic exocytosis (Martinez et al., 1997;
Liberman et al 2016; Kalluri et al., 2017). Additionally, there are variations in the densities
of calcium channels at each ribbon active zone. Calcium channel density directs the
3
probability of synaptic vesicle release which may also determine how much and when
synaptic transmission occurs (Ohn et al. 2013). These observations together suggest that
SGN could be differentially activated by the concentration of synaptic transmission
coming from each ribbon active zone.
There are counter-intuitive relationships that weaken these ribbon active zone-
focused hypotheses. For example, SGN that are pair with large ribbon active zones
display low-spontaneous discharge rates, but display large and fast EPSCs, which is
unlike the relationships in sensory systems such as retinal bipolar neurons. In retinal
bipolar neurons, large ribbon active zones are paired with high spontaneously discharging
neurons (Mehta et al., 2013).
Whether the post-synaptic properties of these synapses contribute to intensity
sensitivity remains poorly understood, although there is some evidence to suggest that
this might be the case. For example, the expression of glutamate receptors and the
kinetics of the excitatory post-synaptic currents (EPSC) are different across synapses
(Liberman et al. 2011; Goutman and Glowatzki, 2011; Grant et al., 2010). The magnitude
and summation of mEPSCs could lead to differences in neuronal responses, but no one
has yet linked post-synaptic responses to the established sound intensity SGN
subgroups.
What remains untested is whether the ion channel properties of SGN play a role
in sound intensity coding. The ion channel properties determine the threshold and
excitability of a neuron, and therefore, are a vital piece of information when considering
how the sound intensity is encoded at the ribbon synapses of the inner hair cell. There is
more focus on how the pre-synaptic ribbon active zone solely instructs the SGN. The
4
field’s conception is that the ion channel properties of SGN are homogenous and most
likely do not play a role in differentiating the SGN subgroup’s physiology.

In this dissertation, my objective is to test whether the ion channel properties play
a role in shaping the diverse physiology of SGN. I hypothesize that the ion channel
properties of SGN are diverse and can be systematically linked to their function to encode
different sound intensities. By understanding how the ion channel properties play a role
in shaping the function of SGN, we can better understand how sound intensity is encoded
at the auditory nerve. Furthermore, a better understanding of how SGN work may provide
insight into future therapies of deafness, such as hidden-hearing loss, which is selectively
degenerates low-SR SGN.
In chapter 1, I provide a literature review of the concepts and previous experiments
that have led me to develop my hypotheses. I provide an in-depth summary of the
neuroanatomy and physiology of the cochlea, organ of Corti, and the inner hair cell active
zones. In addition, I provide a summary of how studying this question can aid in the
shared public health interest to learn about specific types of deafness. My goal in this
chapter is to show my conceptual understanding of the literature and to provide insight
for the next person who may pick up where I left off when studying this question.
Chapter 2 is the focal point of the dissertation and presents my most significant
contributions to the field. In this chapter, I present the results from a series of experiments
where I perform simultaneous patch-clamp recordings and single cell labelling on acute
semi-intact preparations of cochleae. These results link the cellular biophysics of spiral
ganglion neurons to their putative intensity-coding subgroup, thus, filling our gap in
5
knowledge in how the biophysical properties of SGN play a role in sound intensity
encoding. Also, we provide an inference statistical model that predicts SGN subtypes
solely on the biophysical properties of the neuron. Not only does this model provide a tool
to predict SGN subtypes, the model concisely defines the relationship of how the
biophysical properties are shaped among the SGN subtypes.
Chapter 3 is a deeper analysis of the electrophysiological measurements
presented throughout the dissertation. The bipolar morphology of SGN in the semi-intact
preparation raises the possibility of inadequate space-clamp and nonisopotential
behavior. It is important to carefully evaluate this possibility since non-isopotential
behavior would mean that the results presented in chapter 2 could stem from unknown
combination of variations in the spatial extent of voltage clamp, somatic ion channel
densities as well as the size of the currents shunting through the axial conductance and
adjacent dendritic compartments. The analysis shows that the SGN recorded in the semi-
intact preparation largely have as isopotential compartments, and therefore, lead us to
not have significant concern for our biophysical assessment.
Chapter 4 expands upon the prediction models introduced in chapter 2. Here, I
reduced the demand of the model to predict just one of two categories of SGN (putative
high-SR or low-SR SGN) based on their biophysical properties. As a result of these loosen
model constraints, I was able to use methods to validates the model’s accuracy that were
not available in the initial model fitting process. The bimodal classifier prediction model
presented here would benefit the field’s progress in easily separating and studying these
two subgroups of SGN.
6
Chapter 5 details my contribution of the collaborative project performed by
members the Segil, Ichida, and Kalluri laboratories. The collaboration focused on
providing an experimental protocol to directly reprogrammed somatic cells into cochlear
hair cell-like cells. Here, I show how the electrophysiological properties of these hair cell-
like cells are similar to primary hair cells. The results produced in this chapter add to a
larger body of work showing that the molecular, morphological, and physiological
properties of the direct reprogrammed cells match those of normal hair cells. With this
study, we contribute to the field of restorative and regenerative hearing science by
providing a method to produce large quantities of hair cell-like cells to test the lethality of
potential ototoxins and preventative treatments.
Finally, in chapter 6, I discuss the overall conclusions of the dissertation including
how the biophysical properties of spiral ganglion neurons play a role in sound intensity
coding. I provide a summary of the limitations and setbacks I encountered during the
course of performing these experiments. Lastly, I detail future experiments I believe would
be beneficial to the field that directly stem for the ideas and results provided in this
dissertation.

7
Chapter 1.
Literature review on how sound intensity coding at the auditory nerve
Overview of auditory transduction
Sound has two components, frequency and intensity, and each component has a
perceptual correlate. Frequency is the rate at which the pressure of sound waves
changes in a second. When one perceives a frequency, they hear the pitch of the sound.
Sound waves with high frequencies have a high pitch, and sound waves with low
frequencies have a low pitch. Humans are able to hear a wide range of frequencies from
20 to 20,000 Hz, separately and simultaneously, which is important for our ability to hear
harmonics of speech and music (Plack et al., 2014). At any particular frequency, the
intensity of the sound stimuli varies. High intensity, large amplitude pressure waves are
perceived as loud sounds, while low intensity, small amplitude pressure waves are
perceived as soft sounds. The auditory system is uniquely set up to take in the pressure
wave stimuli of sound and encode its frequency and intensity in a process known as
auditory transduction. Here, I will begin with a brief overview of auditory transduction of
the peripheral auditory system.
Auditory perception begins when sound waves travel to the outer ears. The outer
ears consist of cartilaginous structures that provide a cavity to capture and modify the
acoustic signal by amplifying the sound pressure onto the tympanic membrane. The
tympanic membrane is connected to a series of bone structures in the middle ear that
serve as an apparatus to transfer the sound energy into a mechanical force. This is
accomplished by matching the low-impedance air to the high-impedance fluid-filled inner
ear organ, wherein little to no energy is reflected. The mechanical force of sound then
8
strikes the sensory organ of the inner ear, the cochlea, where pitch is first represented as
a traveling wave.
Sound energy displaces the basilar membrane of the cochlea based on the
frequency of the stimulus. For any particular frequency there is an area of the basilar
membrane that has a maximum displacement, which is established by the collective
mechanical properties of the cochlea (i.e. arrangement of collagen/elastin). Von Békésy
originally observed this in a series of experiments measuring the movement of the
cochlea in human cadavers (Békésy, 1960). The cochlea’s ability to confine maximum
displacement at particular areas of the basilar membrane allows for frequencies to be
separately processed along the length of the cochlea. This process is known as tonotopy,
and it is seen in human and numerous animal models. The tonotopy of the cochlea allows
high frequencies to be processed at the base of the cochlea and low frequencies to be
processed at the apex. Furthermore, the cochlea has non-linear functionality to heighten
the displacement of the basilar membrane, which further sharpens its frequency
selectivity (Robles et al, 2001).
The cochlea consists of a coiled, hollow tube that consists of three fluid-filled
chambers: the scala vestibuli, scala media, and scala tympani. Within the scala media,
the basilar membrane along with a series of connecting membranes mechanically
activate the primary neural circuitry of the cochlea, known as the organ of Corti. The organ
of Corti consists of specialized hair cells and supporting cells that are mechanically
activated by the traveling wave of sound energy (Hudspeth, 1982). The displacement of
the basilar membranes triggers the stereocilia of hair cells to deflect back and forth. The
deflection increases the tension of tip-links between the stereocilia which results in
9
opening mechanically gated ion channels embedded at the ends of the stereocilia. The
opening of these ion channels activates the hair cells. Hair cell activation leads to synaptic
transmission with the first neuronal cell type in the auditory system, the spiral ganglion
neuron. Spiral ganglion neurons (SGN) make up the auditory nerve, the site where sound
is first encoded as a neuronal signal. SGN are bipolar neurons whose cell bodies are
located in ganglia within the organ of Corti distal to the hair cells. From the cell body, each
SGN have two projecting neurites, one peripheral neurite that synapses onto hair cell(s)
and one central neurite that projects towards the brainstem via the eighth cranial nerve.
Because SGN are the primary neuronal cell type in the peripheral auditory system, it is
the primary site of auditory encoding, and therefore, understanding the underlying
mechanisms that allow SGN to encode sound are of high interest.
Neuroanatomy and physiology of the peripheral auditory system
The neural circuitry of the peripheral auditory epithelium consists of three rows of
outer hair cells and a single row of inner hair cells. Distal to the hair cells, bipolar spiral
ganglion neuron (SGN) cell bodies are located within a ganglion and project one neurite
to the hair cells of which they form a synapse and one neurite towards the brainstem via
the eighth cranial nerve. There are two distinct subtypes of SGN within the ganglia: Type
I SGN represent 95% of the population and synapse exclusively with inner hair cells; type
II SGN represent ~5% of the population and project past the inner hair cells, radially turn,
and synapse with multiple outer hair cells (Spoendlin, 1969; Kiang et al., 1984; Liberman
and Oliver, 1984). Efferent innervation including lateral olivary complex (LOC) fibers
synapse onto SGN afferent terminals and medial olivary complex (MOC) fibers terminate
throughout the outer hair cell region (Liberman et al., 1990) (Maison et al., 2003).
10
As the travelling wave of sound energy moves throughout the epithelium, the outer
hair cells function to amplify the displacement of the basilar membrane to a specific
location of the tonotopic map of the cochlea. Once mechanically active, the inner hair
cells trigger synaptic transmission with type I SGN. Thus, sound encoding information is
initiated by the inner hair cell induced activity of type I SGN. Overstimulation of the
cochlea is compensated by limiting the amplification of basilar membrane displacement
by the outer hair cells by the activation of MOC fibers (Liberman, 1988). Although not fully
understood, it is hypothesized that the type II SGN afferent function to sense
overexcitation of the cochlea, which in turn may activate MOC fiber (Heil et al., 2015;
Reijntjes et al., 2016; Maison et al., 2003). Furthermore, the role of LOC efferent fibers
remains poorly understood; however, their location on the type I SGN terminals speculate
a role in controlling the activity of type I SGN during synaptic transmission with its pre-
synaptic hair cell (Arnold et al., 1998; Felix et al., 1992; Groff et al., 2003).
Primary auditory neurons encode for sound frequency and intensity
Frequency encoding is performed along the length of the cochlea. High-frequency
sound information is encoded at the basal end, while low-frequency information is
encoded at the apex. The structural organization of the cochlea to encode frequency
along the length of cochlea is known as the tonotopic map. SGN are also tonotopically
activated and its characteristic frequency (CF) is the frequency at which an SGN
preferentially responds. Pioneer studies using sharp electrode in vivo recordings
characterized the SGN population frequency selectivity into discrete tuning curves (Kiang,
1965; Kiang et al., 1967; Liberman and Kiang, 1978). The tuning curves are a
measurement of the neurons’ action potential (spike) behavior over a range of sound
11
levels (dB) as a function of sound frequency. As the experimenters decreased the sound
level, the rate of change in the neuron’s spike pattern changed for a narrow band of
frequencies until the CF was observed at its threshold activation. The tuning curves
reveal how frequencies are represented in the neural code of the auditory nerve. Each
SGN selectively codes for a different frequency and represent this selectiveness in their
firing patterns.
Kiang et al observed that SGN with the same CF displayed tuning curves with
differential threshold activation (Kiang, 1965). At any particular CF, there are three
physiologically distinct subgroups of type I SGN that differ in their threshold, spontaneous
discharge rate, and dynamic range of sound level activation (Kiang 1965; Liberman and
Kiang 1978). These subgroups of SGN are found in numerous mammalian models
including cat, mouse, rat, gerbil, chinchilla, and macaque monkey (Liberman, 1978;
Temchin et al., 2008; Borg et al., 1988; Huet et al., 2016). In vivo sharp electrode
recordings were performed to characterize the electrophysiological properties of these
subgroups of type I SGN in response to increasing levels of sound intensity. When there
is no stimulus presence, there is one subgroup of SGN that have a relatively high
spontaneous discharge rate (SR) which range is species-dependent (cats = >20Hz).
There is a second subgroup of SGN that have a low spontaneous discharge rate (cats 0-
9Hz), and a third that has a medium spontaneous discharge rate (cats 10-30Hz).
Currently, the convention in the cochlea neurophysiology field is to refer to these
subgroups of SGN as high-, medium, and low-SR type I SGN. The distribution of high-,
medium, and low-SR SGN are species dependent. In rat and cat, high-SR SGN are the
12
most abundant making approximately 60% of the total type I SGN population (Liberman,
1978; Taberner et al., 2005).
As sound intensity increases, high-SR SGN are the first subgroup to increase their
firing rate, and therefore, have the lowest thresholds (Liberman, 1978; Relkin et al., 1997;
Heinz, 2004). As sound intensity continues to increase, medium-SR SGN increase their
firing rate, followed by low-SR SGN. This behavior showcases the relationship between
the spontaneous discharge rate and in vivo thresholds of the three SGN subgroups: high-
SR/low-threshold, low-SR/high-threshold, medium-SR/medium-threshold. The dynamic
range of activation, limited by maximum firing rate of a single neuron, differs between the
subgroups. High-SR SGN are the first to saturate their firing rate, followed by medium-
SR, and lastly low-SR SGN. The sound level at which each SR-subgroup saturates is
also species dependent. In mouse, high-SR units saturate after 20dB, medium-SR units
saturate around 80dB, while low-SR units are not seen to saturate even after 80dB of
sound stimuli.
Normal hearing enables humans to hear a vast dynamic range of sound intensities
from 0dB to 120dB (Hellman 1990). In relative terms, that range is from a whisper (0-1dB)
to a roaring jet engine (120dB). The overarching theory is that the population code of all
three SR-subgroups of SGN are needed to encode the full dynamic of sound intensities.
However, the underlying cellular and molecular mechanisms that differentiate the three
SR-groups remain poorly understood. This dissertation focuses on novel research to
provide insight on differences associated with the three SR-subgroups of type I SGN.
13
Ribbon synapse properties are spatially organized along the basal pole of the inner hair
Studies that performed in vivo sharp electrode recordings followed by single cell
injected of horseradish peroxidase reported that SR-subgroups somatic and terminal
morphology displayed in various spatial axis within the ganglia and along the base of the
inner hair cell where SGN terminate (Liberman, 1978). For example, within the ganglia,
low-SR SGN were located towards the scala vestibuli while high-SR SGN were located
towards the scala tympani. SGN bipolar peripheral neurites maintain this organization as
they project towards to the inner hair cell. High-SR SGN exclusively synapse on the
adneural face of the inner hair cell, adjacent to the pillar cells of the organ of Corti, while
low-SR SGN are solely found on the abneural face of the inner hair cell, proximal to the
modiolus of the cochlea (Merchan-Perez and Liberman, 1996). These observations have
been used to establish a classification scheme along the modiolar to pillar axis of the
inner hair cell.
Continuing studies within the field have focused on the synaptic physiology and
anatomy of the inner hair cell-SGN synapse. Each inner hair cell establishes synapses
with approximately 30 SGN in a one-to-one pairing, where one inner hair cell is the sole
driver of each SGN’s activity. An inner hair cell-SGN synapse is a specialized ribbon
synapse, where there is a pre-synaptic active zone that singularly contacts the terminal
of a single SGN. Ribbon synapses function to aggregate large quantities of synaptic
vesicles for high-output exocytosis, and therefore, there has been a large focus on
studying the anatomy and physiology of these ribbon synapses in part to discovering
mechanism that underlie differences in the SR-subgroups of type I SGN. The pre- and
post-synaptic components of the ribbon synapse have been correlated with the modiolar-
14
pillar spatial axis and have been used to develop numerous hypotheses on how
differential synaptic components lead to differential SR-subgroups.

Pre-synaptic and post-synaptic components varies along the modiolar-pillar axis of the
base of the inner hair cell:
1. Ribbon size
The pre-synaptic ribbon is a structure composed on cytoskeletal proteins including
RIBEYE and Bassoon. The function of the ribbon is to provide a tethering structure for
large amounts synaptic vesicles at a specific location. The ribbon permits high rates of
exocytosis and replenishes a single inner hair-SGN synapse with vesicles for sustained
synaptic transmission. Previous studies have quantified the size of the ribbon via immuno-
labelling for RIBEYE and Bassoon. In the mature mouse and rat, larger ribbons are found
on the modiolar-side of the inner hair cell, while small ribbons are found on the pillar side
(Liberman et al., 2016; Kalluri et al., 2017). This observation has led to the hypothesis
that large modiolar-side ribbons have the structure to hold more vesicles than smaller
pillar-side ribbon, and therefore, modiolar-side ribbon synapse may provide larger
amplitude and/or rate of glutamate quanta.
2. Calcium channel density
Pre-synaptic calcium ion channels are clustered at each ribbon synapse and
function to trigger voltage-dependent exocytosis of ribbon-tethered synaptic vesicles.
Larger densities of calcium channels are located on the modiolar-side than on the pillar-
side of the inner hair cell (Ohn et al., 2016). This observation leads the hypothesis that
15
with more calcium channel densities, modiolar-side ribbon synapses have a higher
probability of initiating exocytosis than pillar-side ribbon synapses.
3. Glutamate receptor density
Glutamate receptors are located on the terminals of SGN. Immuno-labelling has
shown in mature mice models larger concentrations of glutamate receptors (GluR2/3) on
pillar-contacting SGN than modiolar-contacting SGN (Liberman and Liberman, 2016).
Contrary findings have been reported in rat models where no differences in GluR2/3
signal was measured along the modiolar-pillar axis of SGN terminals.
4. PSD signaling/GluR-subtypes
Liberman et al reported a concentration gradient of GluR2/3 signaling along the
modiolar-pillar axis, these investigators reported an equal distribution of post-synaptic
density marker (PSD95) among modiolar- and pillar-contacting SGN (Liberman et al.,
2016). PSD95 is a tethering protein for glutamate receptor subtypes, and therefore, it is
possible for other subtypes of GluR, such as GluR4, to populate the modiolar-contacting
SGN at an equal concentration that GluR2/3 populated pillar-contacting SGN. GluR4 is
not easily quantified via immuno-staining and therefore could not be measured in the
same matter as GluR2/3.
5. Spike initiation zone
The spike initiation zone of SGN is located proximal to the ribbon synapse where
large concentrations of voltage-gated sodium channels are positioned near the habenula.
When synaptic transmission is initiated, the activation of the ribbon synapse releases
glutamate into the synaptic cleft and bind with GluR of the SGN terminal. GluR activation
induces excitatory post synaptic currents (EPSC), followed by an excitatory post-synaptic
16
potential (EPSP). The EPSP travels down the terminal to the spike initiation zone which
activates voltage-gated sodium channels and triggers an action potential. In addition to
sodium channels, voltage-gated ion channels, such as potassium (K), hyperpolarization-
activated mixed cationic channel (HCN), and leak channels, are located at the terminal.
The composition and function of these ion channels to the encoding performed at the
inner hair cell-SGN synapse are poorly understood. However, K-channels and HCN-
channels are known to control the thresholds and excitability of a neuron, and therefore
are hypothesized to underlie the function of the SGN. In this dissertation, I set out to
characterize the ion channel composition of individual SGN and test whether the ion
channel composition of SGN systematically differentiate along the modiolar-pillar axis.

Ion channel properties of auditory neurons are highly diverse
The ion channel properties of spiral ganglion neurons (SGN) have been
characterized via patch-clamp electrophysiological recordings in cultured, dissociated
ganglia preparations (Mo and Davis 1997; Liu et al., 2014; Lv et al., 2010). In the
dissociated preparations, the SGN is removed from its pre-synaptic partner of which
drives the behavior of the neuron. These foundational patch-clamp recordings showed
that the ion channel properties of SGN are heterogeneous in their spike patterns and
underlying ion channel composition. Therefore, there is a large amount of diversity in
post-synaptic features that are independent of pre-synaptic influence from the inner hair
cell.
This heterogeneity is best illustrated by visualizing the action potential behavior in
response to a step of injected current. The spike patterns can be generally segmented
17
into groups based on the time the spike behavior takes before reaching a steady state,
as known as the accommodation. Rapidly accommodating SGN will fire a single action
potential in response to a step of injected current. Second, slow-accommodating SGN will
firing multiple action potentials in response to a step of injected current before reaching a
steady-state potential. Third, intermediately accommodating SGN will fire one-to-two
action potentials in response to a step of injected current.
The prevalence of each accommodation-subgroup of SGN is dependent on the
tonotopic location of where the SGN is located and the holding potentials one applies to
during the patch-clamp recordings. Previous studies have recorded from SGN from the
base, middle and apical turn of the cochlea and have reported that rapidly
accommodating SGN are primarily found in the basal turn, while slow-accommodating
SGN are primarily found in the apical turn (Crozier and Davis, 2014; Adamson et al. 2002).
The middle turn consists of the most even distribution of accommodation-subgroups.
Additionally, the distribution changes depending on the holding potential one applies
during the recordings. When Davis et al hyperpolarized the holding potential from -60mV
to –80mV, they report a rise in the prevalence of intermediately- and slowly-
accommodating SGN. This increase in prevalence is most likely due to the more
hyperpolarizing holding potential relieving inactivation of voltage-gated sodium channels.
Furthermore, the distribution of accommodation-subgroup is maintained by the
neurotropic factors that are implemented by the culturing conditions of the experiments
and that flow through the cochlea in vivo (Flores-Otoro et al. 2007; Sun and Salvi 2009).
Two neurotropic factors needed for the normal development of the cochlea are
neurotropic factor 3 (NT-3) and brain derived neurotropic factor (BDNF). NT-3 and BDNF
18
are excreted by the cells of the organ of Corti and are speculated to regulate the neural
circuitry of the cochlea (Fritzsch et al., 1997). There is a spatial gradient in the
concentrations of NT-3 and BDNF along the apical-basal axis of the cochlea. In
embryonic stages, the cochlea has higher concentrations of NT-3 in the apex than at the
basal. The effects of NT-3 are most salient in NT-3 knockout mice where the basal turn
of the cochlea does not generate, while parts of the middle and apical turn develop
(Fritzsch et al., 1997). Over the course of post-natal development, the concentration of
NT-3 decreases in a spatial gradient from the apex to basal turn. Similarly, BDNF is also
displayed in a spatial gradient along the apical-basal axis, but in the reverse direction.
Larger concentrations of BDNF are found in the basal turn than in the apex.
NT-3 and BDNF have been found to be essential to the development and
maintenance of the cochlea. Flores-Otoro et al added these neurotropic factors to the
culture media that used in the cultured, dissociated ganglia preparations; however, these
neurotropic factors have been shown to influence ion channel properties of SGN (Flores-
Otoro et al. 2007). As noted, there is a spatial gradient of NT-3 and BDNF and a spatial
gradient in the distribution of accommodation-subgroups of SGN along the apical-basal
axis. In vitro experiments where apical turn SGN were cultured in a media that mimicked
the neurotropic conditions of the basal turn, transformed the spiking patterns of apical
SGN into the spiking phenotype of basal SGN (Adamson et al., 2002). These
observations show that the extrinsic factors that the SGN are influenced via in vivo and
in vitro conditions highly impact the biophysical properties of SGN.
Diversity in the spike patterns are not only seen along the apical-basal axis, but
also observed in local regions of the cochlea. The middle turn of the cochlea is where
19
spike properties are the most diverse (Adamson et al., 2002). Also, in the middle turn, a
similar scale of diversity of SR-subgroups is observed in the middle turn of the cochlea,
supporting the idea that the post-synaptic biophysical properties of SGN play a role in
establishing the SR-subgroups (Liberman, 1978). Middle turn SGN display a wide range
of biophysical properties including thresholds, resting potentials, and membrane time
constants. Hypotheses on how the biophysical properties of SGN could elicit different SR-
subgroup include having neurons with more depolarized resting potentials needed
smaller amounts of stimuli to induce an action potential. This depolarized resting potential
with low-threshold behavior in combination with a slow time constant would enable a
neuron to produce high-SR firing behavior. Low-SR firing behavior could therefore by
established with the reverse relationship: hyperpolarized resting potentials needing larger
amounts of stimuli to produce an action potential (high-thresholds) with fast time constant
could produce low-SR firing behavior. Heterogeneity in these biophysical properties
among the SGN population could therefore establish a full range of responses (i.e.
medium-SR behavior) that are not completely dependent on the pre-synaptic influence of
the inner hair cell.
Previous experimenters have characterized the specific ion channels that
contribute to the diverse firing patterns of SGN via pharmacological studies. In these
studies, Liu et al applied ion channel antagonists to block the currents flowing from
specific ion channels and measured the over size of the currents and the effects of the
currents to the action potential behavior (Liu et al., 2014a; Liu et al., 2014b). In particular,
ion channels that affect the resting potential and threshold of the SGN were of particular
focus due to their potential biophysical role in shaping the SR-subgroups. Low-voltage
20
gated currents are composed of multiple type of ion channels including Kv1 and HCN.
When these ion channels are pharmacologically blocked in vitro, there is a change in the
thresholds, resting potentials, and overall biophysical behavior. Blocking Kv1 depolarizes
the resting potential and lowers the amount of current needed to elicit an action potential
(lowers thresholds) (Liu et al., 2014). Blocking HCN, hyperpolarizes the resting potential,
bringing it further away from its original voltage-threshold (Liu et al., 2014). The role of
low-voltage gated potassium currents (IKL) and hyperpolarization gated mixed cationic
current (Ih) in regulating the threshold and excitability of a neuron is well characterized in
octopus cells of the cochlear nucleus (Goldings et al., 2012). These experiments set the
stage for a hypothesis that low-voltage gated currents, driven by Kv1 and HCN channels,
underlie the excitability of the SGN, and therefore, differences in concentrations of Kv1
and HCN at a local region of the cochlea could establish a post-synaptic role in producing
SR-subgroups.

The role of biophysical properties of auditory neurons in underlying differences in
thresholds and spontaneous rate remains poorly understood.

There have been extensive studies on the ion channel properties of SGN through
patch-clamp recordings, immuno-labelling, and single-cell RNA-sequencing experiments;
however, there has been no study that directly links the diversity of ion channel properties
to the SGN SR-subgroups. Davis et al has performed numerous patch-clamp experiments
to characterize the ion channel properties of SGN; however, because these experiments
were performed in cultured, dissociated preparations, there is no way to associate specific
21
ion channel properties to any SR-subgroups. Furthermore, these experiments lack a
control for the influence of neurotropic factors on the ion channel properties of SGN during
the culturing period. A similar gap in knowledge is seen in recent single-cell RNA-
sequencing studies that have used molecular expression levels of post-synaptic features
to subdivide type I SGN into three subgroups. Although, these datasets provide valuable
information on the relative expression levels of RNA specifically tied to specific post-
synaptic properties that can underlie functionality, there is no way to link these RNA-
subdivided groups to the behavior of the SGN and/or the spatial organization of the SGN
innervation that the SR-subgroups align themselves in the organ of Corti.
More recently, the use of acute semi-intact preparations of cochleae have been
used to control for neurotropic influence and to take advantage of in situ spatial
organization of SR-groups along the modiolar-pillar axis along the base of the inner hair
cell. In experiments using semi-intact epithelial preparations, Rutherford et al. (2012)
measured the ion channel properties of SGN terminals (most likely) from the modiolar
face of the inner hair cell. Although this approach preserved some information about the
position of the terminals on the intensity coding axis, the difficulty in accessing pillar
contacting terminals means that they were unable to collect a representative sampling
from all types of terminals (Rutherford et al., 2012).
In order to measure the ion channel properties of SGN along the intensity-coding
axis, I will use an approach that combines advantages of a semi-intact epithelium
approach, where the modiolar to pillar intensity coding axis is preserved, with a somatic
approach where we can record from all types of SGN.
22
A better understanding of how auditory neurons work may provide insight into future
therapies for deafness

Noise-induced hearing loss (NIHL) can be defined as damage to the cochlea, the
sensory epithelium, and/or the spiral ganglion neurons (SGN) after exposure to sound
stimulus. Damage to any of these areas can result in disrupting auditory synaptic
transmission and perception (Kerr and Byrne 1975; Liberman, 2016). One type of NIHL
that overlaps with my dissertation is hidden-hearing loss, a type of NIHL that is caused
by damage to SGN synapses via cochlear synaptopathy. People who have hidden
hearing loss exhibit a limited range of sound intensity perception (Furman et al., 2013;
Yevgeniya et al., 2013; Kujawa and Liberman, 2015; Moser and Starr, 2016). The work
presented in this dissertation provides insight into understanding the mechanisms by
which SGN encode sound information in people with normal hearing which we may
provide insight into future therapies for deafness that hope to protect and/or restore the
function of these neurons.
The markers for hidden-hearing loss are not seen using the typical diagnostic tool
to measure hearing loss, otoacoustic emission test (OAE), which measures the cochlea’s
structural health (Kemp, 2002). In hidden-hearing loss, there is no hair cell damage, and
therefore, the OAE reports the patient as normal; thus, providing its ‘hidden’ namesake
(Liberman, 2009; Furman et al., 2013). In addition to the pathology being ‘hidden’ to OAE,
ABR threshold tests can also report false negatives due to the noise-induced damage
only affecting the high-threshold neurons (aka low-SR SGN) (Furman et al., 2013). The
low-threshold SGN that survive would mediate a normal response rate of the ABR, and
23
only when sound level pressures are above 80 dB would the ABR report any abnormal
changes.
Spiral ganglion neuron degradation is a slow process compared to the hair cells.
Therefore, it is referred to as a secondary degradation. Although the cell bodies remain
intact for a couple of weeks after noise-induced trauma, the synapses between spiral
ganglion neurons and hair cells are impacted 24 hours after acoustic trauma (Liberman,
1980; Robertson, 1983; Liberman, 2013). When observing the terminals of spiral ganglion
neurons at the inner hair cell, investigators have reported large swellings at the synaptic
active zones. The swellings are followed by degradation of synapses (Kujawa et al., 2009;
Furman et al., 2013). The synaptopathy has also been seen independent of hair cell
damage. Furthermore, there is an organization to the synaptopathy where afferent fibers
that code for high intensities are more susceptible to degradation than those that code for
low intensities. The field is reconsidering its understanding of a “secondary degradation”
to expand upon the perceptual effects of somatic versus terminal degeneration. Overall,
there is currently a large focus in better understanding what causes synaptopathy and
why one subgroup of SGN (low-SR/high-threshold SGN) are more susceptible to this type
of degeneration.
The leading hypothesis is that hidden hearing loss is caused by glutamate
excitotoxicity, which leads to swelling and destruction of cochlear afferent terminals
(Liberman 2009; Robertson, 1983; Yamasoba et al., 2005). Inner hair cells release
glutamate as an excitatory neurotransmitter onto the afferent terminals during synaptic
transmission. As sound levels increase, the amount of glutamate release also increases,
which can lead to overstimulation of post-synaptic receptors. This overstimulation can
24
lead to toxic levels of neurotransmitter release, causing deleterious sodium, potassium,
and chlorine ion flux within the post-synaptic terminal. The ion flux causes an osmotic
shift, which results in increasing amounts of fluid into the terminal, followed by swelling,
and, ultimately, rupturing of the plasma membranes of afferent terminals (Le Prell et al.,
2001). In support of this mechanism, when experimentally applying glutamate blockers to
the cochlea, terminal swelling and destruction, and thresholds shifts are reduced
(Henderson, 2006).
Maintaining the ion concentrations within the cochlear fluids is critical for normal
hearing, and therefore, a disruption ion balance causes hearing loss (Wang et al., 2002).
The organ of Corti is bathed in two cochlear fluids, endolymph and perilymph. The ion
concentration of these fluids is maintained by the circulation of fluids with the lateral wall
of the organ of Corti and the stria vascularis. Damage to the stria vascularis can cause
hearing loss, but there is a second effect disrupting the levels of potassium ions in the
endolymph. The endolymph osmolarity and ion concentration are critical to the hair cells’
ability to function. Noise-induced trauma can lead to supporting cell loss in the lateral wall
and stria vascularis, which can result in destruction of sodium-potassium pumps within
these structures (Wang et al., 2002; Henderson, 2006). The disrupted pumps, which are
vital to the stability of the endolymph, could play a major role in the underlying
mechanisms of NIHL in hidden hearing loss, but their role still remains unclear.
As previously mentioned, cochlear synaptopathy observed in hidden-hearing loss
selectively degenerates low-SR/high-threshold SGN (Furman et al., 2013). Within this
dissertation, I provide a tool to identify SGN subtypes that can be used to discover the
reasons why one subtype of SGN is more susceptible to noise induced damage.
25
Moreover, by providing an experimental platform to identify SGN subtypes, future
experiments can test for therapies to restore hearing.
26
Chapter 2.
Functional diversity in the auditory nerve emerges from a maturational gradient in
biophysics and morphology

Introduction
The bipolar afferent neurons of the auditory system are the principle conduits for
information transfer from the sensory periphery to the brainstem. In mature mammals,
one inner hair cell provides input to type I spiral ganglion neurons (SGN) with different
response properties (Liberman, 1982). Some SGN have high rates of spontaneous
discharge and respond sensitively to low-intensity sounds (high-SR group), whereas
others have low rates and respond most sensitively to high-intensity sounds (low-SR
group). Together these spontaneous rate groups (SR groups) convey the vast range of
sound intensities needed for normal hearing.
Despite their fundamental importance to sound encoding, the biophysical
mechanisms defining sound intensity sensitivity remain unknown. Decades of research
focusing on this question have led to multiple classification schemes based on in vivo
physiology and active zone morphology (Kawase and Liberman, 1982; Liberman and
Oliver, 1984; Merchan-Perez and Liberman, 1996). Specifically, these studies report an
association between synaptic position on inner hair cells and intensity sensitivity; wherein
high-threshold, low-SR SGN preferentially synapse on the modiolar face of an inner hair
cell, and low-threshold high-SR SGN synapse on the pillar face.
Several anatomical features are correlated to synaptic position. These include
differences in the type, density, and voltage dependence of pre-synaptic Ca
2+
channels
27
and Ca
2+
sensors (Ohn et al., 2016; Wong et al., 2014), the relative complexity of the
synaptic ribbon (reviewed in Moser et al., 2006; Safieddine et al., 2012), the expression
of post-synaptic glutamate receptors (Liberman et al. 2011) and the kinetics of excitatory
post-synaptic currents (Goutman and Glowatzki, 2011; Grant et al., 2010; Keen and
Hudspeth, 2006; Li et al., 2009). Many of these observations are counterintuitive and
inconsistent with expectations based on other systems. For example, the pre-synaptic
active zones opposing high-SR SGN have smaller ribbons and calcium currents than
those opposing low-SR SGN (Merchan-Perez and Liberman, 1996). This stands in
contrast to large ribbons generating faster EPSC rates in retinal ganglion cells (Mehta et
al., 2013). Consequently, the key factors responsible for defining each SR-subgroup and
their differing sensitivity to sound intensity remain elusive.
Here, we ask whether cell-intrinsic diversity among SGN contributes to sound-
intensity sensitivity. Previous studies in cultured spiral ganglion explants established that
SGN are rich in their complements of ion channels and respond to injected currents with
diverse firing patterns (Mo and Davis, 1997; Davis, 2003, Lui et al., 2014). Systematic
variation of somatic ion channels along functionally relevant spatial axes would suggest
that such variation is relevant for neuronal computations. For example, a previous study
using semi-intact cochlear preparations reported that type I SGN (which contact inner hair
cells and are the primary conduits for sensory information) can be biophysically
distinguished from type II SGNs (which contact the electromotile outer hair cells) by the
kinetics of their potassium channels (Jagger and Housley, 2003). Single cell RNA-
sequencing studies report that type I SGN can be further divided into genotypic subgroups
based on RNA expression levels for a variety of proteins including ion channels, calcium
28
binding proteins and proteins affecting Ca
2+
influx sensitivity (Shrestha et al., 2018; Sun
et al., 2018; Petitpre et al., 2018; Sherill et al., 2019). However, no study has tested
whether differences in type I SGN intrinsic biophysical properties are logically aligned to
the in vivo SGN groups (i.e. SR-groups).
Here, we used simultaneous whole-cell patch clamping and single-cell labelling of
SGN in acute semi-intact cochlear preparations. By keeping hair cells and neurons
connected, this preparation allows tests for links between post-synaptic cellular
biophysics and putative SR-subgroups (as inferred from synaptic location). Our data
show strong correlations between SGN biophysics and synaptic position throughout the
first two weeks of post-natal development. This is the first study to demonstrate such an
alignment between the intrinsic properties of spiral ganglion neurons and their function.
Moreover, the strength of the correlations enabled us to predict synaptic position based
solely on easily measured biophysical properties, suggesting a route to identify putative
SR-subtypes in SGN.

Methods

Preparation
Data were collected from spiral ganglion somata in semi-intact cochlear preparations from
Long-Evans rats (RRID:RGD_2308852) on post-natal day (P)1 through P16. Animals
were handled in accordance to the National Institutes of Health Guide for the Care and
Use of Laboratory Animals and all procedures were approved by the animal care
committee at the University of Southern California.
29
Temporal bones were dissected in chilled and oxygenated Liebovitz-15 (L-15)
medium supplemented with 10mM HEPES (L-15; pH 7.4, ~315 mmol/kg). The semi-
circular canals were removed, as was the bone surrounding the cochlea’s sensory
epithelium. The cochlea was then cut into three turns, with the middle turn used for all
experiments. The middle turn was further prepared by removing the Reissner’s
membrane, stria vascularis, and tectorial membrane. The cochlear turn was then
mounted under two stretched nylon threads on a glass coverslip. The entire coverslip
with cochlea so mounted was incubated in an enzyme cocktail containing L-15, 0.05%
collagenase, and 0.25% trypsin for 15-20 minutes at 37 degrees C. The digested tissue
was washed with fresh L-15 and mounted under the microscope objective. The
preparation was then continually perfused with fresh oxygenated L-15.

Electrophysiology
Recordings were made between one and five hours after dissection. Preparations
were viewed at X630 using a Zeiss Axio-Examiner D1 microscope fitted with Zeiss W
Plan-Aprochromat optics. Signals were driven, recorded, and amplified by a Multiclamp
700B amplifier, Digidata 1440 board and pClamp 10.7 software (pClamp,
RRID:SCR_011323).
Recording and cleaning pipettes were fabricated using filamented borosilicate
glass. Pipettes were fired polished to yield an access resistance between 5-7 MΩ. The
tip of each recording pipette was covered in a layer of parafilm to reduce pipette
capacitance. Large bore cleaning pipettes were used to mechanically remove excess
tissue debris surrounding the somatic area. A second cleaning pipette was used to apply
30
suction to remove myelinating glial cells and layers of myelin from the spiral ganglion
neuron.
Recording pipettes were filled with the following standard internal solution (in mM):
135 KCl, 3.5 MgCl2, 3 Na2ATP, 5 HEPES, 5 EGTA, 0.1 CaCl2, 0.1 Li-GTP, and titrated
with KOH to a pH of 7.35. This yielded a total potassium concentration of 165 mM with a
total osmolality of 300 mmol/kg. Voltages are reported without correcting a junction
potential of approximately 3.8 mV (calculated by JPCalc as implemented in pClamp 10.7,
Barry 1994).
In this study, we only used recordings in which the neuron formed a giga-ohm seal
and presented a stable resting potential. After recording whole-cell currents and voltage
responses in current clamp, we stimulated the cell with current pulse trains which we
empirically determined helped drive the biocytin toward the peripheral terminal. Neuronal
terminals were successfully filled in ~50% of recordings. Some neurons did not fill if the
soma lysed after the recording pipette was pulled away. We found that pipette resistances
of 5-7 MΩ were best for successfully sealing, driving biocytin, and cleanly pulling away
from the soma.
Analysis of electrophysiology

All data were analyzed with pClamp 10.7 software (pClamp, RRID:SCR_011323).
In current-clamp mode, we measured various whole-cell properties of the neuron
including the resting potential, current threshold for spikes, action potential latency, and
the after-hyperpolarization time constant. In voltage-clamp mode, we measured the
approximate steady-state value of outward currents (~taken 400 ms after stimulus onset)
31
in response to a family of voltage steps. In many cells, the outward current inactivated
for long-duration voltage step. We quantified the time-course for inactivation, 𝜏
!"#$%
, by
fitting a single exponential curve from the peak outward current to end of the 400 ms.
This was computed in in all cells for the largest command voltage applied (+ 70 mV)
(Figure 4C, inset). In some neurons (n = 39 of 128), a single exponential curve did not
provide a good fit or the 400 ms protocol was too short to adequately estimate the time
course; these cells were not included in subsequent multiple variable regression
analyses.
Cell capacitance and series resistance were estimated using the membrane-test
protocol on pClamp and/or by analyzing a recording of the capacitive transient in
response to small hyperpolarizing voltage steps. No online series resistance correction
was applied during recordings. All whole-cell currents and stimulus voltages in the main
text and figures are reported without correcting for series resistance and without
normalizing by cell capacitance. As we discuss in the supplement section, these
corrections do not change our interpretations.
Immunohistochemistry

Following the electrophysiology, the preparation was processed to
immunohistochemically label hair cells and to attach a fluorescent label to the biocytin
filled peripheral processes. The preparations were fixed in 4% paraformaldehyde for 15
minutes at room temperature followed by three rounds of washing in 5% phosphate-
buffered saline (PBS) for 10 minutes each on a shaker. The preparations were placed
for one hour in a blocking buffer consisting 16% normal goat serum, 0.3% Triton-X, 450
32
mM NaCl, and 20 mM phosphate buffer. Next, preparations underwent three rounds of
wash in 5% PBS for 5 minutes each. The preparation was then sequentially incubated in
the following primary and secondary antibodies dilutions. Primary antibodies are added
and incubated for 12-18 hours at room temperature. Secondary antibodies are added and
incubated for one hour. Following both primary and secondary incubations, we washed
the samples for three 15 minutes rounds in 1x PBS.
Primary antibodies include (1) anti-myosin-VI-rabbit polyclonal (Proteus
Biosciences Cat# 25-6791, RRID:AB_10013626) to label the cytoplasm of hair cells, (2)
streptavidin Alexa Fluor 488 conjugate (Molecular Probes Cat# S32354,
RRID:AB_2315383) to label biocytin filled neurons, and (3) anti-peripherin (Millipore Cat#
AB1530, RRID:AB_90725) to label type II spiral ganglion neurons. Secondary antibodies
include Alexa Fluor 594 anti-rabbit (Thermo Fisher Scientific Cat# A-11080,
RRID:AB_2534124).

Imaging
The immunolabeled preparations were mounted under a glass coverslip and onto
glass slides with the fade-protectant medium Vecta-shield (Vector Laboratories Cat# H-
1000, RRID:AB_2336789). Hardened nail polish dots were placed under the coverslip to
prevent it from crushing the cochlear sections. We generated z-stack images of each
preparation one of the following two confocal microscopes: (1) Olympus FV1000 laser
scanning confocal microscope and (2) Zeiss LSM 800 with a 10-60x, 1.42 numerical
aperture, oil immersion objective, and a threefold zoom (on IHC-SGN synapse). The
scanning format was set to collect 1024x1024 pixels yielding a sampling of 0.069 µm/pixel
33
in the lateral (XY) dimension. Sections were acquired with z-steps of 0.49 µm with
pinholes set at 1 Airy unit. We scanned the different fluorescent signals in separate
channels in sequence to ensure optimal imagining of each structure.
Analysis of morphology

Images were processed using Imaris software (Imaris, RRID:SCR_007370) in
order to analyze the connectivity pattern of the SGN afferent projection and the hair cells.
In the XY plane of the z-stack images, SGN were primarily classified based on whether
the afferent projection contacted the inner hair cell region or projected past the inner hair
cells, turned radially and projected into the outer hair cell area. Three-dimensional
reconstructions of the afferent projection were produced in Imaris by using the
FilamentTracer tool in Imaris (FilamentTracer, RRID:SCR_007366). The tool reconstructs
the fluorescent signal by tracing the fiber (and branches when present) and fills the area
with sequential spheres whose diameters best fit the diameter of the fiber at a specific
point. Imaris provides multiple measurements of the fiber including the length, diameter,
and volume of the afferent fiber on average and/or at a region of interest. The number of
branches were counted manually by counting the number of projections extending from
the initial parent fiber. The diameter varied along the length of the extending branch from
the soma to hair cell. We measured and reported the average diameter of the extending
branch as the fiber diameter. In fibers with multiple branches contacting the hair cell, we
took the contact point that was closest to the cuticular plate as the primary constant.
34
Statistical Analysis
For our statistical analysis, we used a combination of pClamp (pClamp,
RRID:SCR_011323), Matlab (MATLAB, RRID:SCR_001622), JMP (JMP,
RRID:SCR_014242), Origin Pro (OriginPro, RRID:SCR_015636), Prism 8
(RRID:SCR_002798), and Imaris (Imaris, RRID:SCR_007370) software packages.
pClamp software was be used to gather and quantify raw data from electrophysiological
recordings. Imaris and MATLAB was used quantify the morphology of the SGN. All
variables were confirmed as being normally distributed using a Shapiro-Wilk test for small
datasets. To compare the distributions of two parameters, we used a Student’s t-test.
For comparisons across multiple conditions, we applied one-way ANOVAs followed by
individual comparisons using post-hoc t-tests. The strength of correlations between two
continuous variables is reported by Pearson’s r. We used an alpha level of 0.05 for all
statistical tests. Whenever possible, the distribution of data is displayed by graphing
individual points.
Model Making
We built models based on multiple-linear-regressions to predict normalized-basal position
based on the biophysical properties of type I SGN (see Results). Given their small
numbers, type II SGN were excluded from the data set used for modeling. The
regressions computed a predicted basal position (𝑦 #
!
) based on a linear combination of
multiple input variables (i.e. biophysical features from current clamp and voltage clamp):
𝑦
!
=𝛽
"
+𝛽
#
∗𝑥
#,!
+𝛽
%
∗𝑥
%,!
+𝛽
&
∗𝑥
&,!
+𝛽
'
∗𝑥
',!
+𝜖
!
(2)
𝑦
!
=𝑦
(
( +𝜖
!

35
The input variables (𝑥
&,(,),"
) contributing to the regression were those with the most
significant correlations with normalized basal position (Table 1). Next, the regression
coefficients and intercepts (𝛽
&,(,),"
𝑎𝑛𝑑 𝛽
*
) were optimized to minimize the sum of the
squared error between the predicted basal position (𝑦 #
!
) and the actual basal position (𝑦
!
):

) 𝜖
!
%
!
=) (𝑦
!
−𝑦
(
()
%
!

(3)

To test for redundancy across variables we used equation 4 to compute variable
inflation factors (𝑣𝑖𝑓) :

𝑣𝑖𝑓(𝛽
'
)=
1
1−𝑅
'
%

(4)

where 𝑅
"
(
is the 𝑅
(
obtained when the n
th
variable (𝑥
"
) is regressed against the remaining
variables. Variables with 𝑣𝑖𝑓𝑠 greater than 4.0 were deemed to be highly collinear and
thus were eliminated as containing redundant information. We used a backward stepping
procedure to successively eliminate input variables such that the fewest possible input
variables provided the optimal adjusted-R
2
and root-mean-squared error (RMSE). The
coefficients β and 𝑣𝑖𝑓 values for four models corresponding to four data subsets are
reported in Table 2.
36
Results
Developing spiral ganglion neurons can be classified based on their pattern of
connectivity with hair cells.

We characterized the morphology and electrophysiology of spiral ganglion
neurons (SGN) using whole cell patch-clamp recordings combined with single-cell
labeling from the middle turn of rat cochleae. Labelled SGN were classified in the XZ
plane of the organ of Corti, where both sides of the inner hair cell, the tunnel of Corti, and
three rows are outer hair cells are identifiably organized for classification (schematized in
Figure 1A). The dendritic morphology of individual SGN was characterized by introducing
biocytin into the soma via the patch pipette. The biocytin diffused to fill the peripheral
terminal (Figure 1B). The biocytin filled neuron was visualized after fixation and immuno-
processing using fluorophore conjugated streptavidin (Figure 1B; see methods).
We identified two subtypes of SGN based on their connectivity to cochlear hair
cells. Type I SGNs contacted inner hair cells (Figure 1D-1F; n= 50), while type II SGNs
(Figure 1G; n = 9) projected underneath the inner hair cells, turned towards the cochlear
base and made multiple contacts under the outer hair cells (ranging from 6 to 15
contacts). Of the type I afferents, many reconstructed afferent fibers had terminal
branches that did not touch on any hair cells but projected into neighboring areas,
proximal to the hair cells (Figure 1D.2, 1E.2, 1F.2). These fibers project on areas that are
not visualized with Myosin VI signaling. In other cases (n=2), type I SGN terminals extend
to a neighboring inner hair cell. Finally, in one case, the fiber projected into the outer hair
37
cell layer. We classified this fiber as a Type I fiber because it was negative for the Type
II marker peripherin (see Methods).
In this age range, the branching patterns of Type I fibers are highly polarized to
either the modiolar or pillar face of a hair cell. In other words, for those fibers that
contacted the modiolar face, nearly all the branches of the fiber are also located on the
modiolar side of the hair cell (whether they touched the hair cell or not). To quantify this
apparent polarization, we defined a metric that we refer to as the ‘polarization vector’. To
do this, we assigned a unitary value and sign to each branch depending on polarization
(+1 for modiolar and –1 for pillar). The net polarization of the fiber was computed by
taking the sum of the individual polarizations divided by the total number of branches
(Figure 1J). A polarization vector value of 1 indicates that all of the branches of a fiber
are on the modiolar side of the inner hair cell, –1 indicates that all branches are on the
pillar side, and zero means that the branches were equally distributed on both sides. Most
fibers are unambiguously modiolar- or pillar-contacting. In three pillar-contacting SGN,
some branches slightly crossed to the modiolar-side yielding polarization vector values
that were slightly more positive than -1 (i.e. Figure 1F.2); however, terminal branches did
not equally project to both sides of the inner hair cell. Whether this indicates the presence
of selective guidance cues for modiolar versus pillar-contacting fibers remains to be
tested. Although the polarization vector allowed us to confidently assign fibers as being
either modiolar- or pillar-contacting, we moved away from this dichotic classification in
preference for a more continuous scale as described below.
We defined a continuous position scale, which we refer to as ‘Normalized Basal
Position’ (NBP), to quantify where SGN terminals contact along the base of inner hair
38
cells (Figure 1H). The ‘modiolar’ and ‘pillar’ halves of the hair cell were first defined by
drawing a bisecting plane from the basal pole to the cuticular plate of the hair cell in the
XZ plane of the confocal scan. The origin (NBP = 0) was set at the basal pole of the
bisected inner hair cell. NBP was taken as the distance from the basal pole of the hair
cell to the position of the SGN fiber contact (c) relative to the length of the inner hair cell
(L) (basal pole to cuticular plate) (Figure 1H). The normalization by hair-cell length
standardized the position scale across preparations. Positive NBP values contacted the
inner hair cell from the basal pole and up the inner hair cell towards the cuticular plate on
the modiolar-face of the inner hair cell. Negative NBP values were for contact positions
ranging from basal pole up towards the cuticular plate on the pillar-face of the inner hair
cell.
NBP measurements were only made when the preparation was in relatively
pristine condition after electrophysiological manipulations and immunostaining.
Therefore, only a subset of the data had this measurement (n = 38 of 50). If there were
multiple branches that contacted the inner hair cell, we used the most extreme contact
position relative to the basal pole for our analyses. NBP values ranged from -0.32 to
+0.52 (n = 38) (Figure 1I). In the preparations where we could not measure the NBP, we
could classify the position on a bimodal system as either modiolar-contacting (NBP > 0)
or pillar-contacting (NBP < 0). In total, we classified 34 fibers as modiolar-contacting, 16
fibers as pillar-contacting, and 9 as type II SGN. The non-uniform distribution between
Type I, Type II and subtypes of Type I fibers is in line with the pattern of innervation
previously reported in rat and cat animal models, where the vast majority of SGN fibers
39
appear to land on the modiolar-face of the inner hair cell (Oliver et al, 1982, Kalluri et al.,
2017).

40

Figure 1:
Semi-intact preparation combines SGN biophysics with identification of pattern of connectivity with hair cells. A.
Schematic of organ of Corti innervation in the XZ plane to show how type I SGN innervate two sides of an inner hair
cell. B. Confocal image of two type I spiral ganglion neurons (SGN) injected with biocytin in the acute semi-intact
preparation of rat middle-turn cochlea. Hair-cells and biocytin labelled fibers are visualized by immunolabeling for
Myosin VI (red; hair cells) and streptavidin conjugated to Alexa Flour 488 (green, fiber). C. Myelinating Schwann cells
are mechanically removed by suction pipettes to make SGN somata accessible to recording electrodes. D.1-F.1.
Confocal images in the XY plane show type I SGN afferent fibers making exclusive connections with inner hair cells.
D.2-F.2 The synaptic position of the afferent fiber onto the inner hair cell is assessed by examining the cross-section
of the hair cell in XZ plane. G. Type II SGN projecting radially and turn within the outer hair cell region. H. A schematic
of an individual inner hair cell with guides showing how the normalized basal position (NBP) is measured for each type
I SGN. NBP is measured by the ratio distance from the base to the synaptic position along the radial axis of the hair
cell (C) divided by the maximum length of the inner hair cell (L). Positive NBP values indicate that the synaptic position
is on the modiolar-side of the inner hair cell, while negative NBP values indicate that the synaptic position is on the
pillar-side of the inner hair cell. I. The distribution of type I SGN contacts as a function of NBP. NBP values averaged
at 0.11 +/- 0.23 with a range from -0.27 to +0.52 (n = 38). J. Type I SGN are strongly polarized to either the modiolar
(polarization vector = 1) or the pillar (polarization vector = -1) side of the inner hair cell.
41

Biophysical properties of developing spiral ganglion neurons change with normalized
basal position

Diversity of firing patterns in current clamp

In current clamp mode, we measured the firing patterns evoked from SGN somata
in response to injected steps of current (Figure 2). Firing patterns were diverse and
qualitatively binned into four groups based on degree of accommodation. Rapidly
accommodating firing patterns had a single action potential at the onset of a step of
current (Figure 2A). Intermediate-accommodating firing patterns had more than one
action potential (Figure 2B) but accommodated more quickly than slowly accommodating
firing patterns which typically had multiple action potentials throughout the duration of the
stimulus (Figure 2C). The fourth response category did not have clearly identified action
potentials and instead had graded depolarizations in response to a family of current
injections (Figure 2D). Although neurons with graded responses did not fire action
potentials when perturbed from their natural resting potential, they were capable of firing
action potentials when they were first artificially held at -80 mV (Figure 2D inset); the more
negative holding potential presumably relieves sodium channel inactivation.
The four patterns were not encountered in equal proportions; rapidly
accommodating firing patterns were the most prevalent (n=49), followed by intermediate-
accommodating firing patterns (n=19). Slow-accommodating patterns were the least
prevalent (n=5), with all observations occurring between P1 and P3, suggesting that they
42
might be an immature phenotype in acute preparations. Graded-firing neurons were
observed at all ages (n=55). All firing patterns were observed in Type I SGN, while only
rapidly-accommodating (n=2) and graded-responses (n=7) were observed in Type II
SGN. Therefore, we did not find a simple association between firing pattern and Type I
and Type II fiber morphology.
To move beyond a qualitative classification that forces responses into discrete
firing pattern groups, we measured an array of response features to examine subtle
differences. Our reasoning was that variability in detailed response features also reflects
variability in the intrinsic biophysical properties of a neuron, including ion channel
composition (Liu et al., 2014a, Liu et al., 2014b). This shift towards quantification was
especially important because rapidly-firing, intermediate-firing and graded-responses
were not always easily discriminable, and slowly accommodating firing patterns were
rarely encountered.
In spiking neurons (i.e. rapidly-, intermediate- and slowly-accommodating), we
measured features that are directly related to an action potential (Figure 2E inset).
Current thresholds (F(2,71) = 9.67, p = 0.0002, ANOVA), voltage thresholds
(F(2,71)=3.75, p = 0.0283, ANOVA), and after hyperpolarization time-constant (F(2,71) =
13.36, p < 0.0001, ANOVA) varied by spike-pattern. Current thresholds were larger,
voltage thresholds were more hyperpolarized, and after-hyperpolarization times (AHP)
were faster in rapidly accommodating neurons than in intermediate- and slowly
accommodating neurons (Figure 2E). Overall, the action potential features were more
broadly distributed for the rapidly accommodating firing group. Although there were no
significant differences in resting potential among spiking neurons (potential measured in
43
the absence of current injection), graded-response neurons (𝑉𝑚
33333
= -51.77 +/- 1.14 mV)
had significantly more depolarized resting potentials than did spiking-neurons (𝑉𝑚
33333
= -
56.86 +/- 0.98 mV) (p = 0.0010, t-test), which may account for why the graded-response
neurons spiked only after first being hyperpolarized to -80 mV (Figure 2D, inset).
44

45
Figure 2:
SGN can be qualitatively grouped by firing patterns in response to step current injections: A. Rapidly-accommodating
neurons (red dots) fire a single action potential, B. Intermediately-accommodating neurons (blue dots) fire more than
one action potential and eventually reach a steady-state potential, C. Slowly-accommodating neurons (black dots) fire
multiple actions and do not reach a steady-state potential, and D. Graded-response neurons (triangles) do not produce
an action potential, but produce graded-depolarization in response to incrementing current steps. These neurons are
capable of firing action potentials after their membrane potential are held at -80 mV (inset). E. Action potential features
vary among spiking neurons. Current- (F(2,71) = 9.67, p = 0.0002) and voltage-thresholds (F(2,71)=3.75, p = 0.0283)
and after-hyperpolarization potential (AHP) time constants (F(2,71) = 13..36, p < 0.0001) were significantly different in
rapidly-accommodating firing neurons than in intermediate- and slowly-accommodating neurons. F. Action potential
features were significantly correlated with normalized basal position (NBP). As NBP values increased, current
thresholds increased (R2 = 0.47, p< 0.001), voltage thresholds hyperpolarized (R2 = 0.20, p = 0.045), and AHP time
constants became faster (R2 = 0.39, p< 0.001). G. The distribution of firing pattern subgroups as a function of NBP
shows a significant relationship between the firing-pattern subgroup and NBP (p = 0.0097), with rapidly accommodating
neurons found primarily on the modiolar-face and intermediately accommodating neurons found on the pillar-face of
the inner hair cell. Graded-firing neurons are found throughout the NBP scale on both the modiolar and pillar faces of
the hair cell.

Current-clamp features of spiking-neurons are correlated with normalized basal position

Next we tested for and found significant correlations between action potential
features of spiking-neurons and normalized basal position (NBP) (Figure 2F). Larger
currents were needed to initiate spiking as contact positions moved from the pillar face
(NBP < 0) to the modiolar face (NBP > 0) (r(20)=0.68, p= 0.0007). Voltage thresholds
hyperpolarized (r(20) = -0.45, p = 0.045) and after-hyperpolarization times became
shorter with increasing values of NBP (r(20)= 0.62, p<0.0001). These correlations
between individual action potential features and basal position are similar to the gross
differences between rapidly- and intermediate-accommodating spike patterns.
Intermediately accommodating neurons, with their tendency for smaller current
thresholds, depolarized voltage thresholds, and slow AHP time courses, were primarily
found on the pillar-side of the inner hair cell (𝑁𝐵𝑃
333333
= -0.08 +/- 0.7) (blue circles, NBP < 0,
Fig. 2G). Rapidly accommodating neurons were found throughout the NBP scale but were
more prevalent on the modiolar face (𝑁𝐵𝑃
333333
= 0.18 +/- 0.5) (p=0.0097, test-t) (red circles,
NBP > 0, Fig. 2G).
46
Current-clamp features of non-spiking neurons are also correlated with normalized basal
position

In spiking neurons, we showed that current threshold was significantly correlated
with basal position, but non-spiking neurons with their graded responses did not have
thresholds. Given that many labeled neurons had graded responses, we examined other
current-clamp features not related to spiking and found that response latencies of spiking-
and non-spiking neurons (described below) were also significantly correlated with basal
position. This ultimately allowed us to pool the spiking- and non-spiking neurons into one
group for subsequent analysis.
In spiking neurons, we measured response latency as the time delay between the
onset of the current injection and the first spike induced by threshold-level current steps
(first-spike latency, tLatency). First-spike latencies were shorter in a representative
modiolar-contacting neuron than in a pillar-contacting neuron (Figure 3A). On average
current thresholds in spiking neurons were inversely proportional to latencies (Figure 3B)
so that the relationship was well fitted by a function of the form:

𝐼
)*+,- *
∝
1
𝑡
./),'01
6
(5)

In non-spiking neurons we measured response latency as the average time
needed to reach peak depolarization for a family of current steps. This is shown for two
example neurons in Figure 3C. The depolarization of the modiolar-contacting neuron
reached peak potential earlier than did that of the pillar-contacting neuron. Graded-firing
neurons’ response latencies showed similar correlations to normalized basal position as
47
did first-spike latency in spiking neurons (Figure 3C & 3D). Response latencies in both
cases become faster as fiber-contact positions moved from the pillar to modiolar face of
the inner hair cell (r(34) = -0.74 , p< 0.0001) (Figure 3D).

Figure 3:
Response latencies are alternate predictors for normalized basal position. There is a significant inverse relationship
between first-spike latency and current thresholds. Black line shows a fit Ithresh = constant/latency. Color transition
from red to blue indicates position on NBP scale. Type II SGN are colored black. Unlabeled and therefore unclassified
SGN are in gray. A. First spike latencies compared between three spiking SGN with different NBP values. Latencies
are faster for more positive NBP B. In non-spiking neurons, response latency is computed by averaging the time from
the onset of the stimulus until peak depolarization potential over a series of current steps. Similar to the dependence
of first-spike latency, average response latency is faster for modiolar-contacting SGN (NBP>0, red) than for pillar-
contacting SGN (NBP<0, blue). C. Response latencies for all spiking (black dot) and non-spiking (triangle) neurons
plotted against NBP. In both spiking- and non-spiking SGN, response latencies become faster as NBP values increase
(R2 = 0.55, p< 0.001). Latencies are highly variable across type II fibers (gray symbols).

48
Diversity in whole-cell currents in voltage-clamp

Next, we discuss the dependence of whole-cell currents (measured in voltage-
clamp) on normalized basal position. In Figure 4A-4C, we show the currents measured in
response to a family of voltage steps (from -120 mV to +70 mV) in three example somata;
each identified as a modiolar-contacting, pillar-contacting, and type II spiral-ganglion fiber,
respectively. The magnitude of whole-cell currents was highly variable. Each response
consisted of net inward (negative) and outward (positive) currents. The inward currents
were largely transient, activating and inactivating quickly, consistent with our expectations
for currents conducted by sodium channels. The outward currents inactivated relatively
slowly to produce a nearly steady net current around 400ms after the onset of the current
step.
Previous studies have shown that SGN have both high-voltage- and low-voltage-
gated potassium currents in diverse amounts which can lead to differences in thresholds
and excitability (Liu et al., 2014; Lv et al., 2010). Here we assume that the steady-state
outward currents we measure in voltage-clamp are largely dominated by potassium
currents. To quantify the size and voltage dependence of these currents, we converted
them into approximate whole-cell conductance by dividing the steady-state current (Iss)
by the driving force for potassium (Vm-EK), where Vm is the membrane voltage and EK is
the reversal potential for potassium for our recording conditions. The resulting whole-cell
conductance curves were fit with Boltzmann curves from which we characterized three
features: 1) the value at -30 mV (g-30) as a rough gauge for the size of low-voltage gated
potassium conductances, 2) the maximum value (gmax) to gauge total potassium
49
conductance, and 3) the voltage for half-maximum activation (V1/2) to gauge the relative
balance between low-voltage and high-voltage conductance. In several cases (n=10 out
of 128), the conductance curves did not saturate. These SGN were omitted from
subsequent analyses. In addition to characterizing the voltage dependence and size of
the steady-state outward conductance, we measured the time constant over which the
outward current inactivates at +70 mV, referred to here on as 𝜏
!"#$%
(Figure 4C, inset).

Whole-cell conductances grow as a function of normalized basal position.
As we showed in Figure 2 and 3, current thresholds and response latencies were
strongly correlated with normalized basal position. We hypothesized that voltage-clamp
responses of whole-cell currents would be similarly correlated with basal position. Indeed,
we can see qualitatively that modiolar-contacting neurons have larger maximum whole-
cell conductance compared to pillar-contacting neurons (as indicated by the color
transition from blue to red as normalized basal position becomes more positive (Figure
4E).
Individual conductance curve features show significant correlations with
normalized basal position. Maximum whole-cell conductance (gmax) increased as a
function of NBP (r(34)= 0.54, p = 0.0003) (Figure 4F). Neurons terminating on the
modiolar-face also had large g-30, suggesting that they may have had larger low-voltage
gated components (r(34) = 0.64, p < 0.0001) (Figure 4G). However, V1/2 was not
significantly correlated to basal position (r(34) = -0.23, p = 0.17) (Figure 4H). Inactivation
time constant was correlated with normalized basal position; with the largest time
50
constants found for neurons on the modiolar face of the of the inner hair cell (r(25) = 0.50,
p = 0.0087).
51

52
Figure 4:
Net whole-cell conductances increase as SGN synaptic position moves from pillar to modiolar-face of the inner hair
cell. A-C. Voltage-clamp responses of a modiolar-contacting (A), pillar-contacting type I (B), and a type II SGN (C). For
each cell, we held the potential at -60 mV, followed by a series of voltage 400 ms steps from -120mV to 70mV, and
then finally back to the holding potential. For each cell we measured current-voltage relationships at approximate
steady-state (400 ms indicated by the arrow). We measured the time course for outward current inactivation by fitting
a single exponential curve to the current evoked by a +70 mV voltage step (inset). D. Steady-state conductance as a
function of command voltage of all SGN recorded with three measurements: the magnitude of conductance at -30 mV
(g-30), the maximum conductance (gmax), and the half-activation potential (V1/2) via Boltzmann function. E. Steady-state
conductances of identified type I SGN as a function of normalized basal position. F-I. Voltage clamp properties as a
function of normalized basal position (NBP) (Spiking, black dot; non-spiking, triangle). Linear regression models were
fitted to each parameter with statistical significance test of p < 0.05. All parameters were significant except for V1/2.

Current- and voltage-clamp features together predict normalized basal position

As we showed in the previous section, we found significant correlations between several
biophysical properties and normalized basal position. In this section, we generate
prediction models for normalized basal position based on a linear combination of multiple
biophysical features. The full details of the model building can be found in the methods.
Briefly, the process began by pooling all the current-clamp and voltage-clamp features
that are significantly correlated with basal position. To prevent overfitting, the number of
variables retained in the model was reduced using a backward stepping process, with
highly collinear (i.e. redundant) variables removed. By analyzing the collinearity between
biophysical properties, the model building process also pointed at the underlying
relationships between variables. In Figure 5, we show four prediction models
corresponding to the four ways in which the data were divided in different subsets. The
first model is based only on the subset of neurons that spiked in response to current steps.
These neurons were pooled across a wide age range (from P3-P10). The second model
is over the same age range but also includes non-spiking neurons. By including non-
spiking neurons, the model limits the current-clamp variables to those that can be
measured in the absence of action potentials. The third and fourth models include spiking
53
and non-spiking neurons but filter the data into narrow age bins. By looking at narrow
age ranges we test if spatial gradients are present at multiple developmental time points.

Figure 5:
Combining voltage-clamp and current-clamp features in multiple-variable regression- models to predict normalized
basal position (NBP) based on biophysics. A. Prediction model for spiking neurons (P3-P10) relies on Ithresh, gmax and
tinact. B. Prediction model that pools spiking and non-spiking neurons (P3-P10) is successful by using latency in place
of current threshold. C,D respectively. Biophysical gradients are still successful at predicting spatial position even
after filtering data into narrow age bins. P3-P5 in C and P6-P8 in D. Predictor variables used in each model are
presented at the bottom right of the predicted vs. actual plot. E. Current threshold increase with increases in g-30. F.
Response latency as a function of g-30. Fit is a function of the form latency = constant/ g-30.

54

Model 1: Spiking neurons only (P3-P10)
The first multiple regression model was on spiking-neurons between P3 and P10. The
following seven current- and voltage-clamp features were significantly correlated with
normalized basal position: current threshold, voltage threshold, first-spike latency, AHP
time constant, resting potential, g-30, gmax, and 𝜏
!"#$%
(Table 1, see also Figure 2). After
reducing the number of input variables by accounting for variable inflation and via the
backward stepping process, the best prediction model for spiking-neurons included three
variables: current threshold, gmax, and 𝜏
!"#$%
(Figure 5A). Current-threshold was the most
significant variable predicting the normalized position (recall that it accounted for nearly
47% of the variance on its own, Figure 2F). This was followed by gmax and 𝜏
!"#$%
. Note
that other variables like latency and g-30 were individually more significantly correlated
with basal position than gmax or 𝜏
!"#$%
, but they were strongly collinear (or redundant) with
current threshold and thus were eliminated during variable reduction.
The performance of the prediction model is illustrated by plotting predicted basal
position against actual position (Figure 5A). If the model were perfectly accurate, each
point (representing an individual neuron) would lie on the 45-degree trend line. The more
spread the points are from the line, the less accurate the prediction. The model (Table
2) accounted for 75% of the variance in the data (adjusted- R
2
= 0.75, p = 0.0004) and
accurately predicted the basal position within approximately +/-9 % margin of error
(RMSE = 0.092).
55
Model 2: Spiking plus non-spiking neurons
In the absence of current threshold, the most influential predictor variable for non-spiking
neurons is response latency. Note that although response-latency is strongly correlated
with basal position for both spiking- and non-spiking neurons, it is also highly correlated
with current threshold (r = -0.64, p < 0.0001) (Figure 3B). This collinearity between
current-threshold and response latency was significant enough to inflate the models if
both variables were included. Thus, none of the models use both current threshold and
latency. The best model after pooling spiking- and non-spiking neurons used response
latency and 𝜏
!"#$%
to best predict NBP. This regression model accounts for 61% of the
total variance (adjusted-R
2
= 0.61, p < 0.0001) and accurately predicts the basal position
within a 13.6% margin of error (RMSE = 0.136) (Figure 5B).
Models 3&4: Spatial gradients in biophysical properties are present in early post-natal
days and maintained until the onset of hearing.

Next we tested whether spatial gradients in biophysical properties were still
present if the data were filtered into two narrow age ranges; P3-P5 and P6-P8. We ran
the same model building procedure as in the previous sections. By binning spiking- and
non-spiking neurons, we had enough neurons in each age group to maintain significant
power. Since non-spiking neurons were included, only the current-clamp features that
were not related to spiking were included (similar to Figure 5B). As illustrated by
performance of the prediction models in Figure 5C and 5D, modiolar to pillar gradients in
biophysical properties are present even within a narrow age range, confirming that the
gradients are not sampling errors resulting from pooling the data over wide age ranges.
56
This is reassuring since the biophysical properties of SGN change with maturation (as we
show in the next section) and our dataset is sparse.
Consistent with the overall data set, even in this narrow age group, we found
significant collinearity between current-threshold and response latency (r = -0.65, p =
0.0007). Using response latency as the sole input variable for the model at P3-P5
accounts for 55% of the variance in the data (R
2
= 0.55, p = 0.0010) and predicts the basal
position within a 12.3% margin of error (RMSE = 0.123) (Figure 5C).
At P6-P8, response latency, resting potentials, g-30, and gmax were all significantly
correlated with the basal position (Table 1). Response latency still accounted for 55% of
the variance (R
2
= 0.55, p = 0.0003) and predicted the basal position within a 14% margin
of error (RMSE = 0.140) (Figure 5D). These narrow-age-binned models show that
significant spatial gradients in biophysical properties exist in early post-natal days and are
maintained until the onset of hearing.
In summary, we employed current- and voltage-clamp features in different
combinations within multiple-linear-regression models to describe the systematic
variation of SGN electrophysiology as a function of terminal contact position on inner hair
neurons. The common underlying factor relating current threshold and latency appears
to be the net input conductance of the neurons (as illustrated in Figure 5E and 5F), as
probed here by g-30 in voltage clamp. As net conductance increased, larger currents were
needed to depolarize the SGN somata to threshold, as would be consistent with Ohm’s
Law (Figure 5E). In a similar broad view, larger conductance would also likely translate
into faster membrane time constants (as described by time constant; = 𝑅𝐶 =
+
,
). Faster
membrane time constants are likely to yield shorter first spike latencies. In Figure 5F, we
57
plot response latency for all the labeled and unlabeled neurons from the present study
against g-30. Consistent with the above suggestion, response latency is inversely related
to net conductance, as expected if latency is proportional to membrane time constant.
These results suggest that spatial gradients in SGN biophysical properties are ultimately
related to variations in the net conductance seen by the recording electrode.
Spatial gradients in biophysical properties may be maturational gradients.

Next we examined age-dependent trends in SGN electrophysiology. Although we
do not have morphology above P10, we have several somatic recordings between P10-
P16 (n=13). Key predictors for NBP (i.e., current threshold, response latency, and g-30)
remain highly variable even at older ages (Figure 6). For example, the size of the near
steady-state outward current (at 400 ms) as a function of the command voltage is shown
binned into three age groups (P1-5, P6-9, and P10-16; Figure 6A.1 through 6A.3). The
currents are smaller on average for the younger group than for the older groups. This is
consistent with the idea that neurons acquire larger currents as they mature. However,
the overall range in current amplitudes is large, even within any particular age group. For
example, many SGN at older ages have currents as small as those at younger ages. This
suggests that not all neurons are maturing at the same rate.
Age-dependent changes in biophysical properties are qualitatively similar to the
local spatial gradients (modiolar-to-pillar) we described in the previous section. With age,
average latencies become faster, current thresholds increase, and conductance values
become larger (Figure 6B-6E). And yet, within any age range we saw spatial gradients,
with modiolar contacting fibers having faster latencies, larger thresholds and larger
58
conductances (note the gradation in color from blue to red indicating an increase in
normalized basal position within each age bin in Figure 6B through 6E). We argue that
spatial and maturational gradients are qualitatively similar, raising the possibility that
diversity in SGN biophysical properties reflects a local spatial gradient in maturation.
To illustrate the idea that spatial gradients are maturational gradients, we plotted
the age dependence of gmax, current threshold, and latency (Figure 6F, 6G, and 6H,
respectively) for fibers contacting the modiolar face (NBP>0; red symbols) and for fibers
contacting the pillar face (NBP<0; blue symbols). Current threshold and gmax start with
similar values in modiolar-contacting and pillar-contacting fibers in early post-natal days
and diverge as the neurons develop (Figure 6C, 6D, 6E). On the modiolar face, gmax
grows nearly twice as fast with age than it does on the pillar face. This is indicated by the
steeper slope of the regression line fit (red dashed line) through the data points for the
modiolar face. Current thresholds also change more rapidly on the modiolar face than on
the pillar face (although the data for threshold is sparse), consistent with expectations
based on Ohm’s Law.
Although the variance in gmax and thresholds increases with age, the opposite trend
is seen for latency (Figure 6F,G). This is because latency is inversely related to
conductance (Figure 5). Thus, as gmax increases with age, latency decreases. In Figure
6H, we plot latency as a function of age. The decrease in the variance in the latency (L)
measurement should not be interpreted as a reduction in biophysical heterogeneity. It
simply reflects latency’s inverse dependence on gmax and current threshold.
The red and blue dashed lines overlaid on the data points (Fig. 6H) are two curves
defined by the functions,
59

𝐿
2
=
𝐾
𝑔
3/4
2 6

(5)

𝐿
5
=
𝐾
𝑔
3/4
5 6

(6)
where K is a proportionality constant and 𝑔
-#.
/
and 𝑔
-#.
0
are the linear fits (from Figure
6F) that best describe the age dependence of gmax on the modiolar and pillar faces,
respectively. The fit for the pillar face was constrained to intersect that of the modiolar
face between P0 and P3. The curves defined by equation 5 and 6 are successful at
describing the difference in latencies between the two groups; pillar-contacting fibers
have longer latencies than do the modiolar-contacting fibers and modiolar-contacting
fibers are developing faster than are pillar-contacting fibers. By P10, LP is about 13 ms,
making the fibers contacting the pillar face about 6 days delayed in maturation on average
compared to those on the modiolar face.
60

Figure 6:
61
Biophysical properties of SGN change with age but remain diverse throughout three weeks of post-natal development.
A. Steady-state current/voltage curves show that net outward currents grow as the animal develops. Even within the
P10-P16 age bin, some neurons have small currents, similar to that at P1-P5. B-D. Log-normal plots showing the
distribution of biophysical properties at different age bins. Average steady-state conductance (B) and current thresholds
(C) grow and become more varied as the animal develops. Latencies decrease and become less varied (D). Within
each age bin, biophysical properties of type I SGN change as a gradient along the NBP scale (red through blue
indicating transition from modiolar to pillar faces). E-F. Maximum conductance, current threshold, and latencies plotted
as a function of post-natal age. Conductance and current threshold diverge with age. Although average latency
decreases, they latencies of modiolar and pillar-contacting fibers do not converge. Symbols are colored to indicate
modiolar-contacting fibers (red, NBP>0) and pillar-contacting fibers (blue NBP<=0). The equations defining the fit in E
are 𝑔
!"#
$
= 4(𝑎𝑔𝑒)+7; 𝑔
!"#
%
= 2 (𝑎𝑔𝑒)+10. Dashed lines in F describe the predicted latencies (see text) based on
diverging rates of conductance growth described by the linear fits in E. For comparison, the dotted line shows the
predicted latency if pillar face was two days delayed at all ages.

Next, we describe fiber morphology as supporting evidence for modiolar-
contacting neurons developing at a faster rate than pillar-contacting neurons. Our
hypothesis was that morphological maturation parallel biophysical maturation. The
afferent terminals of SGN are highly branched during early development and are pruned
during maturation to form a single bouton connection between the SGN and inner hair
cell (Druckenrod et al., 2015, Huang et al., 2007. Echteler et al., 1992). Consistent with
previous reports, we also found that the number of terminal branches decreases as a
function of maturation (p < 0.001). (Figure 7A, 7B). We observed mature one-to-one
connections as early as P5 on the modiolar face. In contrast, none of the pillar-contacting
fibers had mature terminal morphology, even by P10 (Figure 7C). We quantified
differences in the pruning rate of modiolar- and pillar-contacting fibers from P5-P10 and
found that modiolar-contacting fibers’ have fewer branches than do pillar-contacting fibers
(1.9 +/- 0.24, n = 12 vs. 3.2 +/- 0.32, n = 7, respectively) (p = 0.030, t-test) (Figure 7C,
inset). Since, modiolar-contacting fibers appear to prune at a faster rate than pillar-
contacting fibers, these data are consistent with the biophysics in suggesting that neurons
on the modiolar face are maturing faster than those on the pillar face.
In mature auditory nerve, fiber diameters are larger for high-SR fibers, which are
the fibers primarily found on the pillar side of the inner hair cell (Merchan-Perez and
62
Liberman, 1996). We measured fiber diameters of fluorescently labelled SGN.
Consistent with previous in vivo labelling studies, we found that type I SGN have
significantly larger fiber diameters than do type II SGN (1.28 +/- 0.05 µm, n = 37 vs. 0.59
+/-0.12µm, n=6; p < 0.0001, t-test) (Figure 7D). Fiber diameters are not significantly
different between modiolar-contacting SGN and pillar-contacting SGN (1.25 +/- 0.29 µm
vs. 1.11 +/- 0.22 µm; p = 0.25, t-test) (Figure 7D). These results are consistent with reports
in mice where fiber diameters begin to differentiate around the onset of hearing (Coates,
personal communication).
63

Figure 7:
A. Three-dimensional reconstructions of SGN terminals are made to quantify the number of branches at each terminal
and the approximate diameter of the fibers. B. The number of branches exponentially decreases over the course of
development. One-to-one connections with the inner hair cell can be seen at P5. In some cases, terminals have multiple
branches at P10. C. Pillar-contacting SGN are not observed to have the mature one-connection morphology. D. Fiber
diameters do not differ between type I SGN along the NBP scale. Type II SGN have significantly smaller fiber diameters
than type I SGN.

64
Discussion

Biophysical properties of SGN vary as spatio-temporal gradients along the base of the
inner hair cell.

Our results show striking correlations between the biophysical properties of SGN
somata and the position where peripheral dendrites contact hair cells. Within the type I
population, current thresholds decreased, first-spike latencies increased, and the size of
net outward conductance decreased in a gradient as contact positions moved from the
modiolar to pillar face of inner hair cells. Although the ion channel properties of SGN
somata are known to be diverse in vitro, this is the first clear demonstration that such
diversity is systematically organized about the auditory nerve’s effective map for
‘intensity-sensitivity’. Specifically, the more excitable neurons with low current thresholds
in vitro are found on the pillar side of the inner hair cell, where in vivo auditory nerve fibers
have high spontaneous rates and are most sensitive to low intensity sounds (Kiang, 1965;
Liberman, 1978). Greater intrinsic excitability may contribute to pillar-contacting fibers’
preferred sensitivity to low-intensity sounds in vivo. Thus, the direction of in vitro
biophysical gradients qualitatively aligns with mature in vivo physiology and hints at a
possible post-synaptic contribution in controlling sound-driven thresholds and excitability.
The biophysical gradient we observed under inner hair cells is likely to be a
maturational gradient, with modiolar-contacting SGN maturing faster than pillar-
contacting SGN. This suggestion is based on a combination of observations. First, we
saw spatial gradients in biophysical properties that are qualitatively similar to the average
65
maturation of SGN biophysics. SGN acquired larger net outward conductance, higher
current thresholds and shorter first-spike latencies as they matured. Second, the pruning
of terminal branches to form a one-terminal-contact is a well-described morphological
hallmark of maturing auditory neurons (Echteler, 1992; Barclay et al., 2011; Huang et al.,
2007). Here, we found that the terminal branches of many modiolar-contacting SGN
fibers are pruned by an earlier age than those of pillar-contacting fibers. This is another
line of evidence supporting the idea of a local maturational gradient under the inner hair
cell. Third, modiolar-contacting fibers appear to be less enriched for the calcium-binding
protein, calretinin, than pillar-contacting fibers in mature rats (Kalluri and Monges-Aviles,
2017) and mature mice (e.g., Shrestha et al., 2018). This difference in protein expression
(among others) emerges only after a period of maturation (Shrestha et al., 2018; Sun et
al., 2018; Petitpre et al., 2018). Perhaps the expression of calretinin in pillar-contacting
fibers is a mark of residual immaturity. Together these data suggest that spatial gradients
in the biophysical properties of SGN reflect a local gradient in maturation.
Biophysical diversity is unlikely to be a transient feature of immaturity and likely
persists into maturity. Here, we showed that the distribution of whole cell conductance
and current thresholds becomes broader as SGN mature. At P10, we estimated that on
average modiolar-contacting fibers were nearly 6 days more developed than pillar-
contacting fibers. Consistent with this, single-cell RNA sequencing recently
demonstrated that the molecular and transcriptomic profile of SGN become more diverse
as the neurons mature and resolve into distinct sub-types (Petitpre et al., 2018; Shrestha
et al., 2018; Sun et al., 2018). Our results suggest that this would happen if SGN continue
to diverge until a critical period for maturation is closed.
66
Together these results suggest that local gradients in the maturation of spiral
ganglion neurons likely resolves into a spatial gradient in biophysical properties auditory
neurons. Given the relationship between synaptic position and intensity sensitivity, our
work suggests that variations in intrinsic biophysical properties contributes to the auditory
nerve’s well-known diversity in intensity sensitivity.
SGN diversification during maturation
The maturational gradients we describe at the local scale under an inner hair cell
are reminiscent of previously described global (i.e. along the length of the cochlea)
gradients in the maturation of SGN biophysics. Across the entire spiral ganglia, firing
patterns become more rapidly accommodating and first-spike latencies become faster as
the animal matures (Crozier and Davis, 2014; Adamson et al. 2002). By the end of the
second post-natal week, firing patterns are more rapidly accommodating and first-spike
latencies are on average faster in the SGN contacting the high-frequency-encoding basal
turn of the cochlea than in the low-frequency-encoding apical turn (Crozier and Davis,
2014; Adamson et al. 2002). These biophysical gradients likely reflect a global cochlear
base-to-apex maturational gradient similar to that previously described in cochlear hair
cells (Waguespack et al. 2007; Lelli et al. 2009). Our data from the middle-turn of the
cochlea suggest that the global pattern of spatio-temporal maturation is being
recapitulated on a local scale under an individual inner hair cell, where modiolar-
contacting SGN mature faster than pillar-contacting SGN.
The biophysical properties of SGN appear to depend on extrinsic factors, at least
in part. In cultured SGN, the presence of growth factors such as BDNF and NT-3 (typically
released by hair cells and supporting cells in situ) has a profound influence on firing
67
patterns and expression of proteins for ion channels and receptors (e.g., Flores-Otoro et
al. 2007; Sun and Salvi 2009). The influence of these growth factors also follows a global
cochlear-base to cochlear-apex gradient; with higher levels of BDNF promoting rapidly
accommodating firing patterns in the base of the neonatal cochlea. In contrast, NT-3 is
found at higher levels in the apex of the cochlea where firing patterns are more slowly
accommodating. That extrinsic signals may influence SGN diversity is further
strengthened by recent findings that post-natal diversification fails in the absence of
synaptic input (Shrestha et al., 2018; Sun et al., 2018).
Although the molecular diversity of SGN appears to be partly dependent on input
from the hair cell, the interactions may be reciprocal. In zebrafish, the presence of afferent
fibers is crucial for the formation and stabilization of synaptic ribbons in the hair cell (Suli
et al., 2016). In mice, Sherril et al. (2019) recently reported that spatial gradients in the
voltage dependence of pre-synaptic calcium channels is controlled by the enhanced
expression of the Pou4f1 transcription factor in modiolar-contacting SGN. This would
mean post-synaptic heterogeneity maybe important for driving pre-synaptic
heterogeneity. These results suggest that functional heterogeneity of auditory nerve
responses in vivo may be determined by complex reciprocal signaling between hair cells
and spiral ganglion neurons during post-natal development. The more mature
electrophysiological phenotype of modiolar contacting fibers may be a reflection of this
reciprocal interaction between the neuron and hair cell.
Comparison between Acute and Cultured Preparations

68
SGN electrophysiology in the acute semi-intact preparation is similarly diverse to
that reported in cultured ganglionic preparations. We found that SGN in the acute
preparation were diverse in action potential properties and size of whole-cell
conductance. SGN with large conductances tended to be rapidly accommodating,
produced action potentials with fast latencies and in response to large current injections.
These neurons tended to contact the modiolar-face of the hair cell. These findings are
consistent with SGN biophysics in cultured neurons, where rapidly accommodating
neurons have faster latencies, higher current thresholds and larger low-voltage gated
potassium currents (Mo and Davis, 1997, Chen and Davis 2006; Lui et al., 2014; Lv et
al., 2010).
Our results are also consistent with the idea that modiolar-contacting SGN may
have larger concentrations of low-voltage-gated potassium conductances (gKL) than pillar-
contacting neurons. This suggestion is based on our observation that the neurons have
larger whole-cell conductance values at negative potentials (i.e. g-30 values). Although
we did not isolate specific ion channel currents (e.g. low-voltage gated currents conducted
by channels such as Kv1 and KCNQ), our results are consistent with previous studies in
which threshold and spike-train accommodation of SGN firing patterns were controlled by
these channels (Liu et al. 2014, Lv et al. 2010). Recent single cell RNA-sequencing
reports are consistent with our results in showing enrichment for potassium channels
(including high-voltage-gated potassium channels (Kv4)) is greater in modiolar-contacting
SGN (Shreshta et al., 2018; Sun et al., 2018; Petitpre et al., 2018).
The inclusion of the graded-response group and low-incidence of the slowly
accommodating group may be a difference between the present study and previous in
69
vitro studies of cultured SGN. Graded responses have also been observed in two other
studies where SGN were recorded from without a significant period in culture (Jagger and
Housley, 2003; Santos-Sacchi, 1993). One possibility is that acute demyelination may
not provide adequate time for sodium channels to redistribute and adequately populate
the newly demyelinated somata (as described in CNS by Shrager, 1993). Acute
demyelination may also push sodium channels to inactivate, requiring time for those
neurons to recover (Shrager, 1993). It is hard to guess if one or both of these scenarios
are at work in the acute preparation. Given the difficulty of recording in the semi-intact
preparation we opted to keep these cells in the analysis population. As our results show,
key membrane properties such as response latencies were similar in both graded and
spiking neurons, allowing us to pool these cells into a unifying analysis group.
Other potential differences between this and previous in vitro studies stems from
our recording in semi-intact preparations where the presence of peripheral dendrites
could make recordings non-isopotential. We excluded this possibility in finding that the
capacitive currents in response to small voltage steps were decaying with single-
exponential time courses. This suggested to us that we were looking at largely
isopotential compartments mostly likely confined to SGN somata (see supplement).
Whether developmental changes in myelination influence this interpretation will require
additional work characterizing the formation of myelin around the somata. Thus, aside
from differences that might arise from culturing, we believe that the interpretation of ion
channel properties is similar in this semi-intact preparation as for recordings from isolated
ganglionic preparations.
70
The biophysical properties of type II SGN compared to type I SGN
Type II SGN represent ~5% of the SGN population, make connections with multiple
outer hair cells, and have a yet unknown role in the neuronal encoding of sound
(Spoendlin, 1969, Kiang et al., 1984; Liberman and Oliver, 1984). In this study, type II
SGN were 9 of 59 labeled fibers ranging in age from P3 to P8. The small number of
neurons over a wide age range limits our ability to analyze and attribute a statistically
significant biophysical phenotype to type II. Also, there may be more than one sub-type
of type II fibers (Vyas et al. 2019), making the scarcity of type II’s even more limiting for
statistical evaluation. However, we found two biophysical trends that distinguish type II
SGN from the majority of type I SGN. Type II fibers tended to be on the extreme end of
the distribution in having small steady-state outward potassium currents that inactivated
rapidly (Figure 4). The values of both of these voltage-clamp features have a large
variance, with type II SGN positioned at the extrema of the two distributions. That the
currents of type II fibers inactivate faster than for type I fibers is consistent with the
previous reports of Jagger and Housley (2003). Our results extend those previous results
by now showing the inactivation time constant of type I SGN varies as a gradient from the
pillar to modiolar face. Although the voltage-clamp responses showed a trend relative to
type II and type I SGN, we did not find significant differences in response latencies in
current clamp for the type II SGN. The latencies of type II SGN were not significantly
different from the mean latencies of type I SGN. Because of this overlap, we were not
able to use these variables to predict differences between type I and type II SGN.

71
The relevance of somatic recordings

In vivo, the first spike-initiation zone is located closer to dendritic terminals than to
SGN somata. Whether gradients like those reported here at the somata are also seen at
the terminal remains to be tested. There is ample evidence showing that similar groups
of ion channels and neuro-transmitter receptors are seen at the soma and peripheral
neurite (Rutherford et al., 2012; Yi et al 2010; Hossain et al., 2005). However, recordings
from SGN terminals show little to no spike failure in response to EPSCs (Rutherford et
al., 2012; Wu et al., 2016), leading to the argument that SGN firing rate is set exclusively
by presynaptic mechanisms. Difficulty in access may limit terminal recordings to the
modiolar face (Rutherford et al., 2012), preventing those studies from seeing the full
biophysical gradient seen here.
Somatic ion channels may also have a function in of themselves. According to
models, the impedance mismatch between the peripheral dendrite and the large soma
can dramatically slow down and attenuate spikes, thereby increasing the probability of
spike failure (e.g. modeling by Rattay et al, 2013; Hossain et al., 2005). In this scenario,
the soma could independently filter spike trains as they sweep towards the brainstem
(Davis et al., 2015), and the nature of that filtering would be determined by somatic ion
channels.
Regardless of the ultimate functional consequence of gradients in somatic ion
channel properties, the observed link between in vitro cellular biophysics with putative
SR-subgroups opens new opportunities for exploring the mechanisms driving SGN
function. We’ve described a prediction model which uses easily measured biophysical
72
properties such as current threshold and first-spike latency to classify SGN into putative
SR-subgroups. Moreover, the modeling strategy systematically identified
covariant/redundant variables among a larger group of easily measured biophysical
variables. This identified gradients in potassium conductance as the most parsimonious
explanation for the group of biophysical gradients observed (Figure 5). As a result, highly
covariant variables such current threshold and latency could be used interchangeably to
predict basal position. By combining non-redundant biophysical variables (such as
current threshold and outward current inactivation), our regression models predict basal
position to within a +/- 9% margin of error. This is remarkable resolution given that the
inputs to the model use relatively simple biophysical measurements. The success of the
model is a reflection of the strength of these underlying relationships.
A final practical outcome of these results is that we can apply the prediction model
to study other properties of SR-subgroups by using technically favorable dissociated
somatic preparations. These would include studies focused on understanding the specific
ion channels expressed by different SGN subgroups, the sensitivity of these neurons to
different efferent neurotransmitters and the vulnerability of SGN subgroups to
degeneration (Furman et al. 2013). This work complements the search for molecular
markers for SGN subgroups because these approaches are feasible in neonatal animals
before molecular expression has matured and in non-mouse species where transgenic
models are not readily available.

73
Chapter 3.
SGN somatic recordings in the acute semi-intact preparations of cochleae display
isopotential behavior despite their bipolar morphology
Introduction
SGN have a bipolar morphology in which their somata rest in ganglia proximal to
the modiolus of the cochlea. Each SGN has two neurites that project from the soma with
one proximal neurite that synapses onto a single inner hair cell and a central neurite that
projects towards the cochlear nucleus via the brain’s eighth cranial nerve. In this
dissertation, I performed simultaneous whole-cell patch-clamp recordings and single cell
labelling in acute semi-intact preparations of cochleae and showed that the biophysical
properties of SGN systemically vary along the base of the inner hair cell. However,
because these patch-clamp recordings were performed on SGN with extending neurites,
there is a chance for the patch-electrode to have poor space-clamp. Poor space-clamp
conditions in this preparation would result in the SGN to display nonisopotential behavior.
Therefore, the conventional interpretation of patch-clamp recordings directly inferring the
underlying whole-cell conductance would have to be substituted for a wider interpretation
of the overall neuronal impedance that considers the combination of the conductances,
capacitive reactance, and inductive reactance in the acute semi-intact preparation.
In a series of in-depth biophysical assessments, we tested whether SGN recorded
in the acute semi-intact preparations of cochleae displayed nonisopotential behavior. In
doing so, we investigated how the Multiclamp 700B amplifier estimates and compensates
transient capacitance currents needed to correct series resistance errors and estimate
whole-cell capacitance. To our surprise, SGN in the semi-intact preparation largely
74
behaved like isopotential compartments of about the same size. As a result, we suggest
that the gradients we reported here reflect gradients in the ion channel densities on SGN
somata.

Results
A circuit diagram of the patch-clamp recordings of SGN in the semi-intact preparation of
cochleae we performed in these experiments is displayed in Figure 8A. The patch
electrode is connected in series to the neuron’s soma which is connected in series to an
axial resistor and then followed by series of compartments along the projecting neurite.
The impedance of the neuron, the combine biophysical effects of the conductances,
capacitive reactance, and inductive reactance, all impact measurements made both
current- and voltage-clamp. Furthermore, the precision in measuring the impedance of
the SGN becomes more difficult in the acute semi-intact preparation where
nonisopotential behavior can be caused by extending neurites and the variability in
myelination of the soma and/or the neurite. Therefore, we considered whether our
recordings displayed isopotential behavior in order to determine whether the differences
in biophysical properties measured in this study can be caused by factors other than the
underlying conductances.
Our arguments are summarized in the four paragraphs below. First, we exclude
the concern about non-isopotential behavior by examining the time-course over which the
capacitive transient current decays in voltage clamp. Second, we calculate the size of
the “well-clamped” portion of the membrane by estimating the effective capacitance, also
in voltage clamp. Third, we conclude that differences in axial conductance did not
75
produce spatial gradients in input conductance by testing for correlations between
dendritic diameter and net-input conductance for Type I fibers. Finally, we exclude the
possibility that systematic errors in series resistance could account for the gradients in
biophysical properties reported here.

1. Isopotential behavior
We tested for non-isopotential behavior by analyzing high resolution recordings of
capacitive transient currents (ICV) in voltage clamp. These measurements were only made
in a subset of recordings. The capacitive transient current was measured by
hyperpolarizing the neuron from -60 mV to -65 mV (Figure 8B). This produced a small
inward current that decayed. The capacitive portion of the current was isolated by first
removing the steady-state current (shaded area under the current). In non-isopotential
cells, the decay of the capacitive current can best be described by a sum of exponential
decays (e.g. equation (1)).

𝐼
+1
= 𝐴
&
𝑒
2
%
t
!

+𝐴
(
𝑒
2
%
t
"

+𝐶 (1)

For example, the capacitive currents in retinal bipolar cells and the Purkinje cells
are best fitted with a bi-exponential function, with each exponent term representing the
sequential decay across two compartments representing the soma and neurite
(Golowasch et al 2009; Taylor 2012; Rall 1969). In contrast, the decay from an iso-
potential cell is adequately described by a single exponential term.
76
Here we fit the capacitive transient currents with equation (1) and compared the
coefficients (A1, and A2) of the two exponential terms where τ2 < τ1 (Figure 8C). The ratio
between the fast and slow terms (A2/A1) shows that the fast component represented more
than 80% of the transient current in 11 out of 16 neurons and was observed in both pillar-
and modiolar-contacting SGN (Figure 8D). Since the fast component dominates, this
suggests that our recordings are largely isopotential.

2. Size of well-clamped membrane
We estimated the extent of the plasma membrane clamped under voltage-clamp
by computing the effective capacitance (Taylor et al., 2012). Capacitance values
averaged at 7.8+/- 0.35 pF (n=35) and were not correlated with normalized basal position
(r = 0.05, p = 0.85) (Figure 8E). These measurements suggest that there was no
systematic modiolar to pillar gradient in the cell surface area clamped by the recording
electrode.
Based on confocal scans in the ganglion, we estimate that the SGN in our
preparation had an average diameter of around 12.7 +/- 0.3 µm, n=28. This observation
is consistent with measurements made in other mammalian species where somatic
diameters range between 10-15 µm (Tsuji and Liberman, 1997; Berglund and Ryugo,
1991; Romand et al., 1987; Nadol et al., 1990). Assuming an unmyelinated spherical
soma (with a specific capacitance around ~1µF/cm
2
), the somatic capacitance would be
around 5.1 +/- 0.2 pF. Based on the similarities in values between the capacitance as
estimated from voltage clamp and from estimated cell diameters, we conclude that we
are largely clamping and measuring the currents local to the SGN somata.
77
In this study, we were not able to measure the extent of the myelin that may be
present on the neuron, and therefore cannot assess how differences in myelination and
consequent variations in specific capacitance influence our measured capacitance.
Because we remove myelinating satellite cells before recording from the SGN soma, we
can assume that there is some myelination of the cell bodies and/or neurite. However, we
cannot account for SGN having differences in the extent of their myelination which would
affect the specific capacitance and total surface area of the neuron. Furthermore,
myelination of the cell bodies and/or the neurite may increase the length constant of
voltage changes along the neuron and could effectively make the neuron display
isopotential behavior. Further studies accessing the passive membrane properties of
SGN in the acute semi-intact preparation are needed to determine whether the measured
capacitance of the SGN changes when different degrees of myelin are present.
Note that the capacitance or membrane surface area of a passive membranes can
also be measured in current-clamp; however, SGN have voltage-dependent
conductances that activate near resting potential (~-60mV). The distortions caused by
the activation of these conductances make it difficult to accurately estimate capacitance
in current clamp.

3. Axial conductance
Differences in biophysical behavior measured in voltage-clamp did not arise from
differences in the axial conductance, where neurons with larger diameters would have
larger axial conductance. Based on our fiber diameter measurements (Figure 7), the axial
resistance of type I SGN does not change significantly as SGN contact position moves
78
along the modiolar-pillar axis. In contrast, type II SGN had significantly smaller fiber
diameters than type I SGN but had similar net conductance values as that of pillar-
contacting type I SGN, suggesting that variations in axial diameter did not translate into
significant variations in input conductance.

4. Series resistance
Series resistance estimates were not significantly different in SGN along the modiolar-
pillar axis, suggesting that there were negligible errors in measuring the magnitude of
currents in voltage-clamp. Series resistance estimates of identified type I SGN ranged
from 2.6 to 11.0 MΩ with an average of 6.6 +/- 2.3 MΩ (n=20). This series resistance
estimate was measured via the Multiclamp 700B membrane seal test by analyzing the
capacitive transient current. In some cases, these estimated series resistances were less
than the pipette resistance, which lead to us question the accuracy of the measurement.
This possible inaccuracy of estimating the series resistance led to our decision to report
our voltage-clamp measurement uncorrected for series resistance. Series resistance
estimates as a function of basal position displayed a slight negative linear trend (Figure
8F). However, when we performed series resistance corrections on our voltage-clamp
measurements, our initial assessment is maintained with maximum magnitudes of
steady-state currents significant correlated with basal position (R
2
= 0.29, p = 0.0003)
(Figure 8F, red trace). To further assess how much error in the series resistance
estimation is needed to collapse the trends of steady-state currents with basal position,
we simulated increasing levels of correlations of the series resistance estimations to the
normalized basal position scale (Figure SG, blue and black traces). Our analysis shows
79
that there would need systematic errors 6 times stronger than the amount we measured
in our dataset to collapse these trends (Figure 8G, black trace). Because series
resistance of a patched SGN was not correlated with whether the neuron contacted the
modiolar- or pillar-face of the inner hair cell, magnitude of currents measured in the
voltage-clamp were not significantly impacted to cause concern for our biophysical
assessment.
Here we did not compensate for or correct voltage errors resulting from the series
resistance as we are accustomed to do in isolated neurons. We chose to report raw
whole-cell currents to avoid inadvertently adding variability from cell to cell that may result
from incorrect estimates of series resistance. Our decision stemmed from our initial
concerns that space-clamp errors in the semi-intact preparations would make it difficult
to accurately estimate series resistance using the built-in capacitance neutralization
function in the Multiclamp 700b. The neutralization procedure assumes a single
exponential decay for an iso-potential cell, as would be reasonable for the compact
morphology of small isolated somata (see above). We were concerned that imperfect
neutralization of capacitance transients would produce errors in both series resistance
and membrane capacitance estimates. To avoid adding additional sources of variability
to our voltage-clamp estimates of currents, we chose not to apply series resistance
corrections or to normalize currents by cell capacitance.

80

Figure 8:
A. Circuit diagram of an acute SGN in the semi-intact preparation clamped in whole-cell configuration. This diagram
shows the series resistance (Rpip), membrane resistance (Rm), axial resistance (Ra), and a secondary compartment
resistance from the neurite (Rcomp) with corresponding capacitors (C). B. Voltage-clamp response of the transient
capacitance current of an SGN in the semi-intact preparation fitted with a bi-exponential function (green trace). The fast
(pink trace) and slow (blue trace) components of the bi-exponential function are plotted to show the contribution of the
current coming for the somata (fast/pink) and neurite (slow/blue). C. The relative size of the magnitude of the fast (A2)
and slow (A1) components of the trace, indicating that the fast component dominates the fit and currents are measured
primarily from the soma. D. The fast component of the bi-exponential (A2) relative proportion to the sum of the bi-
exponential function (A1+A2). A2 relative proportion dominates in both modiolar(red) and pillar (blue) contacting SGN.
E. Cell capacitance, measured as the integral of the capacitance current, is not correlated with basal position (r=0.05,
p = 0.85). F. Series resistance as a function of normalized basal with the mean trend line is shown in red. G. After
performing series resistance error corrections per cell, we see that the position correlation of maximum outward current
(Imax) and NBP does not collapse, and remains correlated (R
2
= 0.23, p = 0.0003). If we simulate increasing levels of
81
systematic series resistance errors by three (blue trace) and six (black trace) times the amount we measured, trends
along NBP do not collapse.

Discussion
When studying the whole-cell biophysics of SGN, we analyzed the relationship
between low-voltage activated (g-30) and high-voltage activated (gmax) potassium
conductances to gauge the ion channel composition of each SGN. The ratio of g-30 to
gmax was correlated with the Vmid of the whole-cell conductance curves, which
supported our analytical view that the more currents activated at -30mV would shift the
whole-cell conductance curves in a hyperpolarized direction. Our data showed SGN with
relatively high g-30/gmax values had a more hyperpolarized Vmid and produced action
potentials at high-thresholds and with short latencies. These findings are consistent with
studies that pharmacologically blocked specific currents (i.e. Kv1, Kv3, HCN) which
depolarized resting potentials, decreased voltage thresholds, and shorten action potential
latencies (Chen and Davis 2006; Lui et al., 2014). Therefore, we can interpret our findings
that modiolar-contacting neurons, with larger magnitudes of g-30 and gmax conductance
values, have larger concentrations of low-voltage potassium currents (IKL) than pillar-
contacting neurons. Furthermore, the larger concentration of low-voltage gated
potassium conductance of modiolar-contacting SGN drives the firing patterns of these
neurons to have larger thresholds with fast latencies.
Consistencies between whole-cell recordings from the acute semi-intact and the
cultured ganglia preparations are valid if we interpret each recording as an isopotential
cell. Isopotential neurons can be represented as a simple RC circuit in which the total
whole-cell conductances and the capacitance of the neuron will drive the biophysical
82
behavior of the neuron. Soma diameter in both acute semi-intact and cultured ganglia
preparations does not significantly vary, and therefore, the whole-cell capacitance of
these neurons’ somata are relatively equivalent and the spatial gradient of biophysical
properties would be driven by the increase in net potassium conductance along the
modiolar-pillar axis. However, whole-cell recordings of SGN from the semi-intact
preparation may not be isopotential due to the axonal process that extends from the
soma. Whole-cell recordings of nonisopotential neurons are prone to space clamp
variability and series resistance compensation errors due to the presence of the axonal
process.
Our initial analysis of the data considered all current measurements coming from
the clamped SGN soma. However, space clamp variability between recordings could elicit
differences in whole-cell current measurements by measuring the currents from the whole
soma and/or the soma with a secondary compartment. The conventional method of
accessing space clamp is to estimate the capacitance of the patched neuron in either
voltage- or current-clamp configuration by measuring a passive membrane response to
a small step of stimulus. A feature in nonisopotential neurons is a mismatch in
capacitance estimates measured in current- and voltage clamp (Nadim et al., 2009;
Taylor, 2012). Current-clamp effectively measures the capacitance of the entire neuron,
soma and its processes, while voltage-clamp measures the capacitance of the portion of
the neuron that is sufficiently clamped with the recording electrode. Our data showed this
mismatch, primarily based on our inability to isolate passive membrane responses in
current-clamp; and therefore, we could not acquire accurate capacitance measurements
in current-clamp configuration.
83
The correlations of biophysical properties with the capacitance would be easily
explained if we assess our recordings as isopotential cells (RC circuits) where larger
capacitance measures would increase the membrane time constant, however, a more
complex interpretation can be considered when analyzing the biophysics of the
instantaneous currents produced by nonisopotential neurons with differential axonal
processes. Impedance differences of SGN along the modiolar-pillar axis can be due to
various properties such as the structure of the soma and neural process, the addition of
myelin, and the ion channel composition. However, the counter-intuitive relationships of
between the biophysical properties measured in this study and what is known about the
overall structure and maturation of SGN along the modiolar-pillar axis leads us to suggest
that impedance differences we present are primarily due to differences in conductance
values. For example, previous studies have shown that in the mature animal, pillar-
contacting SGN have longer and thicker axonal processes than modiolar-contacting SGN
(Merchan Perez and Liberman, 1982; Dunkenrod and Goodrich 2010). Our data shows
that in the first weeks of post-natal development, branch diameter does not differ between
type I SGN along the modiolar-pillar axis. If there were fiber diameter difference in this
post-natal development period, thicker axons would establish longer length constants and
faster time constants for these neurons, which would cause induce faster voltage changes
(i.e latencies) (Rall 1957). However, pillar-contacting SGN display longer latencies than
modiolar-contacting SGN, suggesting the differences in axonal structural do not account
for the impedance differences. Second, differences in myelination of the SGN soma and
neural process may cause impedance differences due to the myelin sheets reducing the
amount of leak currents flowing from the neuron. The maturation process of myelination
84
may likely follow the maturation process of the neurons, which our data suggest that the
modiolar-contacting neurons develop at a faster rate than pillar-contacting neurons.
However, this is counter-intuitive to our measurements showing modiolar-contacting SGN
displaying more leak and overall larger steady-state currents. Structural properties of the
SGN and its neural process do not account for the spatial gradients in impedance we
measured along the modiolar-pillar axis, and therefore, impedance differences are
primarily due to differences in whole-cell conductance.

85
Chapter 4.
Classifying neurons based on simply measurable biophysical properties
Introduction
The synaptic components of the inner hair cell’s ~30 ribbon synapses are spatially
organized into two groups: the modiolar and pillar ribbon synapses. Modiolar ribbon
synapses have larger pre-synaptic ribbons and calcium clusters than pillar ribbon
synapses (Kalluri and Monges-Hernandez 2017; Liberman and Liberman 2016; Ohn et
al., 2016). In addition to the ribbon synapse components, the in vivo electrophysiology of
modiolar- and pillar-contacting spiral ganglion neurons (SGN) differ in their spontaneous
discharge rate and in vivo thresholds (Merchan-Perez and Liberman, 1996). As more data
reveals how modiolar- and pillar-contacting SGN and their respective ribbon synapses
differ from each other, there has been a growing convention to split the SGN population
into a bimodal classification scheme. However, based on the data presented in this
dissertation, a bimodal distribution may not be the best way of explaining how the inner
hair cell’s ribbon synapses are functionally organized.
In chapter 2, linear regressions were performed to predict the normalized basal
position based on biophysical properties with our two best models predicted the basal
position within a 10-14% margin of errors. We chose to use a linear regression because
of the striking linear correlations observed between current threshold and response
latency with the basal position of type I spiral ganglion neurons. A limitation of these linear
regressions is that more data is needed to validate the model’s prediction accuracy and
acquiring this data is experimentally challenging. The challenge of collecting data stems
from the fragile nature of the hair cells after recording and immunoprocessing. In some
86
cases, the hair cells were damaged and therefore we could not sufficiently determine a
normalized basal position. However, in the same cases, we could accurately determine
whether the fiber contacted the modiolar or pillar face. Thus, by reducing the demand of
the model to predict just whether the SGN is modiolar- or pillar-contacting, increases our
sample population and enables the use methods to validate our model’s accuracy that
were not available to us in the initial model fitting process.
Here, I will perform logistic regression model fitting to categorically classify and
predict spiral ganglion neuron subtypes as modiolar- or pillar-contacting. I will begin with
an overview of logistic regression by performing exploratory data analysis on how well
the two best predictors used in the linear regression models, current threshold and
latency, predicts modiolar- and pillar-contacting SGN. I will provide guidelines in how we
can validate our model and ensure that we do not overfit the model. Next, the best
prediction model is made via a multivariant logistic regression that considers how multiple
biophysical properties can be used to most precisely predict modiolar-contacting SGN.
SGN subtypes can be classified to the field’s preferred conventional of modiolar- versus
pillar-contacting by using easily measured biophysical properties. The bimodal classifier
presented here can be used in numerous studies to precisely identify modiolar-contacting
SGN from the total population.
Methods

The logistic regression determines the probability (P) of an input (X) being a part of
dichotomous group (i.e. yes = 1, no = 0). The linear equation for this model is
𝒍𝒐𝒈
𝒏
𝑷
𝟏−𝑷
= 𝒍𝒐𝒈
𝒏
𝒆
(𝜷
𝒐
+𝜷
𝟏
𝑿)
= 𝜷
𝒐
+𝜷
𝟏
𝑿 (1)
87
where the coefficient(s)(ß) of the model represent the rate in change in logit P for a
change in the X variable. This model is used to measure the probability of neuron being
classified as either a modiolar-contacting versus a pillar-contacting neuron.
When performing the linear regressions in chapter 2, we determined whether there
was redundancy within model variables by evaluating the value inflation factors (VIF). We
limited our model to have a VIF less than 4 and removed any variables that had a VIF
over this threshold. For this validation train, test, split model fitting paradigm, I used a
different method to penalize redundant or nonimportant variables that would inflate and/or
not aid the model in its prediction. With the same goal of reducing the model’s complexity
and reduce over-fitting the model, I used Lasso (L1) regularization to penalize features
that would inflate or not aid in model prediction. Lasso regularization ultimately penalized
error-prone and collinear features by reducing the coefficients to zero, thus some features
are completely neglected for the evaluation of the model output and leads to the reduction
of overfitting.
In addition to the analysis detailed in the methods of chapter 2, here, I sorted cells
into spiking and non-spiking responses. To make this sorting easier, we defined a metric
that allowed us to consider the current-clamp responses to a family of current steps (Istep)
ranging from 0 to 150pA. Graded index was quantified as the maximum rate at which
peak depolarization (Vp) changes in response to current steps (𝑔𝑟𝑎𝑑𝑒𝑑 𝑖𝑛𝑑𝑒𝑥 =
max

>1?
>@A%B?
,). The graded index provides a continuous variable to measure the degree of
the graded-depolarization. Large graded index values indicate that the neuron produced
a large spike ( graded index > 1), while small graded index values indicate that the neuron
did not produce a spike (graded index < 1).
88

Results
Univariate Data Analysis to predict SGN subtypes using Logistic Regression

Response latency and current threshold were determined to be the two best
predictors of basal position based on their strong linear correlation. When we split the
data into modiolar- and pillar-contacting SGN, the distributions significantly overlap with
one other (Figure 9A, 9D). Even though a student t-test determines that modiolar- and
pillar-contacting SGN have significantly different current thresholds (t= -3.16, p = 0.004)
and latencies (t = 2.6, p = 0.013), the overlap in the predictors’ distributions may provide
challenges in fitting a model that classifies SGN as either one of the morphological
categories. I performed a logistic regression to test whether these two variables could
independently predict whether the SGN was modiolar-contacting.
89

Figure 9:
Logistic regressions to classify SGN as modiolar (1) or pillar (0) contacting neurons based on biophysical properties.
A. The distribution of response latencies from modiolar (blue) and pillar (orange) contacting SGN.B. Logistic regression
with response latency as the predictor variable. C. The area under the ROC curve is 0.87, which indicates an 87% rate
of predicting true positives. D. The distribution of current thresholds for modiolar and pillar contacting SGN. E. Logistic
regression with current threshold as the predictor variable. F. The area under the ROC curve is 0.93, indicating a more
accurate logistic model.

In Figure 9B and 9E, I show two logistic regressions to predict whether a neuron
contacted the modiolar-side (1) or the pillar side (0) (Figure 10). In Figure 9B, the function
can be interpreted as when the neuron displays a latency of ~20ms, there is a 50%
probability of it contacting the modiolar side of the inner hair cell. As latencies become
faster, the probability of the neuron contacting the modiolar face increases. Similarly, in
Figure 9E, for current threshold, there is a 50% probability of being classified as modiolar
contacting when the displayed current threshold is ~40pA.
90
In order to validate the model’s accuracy, I established training and test group sets
by splitting the dataset into randomly assigned training datasets (75%) and test datasets
(25%). The training set is optimally fitted by the model parameters, in this case via logistic
regression, and establishes a function to predict modiolar- and pillar-contacting neurons
based on the biophysical properties. We then used this function to assess how accurate
the model is at predicting the test dataset, which has not been used in the original model
fitting, and therefore, can assess without biases. Overfitting can be determined when the
model created by the training dataset fits the training set well but cannot predict the test
dataset with balanced accuracy. Therefore, a model that can accurately classify a
neuron’s as modiolar- or pillar-contacting on the test dataset will validate our model.
To evaluate the model, we analysis how well the model predicts true-positives and
how frequently does the model predict false-negatives on the test dataset. We can
visualize this prediction accuracy by assessing the confusion matrix. The confusion matrix
for this model is a 2x2 grid that shows the true positives, false positives, true negatives,
and false negatives for this model’s ability to correct predict modiolar-contacting neurons.
True positives indicate that the model correctly predicted a modiolar-contacting neuron,
while false positive indicate that it failed to classify an actual modiolar-contacting, a
modiolar-contacting neuron. Therefore vice versa, true negatives indicate the model’s
ability to accurately predict pillar-contacting neurons and false positives indicate a pillar-
contacting neuron that was classified as modiolar-contacting neurons.
Actual Positives (Modiolar) Actual Negatives (Pillar)
Predicted Positives (Modiolar) 7 0
Predicted Negatives (Pillar) 4 0

91
In this test dataset, the model classified 11 SGN, and correctly classified 7 SGN
as modiolar-contacting and incorrectly assigned 4 pillar-contacting SGN as modiolar-
contacting. The model failed to determine the boundaries that separate pillar-contacting
SGN from the modiolar-contacting SGN as seen by having 0 actual negatives predicted.
I performed cross-validation of the test dataset, splitting the datasets, fitting the
model and performing the predicting process multiple times, which established different
training and test sets and averages the accuracy scores of the model as shown. Cross-
validation ensures that every data point is used in both the training and test set at least
once and will provide a better estimate of the model’s accuracy. Cross-validation is
especially important for this dataset where pillar-contacting neurons represent only a
small population of the total innervation. After performing cross-validation 15 times, the
average accuracy score for the logistic model is 0.70 for latency and current threshold,
which is more representative of the true accuracy of the model.
The innervation at the inner hair cell is not symmetrical. Modiolar-synapses
represent more than 70% of the total population, with only a small portion of synapses
made on the pillar-face. Therefore, it would be incorrect to assume that there is a 50%
probability that a synapse would be on the pillar-face. A more accurate threshold of the
where the cut-off point is for pillar-contacting SGN may reduce the number of false-
negatives predicted in the model. We can shift and evaluate moving the cut-off point by
shifting the threshold value of the logistic regression. The ROC curves display these
results and can be interpreted as the probability of prediction true-positives (1), as we
move the threshold to predict false-positives (0). To gauge how well the model fits the
data, we can analysis the area under the ROC curve (AUC) (Figure 9C, 9F). For response
92
latency and current threshold, AUC is 0.87 and 0.93, respectively, which means there is
an 87% and 93% accurate prediction to classify the neuron as modiolar-contacting for
latency and current threshold. Therefore, by moving the threshold to exclude more pillar-
contacting neurons, a more accurate model is made to predict modiolar-contacting
neurons from the population with only current threshold or latency as the predictor
variable.
Multivariant analysis boosts the performance of the model to predict SGN subtypes

Next, I performed multivariant analysis to determine whether the combination of
biophysical properties could more accurately classify modiolar- and pillar-contacting
neurons. In previously presented analyses in Chapter 2, a more accurate model was
established using multiple variable, and therefore, our objective is to validate these
models. For this analysis, I used the same variables I used in the linear regressions
including response latency, current and voltage threshold, graded index, resting potential,
conductance magnitudes at -30mV (g-30) and maximum conductance (gmax) to predict
whether the SGN contacting the modiolar- or pillar-face of the inner hair cell.
Lasso regularization of the model variables penalized the weights of the variables
to negligible amounts (0.0), and only allowed the most useful variable to be implemented
in the model. The variables with applicable weights were latency, graded index, resting
potential, and net conductance (gmax). The variables coefficients are displayed in Figure
10C. These variables are similar to what was presented in the linear regression model
and validate our criteria for eliminating variable with VIF score greater than 4. The addition
93
of graded index and gmax, increased the number of true negatives accurately predicted,
which was missing the univariate analysis and made the final model more accurate.

Figure 10:
A. The distribution of biophysical properties that were applicable for model fitting after lasso regularization of
modiolar (blue) and pillar (orange) contacting SGN. B. Multivariant logistic regression ROC curve with an AUC of 0.875,
indicating that the model has an 87.5% accuracy rate of classifying true positives. C. A table of the coefficients for the
94
logistic regression function. Note that lasso regularization penalized non-applicable and redundant variables by making
the coefficients equal to zero.

The distributions of the variables that were not penalized by lasso regularization
methods have areas where modiolar- and pillar-contacting SGN have areas where they
do not overlap, and therefore, can play a role in the model’s goal to classify them into
separate groups (Figure 10A). The multivariate logistic regression outperformed the
univariate analysis and validated the findings presented in the linear regressions
presented in chapter 2 with a prediction score of 0.91:
Actual Positives (Modiolar) Actual Negatives (Pillar)
Predicted Positives (Modiolar) 7 0
Predicted Negatives (Pillar) 1 3

In this test dataset, the model classified 11 SGN, and correctly classified 7 SGN as
modiolar-contacting, 3 SGN as pillar-contacting and incorrectly assigned 1 pillar-
contacting SGN as modiolar-contacting. Overall, we can see that the model correctly
predicted the SGN subtype 10 out of 11 times, and therefore, the accuracy score for this
model would be 0.91. Furthermore, we see that the model is highly precise in classifying
modiolar contacting neurons with 100% accuracy is detecting actual positives. There is a
tendency for the model to predict pillar contacting neurons as modiolar-contacting, but
this happened only in 1 out of 4 SGN.
Similarly, we shifted the threshold for false positive sensitivity as a way to shift to
the distribution of modiolar- and pillar- synapses to be more representative of the in vivo
innervation. This shift can be evaluated with the ROC and AUC scores. Our final AUC
95
was 0.875 and indicates that we can accurately predict modiolar-contacting SGN from
the population with an 87.5% accuracy rate (Figure 10B).
Discussion
In this chapter, we determined that modiolar- and pillar-contacting SGN could be
accurately predicted via logistic regression classification models. The models performed
here strengthened the findings presented in chapter two because we were able to validate
the models by splitting the datasets into training and test sets and accurately predicted
test data as modiolar- or pillar-contacting. We compared our findings to the original
models presented and found that similar variables were best at predicting type I SGN
subtypes such as latency, current threshold, and whole cell conductance. Overall,
validating our model further signifies that the biophysical properties of SGN are
systematically linked to their synaptic position with the inner hair cell.
There were some variables that were better at predicting a bimodal distribution,
modiolar- versus pillar-contacting neurons, than a continuous distribution along the inner
hair cell such as graded index and net conductance. Graded index is a measurement to
determine whether the neuron produced an action potential. Pillar-contacting neurons
failed to produce a spike more often than modiolar-contacting SGN and was a key feature
in classifying SGN via bimodal classification. Similarly, the net conductance of pillar-
contacting neurons was significantly smaller than modiolar-contacting SGN. Based on the
large magnitude of the graded index coefficient, the logistic regression classifier primarily
analyzed whether the neuron fired an action potential to determine whether it was a
modiolar-contacting SGN.
96
The linear regression model (Figure 5B) produced an accurate prediction of the
normalized basal position within a 13% margin of error, while the multivariant logistic
regression model produced an 87.5% accuracy rate (Figure 10B). Reducing the demands
the model to predict a bimodal classification produced an equal set of error than predicting
along a continuous scale. It is difficult to assess which one of the models is better. Both
models can accurately pull out putative high-SR SGN and low-SR SGN from the
population to study in isolation for future studies.
However, only on the continuous scale does the model consider the linear gradient
of how the biophysical properties change along the base of the hair cell. The current
convention of the field is to bisect and compare the two sides inner hair cell: the modiolar
side and the pillar side. This division has been used to set up a bimodal system to
compare the synaptic components. Although, this established bimodal system has been
useful in starting to explain the complex organization of the innervation and the biophysics
of synapses at an individual inner hair cell, it may be bias in its broad attempts to simplify
the relationships among type I synapses. As shown in chapter 2, type I SGN are not
bimodally distributed in their biophysical properties but are more equally distributed along
a continuous axis along the base of the inner hair cell. When looking at the innervation of
a single inner hair cell, type I fibers primarily contact the modiolar face and basal pole
area, which take up approximately 70% of the available surface area compared to the
pillar face. Therefore, pillar contacting spiral ganglion neurons may be the extrema of the
total distribution and may not be well served as a bimodal comparison to the full modiolar
contacting innervation. For example, terminal recordings made on the modiolar face, but
towards the basal pole, may not be significantly different than “pillar contacting SGN”
97
because they are more proximal to the pillar contacting population. Throughout this
chapter, the likelihood of false positives is likely due to this relationship of basal positioned
synapses having similar features that are not like SGN that synapse towards the cuticular
plate on the modiolar side (i.e NBP > 0.3). Therefore, the linear regression model is most
likely the best model for predicting SGN subtypes because it can parse out the basal pole
region with higher resolution.

98
Chapter 5.
Generation of Inner Ear Hair Cells by Direct Lineage Conversion of Primary Somatic
Cells
Introduction
Hearing loss can be attributed to the damage and/or loss of inner and outer hair
cells in the cochlea. Unlike other somatic cell types, hair cells cannot regenerate, and
therefore, any hearing loss due to degeneration of hair cells is permanent. Hair cell loss
can be due to multiple external factors including noise-induced damage and ototoxins
that selectively target and kill hair cells. A goal in the restorative hearing science field is
to assess medicinal substrates for ototoxicity and test for preventative treatments.
Although methods such as gene therapy and stem cell therapy have been making large
advancements in the field, the inaccessibility of the cochlea to deliver treatments provides
obstacles to perform these assessments. Additionally, in vitro methods that remove the
cochlea produce low yields of viable hair cells for assessment, and therefore, testing the
vast amounts of medicinal products currently available to the public and future
candidacidal products comes with a high financial cost. In order to bypass the high costs
and low yields of primary hair cells, investigators have turned to stem cell methods to
produce a high throughput of cells that mimic the molecular and cellular properties of
actual hair cells.

One of these stem cell methods includes the reprogramming of somatic cells to a
pluripotent cell. Takahashi and Yamanaka (2006) showed that mouse fibroblasts could
become induced pluripotent stem cells (iPSCs) through molecularly manipulating the
99
DNA with specific transcriptions factors. The resulting iPSC could then be molecularly
guided into differentiating into a target cell type. This discovery led to a wave of studies
that produced numerous reprogrammed cells types including induced neuronal cells,
cardiomyocytes, motor neurons, and hepatocytes (Feng et al., 2008; Ieda et al., 2010;
Huang et al., 2011; Son et al., 2011). Along with the expansion of different types of
induced cells, the methodology of reprogramming itself also advanced. These advanced
methods bypassed the pluripotent state of the cell and directly transformed one somatic
cell type to another, a process known as direct lineage reprogramming (Martens et al.,
2016).

Direct lineage reprogramming has been implemented to successfully produce vestibular
hair cell-like cells (Liu et al., 2015); however, cochlear hair cells have not yet been
developed. In a collaborative project with Louise Menendez of the Ichida and Segil
laboratories, I have worked to assess the electrophysiological phenotype of induced hair
cells she has developed for her own thesis project. I will briefly summarize the molecular
biological methods she performed to develop the induced cochlear hair cells. This brief
summary is followed by the results of the electrophysiological measurements I performed
in our collaboration. Based on the molecular characterizations and electrophysiological
recordings, the induced hair cells developed accurately mimic the properties of cochlear
hair cells.

100
Methods
Somatic cells transformed into induced hair cells via direct reprogramming (Menendez et
al)
Mouse embryonic fibroblasts (MEFs) were obtained at E13.5 from a transgenic
mouse line that contained the Atoh::nGFP reporter (f.k.a. Math1-GFP; Lumpkin et al.,
2003). These MEFs underwent direct reprogramming by overexpressing hair cell specific
transcription factors via retrovirus transduction. The transcription factors included Atoh1,
Pou4f3, Gfi1, and Six1. Atoh1 has been identified as an essential hair cell gene, and
therefore, having a GFP reporter for the transcription of Atoh1 is a marker for the success
of the direct programming process. In addition to Atoh1, these four transcription factors
transformed the MEFs to induced hair cells (iHCs).
Next, Menendez et al determined whether the iHCs expressed other cochlear hair
cell-like molecular and morphological properties. First, immunostaining of the iHCs for the
hair cell specific protein MyosinVIIa and Parvalbumin was positive. Second, to test for the
presence of stereocilia and acetylated tubulin, immunostaining for phalloidin and kinocilia
was performed and showed positive expression in the iHCs. These immunolabelling
experiments showed Atoh1::nGFP/MyosinVIIa+ cells with polarized morphologies of
which had immunofluorescent phalloidin and acetylated tubulin at a concentrated section
at one pole of the iHC. This morphological arrangement of the phalloidin/tubulin signal
mimics the arrangement of stereocilia of primary hair cells.
Next, the groups of Atoh1::nGFP+ iHCs underwent differential cell culture
protocols. One group of iHCs were cultured in a monolayer of MEFs at day 14-15 post
transduction. The second was co-cultured with a dissociated P1 organ of Corti. For a
101
control group, a third group of dissociated primary hair cells with Atoh1::nGFP+ were also
cultured. Induced hair cells can be viewed in Figure 11.
Results
Passive membrane properties of induced hair cells in each experimental condition
We performed whole-cell patch clamp recordings on our induced hair cells (iHCs) and on
primary hair cells from a dissociated organ of Corti. We measured the biophysical
properties of our cells in voltage-clamp and current-clamp mode in order to analyze the
voltage-gated currents and passive membrane properties of these cells. Here, we
compared the results of iHCs in two experimental conditions: 1) monolayer iHCs (n=10)
and 2) iHCs that are co-cultured with a dissociated organ of Corti (n=11) with recordings
of primary hair cells (n=5). We found that the biophysical properties of our co-cultured
iHCs are in line with those measured of primary hair cells in both their voltage-dependent
current density values and activation time constants. Furthermore, the passive membrane
properties of these induced hair cells in the co-cultured condition exhibit similar resting
potentials, input resistances, and capacitance values. These similar properties were not
observed in the monolayer iHCs, which have overall differential biophysical properties
than both the co-cultured and primary conditions.
102

Figure 11:
A) Immunostaining of monolayer iHCs show polarized F-actin by Phalloidin labeling (red) and a kinocilium by acetylated
Tubulin labeling (cyan). Merged image includes Hoechst nuclear stain. Scale bar represents 10um. B) Co-cultured
iHCs with dissociated primary E13.5 organs of Corti show F-actin rich cuticular plate by phalloidin labeling (red) and
stereocilia by Espin (grey) labeling. Merged image includes Hoechst nuclear stain. Scale bar represents 20um. C)
MEFs infected with GFP control virus do not accumulate the styryl dye FM4-64. Image taken after 30 second of
incubation with FM4-64. Scale bar represents 50um. D) iHCs expressing the Atoh1::nGFP reporter accumulate FM4-
64. Image taken after 30 seconds of incubation with FM4-64. Scale bar represents 50um. E, E’, E”) Whole cell patch
clamping was performed on P1 primary hair cells from a dissociated organ of Corti, co-cultured iHCs and monolayer
iHCs. Results from current clamp show the change in cell voltage as a response to an applied current. Dashed red line
represents -60mV. Current clamp protocol shows steps of 20pA from -10 to +150pA. Scale bars represent 50mV on X-
axis and 250ms on Y-axis. F) Basic membrane properties were calculated from the current clamp data to report resting
membrane potential (Vm), membrane capacitance (Cm) and input resistance (Rin). G, G’, G”) Results from voltage
clamp shows the current output of the cell as a response to applied voltage for primary hair cells, co-cultured iHCs and
monolayer iHCs. Dashed red line represents 0pA. Voltage clamp protocol shows steps of 10mV from -120 to +70mV.
Scale bars represent 1nA on X-axis and 125ms on Y-axis. H) IV curve plotting current density (normalized for cell size)
103
as a function of applied voltage. I, I’, I”) Exponential fits to the voltage clamp traces were used to calculate the current
activation time constants for primary hair cells, co-cultured iHCs and monolayer iHCs. Dashed red line represents 0pA.
Solid red line shows exponential fit to outward currents when clamped from -120mV to 70mV. Scale bars represent
1000pA on X-axis and 12.5ms of Y-axis. J) Current activation time constants reported for P1 primary hair cells,
cocultured iHCs and monolayer iHCs.(Menendez et al.; Figure 4E-J by A.Markowitz)

We performed current-clamp protocols on each iHC and primary hair cell to measure the
passive membrane properties. In order to measure the cell’s intrinsic responses, we did
not hold the membrane potential to a specific voltage. Examples of the current-clamp
response in each experimental condition is shown in Figure 9E. Here, we see that each
cell responds to the current-clamp protocol by depolarizing or hyperpolarizing its
membrane potential. Cells produce a graded depolarization at the onset of positive
current steps, and in some cases (n=2), the monolayer iHCs exhibited oscillations in the
membrane potential in response to positive steps of currents.

Multiple properties were measured from the current-clamp responses including the
resting potential, input resistance, and capacitance. The resting potential was measured
and averaged at two points. The mean resting potential of primary hair cells is -58.6 +/-
6.9mV, and the resting potential of monolayer iHCs and co-cultured iHCs are -50.8 +/-
2.4mV and -54.8 +/-4.1mV, respectively (Figure 11F). The input resistance was measured
at the current-clamp response to -10pA of injected current. The input resistance was
calculated by measuring the change in membrane potential from rest by the injected
current (~-10pA). Input resistance value infer the total ion channel composition of the cell.
Higher input resistance values indicate the cell may have fewer ion channels to allow
current to flow in and out of the plasma membrane. Monolayer iHCs have the highest
input resistance (3837.4 +/- 648.2MΩ) compared to primary hair cells (1950.3 +/-
755.3MΩ) and co-cultured iHCs (1432.1 +/- 345.0 MΩ). Co-cultured iHCs and primary
104
hair cells have similar input resistance values, which may indicate that they have a similar
number of ion channel inserted into their plasma membrane. Lastly, the capacitance of
the cell was calculated by fitting a single exponential through the membrane potential
response when -10pA was injected. The exponential estimates the membrane time
constant and the capacitance is then calculated from this membrane time constant and
the input resistance. The capacitance can be used to infer the surface area of the cell.
Primary hair cells (8.4 +/- 3.1pF) were larger than monolayer (4.2 +/- 1.2pF) and co-
cultured iHCs (5.6 +/- 2.2pF). The mean values of these passive membrane properties
in each experimental condition is compared in Figure 9F.

Voltage-clamp responses of co-cultured induced hair cells are similar to primary hair cells.
In each experimental condition, we performed voltage-clamp protocols to measure
the total currents that are activated at specific voltages. The membrane was clamped at
-60mV, which was followed by a pre-pulse hyperpolarization step to -120mV. The pre-
pulse step is to relieve any inactivation of sodium channels. This was followed by a family
of voltage steps from -120mV to +70mV. An example of a cell’s response in each
condition is shown in Figure 11G. In response to the protocol, the iHCs and primary hair
cells produce positive-outward currents, and, in some cases, transient negative-inward
currents.
The magnitude of the outward currents varied from cell to cell and between each
experimental condition. Also, there was variation in the activation time course of these
voltage-gated currents. For example, in the monolayer condition, outward-currents
105
quickly activate and inactivate, which is seen by the salient decrease in the size of current
from the onset of the stimulus to the steady-state values ~400ms after the stimulus onset.
Contrary to the voltage-clamp response in the monolayer condition, the primary hair cells
and co-cultured iHCs have more slowly inactivating outward currents.
We measured the steady-state outward current density (at 400ms) as a function
of the command voltage to characterize the voltage-clamp responses of our cells in each
experimental condition (Figure 11H). We normalized our current magnitudes by the
capacitance of each cell to analyze the current density of each cell. The current density
provides a more precise measurement of the total ion channel composition of each cell
and indicates that larger values are due to increase in ion channel density rather than cell
size. We plotted the mean current density and standard error of the mean of each
condition. Here, we can see that the monolayer iHCs have smaller current densities than
primary hair cells and co-cultured iHCs. The co-cultured iHCs and primary hair cells have
overlapping voltage-dependent current densities. In particular, the current densities that
are activated at negative membrane potentials have consistent values in both of these
conditions.

A prominent voltage-clamp feature in primary hair cells is a delayed onset of a
slow-activating outward current (Housley & Ashmore, 1992; Marcotti & Kros, 1999). In
order to measure the kinetic properties of this slow-activating outward current, we fit a
single exponential at the onset of the current (Figure 11I, I’, I”) to compare the mean time
constants when the cells were clamped from -10mV to 70mV (Figure 11J). The delayed
onset current of monolayer iHCs displayed fast time constants. In contrast, the co-
106
cultured iHCs and primary hair cells showed similarly longer time constants, indicating
that their outward currents have similar activation kinetics. Together, these
electrophysiological data suggest that when iHCs are co-cultured with dissociated organ
of Corti, the size, passive membrane properties and ion channel function of iHCs and
primary hair cells are similar.

Discussion
Menendez et al established a high-throughput stem cell method to create cell lines
that mimic the properties of cochlear hair cells via direct lineage reprogramming. The
induced hair cells (iHC) had similar genetic, morphological, electrophysiological, and
ototoxic properties that amalgamated into a sufficient hair cell model. Future experiments
can use these iHC cell lines to test future pharmacological substrates for ototoxic effects
without the costly demand of acquiring actual hair cells.
The degeneration of hair cells is the most cause of hearing loss, and because hair
cells do not regenerate, this type of hearing loss is permanent. The work produced in this
dissertation will contribute to the field of restorative and regenerative hearing science by
providing a method to produce large quantities of hair cell-like cells to test the lethality of
potential ototoxins and preventative treatments. In addition to successful mimicking the
functionality of hair cells, this study also showed interesting observations from the most
successful induced hair cells. When induced hair cells were co-cultured with a dissociated
organ of Corti, the cells in the dish started to self-organize into small islands of cells.
Interestingly, the induced hair cells would be on the perimeter of these islands, which
would form a lumen at the center. Although we know the addition of the dissociated organ
107
of Corti induces these structural organizations to form, the underlying mechanisms that
causes the cells to become connected structures remains unclear. Recordings from these
induced hair cells in the co-cultured condition showed an electrophysiological phenotype
that most accurately mimicked primary hair cells. Therefore, interactions between cells in
the newly formed aggregation of cells could impact the biophysical properties of the
induce hair cells. One possible mechanism could stem from cells secreting neurotropic
factors that are necessary for the development of the hair cells. Cells from the organ of
Corti excrete NT-3, NGF, and BDNF that allow the cells to proliferate and also have been
shown to underlie development of ion channel properties. Future studies need to test
whether reprogrammed induced hair cells need neurotropic factors from other cell types
in order to successfully mimic the properties of primary hair cells. Additionally, it is unclear
whether the induced hair cells produce properly assembled ribbon active zones. The
ribbon synapse is the specialized functional unit of the inner hair cell, and therefore, it is
vital for these induced hair cells to produce ribbon active zones.

108
Chapter 6.
Discussion
How does diversity in post-synaptic biophysical properties of spiral ganglion neurons
contribute to sound intensity coding?
The post-synaptic ion channel properties play one role in a larger synaptic network
that work together to differentially drive the SGN terminals. A current conventional view
is that the post-synaptic ion channel properties of SGN were homogeneous, and
therefore, would not play a role in determining differential responses needed to encode
different sound intensities (Rutherford et al., 2012; Wu et al., 2016). The data presented
in this dissertation is the first demonstration that the diversity of ion channel properties is
systematically organized about the auditory nerve’s effective map for ‘intensity-
sensitivity’. Specifically, the more excitable neurons with low current thresholds in vitro
are found on the pillar side of the inner hair cell, where in vivo auditory nerve fibers have
high spontaneous rates and are most sensitive to low intensity sounds (Kiang, 1965;
Liberman, 1978). Greater intrinsic excitability may contribute to pillar-contacting fibers’
preferred sensitivity to low-intensity sounds in vivo. Thus, the direction of in vitro
biophysical gradients qualitatively aligns with mature in vivo physiology and hints at a
possible post-synaptic contribution in controlling sound-driven thresholds and excitability.
Differences in specific pre- and post-synaptic ribbon synapse properties can
independently lead to differences in SGN responses. Below, I demonstrate how each
component of the ribbon synapse may separately drive differential responses in vivo,
starting with the key findings in this dissertation:

109
1. Differences in net conductance
The results in chapter two showed that there was a gradient in the magnitudes of
whole-cell conductance SGN. SGN that display larger whole-cell conductances have
larger amounts of ion channels inserted into their plasma membrane and will elicit
different responses to the same stimulus compared to SGN with lower net whole-cell
conductances. Strong evidence presented in this dissertation supports this claim such as
the significant correlations between net conductance with response latency and current
threshold. Response latency and current threshold are two characteristics that can serve
as markers for how these neurons would respond in vivo. For example, response latency,
the duration from stimulus onset to peak potential, can serve as a proxy for the membrane
time constant. This is because when SGN are modelled as an RC circuit, the membrane
time constant is equal to the product of the capacitance of the cell and the net input
resistance (1/net conductance). In chapter 2, we showed that response latency is best
explained by the function, 1/gmax, and thus, shows an equivalent relationship between
these membrane properties. Differences in the response latency may implicate
differences in EPSPs in vivo which may underlie the differences in excitability that define
the intensity coding SGN subgroups.
Second, current threshold represents another characteristic that can be used to
test how differential in vivo responses occur based on their post-synaptic biophysics.
Current threshold is the minimal amount of current needed to produce an action potential
and therefore can be translated to in vivo thresholds in response to EPSCs. SGN with
large net conductances needed larger amounts of current in order to fire an action
potential and vice versa. If the ribbon active zone produced equal EPSCs for each SGN,
110
then the SGN could respond differently via differences in their net conductance.
Therefore, the gradient in current thresholds measured may be represented by a gradient
in in vivo thresholds that define the sound-intensity coding SGN subgroups.

2. Differences in specific ion channel currents
A second interpretation of the data in chapter two is provided by the larger
magnitudes of whole-cell conductance activated at low-voltages (-30mV). Low-voltage
activated currents, such as IKL and IH, are voltage-dependent and active when SGN are
clamped at -30mV. Therefore, the whole-cell conductance at -30mV provides a gauge of
the total ion channel composition of these low-voltage gated ionic currents. Because
larger whole-cell conductances at -30mV are displayed in one subgroup of SGN, there
may be larger amounts of IKL and IH in these SGN. The biophysical implications of these
low-voltage gated currents have been shown to underlie thresholds and excitability in
other systems, and therefore, may play a similar role in SGN. For example, in the cochlear
nucleus, the reciprocal relationship of the IKL and IH underlies the function of octopus cells
to integrate multiple EPSPs to function as a coincidence detector (Goldings et al., 2012).
Furthermore, the relative portions of IKL and IH have been shown to underlie the firing rate
and regularity of vestibular ganglion neurons (Ventura and Kalluri, 2019; Kalluri et al.,
2010). In both these systems, activation of IKL dampens the excitability of the neuron by
driving the membrane potential away from threshold, while the activation of IH increases
the excitability of the neuron by driving the membrane potential towards threshold. The
whole-cell conductance measurements presented in this dissertation does not separate
the contribution of individual low-voltage gated currents; however, because the total
111
whole-cell conductance at -30mV is larger in one subgroup of SGN, we can interpret this
value as possibly having larger amounts of IKL due to these neuron’s high current
thresholds.

3. Glutamate receptor density
Differences in the glutamate receptor (GluR) densities at the post-synaptic terminal
of each ribbon synapses may play a role in determining differential SGN responses. If
larger densities of GluR are found on a subgroup of SGN, they may be able to produce
larger amplitudes of EPSCs, and thus initiate each neuron with different inputs.
A concentration gradient of GluR2/3 signaling is seen along the modiolar-pillar
axis in mouse (Liberman et al 2016), however, this observation was not reported in Long
Evans rat (Kalluri and Monges-Anvil, 2017). In the same study, Liberman et al reported
an equal distribution of post-synaptic density marker (PSD95), a tethering protein for
glutamate receptor subtypes, among SGN morphological subgroups (Liberman et al.,
2016). This indicates that there is a strong possibility that other subtypes of GluR, such
as GluR4, may populate the SGN terminals at an equal concentration of terminals with
dense GluR2/3 population. GluR4 is not easily quantified via immunostaining and
therefore could not be measured in the same matter as GluR2/3. Therefore, although the
subtypes of GluR may differ between synapses, the overall response driven by the
concentration of GluR may be the same for each SGN.
4. Pre-synaptic ribbon active zone features
The inner hair cells’ ribbon active zone function is to aggregate large quantities of
synaptic vesicles, cluster and stabilize calcium channels, and drive specific synaptic
112
excitation to one SGN terminal. Differential ribbon active zone function can be implicated
by differences in the number of synaptic vesicles, the probability of release rate, and the
quality of quantum transmission. Previous studies have quantified the size of the ribbon
and showed that the size of the ribbon varies between active zones (Liberman et al.,
2016; Kalluri et al., 2017). Larger ribbons can hold more vesicles, and therefore, may
provide larger amplitudes and/or rate of glutamate quanta than smaller ribbon active
zones.
Pre-synaptic calcium ion channels are clustered at each ribbon synapse and
function to trigger voltage-dependent exocytosis of ribbon-tethered synaptic vesicles.
EPSC amplitude and release rate can be determined by differences in the density of
calcium channels (Ohn et al., 2016, Frank et al., 2009). Larger densities of calcium
channels are located on the modiolar-side than on the pillar-side of the inner hair cell
(Ohn et al., 2016) and therefore, have a higher probability of outputting all of the synaptic
vesicles in the readily releasable pool. The outcome of the gradient in calcium release
may be differences in the amplitude and intervals of EPSCs coming from the ribbon active
zone, which would drive the SGN terminals with different inputs.
There is limited data on the possible differences in how the number of vesicles and
the densities of calcium channel determine the quality of glutamatergic quanta release
inner hair cell ribbon synapses (Grant et al., 2010; Ricci 2019 ARO abstract). Differences
in the rate and quality of glutamate release could result in differential subtypes of
SGN that respond preferentially to either large amplitudes and/or multiple quantal events.
113
Limitations of the experimental design and conclusions
First, whole-cell recordings performed in this study could not isolate specific ionic
currents, and therefore, we could not identify the magnitude or contribution of individual
ion channel types. Voltage-clamp recordings measured all contributing conductances that
are activated when the membrane potential is clamped at a particular value. For this
study, we extrapolated the whole-cell conductance values at -30mV (g-30) to include the
contribution of low-voltage gated potassium conductances (gKL), hyperpolarization
activated mixed-cation conductance (gh), and leak conductance. Because SGN with
relatively larger magnitudes of g-30 exhibited large current-thresholds, we interpreted the
g-30 value as the effects of gKL, a current known to increase thresholds. In order to
confirm this assumption, patch-clamp recordings with pharmacological blockers must be
applied to measure the ion channel composition of each SGN.
Second, patch-clamp measurements were performed on the cell bodies of SGN,
away from the spike-initiation zone which is located on the terminal. We chose to record
from the cell body of SGN in order to sample the entire population of SGN that terminate
onto each inner hair cell. The terminals on the pillar-face of the inner hair cell are highly
obstructed by supporting tissues, and therefore, patch-clamping onto the SGN terminal
would not give us a full representation of the SGN population. The ion channel properties
at the soma are therefore used a proxy for those located at the spike initiation zone. In
support for using the cell body as a proxy is seen in previous studies that showed similar
ion channel composition at the soma that is present at the soma (Rutherford et al., 2012;
Yi et al 2010; Hossain et al., 2005). Furthermore, in dissociated, cultured SGN cell bodies
insert glutamate receptors into their plasma membranes indicating that the soma acts like
114
the spike initiation zone in vitro. The soma of the SGN in vivo rests in the middle of the
signal propagation pattern. Therefore, the biophysical properties of the soma may act as
a filter as signals propagate towards the brainstem. Therefore, we could be capturing two
unique properties of SGN functionality with this study, one as a proxy for the ion channel
composition of the spike-initiation zone, and two, a separate measure of the ion channel
properties of the soma which may play an equally important role in the function of these
neurons.
Third, the ages at which I collected the data span across a large developmental
window for the Long Evans rat. Data for these experiments were performed from post-
natal day (P)1 through P16 when there are many age-dependent changes in properties
of the cochlea, including the onset of hearing. The majority of the data I collected were in
ages right before the onset of hearing and showed how the biophysical properties of SGN
were already systematically organized by the time the animal starts hearing. Furthermore,
the sample of SGN recorded from older animal, after the onset of hearing, suggested that
the organization we measured in early post-natal days was still present. Although our
data suggests that the spatial gradient of biophysical properties persists after the onset
of hearing, more data is required to confirm these extrapolated conclusions.
Fourth, as explored in chapter 3, patch-clamp recordings of SGN in the semi-acute
preparation of cochlea are prone to display nonisopotential behavior. Nonisopotential
behavior is due the patch-clamp electrode poorly controlling the potential across l of the
entire neuron’s surface area, also known as poor space-clamp. Differences in space-
clamp could stem from differences in neuronal properties such as the soma size, axial
resistance, and myelination. As discussed in chapter 3, our analysis shows that soma
115
size and axial diameter (resistance) are homogenous among the SGN population and
therefore likely do not contribute to differences in displayed behavior. However, we cannot
account for differences in myelin-related properties among the SGN. In chapter 3, we
suggested that differences in myelination could result in differences in the amount of leak
current presented in the neuron. If a neuron had more myelin and/or more tightly wrapped
myelin around the neurite, there would be less leak current displayed in the patch-clamp
measurements. Because we mechanically remove the myelinating satellite cell around
the soma before each recording, it is impossible to know whether there are fractions of
the myelin sheet still on the soma and/or differences in the myelin in the neurite.

Future Experiments
Exploring the role of glutamatergic input on SGN subtypes
Future studies are necessary to examine synaptic mechanisms that contribute to
cell-type specific expression of ion channels. In this dissertation, I showed that the
biophysical properties of SGN systematically differentiate, however, the mechanisms that
underlie how the SGN develop differential biophysical properties remains unknown. In
the mammalian cochlea, individual SGN are driven by ribbon active zones that vary in
their morphology and physiology (Merchan-Perez and Liberman, 1996; Liberman et al.,
2011; Liberman and Liberman, 2016; Ohn et al., 2016; Kalluri and Monges-Hernandez,
2017). The ribbon active zone consists of the ribbon, which will aggregate and tether
numerous glutamate-filled synaptic vesicles, and voltage-gated calcium channels. If the
variance of ribbon active zone morphology and physiology ultimately results in differential
release of glutamate (e.g. produce mono- and multi-phasic events), then the differential
116
regulation and release of glutamate provided by each ribbon active zone may drive
differential expression of ion channels amongst SGN (Frank et al., 2010). Therefore, I
hypothesize that glutamate release by the ribbon active zone will induce different whole-
cell electrophysiological phenotypes among the SGN population. If this hypothesis is
supported, then when glutamate release is eliminated in the system, differences in whole-
cell conductances and corresponding action potential features that are seen to
differentiate modiolar- and pillar-contacting SGN will collapse, and the SGN population
will become more homogeneous.
Glutamate release triggering the expression of particular ion channels has been
reported in molecular characterization of SGN via single-cell RNA-sequencing in a
genetically altered mouse model in which the expression of vesicular glutamate
transporter (VGLUT3) is knocked-out. In this mouse model, ribbon synapse vesicles are
empty and do not contain any glutamate. The VGLUT knock-out mice has been
extensively studied in the auditory system and is available for purchase through Jackson
Labs (Seal et al., 2008; Akil et al., 2012). The use of transgenic manipulation in this mouse
model is necessary to tease out specific contributions in the synaptic transmission of inner
hair cell-SGN synapse. Mouse, similar to rat, cat, and guinea pig, have similar distribution
the SR-subgroups (Liberman, 1978; Tsuji and Liberman, 1997; Jagger and Housley,
2003; Taberner et al., 2005; Frank et al., 2010). Pre-synaptic ribbon zone morphology
along the modiolar- pillar axis in mouse have a similar developmental timeline as
compared to rat (Liberman and Liberman, 2016; Kalluri and Monges-Hernandez, 2017).
Also, ion channel recordings from mouse somata show similar heterogeneity and whole-
cell ion channel properties as in rat (Mo and Davis, 1997; Jagger and Housley, 2003).
117
Although the molecular characterization via RNA-sequencing shows that
glutamate is necessary for SGN to develop different RNA expression levels of post-
synaptic features, they do not provide data on the electrophysiological behavior of these
neuron in the VGLUT knockout model. In order to test whether glutamate activity is
necessary for differential post-synaptic biophysical behavior, experimenters can use the
experimental protocol explained in chapter 2 to simultaneously record and labeling SGN
in the semi-intact preparation of cochleae. The data collected from the VGLUT3 knockout
mice would need to be compared to wild-type mice that do not have the knock-out
expression. Wild-type mouse ion channel properties are expected to differentiate into two
distinct electrophysiological phenotypes based on whether they contact the pillar- or
modiolar-side of the inner hair cell, as seen in the data, presented here, in rat. VGLUT3
knockout models are expected to have no distinct electrophysiological phenotypes
amongst the SGN population, regardless of morphology.
Alternatively, future experimenters could apply the prediction model to classify
spiral ganglion neurons along the normalized basal position scale (chapter 2) in order to
collect electrophysiological data more rapidly and with easier methods, such as neuronal
dissociation. Collecting measurements such as response latency, current threshold, and
maximum whole-cell conductance are the best predictors of basal position of the inner
hair cell. I would predict that whole cell conductances, current thresholds, and latencies
would not differ along the normalize basal position in the VGLUT knockout mouse model
because differential biophysical phenotypes are dependent on differential qualities of
glutamate release.

118
Determine whether pre-synaptic ribbon active zone morphology correlates with post-
synaptic ion channel properties of cochlear afferents.
In adult cochleae, ribbon size is inversely correlated with SR (Merchan-Perez and
Liberman 1996). Because ribbons have been shown to organize the pre-synaptic active
zone by aggregating calcium channel clusters and stabilizing synaptic connections,
ribbons (or ribbon associated structures) could also drive the maturation of the post
synaptic terminal (Sheets et al., 2011, Sheet et al., 2012). Also, in the hair cell ribbon
synapses of zebrafish, there is evidence showing that the innervation of hair cells
influences the size and number of ribbons (Suli et al., 2016). These findings suggest that
there an active feedback loop between the hair cell’s ribbon synapse and the SGN
afferents. I propose that the high volume of release of large ribbon active zone instructs
the afferent SGN to produce and insert high volumes of ion channels into its plasma
membrane. I would hypothesize that the SGN that contact large ribbons will have larger
whole-cells currents in effect of this proposed mechanism.
Future experiments could test this hypothesis by examining whether there is a
correlation between the whole-cell current density of SGN and the volume of their
contacting ribbon. My thesis data shows that there is a correlation of whole-cell
conductance along the modiolar-pillar axis of the inner hair cell. Previous studies have
shown that pre-synaptic structures including the ribbon volume exhibits a spatial gradient
along this axis with larger ribbons on the modiolar side than pillar side of the hair cell
(Liberman et al 2016; Kalluri et al., 2017). Therefore, in general, larger whole-cell
conductances of modiolar-contacting SGN are measured where larger ribbons are
generally located. However, the modiolar side of the inner hair cell has a large variance
119
of ribbon volume (Chapter 4; Figure 9,10). This leads to question whether the mechanism
that drives the polarization of pre-synaptic structures into spatial gradients also shape the
biophysical properties of SGN in a similar polarizing matter. A future experiment could be
to determine whether there is a correlation between ribbon size and whole-cell
conductance. A positive correlation will reveal a systemic relationship between the SGN
afferent and its paired ribbon active zone.
I would predict that there is a systematic relationship between the size of the pre-
synaptic ribbon and the post-synaptic ion channel current density. If this hypothesis is
supported, then there will be a positive correlation between ribbon size and SGN current
density. This correlation will be more significant than the gross correlation seen along the
modiolar-pillar axis, where generally larger ribbons are found on the modiolar-face than
pillar-face. In the knockout model, we will collapse the general modiolar-pillar spatial
gradient of ribbon sizes. This collapse is expected to result in a parallel collapse of the
current densities along the modiolar-pillar axis. However, because there is still a large
variance of ribbon sizes within the population of synapses, the correlation between ribbon
size and SGN current density is expected to persist.
These results would suggest to me that ribbons play a role in the development of
post-synaptic ion channel composition. Large ribbons that supply large and fast rates of
synaptic vesicles for exocytosis induce changes in the post-synaptic density, including
the aggregation of ion channels. This observation would lead me to question how large
ribbons could drive a post-synaptic feature. A future study would focus on whether post-
synaptic ion channel properties are ribbon activity dependent and whether or not
disrupting the development of ribbon maturity in the inner hair cell would result in similar
120
post-synaptic changes (i.e by using an efferent knockout model where ribbon sizes
collapse).

Exploring the role of efferent input on SGN subtypes
The afferent SGN population receives two synaptic inputs: 1) glutamatergic input from the
ribbon active zone of the inner hair and 2) efferent input from the lateral olivocochlear
nucleus (LOC) (Liberman et al., 1990). If glutaminergic activity doesn’t instruct the post-
synaptic ion channel properties of the SGN, the efferent LOC projections likely contribute
to the differentiation of SGN ion channel properties. LOC fibers have been reported to
release numerous neurotransmitters including dopamine and acetylcholine (Maison et al.,
2003). The function of the LOC fibers remain poorly understood; however, when
experimenters surgically remove the efferents from the system, the synapse between the
inner hair cell and SGN afferent population significantly changes (Liberman et al. 2011).
Removing the LOC fibers collapses the modiolar-pillar spatial gradient in ribbon active
zone features including the size of the ribbon (Yin et al., 2014). The size of the ribbon
along the modiolar-pillar axis is a dominant morphological feature of the adult cochlea
and is shown to correlate with the function (SR) of the SGN. If the elimination of glutamate
releases does not cause the expected collapse in distinct electrophysiological phenotypes
of SGN subtypes, then the input of the efference system is an alternative candidate
shaping the SGN subtypes. Just as the LOC efferent system is necessary for the
development of ribbon active zones along the modiolar-pillar axis (aka intensity coding
axis), the LOC may shape the differential ion channel properties of SGN along the same
axis. Future experiments can use a transgenic mouse model that eliminates the efference
121
system and perform whole-cell recordings to determine whether distinct
electrophysiological phenotypes are amongst SGN without the input of the efferent
system.

122
References
Adamson, C. L., Reid, M. A., Mo, Z. L., Bowne-English, J., & Davis, R. L. (2002). Firing
features and potassium channel content of murine spiral ganglion neurons vary
with cochlear location. Journal of Comparative Neurology, 447(4), 331–350.
Akil O, Seal RP, Burke K, Wang C, Alemi A, During M, Edwards RH, Lustig LR (2012)
Restoration of hearing in the VGLUT3 knockout mouse using virally mediated gene
therapy. Neuron 75:283–293
Barclay, M., Ryan, A. F., & Housley, G. D. (2011). Type I vs type II spiral ganglion
neurons exhibit differential survival and neurogenesis during cochlear development.
Neural Development, 6(1).
Barry, P. H. (1994). JPCalc, a software package for calculating liquid junction potential
corrections in patch-clamp, intracellular, epithelial and bilayer measurements and
for correcting junction potential measurements. Journal of Neuroscience Methods,
51(1), 107–116.
Berglund, A. M., & Ryugo, D. K. (1991). Neurofilament antibodies and spiral ganglion
neurons of the mammalian cochlea. The Journal of Comparative Neurology,
306(3), 393–408.
Browne, C. J., Morley, J. W., & Parsons, C. H. (2012). Tracking the expression of
excitatory and inhibitory neurotransmission-related proteins and neuroplasticity
markers after noise induced hearing loss. PLoS ONE, 7(3), e33272.
Chen, W. C., & Davis, R. L. (2006). Voltage-gated and two-pore-domain potassium
channels in murine spiral ganglion neurons. Hearing Research, 222(1–2), 89–99.
Coate, T. M., Spita, N. A., Zhang, K. D., Isgrig, K. T., & Kelley, M. W. (2015). Neuropilin-
2/Semaphorin-3F-mediated repulsion promotes inner hair cell innervation by spiral
ganglion neurons. Elife, 4.
Cramer, K. S., & Miko, I. J. (2016). Eph-ephrin signaling in nervous system
development. F1000Res, 5.
Crozier, R. A., & Davis, R. L. (2014). Unmasking of spiral ganglion neuron firing
dynamics by membrane potential and neurotrophin-3. Journal of Neuroscience,
34(29), 9688–9702.
Das, G., Yu, Q., Hui, R., Reuhl, K., Gale, N. W., & Zhou, R. (2016). EphA5 and EphA6:
regulation of neuronal and spine morphology. Cell Biosci, 6, 48.
123
Davis, R. L. (2003). Gradients of neurotrophins, ion channels, and tuning in the cochlea.
Neuroscientist, 9(5), 311–316.
Davis, R. L., & Crozier, R. a. (2015). Dynamic firing properties of type I spiral ganglion
neurons. Cell and Tissue Research.
Echteler, S. M. (1992). Developmental segregation in the afferent projections to
mammalian auditory hair cells. Proceedings of the National Academy of Sciences
of the United States of America, 89(14), 6324–6327.
Echteler, S. M., & Nofsinger, Y. C. (2000). Echteler 2000 (postnatal SG death),
446(June), 1–11.
Flores-Otero, J., Xue, H. Z., & Davis, R. L. (2007). Reciprocal regulation of presynaptic
and postsynaptic proteins in bipolar spiral ganglion neurons by neurotrophins. The
Journal of Neuroscience : The Official Journal of the Society for Neuroscience,
27(51), 14023–14034.
Frank, T., Rutherford, M. A., Strenzke, N., Neef, A., Pangršič, T., Khimich, D., … Moser,
T. (2010). Bassoon and the Synaptic Ribbon Organize Ca2+ Channels and
Vesicles to Add Release Sites and Promote Refilling. Neuron, 68(4), 724–738.
Furman, A. C., Kujawa, S. G., & Liberman, M. C. (2013). Noise-induced cochlear
neuropathy is selective for fibers with low spontaneous rates. J Neurophysiol,
110(3), 577–586.
Glowatzki, E., & Fuchs, P. (2002). Transmitter release at the hair cell ribbon synapse.
Nature Neurosci., 5, 147–154.
Golding, N. L., Ferragamo, M. J., & Oertel, D. (1999). Role of intrinsic conductances
underlying responses to transients in octopus cells of the cochlear nucleus. J
Neurosci, 19(8), 2897–2905.
Golowasch, J., Thomas, G., Taylor, A. L., Patel, A., Pineda, A., Khalil, C., & Nadim, F.
(2009). Membrane capacitance measurements revisited: Dependence of
capacitance value on measurement method in nonisopotential neurons. Journal of
Neurophysiology, 102(4), 2161–2175.
Goutman, J. D., & Glowatzki, E. (2011). Short-Term Facilitation Modulates Size and
Timing of the Synaptic Response at the Inner Hair Cell Ribbon Synapse.
Grant, L., Yi, E., & Glowatzki, E. (2010). Two modes of release shape the postsynaptic
response at the inner hair cell ribbon synapse. The Journal of Neuroscience : The
Official Journal of the Society for Neuroscience, 30(12), 4210–4220.
124
Heidelberger, R., Heinemann, C., Neher, E., & Matthews, G. (1994). Calcium
dependence of the rate of exocytosis in a synaptic terminal. Nature, 371(6497),
513–515.
Hopkins, K. (2015). Deafness in cochlear and auditory nerve disorders. Handb Clin
Neurol, 129, 479–494.
Hossain, W. A., Antic, S. D., Yang, Y., Rasband, M. N., & Morest, D. K. (2005). Where
Is the Spike Generator of the Cochlear Nerve? Voltage-Gated Sodium Channels in
the Mouse Cochlea. Journal of Neuroscience, 25(29), 6857–6868.
Huang, L.-C., Thorne, P. R., Housley, G. D., & Montgomery, J. M. (2007).
Spatiotemporal definition of neurite outgrowth, refinement and retraction in the
developing mouse cochlea. Development, 134(16), 2925–2933.
Hudspeth, A. J. (1982). Extracellular current flow and the site of transduction by
vertebrate hair cells. Journal of Neuroscience, 2(1), 1–10.
Issa, N. P., & Hudspeth, A. J. (1994). Clustering of Ca2+ channels and Ca2+-activated
K+ channels at fluorescently labeled presynaptic active zones of hair cells.
Proceedings of the National Academy of Sciences of the United States of America,
91(16), 7578–7582.
Jagger, D. J., & Housley, G. D. (2003). Membrane Properties of Type II Spiral Ganglion
Neurones Identified in a Neonatal Rat Cochlear Slice. The Journal of Physiology,
552(Pt 2), 525–533.
Kalluri, R., & Monges-Hernandez, M. (2017). Spatial Gradients in the Size of Inner Hair
Cell Ribbons Emerge Before the Onset of Hearing in Rats. Journal of the
Association for Research in Otolaryngology, 18(3), 399–413.
Kalluri, R., Xue, J., & Eatock, R. A. (2010). Ion channels set spike timing regularity of
Mammalian vestibular afferent neurons. Journal of Neurophysiology, 104(4), 2034–
2051.
Kawase, T., & Liberman, M. C. (1992). Spatial organization of the auditory nerve
according to spontaneous discharge rate. The Journal of Comparative Neurology,
319(2), 312–318.
Keen, E. C., & Hudspeth, A. J. (2006). Transfer characteristics of the hair cell’s afferent
synapse.
Kim, J. J., Gross, J., Morest, D. K., & Potashner, S. J. (2004). Quantitative study of
degeneration and new growth of axons and synaptic endings in the chinchilla
cochlear nucleus after acoustic overstimulation. J Neurosci Res, 77(6), 829–842.
125
Kopp-Scheinpflug, C., Dehmel, S., Dörrscheidt, G. J., & Rübsamen, R. (2002).
Interaction of excitation and inhibition in anteroventral cochlear nucleus neurons
that receive large endbulb synaptic endings. J Neurosci, 22(24), 11004–11018.
Lelli, A., Asai, Y., Forge, A., Holt, J. R., & Géléoc, G. S. G. (2009). Tonotopic gradient in
the developmental acquisition of sensory transduction in outer hair cells of the
mouse cochlea. Journal of Neurophysiology, 101(6), 2961–2973.
Lenzi, D., & Von Gersdorff, H. (2001). Structure suggests function: The case for
synaptic ribbons as exocytotic nanomachines. BioEssays, 23(9), 831–840.
Li, G.-L., Keen, E., Andor-Ardó, D., Hudspeth, a J., & von Gersdorff, H. (2009). The
unitary event underlying multiquantal EPSCs at a hair cell’s ribbon synapse. The
Journal of Neuroscience : The Official Journal of the Society for Neuroscience,
29(23), 7558–7568.
Liberman, L. D., Wang, H., & Liberman, M. C. (2011). Opposing gradients of ribbon size
and AMPA receptor expression underlie sensitivity differences among cochlear-
nerve/hair-cell synapses. The Journal of Neuroscience : The Official Journal of the
Society for Neuroscience, 31(3), 801–808.
Liberman, L. D., & Liberman, M. C. (2016). Postnatal maturation of auditory-nerve
heterogeneity, as seen in spatial gradients of synapse morphology in the inner hair
cell area. Hearing Research, 339, 12–22.
Liberman, M. C. (1982). The cochlear frequency map for the cat: labeling auditory-nerve
fibers of known characteristic frequency. The Journal of the Acoustical Society of
America, 72(5), 1441–1449.
Liberman, M. C. (1978). Auditory-nerve response from cats raised in a low-noise
chamber. The Journal of the Acoustical Society of America, 63(2), 442–455.
Liberman, M. C., & Dodds, L. W. (1984). Single neuron labeling and chronic cochlea
pathology. III. Stereocilia damage and alterations in threshold tuning curves.
Hearing Research, 16, 54–74.
Liberman, M. C., & Oliver, M. E. (1984). Morphometry of intracellularly labeled neurons
of the auditory nerve: correlations with functional properties. The Journal of
Comparative Neurology, 223(2), 163–176.
Liberman, M. C., Dodds, L. W., & Pierce, S. (1990). Afferent and efferent innervation of
the cat cochlea: Quantitative analysis with light and electron microscopy. The
Journal of Comparative Neurology, 301(3), 443–460.
126
Liu, Q., Manis, P. B., & Davis, R. L. (2014). I h and HCN channels in murine spiral
ganglion neurons: tonotopic variation, local heterogeneity, and kinetic model. J
Assoc Res Otolaryngol, 15(4), 585–599.
Liu, W., & Davis, R. L. (2014). Calretinin and calbindin distribution patterns specify
subpopulations of type I and type II spiral ganglion neurons in postnatal murine
cochlea. Journal of Comparative Neurology, 522(10), 2299–2318.
LR, Edwards RH (2008) Sensorineural Deafness and Seizures in Mice Lacking
Vesicular Glutamate Transporter 3. Neuron 57:263–275
Lv, P., Wei, D., & Yamoah, E. N. (2010). Kv7-type channel currents in spiral ganglion
neurons: Involvement in sensorineural hearing loss. Journal of Biological
Chemistry, 285(45), 34699–34707.
Martinez-Dunst, C., Michaels, R. L., & Fuchs, P. A. (1997). Release sites and calcium
channels in hair cells of the chick’s cochlea. The Journal of Neuroscience, 17(23),
9133–9144.
Mehta, B., Snellman, J., Chen, S., Li, W., & Zenisek, D. (2013). Article Synaptic
Ribbons Influence the Size and Frequency of Miniature-like Evoked Postsynaptic
Currents. Neuron, 77, 516–527.
Merchan-Perez, A., & Liberman, M. C. (1996). Ultrastructural differences among
afferent synapses on cochlear hair cells: correlations with spontaneous discharge
rate. The Journal of Comparative Neurology, 371(2), 208–221.
Meyer, A. C., Frank, T., Khimich, D., Hoch, G., Riedel, D., Chapochnikov, N. M., Yarin,
Y. M., Harke, B., Hell, S. W., Egner, A., & Moser, T. (2009). Tuning of synapse
number, structure and function in the cochlea. Nature Neuroscience, 12(4), 444–
453.
Mo, Z. L., & Davis, R. L. (1997). Endogenous firing patterns of murine spiral ganglion
neurons. J Neurophysiol, 77(3), 1294–1305.
Moser, T., Neef, A., & Khimich, D. (2006, October). Mechanisms underlying the
temporal precision of sound coding at the inner hair cell ribbon synapse. Journal of
Physiology.
Nadol, J. B., Burgess, B. J., & Reisser, C. (1990). Morphometric analysis of normal
human spiral ganglion cells. Annals of Otology, Rhinology & Laryngology, 99(5),
340–348.
Negm, M. H., & Bruce, I. C. (2014). The effects of HCN and KLT ion channels on
adaptation and refractoriness in a stochastic auditory nerve model. IEEE
Transactions on Biomedical Engineering, 61(11), 2749–2759.
127
Negm, M. H., & Bruce, I. C. (2014). The effects of HCN and KLT ion channels on
adaptation and refractoriness in a stochastic auditory nerve model. IEEE
Transactions on Biomedical Engineering, 61(11), 2749–2759.
Ohn, T.-L., Rutherford, M. A., Jing, Z., Jung, S., Duque-Afonso, C. J., Hoch, G., …
Moser, T. (2016). Hair cells use active zones with different voltage dependence of
Ca2+ influx to decompose sounds into complementary neural codes. Proceedings
of the National Academy of Sciences, 113(32), E4716–E4725.
Petitpré, C., Wu, H., Sharma, A., Tokarska, A., Fontanet, P., Wang, Y., … Lallemend, F.
(n.d.). Neuronal heterogeneity and stereotyped connectivity in the auditory afferent
system.
Rall, W. (1969). Time Constants and Electrotonic Length of Membrane Cylinders and
Neurons. Biophysical Journal, 9(12), 1483–1508.
Rattay, F., Potrusil, T., Wenger, C., Wise, A. K., Glueckert, R., & Schrott-Fischer, A.
(2013). Impact of morphometry, myelinization and synaptic current strength on
spike conduction in human and cat spiral ganglion neurons. PLoS ONE, 8(11).
Relkin, E. M., & Doucet, J. R. (1997). Is loudness simply proportional to the auditory
nerve spike count? J Acoust Soc Am, 101(5), 2735.
Romand, M. R., & Romand, R. (1987). The ultrastructure of spiral ganglion cells in the
mouse. Acta Oto-Laryngologica, 104(1–2), 29–39.
Rutherford, M. A., Chapochnikov, N. M., & Moser, T. (2012). Spike encoding of
neurotransmitter release timing by spiral ganglion neurons of the cochlea. The
Journal of Neuroscience : The Official Journal of the Society for Neuroscience,
32(14), 4773–4789.
Safieddine, S., El-Amraoui, A., & Petit, C. (2012). The Auditory Hair Cell Ribbon
Synapse: From Assembly to Function. Annual Review of Neuroscience, 35(1),
509–528.
Safieddine, S., El-Amraoui, A., & Petit, C. (2012). The Auditory Hair Cell Ribbon
Synapse: From Assembly to Function. Annual Review of Neuroscience, 35(1),
509–528.
Santos-Sacchi, J. (1993). Voltage-dependent ionic conductances of type I spiral
ganglion cells from the guinea pig inner ear. Journal of Neuroscience, 13(8), 3599–
3611.
Sheets L, He XJ, Olt J, Schreck M, Petralia RS, Wang Y-X, Zhang Q, Beirl A, Nicolson
T, Marcotti W, Trapani JG, Kindt KS (2017) Enlargement of Ribbons in Zebrafish Hair
128
Cells Increases Calcium Currents But Disrupts Afferent Spontaneous Activity and
Timing of Stimulus Onset. J Neurosci 37:6299–6313

Sheets L, Kindt KS, Nicolson T (2012) Presynaptic CaV1.3 Channels Regulate Synaptic
Ribbon Size and Are Required for Synaptic Maintenance in Sensory Hair Cells. J
Neurosci 32:17273–17286

Sheets L, Trapani JG, Mo W, Obholzer N, Nicolson T (2011) Ribeye is required for
presynaptic Ca(V)1.3a channel localization and afferent innervation of sensory hair cells.
Development 138:1309–1319
Sherrill, H. E., Jean, P., Driver, E. C., Sanders, T. R., Fitzgerald, T. S., Moser, T., &
Kelley, M. W. (2019). Development/Plasticity/Repair Pou4f1 Defines a Subgroup of
Type I Spiral Ganglion Neurons and Is Necessary for Normal Inner Hair Cell
Presynaptic Ca 2 Signaling.
Sherrill, H. E., Jean, P., Driver, E. C., Sanders, T. R., Fitzgerald, T. S., Moser, T., &
Kelley, M. W. (2019). Pou4f1 Defines a Subgroup of Type I Spiral Ganglion
Neurons and Is Necessary for Normal Inner Hair Cell Presynaptic Ca 2+ Signaling .
The Journal of Neuroscience, 39(27), 5284–5298.
Shrager, P. (1993). Axonal coding of action potentials in demyelinated nerve fibers.
Brain Research, 619(1–2), 278–290.
Shrestha, B. R., Chia, C., Wu, L., Kujawa, S. G., Liberman, M. C., Goodrich
Correspondence, L. V, & Goodrich, L. V. (2018). Sensory Neuron Diversity in the
Inner Ear Is Shaped by Activity Article Sensory Neuron Diversity in the Inner Ear Is
Shaped by Activity. Cell, 174, 1229–1246.
Shrestha, B. R., Chia, C., Wu, L., Kujawa, S. G., Liberman, M. C., & Goodrich, L. V.
(2018). Sensory Neuron Diversity in the Inner Ear Is Shaped by Activity. Cell,
174(5), 1229-1246.e17.
Suli, A., Pujol, R., Cunningham, D. E., Hailey, D. W., Prendergast, A., Rubel, E. W., &
Raible, D. W. (2016). Innervation regulates synaptic ribbons in lateral line
mechanosensory hair cells. Journal of Cell Science, 129(11), 2250–2260.
Sun, S., Babola, T., Pregernig, G., Bergles, D. E., Burns, J. C., & Mü, U. (2018). Hair
Cell Mechanotransduction Regulates Spontaneous Activity and Spiral Ganglion
Subtype Specification in the Auditory System. Cell, 174.
Sun, W., & Salvi, R. J. (2009). Brain derived neurotrophic factor and neurotrophic factor
3 modulate neurotransmitter receptor expressions on developing spiral ganglion
neurons. Neuroscience, 164(4), 1854–1866.
129
Taberner, A. M., Liberman, M. C., & Shofner, W. P. (2005). Response properties of
single auditory nerve fibers in the mouse. Journal of Neurophysiology, 93(1), 557–
569.
Taylor, A. L. (2012). What we talk about when we talk about capacitance measured with
the voltage-clamp step method. Journal of Computational Neuroscience, 32(1),
167–175.
Tsuji, J., & Liberman, M. C. (1997). Intracellular labeling of auditory nerve fibers in
guinea pig: central and peripheral projections. The Journal of Comparative
Neurology, 381(2), 188–202.
Vyas, P., Wu, J. S., Jimenez, A., Glowatzki, E., & Fuchs, P. A. (2019). Characterization
of transgenic mouse lines for labeling type I and type II afferent neurons in the
cochlea. Scientific Reports, 9(1).
Waguespack, J., Salles, F. T., Kachar, B., & Ricci, A. J. (2007). Stepwise morphological
and functional maturation of mechanotransduction in rat outer hair cells. Journal of
Neuroscience, 27(50), 13890–13902.
Wichmann, C., & Moser, T. (2015). Relating structure and function of inner hair cell
ribbon synapses. Cell and Tissue Research, 361(1), 95–114.
Wong, A. B., Rutherford, M. a., Gabrielaitis, M., Pangršič, T., Göttfert, F., Frank, T., …
Moser, T. (2014). Developmental refinement of hair cell synapses tightens the
coupling of Ca2+ influx to exocytosis. EMBO Journal, 33(3), 247–264.
Wu, J. S., Young, E. D., & Glowatzki, E. (2016). Maturation of Spontaneous Firing
Properties after Hearing Onset in Rat Auditory Nerve Fibers: Spontaneous Rates,
Refractoriness, and Interfiber Correlations. The Journal of Neuroscience, 36(41),
10584–10597.
Yevgeniya Sergeyenko, Kumud Lall, Liberman MC, Kujawa SG (2013) Age-Related
Cochlear Synaptopathy: An Early-Onset Contributor to Auditory Functional Decline.
Journal of Neuroscience 21 August 2013, 33 (34) 13686-13694
Yi, E., Roux, I., & Glowatzki, E. (2010). Dendritic HCN channels shape excitatory
postsynaptic potentials at the inner hair cell afferent synapse in the mammalian
cochlea. Journal of Neurophysiology, 103(5), 2532–2543.
Yin Y, Liberman LD, Maison SF, Liberman MC (2014) Olivocochlear Innervation
Maintains the Normal Modiolar-Pillar and Habenular-Cuticular Gradients in
Cochlear Synaptic Morphology. J Assoc Res Otolaryngol 15:571–583
130
Zhang, L. I., Tan, A. Y. Y., Schreiner, C. E., & Merzenich, M. M. (2003). Topography
and synaptic shaping of direction selectivity in primary auditory cortex. Nature,
424(6945), 201–205.

Abstract (if available)

Abstract In this dissertation, my objective is to test whether the ion channel properties play a role in shaping the diverse physiology of SGN. I hypothesize that the ion channel properties of SGN are diverse and can be systematically linked to their function to encode different sound intensities. By understanding how the ion channel properties play a role in shaping the function of SGN, we can better understand how sound intensity is encoded at the auditory nerve. Furthermore, a better understanding of how SGN work may provide insight into future therapies of deafness, such as hidden-hearing loss, which is selectively degenerates low-SR SGN. ❧ In chapter 1, I provide a literature review of the concepts and previous experiments that have led me to develop my hypotheses. I provide an in-depth summary of the neuroanatomy and physiology of the cochlea, organ of Corti, and the inner hair cell active zones. In addition, I provide a summary of how studying this question can aid in the shared public health interest to learn about specific types of deafness. My goal in this chapter is to show my conceptual understanding of the literature and to provide insight for the next person who may pick up where I left off when studying this question. ❧ Chapter 2 is the focal point of the dissertation and presents my most significant contributions to the field. In this chapter, I present the results from a series of experiments where I perform simultaneous patch-clamp recordings and single cell labelling on acute semi-intact preparations of cochleae. These results link the cellular biophysics of spiral ganglion neurons to their putative intensity-coding subgroup, thus, filling our gap in knowledge in how the biophysical properties of SGN play a role in sound intensity encoding. Also, we provide an inference statistical model that predicts SGN subtypes solely on the biophysical properties of the neuron. Not only does this model provide a tool to predict SGN subtypes, the model concisely defines the relationship of how the biophysical properties are shaped among the SGN subtypes. ❧ Chapter 3 is a deeper analysis of the electrophysiological measurements presented throughout the dissertation. The bipolar morphology of SGN in the semi-intact preparation raises the possibility of inadequate space-clamp and nonisopotential behavior. It is important to carefully evaluate this possibility since non-isopotential behavior would mean that the results presented in chapter 2 could stem from unknown combination of variations in the spatial extent of voltage clamp, somatic ion channel densities as well as the size of the currents shunting through the axial conductance and adjacent dendritic compartments. The analysis shows that the SGN recorded in the semi-intact preparation largely have as isopotential compartments, and therefore, lead us to not have significant concern for our biophysical assessment. ❧ Chapter 4 expands upon the prediction models introduced in chapter 2. Here, I reduced the demand of the model to predict just one of two categories of SGN (putative high-SR or low-SR SGN) based on their biophysical properties. As a result of these loosen model constraints, I was able to use methods to validates the model’s accuracy that were not available in the initial model fitting process. The bimodal classifier prediction model presented here would benefit the field’s progress in easily separating and studying these two subgroups of SGN. ❧ Chapter 5 details my contribution of the collaborative project performed by members the Segil, Ichida, and Kalluri laboratories. The collaboration focused on providing an experimental protocol to directly reprogrammed somatic cells into cochlear hair cell-like cells. Here, I show how the electrophysiological properties of these hair cell-like cells are similar to primary hair cells. The results produced in this chapter add to a larger body of work showing that the molecular, morphological, and physiological properties of the direct reprogrammed cells match those of normal hair cells. With this study, we contribute to the field of restorative and regenerative hearing science by providing a method to produce large quantities of hair cell-like cells to test the lethality of potential ototoxins and preventative treatments. ❧ Finally, in chapter 6, I discuss the overall conclusions of the dissertation including how the biophysical properties of spiral ganglion neurons play a role in sound intensity coding. I provide a summary of the limitations and setbacks I encountered during the course of performing these experiments. Lastly, I detail future experiments I believe would be beneficial to the field that directly stem for the ideas and results provided in this dissertation.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Slit/Robo signaling underlies the spatial patterning of spiral ganglion neurons to shape the peripheral auditory circuitry assembly

PDF

Synaptic circuits for information processing along the central auditory pathway

PDF

Synaptic mechanism underlying development and function of neural circuits in rat primary auditory cortex

PDF

Spatial and temporal precision of inhibitory and excitatory neurons in the murine dorsal lateral geniculate nucleus

PDF

Contextual modulation of sensory processing via the pulvinar nucleus

Asset Metadata

Creator Markowitz, Alexander L. (author)

Core Title Physiology of the inner ear: the role of the biophysical properties of spiral ganglion neurons in encoding sound intensity information at the auditory nerve

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 04/22/2020

Defense Date 03/20/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag auditory,auditory system,auditory transduction,biophysical properties,biophysics,cellular biophysics,cochlea,machine learning,neurons,neurophysiology,Neuroscience,OAI-PMH Harvest,peripheral auditory system,postsynaptic biophysical properties,sound,sound intensity coding

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Chen, Jeannie (committee chair), Abdala, Carolina (committee member), Kalluri, Radha (committee member), Segil, Neil (committee member)

Creator Email almarkow@usc.edu,amarkowitz24@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-284757

Unique identifier UC11673411

Identifier etd-MarkowitzA-8304.pdf (filename),usctheses-c89-284757 (legacy record id)

Legacy Identifier etd-MarkowitzA-8304.pdf

Dmrecord 284757

Document Type Dissertation

Rights Markowitz, Alexander L.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA