The fine structure of the intracochlear potential field

MISSING IMAGE

Material Information

Title:
The fine structure of the intracochlear potential field
Physical Description:
vii, 155 leaves : ill. ; 29 cm.
Language:
English
Creator:
Zidanic, Michael, 1957-
Publication Date:

Subjects

Subjects / Keywords:
Cochlear Microphonic Potentials   ( mesh )
Cochlea -- ultrastructure   ( mesh )
Guinea Pigs   ( mesh )
Neuroscience thesis Ph.D   ( mesh )
Dissertations, Academic -- Neuroscience -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1989.
Bibliography:
Bibliography: leaves 147-152.
Statement of Responsibility:
by Michael Zidanic.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001136791
oclc - 24785330
notis - AFN6041
System ID:
AA00009082:00001


This item is only available as the following downloads:


Full Text










THE FINE STRUCTURE OF THE INTRACOCHLEAR POTENTIAL FIELD


by

MICHAEL ZIDANIC















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1989















ACKNOWLEDGEMENTS

There are many individuals that I have interacted with over my years as a

graduate student that I would like to thank, for without them I could not have

made this dissertation a reality.

I thank my mentor Dr. William E. Brownell who has provided guidance

and support throughout all phases of my graduate career. I also have special

thanks for Dr. Paul Manis who was responsible for developing the hardware and

software that got this study underway when I joined the lab. Special thanks also

go to Dr. George Spirou for participating in some of the early experiments in

Gainesville and to Dr. Pavel Dulguerov for helping to set up the lab after the

move to Johns Hopkins.

I thank Drs. William Brownell, Paul Manis, George Spirou, Eric Young,

and Murray Sachs of the Center for Hearing Sciences at The Johns Hopkins

University and Drs. William Luttge, Janet Zengel, John Middlebrooks, and David

Green of my supervisory committee for helpful comments on this manuscript.

I thank Ms. Minnie Ann Smith and Ms. Corrine Strathmeyer for the

histological processing of the cochleas, and I also thank Cori for her help with the

surgical preparation of the animals. Ms. Phyllis Taylor prepared the final

versions of figures 3-1 and 3-23. Figure 2-3 was prepared in the Graphic Arts

department of the University of Florida and figure 2-16 was prepared in the

Graphic Arts department at The Johns Hopkins University.

I most affectionately thank my wife and family for their love and support,

especially to help me through the final phase of my dissertation.


- ii -















TABLE OF CONTENTS

page
ACKNOWLEDGEMENTS ....................................... ii
LIST OF ABBREVIATIONS ....................................... v
A BSTRA CT ............................................... vi

CHAPTERS

I INTRODUCTION ....................................... ..

The Current Density Field ................................ 2
The Electric Field and Potential Gradients .................. 3
Conductivity Considerations ............................... 4
Cochlear Potentials--Terminology ......................... 6

2 DC POTENTIALS AND THE DC ELECTRIC FIELD ............. 8

Introduction ......................................... 8
M ethods ............................................ 10
Animal Preparation ................................... 10
Electric Recording .................................... 12
M icropipet Positioning ................................. 13
Data Analysis--DC Potential Profiles ..................... 16
Data Analysis--DC Potential Gradient Profiles .............. 18
Translation and Averaging of DC Gradient Profiles .......... 19
R results . . . . . 2 1
Scala Tympani DC Profiles ............................. 21
Scala Tympani DC Gradient Profiles ..................... 22
Scala Vestibuli DC Profiles ............................ 24
Scala Vestibuli DC Gradient Profiles ..................... 24
D discussion ............................ ............... 26
Radial DC Gradients in Scala Tympani and Scala Vestibuli ..... 26
Standing Currents through Hair Cells ..................... 29
Variability of DC Gradients across Experiments ............. 30
A Current Density Model for Standing Currents ............. 32
Estimate of Current Output of Marginal Cells .............. 33
(Na+ + K+)-ATPase Density to Support the Standing Current .... 34

3 EVOKED POTENTIAL WAVEFORMS AND
COCHLEAR MICROPHONICS ............................. 53

Introduction ......................................... 53
M ethods ............................................ 56
Animal Preparation ................................... 56
Sound Stimulation .................................... 57


- iii -








Microphone Measurements .............................. 58
Two-phase Lock-in Analysis ............................ 59
Probe Tube Calibration .............................. 61
In Situ Acoustic Calibration ............................ 63
Electric Recording .................................... 65
Signal Averaging ..................................... 67
R results ............................... .............. 69
Scala Media Evoked Potential Waveforms .................. 69
CM and Evoked Potential Waveforms in Scala Vestibuli ........ 70
CM and Evoked Potential Waveforms in Scala Tympani ........ 74
CM Input-Output Functions ............................ 79
CM Frequency Response Curves ........................ 81
Modulation of the Standing Current ...................... 82
Discussion .............. ............. ................. 84
Comparison with Previous CM Studies .................... 84
Relation of Microphonic Current to Standing Current .......... 86
Reciprocal Modulation of Ionic Currents .................. 90
Comparison with Previous CAP Recordings ................ 91
Capacitative Currents through the Walls of Scala Media ........ 93
Longitudinal Currents near the Peak of the Traveling Wave? .... 95

4 SUMMARY AND CONCLUSIONS ........................... 140
Current Density Model of Intracochlear Currents ............ 142
Future Experiments ................................. 143
Perspective ........................................ 145

REFERENCES .............................................. 147

BIOGRAPHICAL SKETCH .................................... 153


- iv -















LIST OF ABBREVIATIONS



CAP compound action potential

CF characteristic frequency

CM cochlear microphonic

EP endocochlear potential

SPL sound pressure level (re 20 Pa)















Abstract of Thesis Submitted to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE FINE STRUCTURE OF THE INTRACOCHLEAR POTENTIAL FIELD

by

Michael Zidanic

May, 1989


Chairman: Dr. Janet Zengel
Major Department: Neuroscience

Tone-evoked and DC potentials were recorded at regular intervals along

radial tracks in scala tympani and scala vestibuli of the second turn of the guinea

pig cochlea. The radial component of the electric field along the track was

computed by taking the first spatial derivative of the field potentials. A radial

electric field in scala tympani and scala vestibuli drives a standing current away

from the modiolus in each chamber and into the spiral ligament. The DC electric

field strength reaches a maximum within or near the spiral ligament.

A model for understanding spatial variations in magnitude and phase of the

cochlear microphonic (CM) is presented that is based on the continuity of the

potential field from scala vestibuli through the spiral ligament and into scala

tympani. Within this model, the classic description of a 1800 phase shift of the

CM across the reticular lamina and between scala tympani and scala vestibuli

requires another 1800 phase shift to occur somewhere along the lateral wall

between scala tympani and scala vestibuli. Below 500 Hz, abrupt, near 1800

phase shifts are recorded along electrode tracks that begin outside the spiral

ligament and end deep in scala tympani. Low-frequency microphonics in scala


- vi -









vestibuli are in phase with the CM on the lateral side of the scala tympani phase

shift but are much larger. Continuity of the potential field is maintained by a

gradual CM decrease from scala vestibuli to the lateral wall bordering scala

tympani, reaching a minimum at the phase shift, and a subsequent CM increase

modiolar to this "virtual ground point".

At low intensities a compound action potential (CAP) is superimposed on

the phase of CM associated with displacement of the basilar membrane towards

scala vestibuli. At high intensities, a second CAP is also superimposed on the

CM about 180o out of phase with the low intensity CAP. The CAPs do not

change appreciably in amplitude, polarity, latency or duration as a function of

electrode depth along radial tracks in scala tympani or scala vestibuli, although

the CAPs adapt with successive stimulus cycles.


- vii -















CHAPTER 1

INTRODUCTION

A bi-directional dynamic conversion of mechanical and electrical energy in

the cochlea has been postulated to be responsible for otoacoustic emissions and the

high sensitivity and frequency selectivity of the mammalian inner ear. In the

aftermath of the initial discovery of stimulated acoustic emissions by Kemp (1978),

a candidate for an electro-mechanical generator was soon found in the outer hair

cells. At first, the evidence was indirect: stimulation of the efferents to the outer

hair cells altered distortion products in the ear canal (Mountain, 1980; Siegel and

Kim, 1982). The outer hair cells were more directly implicated with the

discovery of their motile properties in vitro (Brownell, 1983; Brownell et al., 1985;

Zenner et al., 1985). In light of the great significance that outer hair cell motility

has with respect to its presumed cochlear amplifier action (Davis, 1983), reliable

methods are needed to assess outer hair cell function in vivo.

The standing current through the outer hair cell is a feature of

fundamental importance, because the stereocilia-mediated modulation of current

during acoustic stimulation (Davis, 1953; Hudspeth and Jacobs, 1979) should

directly, or indirectly, change the mechanical properties of outer hair cells. We

have previously applied a current density analysis technique to characterize

standing currents and their modulation by click stimuli in scala tympani of the

third turn (Brownell et al., 1983, 1986). Currents were measured by recording

field potentials at finely spaced intervals and computing first spatial derivatives.

We showed that hair cell currents spread into the extracellular spaces of the

cochlea and modulate the local potential field.


-1-





-2-


In the present study, we explore the DC potential field in scala tympani

and scala vestibuli (chapter 2). We also explore the low-frequency modulation of

the intracochlear potential field in the cochlear chambers of the second turn

(chapter 3). We demonstrate that a standing current is directed into the spiral

ligament from scala tympani, a result predicted by Davis (1953) over 35 years ago.

In addition, we show that an equally strong standing current is directed into the

spiral ligament from scala vestibuli, indicating that a significant leakage pathway

for current out of scala media is through Reissner's membrane. During acoustic

stimulation we show that the standing currents are modulated in opposite

directions in scala tympani and scala vestibuli. The reciprocity of extracellular

currents demonstrates that the hair cells indirectly control the modulation of

leakage currents through the scala-vestibuli pathway. We calculate that the

modulation of leakage current through Reissner's membrane may occur mainly by

capacitative effects in the mid-frequency range (above 600 Hz). We argue that

longitudinal currents may be present in scala tympani near best frequency that

should reveal important features of the cochlear traveling wave.



The Current Density Field

The concepts of current, voltage, and resistance as they relate to electric

circuits are familiar ones. Ohm's law describes the relation among these three

quantities, V=IR. While this formulation of Ohm's law is sufficient for modeling

resistors in electric circuits, an extension of the theory is necessary to relate the

analogous concepts of current density, electric potential, and conductivity in the

three-dimensional environment of a biological system.

The steady currents of the stria vascularis into scala media, receptor

currents of hair cells, dendritic and action currents of the eighth nerve, and

leakage currents through the supporting cells and occluding junctions of the






-3-


endolymph/perilymph barrier represent an ensemble of current sources and sinks

with respect to the extracellular spaces of the cochlea. Passive cellular and

acellular elements within the cochlea, such as bone, spiral ligament, basilar and

tectorial membranes, constitute a resistive matrix to extracellular current relative

to the conductive endolymphatic and perilymphatic fluids. The geometric relations

between the resistive elements and the current sources and sinks determine the

direction and density of current through the extracellular spaces. This flow of

current in a conductive medium can be described by the current density vector

field. In general, a vector field can be thought of as an infinite collection of

vectors, each of which is based at a different point in space. In this application,

the current density vector based at a given point is aimed in the direction of ion

movement and has a length proportional to the magnitude of the current density at

that point.

Current density cannot be measured directly with the instrumentation that

is available to the electrophysiologist today. However, the current density field is

intimately related to the potential gradient field, which may be determined by

sampling the potential field 0 at selected points in space with a microelectrode.

More precisely, the potential gradient Vo(x) at a point x set up by a current

density J(x) is related to the conductivity tensor g(x) by Ohm's law,

J(x) = -g(x)Vo(x). The problem is now reduced to measuring the potential

gradient and the conductivity tensor.



The Electric Field and Potential Gradients

In a three-dimensional Cartesian coordinate system, the electric field E at a






-4-


point x is related to the potential field by



E(x) = -Vo(x) = Xir a(x-iv 0- it
ar r av a-
Er(x)ir + E,(x)i, + E1(x)il (1-1)



where Er, EV, and El are the magnitudes of the electric intensity, and ir, iv, and

il are the unit vectors in the radial, vertical, and longitudinal axes respectively.

The three partial derivatives, ao, (o, and a ,(x form the radial, vertical,
ar av al
and longitudinal components, respectively, of the potential gradient vector Vq(x)

and represent the rate of change of potential with respect to space along each of

the three principal axes. When these three components are vectorially combined,

a potential gradient vector Vq(x) is constructed that points in the direction that

the potential field is maximally increasing. Note that the potential gradient field

V0 and the electric field E can be expressed in the same units, mV/mm, but that

they differ in sign. This is a consequence of the convention that the direction of

current in a conductive medium is taken relative to the movement of positive

charge. Just as electric current sets up a voltage drop across a resistor, current

density in a conductive medium sets up a potential gradient along its direction of

flow. In both cases, potential (or voltage) decreases in the direction of current

flow. This convention thus requires that the electric field and the potential

gradient field be oriented in opposite directions.



Conductivity Considerations

We now turn to a discussion of the conductivity tensor in relation to

Ohm's law, J gE. If the conductivity of the medium is the same everywhere,

and the same in all directions, then current density is directly proportional to the

electric field. However, this is not true in general. Conductivity of nervous





-5-


tissue, in particular, may be different in each of the three dimensions due to the

organization of cells and their processes. Such a medium is referred to as

anisotropic. In addition, conductivity may vary along one or all of the principal

axes, such as might be expected in the radial dimension of the cochlea. In this

case the conductivity of perilymph in scala tympani is likely higher than the

connective tissue of the spiral ligament. A medium with conductivity gradients

along at least one of its principal axes is said to be inhomogeneous. In order to

compute the current density at a point x in an inhomogeneous or anisotropic

medium, Ohm's law is expanded in Cartesian coordinates:



J(x) = g,(x)Er(x)i, + g,(x)E,(x)i, + gl(x)El(x)it

= Jr,(x)ir + Jv(x)i, + Jl(x)it (1-2)



where g, gv, and gl refer to the conductivity and J,, J,, and J, refer to the

magnitude of the current density in the radial, vertical, and longitudinal

directions, respectively.

The conductivity of perilymph can be assumed to be isotropic with a value

in the range 1.4-2.0 mS/mm based on Bdkisy's (1951a) measurements and the

calculations of Misrahy et al. (1958). Precise measurements of the conductivity

tensor of the spiral ligament are not available. An impedance of 1.3 kohm across

the scala-tympani portion of the spiral ligament in the basal turn was measured

by Cannon (1976). Assumptions of the spatial distribution of current density set

up by a current injection across the spiral ligament must be made in order to

estimate a specific resistance of the spiral ligament from these measurements.

Until the conductivity tensor of the spiral ligament is measured, the results of this

type of study are best described in terms of potential gradients.





- 6 -


Cochlear Potentials--Terminology

The potential field 0 within the fluid spaces of the cochlea can be

represented as the sum of several component potential fields. In the absence of

sound stimuli, the DC potential field that exists will be denoted by #dc. During

the presentation of an acoustic stimulus, the DC potential field is modulated. In

particular, tone stimuli modulate the potential field at the stimulating frequency,

generating a potential that is known as the cochlear microphonic (CM) and will be

denoted by Ocm. In addition to the CM, a tone stimulus produces a baseline DC

shift for the duration of the stimulus called the summating potential, denoted by

0-P. Recordings from the perilymphatic spaces of the cochlea can also pick up

neural potentials, denoted by Oap. At low frequencies, these neural responses can

interfere with the microphonic response, producing a nonsinusoidal evoked

potential. However, high frequencies elicit a compound action potential (CAP)

transient that lasts only for the first few milliseconds of the tone burst. While

intracellular excitatory post-synaptic potentials have recently been reported (Siegel

and Dallos, 1986), extracellular potentials related to synaptic currents have not

been observed in the mammalian cochlea. Thus one may write



O(x) a Odc(x) + d cm(x) + 0sP(x) + aP(x) (1-3)



By substituting this expression for O(x) back into equation 1-1, we obtain a

similar formula for the decomposition of the electric field into corresponding

component fields:


E(x) a Edc(x) + Ecmn(x) + ESP(x) + Eap(x) .


(1-4)






-7-


Results are presented in chapter 2 that describe pdc along radial tracks in

scala tympani and scala vestibuli of the second and third turns of the guinea pig

cochlea. The first spatial derivative of the DC profile is calculated to obtain the

radial component of the DC electric field, Edc .

In chapter 3 results are presented that describe tone-evoked field potentials

along radial tracks in scala tympani and scala vestibuli of the second turn. These

field potentials contain a mixture of CM (Ocm), summating potentials (0sP), and

CAPs (OaP). Discrete Fourier transforms of the field potentials are calculated to

extract the fundamental component of the response waveform and to analyze the

spatial variations of Ocm in greater detail.















CHAPTER 2

DC POTENTIAL PROFILES AND THE RADIAL DC ELECTRIC FIELD



Introduction



The study of cochlear electrophysiology began in 1930 with Wever and

Bray's discovery of the cochlear microphonic (CM), a cochlear electrical potential

that can reproduce the waveform of a sound stimulus. Twenty years later,

Bekesy (1950) demonstrated that the electrical energy represented by the CM could

be greater than the mechanical energy in the applied stimulus, and in 1952 he

identified a source of electrical energy in the large endolymphatic DC potential

(EP). Based on Bekesy's discoveries, Davis (1953) established a working hypothesis

that the generator sites of the EP and the CM were distinct. Included in this

theory was a standing current generated by the stria vascularis. This current was

assumed to exit scala media through a mechanically-dependent variable resistance

associated with the stereocilia of the hair cells and to flow radially across scala

tympani, through the spiral ligament, and back to the stria vascularis, completing

a local circuit.

The main features of the Davis model are supported by a variety of

experiments. The unequivocal demonstration that the EP is generated by the stria

vascularis came in 1959. Tasaki and Spyropoulos (1959) showed that the EP is

present in "waltzing guinea pigs", a special breed in which the organ of Corti

rapidly degenerates after birth. Davis et al. (1958) showed a similar result after

administration of streptomycin, one of the many aminoglycosides that have toxic


-8-






-9-


effects on hair cells. Tasaki and Spyropoulos (1959) also drained the endolymph

and perilymph in the apical turn and found that a positive DC potential was only

present on the endolymphatic surface of the stria vascularis.

Mechanically-dependent resistance changes of stereocilia were first

confirmed in bullfrog saccular hair cells (Corey and Hudspeth, 1983; Hudspeth

and Corey, 1977) and more recently in chick cochlear hair cells (Ohmori, 1984),

turtle cochlear hair cells (Art et al.,1986), and cochlear hair cells of mouse

otocysts (Russell et al.,1986). Acoustically-evoked resistance changes have also

been measured in vivo from scala media of the guinea pig (Geisler et al., 1977;

Mountain et al., 1980) and from guinea pig cochlear inner hair cells (Nuttall,

1985; Russell, 1983).

Evidence in support of a standing current through the inner hair cells

comes from recordings of auditory nerve spontaneous activity. When the EP is

lowered after administering the loop diuretic furosemide, the spontaneous activity

of auditory nerve fibers concomitantly decreases (Sewell, 1984), consistent with an

EP-dependent depolarizing standing current through inner hair cells. Equally

convincing are the data of Liberman and Dodds (1984) who exposed animals to

loud sounds that were sufficient to cause threshold shifts. Auditory nerve fibers

were isolated, characterized, and labeled with horseradish-peroxidase. Using a

surface preparation technique, the fibers were traced back to the inner hair cells

that they innervated and the number of stereocilia on the innervated inner hair

cell were counted. Spontaneous activity was correlated with the number of

stereocilia that remained, suggesting that each stereocilium contributes a small

depolarizing current to the inner hair cell.

Evidence for a standing current through the outer hair cells comes from

crossed olivocochlear bundle stimulation experiments. When the crossed

olivocochlear bundle efferents to the outer hair cells are stimulated with a train






- 10 -


of shocks, the EP decreases (Fex, 1967; Teas et al., 1970). This result is

consistent with a standing current through the outer hair cells in silence that is

increased by an efferent-induced conductance increase of the basolateral

membrane of the outer hair cell (Geisler, 1974). The finding that the CM to low-

frequency stimuli increases with efferent stimulation (Fex, 1959; Teas et al., 1970)

also supports the Geisler model.

The only direct attempt to demonstrate the presence of standing currents

was a study by Tasaki et al. (1954) that was unsuccessful. No further attempts

have been made to directly measure standing currents despite many experiments

that support Davis' model of transduction (see Brownell et al., 1986 for a review).

Using a current density analysis technique that was used to demonstrate the

existence of a dark current in the vertebrate retina (Hagins et al., 1970; Penn and

Hagins, 1969), we were able to demonstrate transient click-evoked radial current

densities in scala tympani of the guinea pig cochlea (Brownell et al., 1983).

Current density results are presented in this chapter that represent the first direct

measurements of radial standing currents in the cochlea. Preliminary versions of

these results have been presented in abstract form (Zidanic et al., 1984, 1985,

1986, 1987b).




Methods



Animal Preparation

For this study, 101 healthy guinea pigs weighing 200-500 g were

anesthetized with initial doses of 18 mg/kg sodium pentobarbital and 0.25 ml/kg

Innovar-Vet (fentanyl, 0.1 mg/kg; droperidol 5 mg/kg). Anesthesia was

maintained by administering a sustaining dose each hour, alternating between

pentobarbital and Innovar at one-third the induction doses. All of the surgical






- 11 -


procedures and electrophysiological recordings were conducted in a sound isolation

chamber (Industrial Acoustics Corp., model 1202) heated to 32-350C to prevent

hypothermia of the animal and of the exposed cochlea. A mid-sagittal keyhole-

shaped opening was made in the skull with a #10 blade and ronguers, and a

screw inserted and cemented in place. A metal bar was secured to the screw and

the other end inserted into a clamp that fixed the preparation to the recording

table (Newport Corp., model TS-23). Electrical isolation of the stabilization

apparatus was achieved by inserting a polyethylene tube between the head bar

and clamp. The external auditory meatus was dissected free and the cochlea

exposed by a ventral approach. The tensor tympani and stapedius muscles were

severed to eliminate the possibility of artifacts due to middle ear muscle

contractions. The animal was then paralyzed with Flaxedil (gallamine triethiodide,

5 mg/kg) and artificially respired. Paralysis not only insured that the preparation

remained mechanically stable, but also reduced electrical artifacts related to

spontaneous muscular contractions. The animal was grounded with a silver-silver

chloride electrode placed through the bucca, a location that minimized the EKG

artifact on the recording electrodes.

The physiological state of the cochlea was monitored by measuring tone-

evoked compound action potential (CAP) thresholds from a copper wire placed on

the bone near the round window. Sound pressure in the external auditory meatus

was monitored through a calibrated probe tube microphone (described in detail in

chapter 3). Figure 2-1 shows the results of this initial CAP determination in a

sample of 28 guinea pigs. The average threshold in the frequency range

500-5000 Hz was 40-45 dB SPL (sound pressure level re 20 uPa), although the

most sensitive animals could have thresholds in the 30-35 dB SPL range. The

guinea-pig CAP thresholds reported here are within the range that have been

measured by other investigators (Johnstone et al., 1979; Robertson et al., 1980).






- 12 -


The most sensitive auditory neurons for a particular characteristic frequency (CF)

have thresholds 20 dB lower than the CAP threshold determined for the same

frequency (Johnstone et al.. 1979).

In several experiments, CAP thresholds were first measured through a

small opening in the caudo-lateral part of the labyrinth bulla. After opening the

bulla widely and performing the middle ear surgery, CAP thresholds were

remeasured. These procedures had no significant effect on CAP thresholds below

5 kHz. although thresholds could be raised by 10-20 dB at frequencies higher than

7.5 kHz. A study by Brown et al. (1983b) has also shown that opening the bulla

can result in higher thresholds at frequencies above 16 kHz. but that these

procedures have little or no effect on thresholds below 12 kHz.

A fenestra of 40-100 pm in diameter was then made in the otic capsule to

gain access to scala tympani or scala vestibuli of the second or third turn. The

fenestra was made by alternately hand-drilling with a four-sided awl and a fine-

tipped pick. Bone chips were cleared with a fine cotton wick as the hole was

drilled deeper. The exposure was enlarged until a opening to the spiral ligament

was visualized. Threshold shifts following this type of exposure are shown in

figure 2-2 for a population of 19 guinea pigs. Third-turn exposures (15.5 mm

from the base, CF--0.8-1.0 kHz) could generally be made without losing more than

5 dB, whereas second-turn exposures (11.5.5 mm from the base) usually resulted

in threshold shifts of 5-10 dB at frequencies near CF (2.5-4.5 kHz).



Electric Recording

Micropipets were fabricated with a microprocessor-controlled pipet puller

(Mecanex, BB-CH) that was programmed to pull pipets with shanks of 500-800 pm

that were 1-2 pm at the tip, and widened to no more than 30 pm along the length

of the shank. Tip potentials and ion diffusion were minimized by filling the






- 13 -


pipets with a solution that resembles perilymph (155 mM NaCl. 4 mM KCI).

Connection to the recording preamplifiers was made with pipet holders equipped

with Ag-AgCl pellets (World Precision Instruments, model MEH-lS). The

electrodes were filled with electrolyte at least one hour prior to recording to allow

the metal-liquid junction potential to stabilize. Custom-built, high-impedance,

low-noise, low-bias current, DC-coupled preamplifiers (Analog Devices ADS15 or

AD549 were used as the input operational amplifier) with capacitance

compensation and DC offset control provided low-pass (10 kHz cutoff) signals at a

gain of 1000 for oscilloscope viewing and 12-bit A/D conversion (Data

Translation, model DT3388). This gain level provided a reasonable compromise

between a dynamic range of 10 mV to allow for DC drift and a minimum

resolution of 5 uV to record low-level cochlear microphonics (CM).



Micropipet Positioning

Pipets were advanced and withdrawn with a piezoelectric microdrive

system (Burleigh Instruments, model PZ-555) equipped with an optical measuring

head that signaled spindle position with 2 pm accuracy. The pipet holder was

inserted into an insulated adapter that threaded into the end of the microdrive

spindle, ensuring that the electrode's electrical signal was isolated from the

piezoelectric driver voltages. The microdrive was threaded into a custom-built

housing that was bolted to a micromanipulator (Narishigi, model MM-3). A 1 "

vibration-damped rod (Newport Corp., model 345) was bolted to the recording

table, providing a stable mount for the micromanipulator. This

microdrive/micromanipulator arrangement permitted manipulator movements in the

x, y, and z axes, and microdrive movements in the z direction, but only very

limited rotation about a single axis. The method thus chosen for orienting the

electrode track relative to the cochlea was to first orient the cochlea by rotating






- 14 -


the head holder and clamp assembly while viewing through the dissecting

microscope along a vertical line of sight. The dissecting microscope was then

moved away to view at an angle while the horizontal controls on the

micromanipulator were used to position the electrode directly over the cochlea.

The final placement of the tip of the micropipet in the fluid over the exposure

was made under microdrive control.

The desired orientation of the pipet was always along a radial path from

the lateral wall towards the modiolar axis of the cochlea. The angle of the track

with respect to the modiolar axis depended on the cochlear chamber to be studied.

In scala-tympani experiments, the optimal orientation was along a path parallel to

the basilar membrane, whereas in scala-vestibuli experiments, the apex of the

cochlea was tilted up about 300 in order to achieve a track parallel to Reissner's

membrane. These orientations define the radial component of a local Cartesian

coordinate system in scala tympani and in scala vestibuli (labeled i, in

figure 2-3). The vertical component, iv, is oriented orthogonally to the radial

component and points 'towards the apex of the cochlea. Once the desired

orientation of the cochlea was set, the pipet was placed in the fluid over the

exposure by using the horizontal controls on the micromanipulator.

In 12 experiments the cochlea was removed from the animal after the

experiment was terminated. The specimen was then fixed, dehydrated, embedded

in celloidin, and sectioned approximately parallel to the electrode track. The

example shown in figure 2-3 is traced from the only case where the electrode

track through the spiral ligament was recovered. In this third-turn scala-tympani

experiment the electrode track was at an angle of about 300 from the basilar

membrane. Since this type of reconstruction could not be done for all

experiments, the data from different experiments were not corrected for any

deviations from the defined radial direction (i.e., parallel to the basilar membrane

or parallel to Reissner's membrane).






- 15 -


Also illustrated in figure 2-3 is a fixed pipet (labeled the spiral-ligament

(SL) electrode) which was placed in the drop of fluid that formed over the otic-

capsule fenestration. The formation of the drop results from hydrostatic pressure

within the cochlea that forces perilymph out of the exposure. Eventually,

cerebrospinal fluid replaces the leaking perilymph. The spiral-ligament pipet was

broken back several hundred microns to a tip size of about 10 pm to achieve a

lower impedance and noise level. A second micromanipulator (Narishigi,

model MM-3, mounted on another rod bolted to the recording table) was used to

position the fixed electrode. The fixed electrode served both as a local DC

reference and also as a monitor of the CM.

The depth of the fluid layer from the air-perilymph interface to the

surface of the spiral ligament could vary between 100-200 pm depending on the

shape and location of the exposure. Since small fluctuations of the depth of the

fluid layer are expected, the movable pipet was advanced about 40 pm from the

surface before beginning the 500-700 pm recording track to help ensure that the

pipet would remain in the fluid upon withdrawal to the starting point. This

starting point within 40 pm of the air-perilymph interface is assigned a value of

zero and all depths modiolar to this point are assigned positive values. Since it

was noted in early experiments that the pipet usually bent while penetrating the

spiral ligament, the protocol for the later experiments was to manually advance

the electrode, and then to collect data as the electrode was withdrawn in 20 pm

steps under computer control.

Occasionally, the pipet lost electrical contact with the preparation

100-200 pm before reaching the starting point of the track. Loss of contact

usually meant that the pipet had broken (some pieces of which have appeared in

histological material). Electric recording was also aborted when large positive DC

shifts (5-70 mV) or large negative DC shifts (30-80 mV) were encountered over a






- 16 -


limited distance on some scala-tympani tracks that probably came close to or

penetrated scala media. The magnitude of the large DC shifts varied on

successive excursions along the same track, although the positive DC shifts were

usually stable when the electrode was stopped along the track. The negative DC

shifts were very unstable and were most likely intracellular potentials associated

with hair cells or supporting cells in the organ of Corti. In order to avoid such

large DC shifts, the electrode was withdrawn and repositioned towards the base

(if scala tympani) or apex (if scala vestibuli), and in some cases it was also

necessary to enlarge the exposure.



Data Analysis--DC Potential Profiles

The DC data reported in this chapter were collected along the same

electrode excursions as the CM data reported in chapter 3. At each stopping

point along the electrode track one or more acoustic stimuli were presented to the

animal. The evoked responses picked up by the fixed and movable electrodes

were simultaneously averaged (10-25 repetitions). The acoustic stimulus was

delayed by 2 msec relative to the initiation of A/D conversion in order to record

both the baseline DC potential and evoked response. The A/D values were

averaged over this 2-msec interval to compute a DC baseline for each electrode

and stimulus combination. If more than one acoustic stimulus was presented, then

the DC potential for each electrode was computed by averaging the DC baselines

preceding the evoked responses. In the equations that follow, the averaged DC

value on the movable electrode at depth xi is symbolized DC,m(xi), and the

averaged DC value on the fixed electrode when the movable electrode was at

depth xi, DCfx(xi).

Drift in the junction potential of the ground electrode was eliminated by

subtracting the DC potential recorded by the fixed electrode from the DC potential






- 17 -


recorded by the movable electrode as follows:



DCm2-fx(xi) = DCmv(xj) DCfx(xj) (2-1)



where x1=O pm, x2=20 pm, .... xd=20(d-1) pm. The results of this operation are

illustrated in figure 2-4. Multiple DC profiles are plotted from sequential

excursions along the same track. The profiles superimpose to a significantly

greater degree when the fixed electrode is used as the DC reference (bottom)

instead of the ground electrode (top).

In addition to drift in the junction potential of the ground electrode, there

was also long-term DC drift of the potential between the fixed and movable

electrodes. It was necessary to eliminate this drift by subtracting the DC

potential difference between the two electrodes when the movable electrode was at

the starting point of the track (i.e., when the two electrodes were closest to each

other) from the DC potential differences recorded along the rest of the track as

follows:



odc(x,) DCmx.(xi) DCmv-fx(X) (2-2)



for i=1,2,...,d. This operation effectively defines a DC reference point at the end

of each withdrawal track, a location within 40 pm of the air-perilymph interface.

The DC potential is assigned a value of zero at this location, and all DC

potentials modiolar to this location are shifted by an amount equal to the

deviation of this first point from zero. Unlike the DC-referencing operation

performed in equation 2-1, this operation has no effect on the DC potential

gradient calculations.






- 18 -


Data Analysis--DC Potential Gradient Profiles

In a three-dimensional Cartesian coordinate system, the DC electric field Eac

at a point x is related to the DC potential field edc by



Edc(x) = -Vdc (x) = dr Oc(v (xj

Edc(x)ir + Edc(x)i, + E(x)it (2-3)



where Erfc, Efc, and Efc are the magnitudes of the DC electric field and ir. iv,

and il are the unit vectors in the radial, vertical, and longitudinal axes
aodc(X) aodc(X) aodc (X)
respectively. The three partial derivatives, a and form

the radial, vertical, and longitudinal components, respectively, of the potential

gradient vector VodC(x) and represent the rate of change of potential with respect

to space along each of the three principal axes. When these three components are

vectorially combined, a potential gradient vector Vodc(x) is constructed that points

in the direction that the potential field is maximally increasing.

Before computing the first spatial derivative, the DC potentials are digitally

filtered using the following weighting function:



.dc(xi) = 20dC(x~ ,) + 3 ,dc(x,) + 2dc(x) (2-4)



for i=2..... d-1. Raw data points, not smoothed data, are plotted in the figures in

the results section. The radial component of the DC electric field Edc is then

computed from the smoothed potentials as a finite difference:



Edc(x1) = 1dc(xi_,) dc(x,+) (2-5)
xi+l xi-I


for i=3,...,d-2. This combined smoothing and finite difference algorithm has





- 19 -


previously been referred to as D3 smoothing (Brownell et al., 1983) of the first

spatial derivative following the nomenclature introduced for smoothing formulas of

the second spatial derivative (Freeman and Nicholson, 1975).

Since the origin of the local coordinate system in the cochlea is taken to be

the air-perilymph interface at the site of the otic-capsule fenestra and the positive

radial direction is toward the modiolus, it follows that the radial unit vector of

the electric field is oriented towards the modiolus. Thus, a positive electric field

is a driving force for current towards the modiolus, and a negative electric field

drives current toward the spiral ligament. The terms potential gradient and

current are used interchangably with electric field when describing and discussing

the results of this study. The radial component of the DC current density field

at a point x, J C(x), is related to the DC electric field by Ohm's law,


JdC(x) gdC(x)EdC(x), (2-6)



where gcr(x) is the resistive component of the conductivity tensor in the radial

direction at the point x. If there is a conductivity gradient along the radial track,

i.e., ag (x) 0 0, then the shape of the radial current density profile will be
ar
different from the shape of the radial DC electric field profile. Although

conductivity gradients are not expected within the fluid spaces of the cochlea, it

is likely that the conductivity of the spiral ligament tissue is less than that of

perilymph.



Translation and Averaging of DC Gradient Profiles

In order to compare DC gradient profiles across experiments, it was first

necessary to reduce all of the DC gradient profiles collected during an experiment

to a single representative profile. This was easily accomplished in some






- 20 -


experiments where the DC profiles were quite reproducible (e.g., experiment GP89

shown in figure 2-4, panel B). In these cases, the DC profiles were averaged to

come up with a mean DC profile. However, this degree of reproducibility should

be expected only when electrode excursions along a track are done in rapid

succession, starting and stopping at the same locations in the cochlea. This was

not always possible to achieve. Reproducibility could sometimes be compromised

by miscounting one or two of the 20 Am steps when the electrode was manually

advanced. Changes in the relative position of the movable electrode and the

cochlea could also occur if the animal was accidentally moved while given an

injection or if a broken pipet was replaced.

To compare DC potentials that were collected over the course of several

hours, some DC profiles were translated along the radial axis (i.e., the abcissa) by

a multiple of 20 Am to compensate for presumed changes in the position of the

electrode with respect to the cochlea. This translation procedure is illustrated in

figure 2-5. Four DC potential profiles that were collected over a three-hour

period in a second-turn scala-tympani experiment are shown in panel A

(uncorrected depth). Potential gradients were calculated from the profiles of

panel A and are shown below in panel C. The peak of the DC gradient varies

from depth 60 (circles) to depth 140 (triangles). The average of these DC

gradients would result in a peak wider that any of the individual profiles.

However, when the data are first shifted by a multiple of 20 Am to make the DC

gradient peaks coincide (panels B and D), the profiles are reproducible in the

region of the peak. Standard deviations were calculated for these two sets of

data, shifted versus unshifted. In the region of the peak (0-200 Am), the average

of the shifted DC gradient has a standard deviation of less than 1 mV/mm

compared to 2-3 mV/mm for the unshifted data. Outside the peak region in the

200-600 Am range, the standard deviation is about 1 mV/mm for both the shifted






- 21 -


and unshifted data, indicating that the translation operation makes the average no

worse in this region.

With a recording time of one to three hours, it was possible to perform

5-15 complete excursions along the same electrode track. While all of the DC

profiles recorded in a given experiment had the same general shape, there were

usually a few profiles that either had unusual DC shifts or a much larger or

smaller DC shift than the majority of the other profiles. Two examples of

unusual DC shifts are shown in figure 2-6: (1) the profile represented by circles

in the GPA9 plot in the 440-460 Am depth range and (2) the profile represented

by squares in the GPB2 experiment in the 200-280 Apm depth range. Profiles that

had abrupt DC shifts larger than 200 AV were not included in the DC gradient

analysis. Two to six of the remaining profiles that fell within a median range

after the translation operation were then chosen for display and averaging.




Results



Scala Tympani DC Profiles

The scala-tympani DC profiles presented in this chapter are based on the

analysis of DC potentials collected along 140 radial electrode excursions in 16

guinea-pig cochleas. In 14 of these cochleas a single hole was made in the otic

capsule in either the second or third turn. In the two other experiments (GP83

and GP85) a hole was made over scala tympani of both the second and third

turns. Four of the second-turn experiments are shown in figure 2-6 and four of

the third-turn experiments are shown in figure 2-7. Each graph in these two

figures shows several profiles recorded along the same track. Four experiments

from each cochlear chamber were chosen to illustrate the range of scala-tympani

DC profiles observed. Experiment GP94 represents the most common type of






- 22 -


profile in the second turn. In this experiment the DC potential began to increase

within 100 pm of the air-perilymph interface and then continued to rise over a

200-300 pm distance to a value of over 1 mV. In the second turn this DC

plateau value ranged from 1.0-1.6 mV (mean 1.330.2 mV, n-9). Two of the

experiments shown, GP89 and GPB2, were unusual because the DC continued to

increase over the 600 pm length of the track. Experiment GPA9 was the only

experiment in scala tympani of the second turn where the DC potential decreased

in the 300-500 pm depth range.

Radial DC profiles recorded in scala tympani of the third turn were

similar to those obtained in the second turn, although there was much more

variation in the plateau DC value: range 0.2-2.4 mV, mean 1.430.7 mV, n=8. In

the remaining third-turn scala-tympani experiment that was not included in these

calculations, the DC potential shifted negatively by 1 mV as the electrode was

advanced into scala tympani.



Scala Tympani DC Gradient Profiles

For each experiment shown in figures 2-6 and 2-7, the DC profiles were

averaged and the radial gradient of the mean DC profile is plotted in either

figure 2-8 (second-turn data) or figure 2-9 (third-turn data). Averaged radial DC

gradients are also shown for five other second-turn scala-tympani experiments in

figure 2-8 and for three other third-turn experiments in figure 2-9. These DC

gradient profiles were grouped together for plotting on the basis of the magnitude

of the DC gradient peak. In all experiments shown, the DC gradient had a major

negative peak (implying that current is directed away from the modiolus) about

100-200 pm from the air-perilymph interface. The magnitude of the peak varied

from 3.5-12 mV/mm in the second turn and from 1-18 mV/mm in the third turn

and usually tapered off to near zero around 300-400 um into the track. Three






- 23 -


exceptions were the long sloping DC profiles observed in experiments GP89,

GPB2, and GP32 that resulted in very broad DC gradients that remained negative

over a 500 pm distance. In six of the second-turn experiments a distinct subpeak

of magnitude 1.5-4.0 mV/mm was present 100-200 pm modiolar to the major peak

(e.g., figure 2-8, experiment GP94--depth 340 pm). In contrast to DC gradient

profiles of the second turn, only one third-turn profile had a distinct subpeak on

the modiolar side of the major peak (figure 2-9, experiment GP83--depth 200 pm).

Instead, the third-turn profiles tended to have a broader major peak.

In the scala-tympani experiment (GPA9) where the DC potential decreased

in the 300-500 pm depth range, the DC gradient profile went from negative to

positive 300 pm into the track. This implies that current is directed toward the

modiolus at locations modiolar to the zero crossing and that current is directed

away from the modiolus on the spiral-ligament side of the zero crossing. One

interpretation of this pattern of radial current is that a standing current source

was located above the zero-crossing location in this experiment.

The radial DC gradient data from scala tympani of the second and third

turn are summarized in figure 2-10. All of the DC gradient profiles shown in

figures 2-8 and 2-9 were separated into two groups based on the magnitude of

the negative peak (criterion level 8 mV/mm). The three DC gradient profiles that

had two equal magnitude peaks (GP89, GPB2, and GP83t3) were not included, nor

were the profiles that had the largest (GP82) and smallest (GP84) negative peak,

nor was the single profile (GPA9) whose DC gradient changed direction. Before

plotting these DC gradient profiles on the same graph, the profiles were shifted

along the horizontal axis in order to make the negative DC gradient peaks

coincide at depth 0 pm. The top panel shows five profiles in the 8-12 mV/mm

range; the middle panel shows five profiles in the 3-8 mV/mm range. The

profiles shown in the top and middle panels were averaged and the means plotted






- 24 -


in the bottom panel. Even though the major peaks differ by a factor of two, the

modiolar slopes of the large and small types of profiles are not significantly

different from one another, implying that these two types of profiles are not

simply scaled versions of one another.



Scala Vestibuli DC Profiles

Radial DC profiles in scala vestibuli were obtained along 85 electrode

excursions in 8 guinea-pig cochleas: 5 from the second turn and 3 from the third

turn. Four of the second-turn experiments are shown in figure 2-11 and all of

the third-turn experiments are shown in figure 2-12. The DC profiles in scala

vestibuli look quite similar to profiles in scala tympani. The DC potential

increased along a shallow slope near the air-perilymph interface and then rapidly

increased 1-3 mV over a 100-200 pm distance. However, the DC potential in

scala vestibuli did not reach a plateau value as it usually did in scala tympani.

In all experiments (except GPC8-1) the DC potential continued to increase along a

shallow positive slope for the 500-600 Am length of the track.



Scala Vestibuli DC Gradient Profiles

Average radial DC potential gradients were calculated for all of the scala-

vestibuli experiments and are plotted in figure 2-13 (top panel, second-turn data;

bottom panel, third-turn data). In all experiments the DC gradient had a single

negative peak (implying that current is directed away from the modiolus) located

100-240 Am from the air-perilymph interface. The magnitude of the peak varied

from 10-30 mV/mm in the second turn and from 3-14 mV/mm in the third turn.

The presence of the shallow positive slope of the DC profile at the modiolar end

of the track results in a DC gradient (driving current away from the modiolus)

that remains in the 0.5-2.0 mV/mm range over a 200-400 pm distance in scala

vestibuli.






- 25 -


The DC gradients in scala vestibuli of the second and third turns are

compared in figure 2-14. A mean DC gradient profile was calculated for each

turn by first horizontally shifting the profiles along the depth axis by an amount

that made the negative peaks line up at depth 0 am (i.e., the same operation

described for figure 2-10). The five shifted profiles from the second turn were

averaged and. plotted as squares and the three shifted profiles from the third turn

were averaged and plotted as circles. Even though the magnitudes of the peaks

differ by a factor of four in the second and third turn, the DC potential gradient

deep in scala vestibuli of each turn is 1-2 mV/mm.

The shapes of the DC gradient profiles in scala vestibuli were compared

by normalizing the profiles relative to the negative peak. The results of this

normalization procedure are presented in figure 2-15 (top panel, second-turn data;

bottom panel, third-turn data). The striking similarity of the normalized profiles

implies that the shape of the DC gradient profile in scala vestibuli is independent

of the magnitude of the DC potential gradient. A similar statement cannot be

made of the scala-tympani DC gradients shown in figure 2-10. The modiolar

slopes of the DC potential gradient profiles in scala tympani are similar to one

another, regardless of the magnitude of the peak. No amount of shifting and

scaling would make the scala-tympani DC potential gradients superimpose. In

contrast, all of the DC potential gradient profiles in scala vestibuli of the second

turn are, at least to a first approximation, translated and scaled versions of one

another.

In contrast to the second-turn results, the magnitude of the DC gradient

deep in scala vestibuli of the third turn is 20-40% of the negative peak. On the

other hand, if one ignores the 100-200 pm depth range of experiments GPD2 and

GPC8-1, then the shape of negative peak is actually quite similar in the second

and third turns. In experiment GPD2, the abrupt 100 pV shift in the DC






- 26 -


potential in the 160-200 pm depth range raises suspicion about this part of the DC

profile. The other third-turn profile, GPC8-1, was unusual in that an

endocochlear potential (+70 mV) was encountered at depth 500 jpm. indicating that

the electrode penetrated through Reissner's membrane and into scala media. This

may also explain why this track encountered the largest DC gradient in scala

vestibuli of the third turn and may also explain the absence of a shallow slope of

the DC profile at the modiolar end of the track.




Discussion



Radial DC Gradients in Scala Tympani and Scala Vestibuli

Within a given preparation DC potential profiles and the computed DC

gradients were reproducible for several hours. This argues that during the course

of the experiment significant changes in the standing currents did not occur while

the animal was alive, nor did the electrode penetrations alter the pattern of

current flow. There is a rapid postmortem decrease in the radial DC gradient

recorded from either scala tympani or scala vestibuli, implying that the standing

current is dependent on oxidative metabolism (Brownell et al., 1986; Zidanic

et al., 1985).

The most striking feature of radial DC gradient profiles in scala tympani

and scala vestibuli of the second and third turns is the single large negative peak

within or near the spiral ligament representing a standing current flowing away

from the modiolus and into the spiral ligament. Modiolar to the large peak the

DC gradient tapers off rapidly. The simplest interpretation of the shape of the

DC gradient profiles is that the single large peak represents the potential gradient

generated by a standing current as it flows through the spiral ligament. Large

potential gradients are expected in the spiral ligament for two reasons.






- 27 -


The first is that the conductivity of the spiral ligament is likely to be

lower than that of perilymph based on anatomical considerations. The spiral

ligament is composed of loosely-packed cells in an extracellular matrix of fibers.

Some of the fibers are continuous with those that comprise the basilar membrane,

extending out from the attachment of the basilar membrane in a fan-like manner.

These fibers provide an elastic support for the basilar membrane to the otic

capsule. While gross impedance measurements have been made across the spiral

ligament in vivo (Cannon, 1976), many assumptions must be made to calculate

tissue conductivity from the measurements. Precise measures of interstitial fluid

conductivity have been obtained in other preparations that provide an insight to

the question of spiral-ligament conductivity. In an in vitro isolated rat retina

preparation, Hagins et al. (1970) determined that the extracellular conductivity in

the interstitial space in the outer nuclear layer is five times less than extracellular

fluid. Since the cell packing appears to be much less in the spiral ligament than

in the outer nuclear layer of the retina, the conductivity ratio of perilymph to the

spiral ligament is probably at most two to four. By Ohm's law (equation 2-6), the

same current density will generate a higher potential gradient in a region of low

conductivity (i.e., in the spiral ligament) than in a region of high conductivity

(i.e., in perilymph).

A second reason for large potential gradients in the spiral ligament is

based on electroanatomical considerations. The depth of both scala vestibuli and

scala tympani is large in the middle of the chamber and narrows near the lateral

wall. If all of the current flowing through each chamber is recycled back to the

stria vascularis, then the current density will become concentrated near the lateral

wall and will be maximal within the spiral ligament.

Despite inter-animal variability in the radial DC gradients, the smallest DC

gradients are recorded from scala tympani whereas the largest DC gradients are






- 28 -


recorded from scala vestibuli of the second turn. One explanation for large scala-

vestibuli DC gradients is based on anatomical constraints for current flow from

scala vestibuli into the spiral ligament. There is only a narrow portion of the

spiral ligament, no more than 50 im wide, that extends above the attachment of

Reissner's membrane to the lateral wall. All of the current returning to the stria

vascularis through scala vestibuli must "squeeze through" the spiral ligament at

this location. On the other hand the spiral ligament is 100-150 pm wide at the

level of the basilar membrane. Thus, even if similar current densities are present

in the central portions of both scala tympani and scala vestibuli, a larger radial

current density may be present near the spiral ligament in scala vestibuli since

the current must flow through a smaller cross-sectional area. Another explanation

for the differences in magnitude of the DC gradients in scala tympani versus

scala vestibuli is that the conductivity of the spiral ligament at the scala-vestibuli

border may be lower than its value at the scala-tympani border.

In several scala-tympani experiments there was a small subpeak in the DC

gradient profile 100-200 pm modiolar to the major peak. The location of this

subpeak corresponds to where the cochlear microphonic (CM) is changing most

rapidly, i.e., at the peak of the CM gradient (see chapter 3). Because of this

spatial separation of DC and CM currents in scala tympani, very little current is

modulated in the region of the DC gradient peak. This result suggests that the

net current through the scala-tympani pathway does not change during acoustic

stimulation (perhaps the hair cell currents are reciprocally related to changes in

leakage currents through the external sulcus). However, the presence of

microphonic currents in scala vestibuli oriented in the opposite direction from the

scala-tympani microphonic currents (see chapter 3) argues that at least some of the

current that is modulated by the hair cells is shunted through the scala-vestibuli

pathway.






- 29 -


Standing Currents through Hair Cells

Since the standing currents are modulated by acoustic stimuli, at least part

of the standing current can be assumed to flow out of scala media through the

mechanically-sensitive transduction channels in the stereocilia of the inner and

outer hair cells. The microphonic currents reduce and enhance the standing

current during opposite phases of the stimulus cycle (see chapter 3), consistent

with a resting current flowing through the transduction channels in silence.

A significant radial DC gradient (1-2 mV/mm) representing a standing

current away from the modiolus is present in scala vestibuli over the 600 apm

length of the track. One explanation of this result is that most of the current

that flows through the scala-vestibuli pathway leaks through scala media near the

modiolus, where Reissner's membrane attaches to the spiral limbus. In contrast,

there is an absence of a radial DC gradient at the modiolar portion of the track

in scala tympani. The DC profiles in scala tympani generally reach a plateau

value within 400 pm of the starting point of the track. The absence of a radial

potential gradient at the modiolar portion of the scala-tympani tracks indicates

either (1) that there is no current density at this location, or (2) that the current

has changed direction from a radial orientation near the spiral ligament to a

vertical orientation. The latter possibility is more likely since the modiolar

portion of the track is probably located below the organ of Corti. The current

that flows through the outer hair cells would be expected to be oriented in the

vertical direction near the hair cells as it spreads out into scala tympani. The

current through the single row of inner hair cells (located about 100 Am modiolar

to the three rows of outer hair cells) is probably not large enough to influence the

current density field below the outer hair cells. An alternative possibility is that

some of the current through the inner hair cells may leak into the modiolus and

return to the stria vascularis via a spiral-limbus/scala-vestibuli pathway or

through the vasculature.






- 30 -


Variability of DC Gradients across Experiments

The magnitude of the large DC gradient peak near the lateral part of the

track varied from 3-30 mV/mm. There are several possible sources for the

differences in magnitude of the radial DC gradients across experiments. For the

experiments in scala tympani, electrode orientations of 300 relative to the basilar

membrane and distances of 0-200 pm from the basilar membrane could very well

be possible. A technique has been developed to reconstruct electrode tracks from

histological material (Cousillas et al., 1989). Until this technique is combined with

potential gradient measurements in scala tympani and scala vestibuli, an

assessment of the effect of electrode orientation on the measured potential

gradients is very difficult. The one-dimensional analysis used in this study only

measures the component of the potential gradient vector aligned with the electrode

track. Since the complex geometry of the cochlear chambers and the distributed

nature of the presumed current sources and sinks are expected to generate a

nonuniform electric field in the cochlea, a detailed electroanatomical model of

current flow in the cochlea seems necessary to examine the effect of electrode

track orientation on potential gradients in scala tympani.

The second possible source of variability is trauma associated with the

otic-capsule surgery. In the cochleas that were examined histologically, the spiral

ligament was always detached from the otic capsule for some distance from the

site of the exposure. While some detachment of the spiral ligament could be

observed during the experiment, the detachment may be exaggerated in the

histology because of tissue shrinkage during dehydration. Complete detachment of

the spiral ligament from the otic capsule could permit the standing current to

short circuit around the presumably less-conductive spiral ligament, resulting in

smaller potential gradients recorded along the electrode track.






- 31 -


The amount of cochlear trauma was assessed by measuring compound

action. potential (CAP) thresholds before and after the otic-capsule surgery. Seven

of the nine second-turn experiments had average threshold shifts of more than

5 dB in the 3.0-5.0 kHz range (i.e., the characteristic frequency (CF) range of the

second turn), whereas only five of the eleven third-turn experiments had average

threshold shifts of more than 5 dB in the 0.5-2.0 kHz range (i.e., the CF range of

the third turn). It is well-known that the cochlear base is the most susceptible

part of the cochlea to noise damage and ototoxic drugs, and the results presented

here are consistent with a base-to-apex gradient of susceptibility of the cochlea to

trauma. Alternatively, CAP measurements may be more sensitive in the second

turn where neurons that innervate that region have relatively sharp tuning curves

compared to neurons that innervate the third turn (Liberman, 1978). Thus, it may

require more damage to the third turn than the second turn to raise CAP

thresholds at CF by the same amount.

For the experiments in scala vestibuli, the magnitude of the peak of the

DC gradients differed by a factor of three. The differences in magnitude could

result from either differences in current density or conductance. While such

differences in the conductivity of perilymph would explain the results quite

nicely, it seems rather unlikely that the conductivity of perilymph could be that

different across experiments. On the other hand, if the standing current through

the Reissner's membrane pathway is the same across experiments, and that

differences in the conductivity of the spiral ligament are responsible for the

observed differences in magnitude of the DC gradients, then one would still

expect similar gradients in scala vestibuli. However, the magnitude of the DC

gradient near the modiolus remains a fixed percentage of the magnitude of the DC

gradient peak (about 5%). Thus, differences in conductance are not likely to be

responsible for the DC gradient magnitude differences. The differences in






- 32 -


magnitude more likely result from differences in the amount of current leaking

through the scala-vestibuli pathway.

In histological material from several scala-vestibuli and scala-media

experiments, detachment of the spiral ligament from the lateral wall resulted in

Reissner's membrane sagging onto the tectorial membrane in the region of the

exposure. If such an event is not just an artifact of shrinkage during histological

processing of the tissue but actually occurs while the experiment is in progress,

then a change in the static position of the stereocilia could be induced. This

would lead to a change in the resting current through the transduction channels

(Hudspeth and Corey, 1977), and indirectly lead to a change in current through

the scala-vestibuli pathway. An alternative explanation is that surgical trauma to

the stria vascularis could lead to a decrease of the endolymphatic potential (EP).

A lower EP would drive less current through all pathways for current flow out

of scala media, including the scala-vestibuli pathway.



A Current Density Model for Standing Currents

These results demonstrate that radial standing currents are present in both

scala tympani and scala vestibuli of the guinea-pig cochlea. In each of these two

cochlear chambers, current is directed into the spiral ligament. Since the standing

current flows in the same direction in both scala tympani and scala vestibuli, one

possibility to consider is that the current from one cochlear chamber leaks into

the adjacent chamber of the next turn through the thin bone that separates the

cochlear turns. If this is a significant pathway for DC currents, then frans-coil

currents should also be present during very low-frequency stimulation. However,

at 50 Hz, where there is little or no phase lag of the CM from base to apex

(Dallos and Cheatham, 1971; Oshima and Strelioff, 1983; Zidanic, unpublished

observations), not only are low-frequency microphonics out of phase in scala






- 33 -


tympani versus scala vestibuli (Dallos et al.,1971; also see chapter 3) but the

modulation of radial current is in the opposite direction in scala tympani versus

scala vestibuli (see chapter 3).

Since trans-coil currents now appear unlikely, models for intracochlear

currents should be restricted to local flow in the transverse plane and longitudinal

flow along the coiling scalae. A model of current flow in terms of current

density field lines is depicted in figure 2-16. According to this model, current is

recycled within the same transverse cochlear cross-section. Thus, the current

generated by the stria vascularis exits scala media and returns via scala tympani

or scala vestibuli to be recycled back through the stria. The results of this

chapter support this model since standing currents were not seen directed into the

modiolus in either scala tympani or scala vestibuli. This model is also consistent

with experiments that have shown that potassium ions are actively taken up from

scala tympani to scala media (Konishi et al., 1978) and that the primary source of

endolymph is not from the vascular system (Wada et al., 1979).



Estimate of Current Output of Marginal Cells

An estimate of the total current generated by a given transverse wedge of

the cochlea can be derived from the measurements presented in this chapter.

While potential gradients as large as 10-15 mV/mm were recorded along radial

tracks, the large peak was within the first 200 pm of the track and could be due

to a low conductivity of the spiral ligament. On the modiolar side of the DC

gradient peak, the radial potential gradient is in the 1-2 mV/mm range. Based on

a conductivity of 2 mS/mm for perilymph (Bdkdsy, 1951a), this gradient results in

a current density of 2-4 pA/mm2. The depth of both scala tympani and scala

vestibuli is approximately 200 pm, and if one assumes that the radial flow of

current is uniform throughout the scalae, then a total current of 0.8-1.6 pA flows






- 34 -


through both scala tympani and scala vestibuli in a 1 mm wedge of the cochlea.

Using anatomical data of the stria vascularis provided by Forge et al. (1987), one

may calculate that approximately 4250 marginal cells are present in a 1 mm

wedge of the cochlea. These calculations estimate that the average current output

of a single marginal cell is 188-376 pA and that a current density of

3.5-7.0 piA/mm2 is maintained through the marginal cell membrane into scala

media.



(Na+ + K+)-ATPase Density to Support the Standing Current

This analysis can be taken one step further by considering a specific model

of ion transport by marginal cells. It has recently been reported that marginal

cells have resting potentials 10 mV more positive than the endocochlear potential

(Melichar and Syka, 1987; Offner et al., 1987). This observation favors a model

of passive K+ flow from the cytoplasm of the marginal cell into scala media. The

mechanism of the generation of positive resting potentials in marginal cells is not

clear, but it has been established that marginal cells have a high concentration of

(Na+ + K+)-ATPase (i.e., the sodium-potassium pump) in their basolateral

membrane (Kerr et al., 1982; Kuypers and Bonting, 1970; Mees, 1983). If one

assumes that the current pumped into scala media by the marginal cells is carried

by K+ via (Na+ + K+)-ATPase, then a calculation may be made as to how many

sodium-potassium pumps are required to support this current.

The normal operation of (Na+ + K+)-ATPase is to exchange internal Na+

for external K+ in a 3:2 ratio, although other modes of exchange are possible

(Beaugd and Lew, 1977). A normal turnover rate for the pump is about 20 times

per second, corresponding to 40 K+/sec per pump (Boardman et al., 1972). Thus,

at least 31,000 (Na+ + K+)-ATPases per marginal cell are required to generate a

200 pA K+ current. The occluding junctions between marginal cells at their






35 -


apical membrane are extremely well-developed (Jahnke, 1975), and if combined

with a very low basolateral cell membrane permeability to K+ then a high

efficiency may be achieved. If one assumes a basolateral surface area of about

100 Am2 (a difficult quantity to calculate considering the numerous basolateral

infoldings), then the sodium-potassium pumps need to be packed in the membrane

with a density of 310/pm2, a value that is within the range measured for other

cells (Boardman et al., 1972).





- 36 -


55
-_1
a .............
^ 50


-o 45 ................




40 -



.5 5.0

Frequency (kHz)




Figure 2-1. CAP threshold for frequency range 0.5-5.0 kHz averaged over 28
guinea pigs. Threshold was defined as decibels sound pressure level re 20 pPa
(dB SPL) required to elicit a CAP with a 5 pV peak amplitude. Round window
potentials were averaged in response to 25-50 repetitions of 20 msec tone bursts
with a rise time of 1 msec. The tone stimulus was presented at a random phase
relative to the initiation of A/D conversion to average out the CM. Dotted lines
indicate one standard deviation from the mean.





- 37 -


10 -


Frequency (kHz)




Figure 2-2. CAP threshold shifts after making a 40-100 pm fenestra in the otic
capsule. Squares are the average of nine second turn exposures; circles are the
average of 11 third turn exposures.






- 38 -


Figure 2-3. Camera lucida drawing of a celloidin-embedded section through the
third-turn exposure of an experimental animal (GP82). Scala vestibuli (SV), scala
media (SM), and scala tympani (ST) are labeled. Reissner's membrane separates
SV from SM. The movable pipet is shown at the modiolar end of a reconstructed
electrode track (ST). Since the tissue was sectioned obliquely relative to the
radial path of the electrode track, the pipet crossed this plane of section only at
the fenestra over the spiral ligament. The orientation of the pipet relative to the
basilar membrane was determined by the appearance of a hole in the spiral
ligament on sections that were 40-60 pm closer to the mid-modiolar plane. The
large bone chip that sheared off from the otic capsule extended about 20 pm on
both sides of the section shown. The spiral-ligament pipet that monitored
potentials in the fluid that collects over the otic-capsule fenestra is also shown.
The scale above the ST pipet is in microns. The vectors representing the two-
dimensional local coordinate system are shown deep in ST and in SV. The radial
dimension is parallel to the basilar membrane in ST. The coordinate system in
SV is tilted relative to the ST system so that the radial dimension is parallel to
Reissner's membrane. The third dimension of the coordinate system, the
longitudinal component, projects orthogonal to the plane of the page.





- 39 -


1.2 -

1.0 -

.8

.6

.4 -

.2

.0





1.2-

1.0-

.8 -

.6 -

.4 -

.2 -


200

Electrode


400


Depth (Jim)


Figure 2-4. Effect of using the fixed electrode outside the spiral ligament as a
local DC reference. Data from four electrode excursions along the same track in
scala tympani of the second turn (GP89). DC profiles using the ground electrode
as the DC reference (top). DC profiles using the fixed electrode as the DC
reference (bottom).


0 200 400


.0 -


600





- 40 -


I I200 400
0 200 400


1.2 -



.8 -



.4 -



.0 -


600
600


0



-5



-10 -


F I
0 200


0 200


400


600


Uncorrected Depth (Am)


0 200


Electrode Depth (Am)


Figure 2-5. Shifting operation that compensated for electrode-positioning errors
that may have occurred during the course of some experiments. Data from four
electrode excursions along the same track in scala tympani of the second turn
over a 3-hour period (GPA9). Data on right were shifted by a multiple of 20 pm
that made the DC gradient peaks coincide. Shift factors were as follows (in Am):
o (+40), o (+80), A (0), + (+20). A. Unshifted DC profiles. B. Shifted DC
profiles. C. Unshifted DC gradient profiles. D. Shifted DC gradient profiles.


1.2 -


C.


4
400


I
600


0



-5


-10


600


_ .





- 41 -


Second Turn Scala Tympani DC


GP94


I I I
0 200 400


600
600


I I I4
0 200 400


GP89


0 200


1.0 -


1.2 -


.8 -


.4 -


.0 -


400


600


GPB2


I I I
0 200 400


Electrode Depth (gm)



Figure 2-6. Radial DC profiles from four different experiments in scala tympani
of the second turn. For each experiment, the DC profiles that were subjected to
the DC gradient analysis are shown after they were shifted according to the
procedure described in figure 2-5. The animal number is indicated in each plot.


1.2 -


6
600


600





- 42 -


Third Turn Scala Tympani DC


1.0 -

.8 -

.6 -

.4 -

.2 -


GP32


.0 IJ


0 200


600


.8-


.6


.4 -


.2 -


.0 -


0 200


GPC7


0 200


600


0 200


Electrode Depth (Am)



Figure 2-7. Radial DC profiles from four different experiments in scala tympani
of the third turn. For each experiment, the DC profiles that were subjected to
the DC gradient analysis are shown after they were shifted according to the
procedure described in figure 2-5. The animal number is indicated in each plot.


GP63


2.0 -

1.5 -

1.0 -

.5 -

.0 -


400


1.0


.5 -


.0


400





- 43 -


Second Turn Scala


Tympani DC Gradients

B.


2-


0


-2 -

-4
-4


-6 -


0- GP94
O GPA9


O GP89
o GPB2


I I
200 400


0

-2

-4

-6 -

-8


O GP83t2
o GP99


I I
200 400


600


r I
0 200


Electrode Depth (Am)


Figure 2-8. Radial DC gradient profiles from nine experiments in scala tympani
of the second turn. Each trace represents the average of two to six DC gradient
profiles collected along the same track in a given animal. The profiles have been
grouped together based on the magnitude of the negative DC gradient peak.
A. DC gradients computed for the experiments shown in the top two panels of
figure 2-6. B. DC gradients computed for the experiments shown in the bottom
two panels of figure 2-6. C. Two additional DC gradient profiles in the
10-12 mV/mm range. D. Three additional DC gradient profiles in the
6-8 mV/mm range.


0

-2 -

-4

-6 -

-8

-10


C.


600


200
200


I
600


0

-4
-4 -


-8 -


GP85
GP95
GP96


400


600


_






- 44 -


Third Turn Scala Tympani DC Gradients


0
-5
-5 -


-10 -


-15 -


-20 -





1-

0-


-1 -

-2
-2 -

-3

-4

-5 -


01
0


200


1
200


O GP63
o GP82
A GP85



600












O GP32
o GP83
A GPC7
+ GP84


600


Electrode Depth (Am)


Figure 2-9. Radial DC gradient profiles from seven different experiments in
scala tympani of the third turn. Three experiments with DC gradient peaks in
the 10-18 mV/mm range (top). Four experiments with DC gradient peaks in the
1-5 mV/mm range (bottom).































Figure 2-10. Radial DC gradient profile summary for scala tympani. Each
profile has been shifted along the abscissa so that the peak of the DC gradient is
plotted at depth 0 pm. Top plot shows five profiles with DC gradient peaks in
the 8-12 mV/mm range, three from the second turn (T2) and two from the third
turn (T3). Middle plot shows five profiles with DC gradient peaks in the
3-8 mV/mm range, three from the second turn (T2), and two from the third turn
(T3). Bottom. Four of the five profiles displayed in the top plot were averaged
and are plotted as squares (GP63 was not averaged in because depth increments
of 25 pm were used in that experiment). All five profiles displayed in the middle
plot were averaged and are plotted as circles.





- 46 -


2

0 --

-2 -

-4 -

-6 -

-8 -

-10 -

-12 -

-200


0 200 400


600


O GP85t2
T2 o GP95
A GP96
T+ GPC7
T3 x GP32


I I2
0 200


400
400


6
600


O mean of profiles
in top panel

o mean of profiles
in middle panel


200
200


400
400


6
600


Relative Depth (im)


O GP83t2
T2 o GP99
A GP94
+ GP63
T3 GP85t3


-2 -

-4

-6
-6 -

-8 -


I
-200
2-

0 --

-2 -

-4 -

-6 -

-8 -

-10 -

-12 -

-200





- 47 -


Second Turn Scala Vestibuli DC


0 200


GPA7


600


1.5 -



1.0 -



.5 -



.0 -


0 200 400


GPA6


0 200


400


600


0 200


Electrode Depth (gpm)


Figure 2-11. DC profiles from four different experiments in scala vestibuli of
the second turn. For each experiment, the DC profiles that were subjected to the
DC gradient analysis are shown after they were shifted according to the
procedure described in figure 2-5. The animal number is indicated in each plot.


GP77


3.5 -
3.0 -
2.5 -
2.0 -
1.5 -
1.0 -


1.6 -


1.2 -


.8 -


.4 -


.0 -


600





- 48 -


Third Turn Scala Vestibuli DC


2.0 -


1.5 -


1.0 -


.5 -


200 400
200 400


.0 -


600


GPC8-1


I I
0 200


.8-

.6 -

.4 -

.2 -

.0 -


GPD2


1.2 -

1.0 -

.8 -

.6 -

.4 -

.2 -


I I I
0 200 400


.0 -


600
600


GPC8-2


200


600


Electrode Depth (M/m)



Figure 2-12. Radial DC profiles from four different experiments in scala
vestibuli of the third turn. For each experiment, the DC profiles that were
subjected to the DC gradient analysis are shown after they were shifted according
to the procedure described in figure 2-5. The animal number is indicated in each
plot. Data were collected along two separate tracks in experiment GPC8. A
+65 mV EP was encountered at depth 500 Am of the first track (GPC8-1). The
exposure was opened up towards the apical end of the cochlea and another
electrode track (GPC8-2) was attempted. A +52 mV EP was encountered at
depth 600 pm of the second electrode track.


GPD5


.8 -


.6 -


.4 -


.2


400


600


' I






- 49 -


0 -- -- -



-10 -

0 turn 2 SV
0 GP76
-20 o GP77
A- GPA1
+ GPA6
E X- GPA7
E -30 -
> I I I I
0 200 400 600
i-'
0

0 -o --__ ^ _- _-- _-



-5


turn 3 SV
-10 \- GPD2
0 GPD5
A GPC8-2
+ GPC8-1
-15 -
I I I I
0 200 400 600

Electrode Depth (pm)



Figure 2-13. Radial DC gradient profiles from scala vestibuli. Second turn
data (top). Third turn data (bottom).





- 50 -


0 -----------------------------






8 turn 2 SV
SIo turn 3 SV
0 -12 -


-16 -

-200 0 200 400 600
Relative Depth (pm)




Figure 2-14. Average radial DC gradient profiles from scala vestibuli of the
second and third turns. Before averaging, the DC gradient profiles of figure 2-13
were first shifted by an amount that made the negative peaks line up at
depth 0 pm. The DC gradient profile collected along the first track in experiment
GPC8 was not included in the average for the third turn.





- 51 -


.0 -------------------------

-.2

-.4
turn 2 SV
? -.6 0 y- GP76
E 0 GP77
A GPA1
> -.8 + GPA6
E X GPA7
S-1.0 -

-200 0 200 400 600

0
0
.0 ---------- -------------------
N
c -.2

Z .4

-6 turn 3 SV
SD- GPD2
8 -0 GPD5
A GPC8-2
+ GPC8-1

I I I I I
-200 0 200 400 600
Relative Depth (Am)

Figure 2-15. Normalized DC gradient profiles from scala vestibuli. Before
normalizing the DC gradient profiles of figure 2-13, they were first shifted by an
amount that made the negative peaks line up at depth 0 pm. The DC gradients
were normalized by dividing the magnitude of the DC gradient at each point by
the magnitude of the DC gradient at depth 0 Am. This operation assigns a value
of 1.0 at depth 0 pm for all profiles. Second turn profiles (top). The division
factors were as follows: GP76(14.8), GP77(27.5), GPAI(18.8), GPA6(9.35),
GPA7(12.4). Third turn profiles (bottom). The division factors were as follows:
GPC8-1(12.9), GPC8-2(6.00), GPD2(4.63), GPD5(3.62).






- 52 -


Figure 2-16. Model for standing currents in the cochlea in terms of current
density field lines. The magnitude of current density at any particular location is
determined by the local concentration of field lines. The direction of current is
indicated with arrows along the field lines. The basic assumption of the model is
that all current generated by the stria vascularis (shaded area) exits scala media
and is recycled via scala tympani and scala vestibuli. Little or no current is
presumed to flow through the bone that forms the boundary of each chamber with
the modiolus, separates scala vestibuli and scala tympani of adjacent turns, and
encapsulates the entire cochlea. This local flow of current results in a
concentration of the current lines within the spiral ligament. A major pathway
for current leakage from scala media is through the mechanically-sensitive
transduction channels in the stereocilia of the hair cells. Current also is
postulated to leak through Reissner's membrane and also through (or between) the
supporting cells lateral to the organ of Corti.















CHAPTER 3

EVOKED POTENTIAL WAVEFORMS AND COCHLEAR MICROPHONICS



Introduction



The first measurements of the cochlear microphonic (CM) were made in the

internal auditory meatus of the cat (Wever and Bray, 1930). While these

investigators first thought that the CM was generated by the auditory nerve, it

was determined soon after that the microphonics were generated by the hair cells

of the organ of Corti (Adrian, 1931). In the early days of cochlear physiology,

investigations of the CM were made from the round window or on the nearby

bone (Davis et al., 1934) and it is now well-established that round-window CM

recordings in the normal animal reflect the activity of hair cells located in the

basal coil of the cochlea (Dallos, 1969).

A major improvement in CM recording was made with the classic study of

Tasaki et al. (1952) using differential electrodes. With this technique, a pair of

electrodes are placed in same same cochlear turn, one in scala tympani, the other

in scala vestibuli. A differential amplifier is used to cancel out common

potentials that are generated at a distance from the two recording electrodes. The

differential configuration thus provides information about the physiology of hair

cells located within about a 1 mm longitudinal section of the basilar membrane

(Dallos, 1969). However, the frequency range for which valid data may be

collected is limited to frequencies below the characteristic frequency (CF) of the

recording location (Dallos et al., 1971). To obtain complete frequency response


- 53 -






- 54 -


curves of hair cells, it is necessary to record intracellularly. While our

knowledge of inner hair cell physiology has been greatly enhanced over the last

ten years with successful intracellular recordings from a number of laboratories

(Brown and Nuttall, 1984; Brown et al., 1983a; Goodman et al., 1982; Nuttall

et al., 1981; Russell and Sellick, 1978), intracellular studies from outer hair cells

remain difficult (Dallos, 1985, 1986; Russell and Sellick, 1983; Russell et al.,

1986). Given the inherent difficulties of in vivo intracellular studies and the

relatively low spatial resolution of differential CM recordings, we have developed

a current density analysis that bridges the gap between these techniques (Brownell

et al., 1983).

The approach of this study is to characterize the spatial variation of tone-

evoked field potentials along radial tracks in scala tympani and scala vestibuli.

The use of monopolar electrodes ensures that both the neural and microphonic

components are recorded. By computing the discrete Fourier transform of

averaged response waveforms, the magnitude and phase of CM can be accurately

determined even when the CM response is no larger than the electrical noise

level. Such digital calculations perform a function similar to the analog circuitry

of lock-in analyzers. While the 60-80 dB dynamic reserve of a lock-in analyzer

would permit CM measurements at much lower stimulus levels, the great

advantage of signal averaging is that CM, compound action potentials (CAPs), and

summating potentials are all captured in the evoked response waveform, which

can then be operated on to analyze a particular component in greater detail.

Since the early 1950's it has been recognized that round-window and

intracochlear field potentials contain both auditory-nerve CAPs and CM (Davis

et al., 1950; Tasaki et al., 1952). While round-window CAP thresholds are

routinely used to monitor the physiological state of the cochlea, the CAPs that can

be recorded from intracochlear electrodes have not been characterized in detail.






- 55 -


The main reason for this is that the round-window location is very accessible and

so there has not been a great motivation for recording CAPs elsewhere. However,

the potential significance of intracochlear CAPs has recently been augmented by

the report of Ruggero et al. (1986) showing that very low-frequency stimuli evoke

multiple CAPs at the round window of chinchillas. Based on single unit

recordings, the mode of excitation of inner hair cells in the apical turns of the

cochlea appears to be different from the mode of excitation in the base, the

differences probably being related to cochlear micromechanics (Ruggero and Rich,

1987). Since the mechanism of activation of inner hair cells at threshold is of

great importance, one of the goals of this study is to describe the CAPs that can

be recorded from intracochlear electrodes in the second cochlear turn of the

guinea pig.

While intracochlear electrodes have been routinely used to measure CM,

the CAPs are always cancelled in these CM studies by the use of differential

electrodes. In addition, all of the classic CM studies were done with either low-

sensitivity oscilloscope recordings or with wave analyzers that only record the

magnitude of the fundamental component of the response waveform. More

recently, signal averaging techniques have been used to characterize click-evoked

waveforms (Brownell et al., 1983; Echeverria and Robles, 1983). However, there

have been but a few tone-evoked waveforms published in the literature. Those

that have were recorded from either the round window (Ruggero et al., 1986) or

in scala tympani of the basal turn (Russell and Sellick, 1983).

The results presented in this chapter provide the first detailed

characterization of tone-evoked field potentials in the cochlea. A reciprocal

relationship between the modulation of currents in scala tympani and scala

vestibuli of the second turn during low-frequency acoustic stimulation (50-400 Hz)

is shown. New insights on the pattern of current flow in the cochlea are






- 56 -


presented. In particular, capacitative currents across Reissner's membrane are

calculated that may represent the dominant pathway of leakage current modulation

through the endolymph/perilymph barrier at moderate stimulus frequencies (above

600 Hz). Large longitudinal currents are expected in scala tympani near best

frequency that should reveal important features of the traveling wave in the

upper cochlear turns. Portions of these results have been presented in abstract

form (Zidanic and Brownell, 1986; Zidanic et al., 1985, 1986, 1987a, 1987b).




Methods



Animal Preparation

The data presented in this chapter were collected from the same guinea

pigs that were used for the experiments of the preceding chapter. The surgical

preparation of the animals is described in detail in chapter 2. A brief description

of the experimental procedure follows. Experiments were conducted in a sound

isolation chamber heated to 320-350C. Healthy guinea pigs were anesthetized with

pentobarbital (18 mg/kg) and Innovar-Vet (0.25 ml/kg), tracheotomized, stabilized

relative to the recording table with a head bar, and the bulla widely exposed

through a ventral approach. The tensor tympani and stapedius muscles were

severed, a reference electrode was placed through the bucca, and then the animal

was paralyzed with Flaxedil (gallamine triethiodide, 5 mg/kg). The acoustic

system was calibrated through a probe tube microphone and compound action

potential (CAP) thresholds in the 500-5000 Hz range were determined by

recording from a copper wire electrode placed on the round-window niche (see

chapter 2 for CAP threshold results). A fenestra of 40-100 pm in diameter was

made in the otic capsule over either scala tympani, scala media, or scala vestibuli

of the second turn (see figure 3-1) and the CAP thresholds were remeasured (see

chapter 2 for threshold shifts following otic-capsule surgery).






- 57 -


Two micropipets were placed in the fluid that collected over the otic-

capsule exposure. One was a fixed electrode that remained outside the spiral

ligament for the duration of the experiment and the other was mounted on the

spindle of a piezoelectric microdrive that was advanced and withdrawn along a

600 pm track in 20 jim steps. The desired orientation of the track was in the

radial direction, parallel to the basilar membrane when recording from scala

tympani and parallel to Reissner's membrane when recording from scala vestibuli

or scala media (see chapter 2 for more details). Tone-evoked potentials were

simultaneously averaged from each electrode at the stopping points of the movable

electrode along the track. Multiple electrode excursions along the same track

were made in order to explore a broad range of the frequency/intensity spectrum.

Each electrode excursion required 5-20 minutes to complete depending on the

number of different stimuli presented at each depth.



Sound Stimulation

One of the goals of this study was to determine the relative magnitude and

phase of the cochlear microphonic (CM) in each of the chambers of a single

cochlear turn. Since measurements in scala tympani, scala vestibuli, and scala

media were not made from the same cochlea, it was necessary to have the

acoustic system calibrated to compare results from different experiments. A two-

phase lock-in amplifier was used to calibrate both the magnitude and phase

response of the acoustic system. The acoustic calibration procedure permitted

attenuation values to be precisely converted to sound pressure levels. Such

procedures also permitted the relative phase of the CM recorded from the

different cochlear scalae to be determined by plotting CM phase relative to a

common reference, i.e., peak rarefaction at the tympanic membrane.






- 58 -


Sinusoidal stimuli were generated by a programmable oscillator (Krohn-Hite,

model 4031 R) gated through an electronic switch (Grason-Stadler, model 1278B)

and attenuated (Grason Stadler, model 1284) under computer control (DEC PDP-

11/83). The 600 92 output impedance of the attenuator was approximately

matched by placing a 560 12 resistor in series with a Beyer DT-48 25 2 speaker.

The speaker was encased in an aluminum housing that mated to an acoustic

coupler mounted on a post on the recording table. Connection to the external

auditory meatus of the animal was made with a polyethylene tube pressure fit to

the acoustic coupler and sealed to the meatus with cyanoacrylate (Krazy Glue).

This ensured that the only metallic connection to the animal was with the ground

electrode. A 5" electret-condenser microphone (GenRad model 1962-9610,

GenRad model 1972-9600 preamplifier, custom-built 18 V power supply)

monitored sound pressure through a 1 mm i.d. calibrated probe tube, 3.4 cm in

length, the tip of which was 2 cm from the tympanic membrane.



Microphone Measurements

A sinusoidal electrical input to a speaker generates sound pressure

variations, p(t), at the stimulating frequency:



p(t) V22 Prms sin (2n/ft + Op). (3-1)



where Prms is the root-mean-square magnitude and Op is the phase (re electrical

input to the speaker) of the fundamental component of the sound pressure level.

A microphone placed in this sound field generates a voltage, v(t), that is also

modulated at the same stimulus frequency:






- 59 -


v(t) = v/2 Vrms sin (21rft + 0,), (3-2)



where Vrms is the root-mean-square magnitude and 6V is the phase (re electrical

input to the speaker) of the fundamental component of the microphone voltage.

According to manufacturer specifications, the probe tube and reference

microphones used in this study have very negligible phase shifts of their voltage

response relative to pressure (B & K model 4136, 100 at 5 kHz). Thus we can

define a single phase variable, 0, such that



0 : ? p 0a (3-3)



Microphone voltages were converted to decibels sound pressure level re

20 PPa (i.e., 0 dB SPL) according to the following equations:



Prms = Vrms (3-4)


dBSPL 20 logo 2 Prms (3-5)


where a is the sensitivity of the microphone in mV/Pa. Microphone sensitivities

were determined with a calibrated 250 Hz 124 dB SPL sound source (B & K

model 4220 pistonphone; 1" B & K sensitivity: 0.669 mV/Pa; 5" GenRad

sensitivity: 6.69 mV/Pa).



Two-phase Lock-in Analysis

The root-mean-square voltage output (denoted Vrms) of the reference and

probe microphones during the acoustic calibration frequency sweep were

sequentially recorded with an Ortec 9505 two-phase lock-in analyzer under






- 60 -


computer control according to the algorithm described below. While most of the

CM data presented in this chapter were collected by averaging evoked potential

waveforms, the data shown in figures 3-15 and 3-21 were collected with the lock-

in technique.

The two-phase lock-in analyzer generates two signals, X and Y, such that



X Vrms cos 0 (3-6)
and
Y Vrms sin 0 (3-7)


where Vrms and 6 are defined as in equations 3-2 and 3-3. An iterative

algorithm was used to average the X and Y lock-in signals. In a steady-state

situation these outputs are constant, but they require a certain amount of time to

settle after the calibration tone is enabled by the electronic switch. This settling

time is governed by the lock-in time constant, r, which was remotely controlled

by the calibration program. In addition, the time constant controls the effective

bandwidth of the lock-in amplifier, becoming narrower as the time constant

increases. While a response with a high signal-to-noise ratio can be measured

quickly with a very short time constant (100 msec), a low-level signal in a noisy

background will require a long time constant to obtain an accurate reading. The

following algorithm was designed to make fast determinations of high-level signals,

but also accurate measurements for low-level responses. A typical acoustic

calibration run covering the 50-5000 Hz range required about 5 minutes to

complete.

The time constant was initially set to 100 msec. After waiting two time

constants from the gating of the electronic switch, the two lock-in amplifier

signals were digitized 100 times with a sampling interval of T/100. The mean and

standard deviation of these 100 points were computed. If the signal-to-noise ratio

of either the X or Y signal was less than 20 dB, then another 100 points were






- 61 -


averaged over another time constant. If an average of the X and Y signals was

not obtained within the criterion signal-to-noise ratio after 5 time constants from

the time the calibration tone was enabled, then the tone was turned off, the time

constant was increased (maximum 3 sec), and the algorithm was repeated.

The root-mean-square magnitude and phase of the microphone voltage was

then computed from the averaged X and Y lock-in signals with the standard

rectangular to polar coordinate transformation:



Vrms = VXA2 + Y2 (3-8)


0 tan-1 (3-9)



Sound pressure level (dB SPL) was then computed by substituting Vrms into

equations 3-4 and 3-5.



Probe Tube Calibration

The acoustic characteristics of the probe tube and speaker were calibrated

every two or three months. A 1" reference microphone (Bruiel and Kjaer,

model 4136 with GenRad model 1995-3410 preamplifier supplying the 200 V

polarization voltage) was placed 3 mm from the end of the probe tube in a closed

cavity. The spectrum of the reference microphone response is shown in panel A

of figure 3-2 and the spectrum of the probe microphone response is shown in

panel B. Below 300 Hz, both microphones pick up a maximum pressure of

104-106 dB SPL. Resonances are evident in the response curves at

0.6, 1.4, 2.6, 3.6, and 4.5 kHz. As expected, the magnitude peaks are associated

with rapid phase shifts of about r radians at the resonant frequencies. The

difference of the two curves gives the probe tube calibration curve (figure 3-3)






- 62 -


which only has two resonances at 0.6 and 4.5 kHz. Otherwise, the probe tube

transfer function is low-pass.

At very low frequencies, the B & K output is at ir radians (1800) relative to

the electrical signal into the speaker. In contrast, the probe microphone response

is in phase with the electrical signal into the speaker. This results from the

different design characteristics of electret-condenser versus +200 V polarized air-

condenser microphones. The B & K output is negative in response to positive

pressure, whereas electret-condenser microphone responses are positive-going for

positive pressure at low frequencies. Thus, a positive electrical input to the

Beyer DT-48 speaker results in positive pressure in this small closed cavity

without any phase lag up to 250 Hz, a 300 lag at 500 Hz, and a 600 lag at

1000 Hz.

The standard deviation of magnitude and phase responses of the reference

and probe microphones are shown in squares on expanded scales below the

individual response traces (figure 3-2). The magnitude data were generally

reproducible to within l dB and the phase data to within 0.1 radians (~6).

However, in frequency regions near the magnitude peaks and rapid phase shifts

the standard deviations could be as high as 2 dB SPL and 0.4 radians (~23).

The large standard deviation (4 dB SPL) at 5 kHz for the probe microphone is

due to the combined effects of the speaker response (90 dB SPL maximum output),

the 20 dB attenuation of the probe tube, and the additional 20 dB attenuation used

for the calibration procedure. This results in an absolute sound level of

50 dB SPL at the probe microphone, a level near the noise floor of the probe

microphone/preamplifier combination.

The standard deviation profile of the probe tube frequency response shows

that the magnitude was reproducible to within 1.3 dB except above 4.4 kHz

(figure 3-3). The phase characteristics of the probe tube were reproducble to






- 63 -


within 0.1 radians, except in the 500-750 Hz range and again in the 3-5 kHz

range where the rapid phase shifts occurred in the probe tube and speaker

responses. Note that the large variability seen in the reference and probe

microphone responses at 3.6 kHz was not present in the probe tube calibration

curve. This probably resulted from small changes in the resonances of the closed

cavity from one calibration run to the next.



In Situ Acoustic Calibration

Sinusoidal stimuli were delivered to the speaker at a constant 20 or 40 dB

attenuation (re 1.0 Vrms) as frequency was swept from 50 to 5000 Hz. Tone

duration varied from 1-10 seconds depending on the signal-to-noise ratio of the

microphone response. At 20 dB attenuation, the actual sound level delivered to

the animal during the calibration procedure was below 90 dB SPL, except in the

1.6-3.6 kHz range where sound levels varied from 90-99 dB SPL. In later

experiments, a standard guinea-pig ear calibration table was preloaded into the

acoustic calibration program so that the attenuators could be automatically

adjusted during the frequency sweep to keep the tone level at about 80 dB SPL.

The output of the probe microphone was measured with a lock-in amplifier

in a manner identical to that used for the probe tube calibration. The results

from six guinea-pig ears are shown in figure 3-4. The magnitude of the probe

microphone response was corrected for the attenuation of the electrical input to

the speaker (20 or 40 dB) and plotted on the left graph of panel A. The phase

data are plotted on the right graph of panel A in radians relative to the electrical

input to the speaker. Note that the low-frequency response (< 800 Hz) is nearly

identical to the probe microphone response in the closed cavity (see figure 3-2,

panel B).






- 64 -


The probe microphone responses were then corrected for the transfer

characteristics of the probe tube and the results plotted in panel B. Below

400 Hz, the maximum sound level that could be presented was the same as that

achieved in the closed cavity: 104-106 dB SPL. In the 400-1000 Hz range, the

maximum sound level was a few dB lower than the closed cavity results. In the

1-5 kHz range, the resonant peaks that were present at 1.4, 2.6, and 3.6 kHz in

the closed cavity were not present in the guinea-pig ear, although a notch in the

magnitude response appeared at 4.4 kHz.

The phase data on the right graph of panel B identify the phase relative to

the electrical input to the speaker at which maximum rarefaction occurs at the tip

of the probe tube. Since the tip of the probe tube was about 2 cm from the

tympanic membrane, the phase values plotted do not precisely represent the phase

of maximum rarefaction at the tympanic membrane. It can be estimated that the

phase values plotted are in error by up to 340 at 1600 Hz and 1040 at 5000 Hz,

the highest frequency used for CM recordings in this study. Since these phase

errors are small with respect to the phase lags introduced by the cochlear

traveling wave, all CM phase data are plotted relative to rarefaction at the tip of

the probe tube.

Six other guinea-pig ear calibrations are shown in figure 3-5 that were

somewhat different from those shown in figure 3-4. The maximum sound

pressure that could be delivered to these ears was reduced 2-15 dB at the lowest

frequencies. Correlated with this low-frequency rolloff was a phase lead of up

to 750 (at 50 Hz) in the pressure response in the external auditory meatus relative

to the calibrations shown in figure 3-4. This low-frequency rolloff was

eliminated in several experiments by resealing the polyethylene tube to the

external meatus and placing vaccuum grease around the tube. Thus a partially

open acoustic system was probably responsible for this behavior.






- 65 -


Electric Recording

Micropipets with tips 1-2 um in diameter and shanks 500-800 pm in length

were fabricated with a microprocessor-controlled pipet puller (Mecanex, BB-CH)

and filled with a solution that resembles perilymph (155 mM NaCl, 4 mM KCI--

for scala-tympani and scala-vestibuli recordings) or endolymph (156 mM KC1,

1.4 mM NaCl--for scala-media recordings). The pipet used as the fixed electrode

in the fluid over the exposure was broken back to have a tip diameter of about

10 Am to achieve a low impedance and lower noise level. Capacitance

compensation was achieved with custom-built, high-impedance, low-noise, DC-

coupled preamplifiers. Subsequent stages provided amplification (xlOOO for scala-

tympani and scala-vestibuli recordings, xl00 for scala-media recordings) and low-

pass filtering (10 kHz cutoff) of the electrode signals before digitization (Data

Translation DT3388--12-bit A/D converter). The simultaneous sample-and-hold

(5 nsec) feature of this A/D converter ensured that no phase errors were

introduced when multiple electrode signals were digitized.

The recording system was calibrated by superimposing an electrical signal

on the animal (either a sinusoid or a square wave) that could be recorded from a

microelectrode placed in the cochlea. Sinusoidal stimuli were generated and

attenuated as described for sound stimulation. The only difference was that the

attenuator output was directed to the preparation (which was lifted from ground

with a 544 12 resistor) instead of the speaker. Frequency response curves were

measured by directing the output of an electrode into the lock-in analyzer.

Square waves were generated with a calibrator (Bioelectric Instruments,

model CA5) synchronized to the initiation of A/D conversion with a constant

delay. One to two hundred square-wave responses were averaged.

Figure 3-6 shows the results of an in vivo recording system calibration.

The magnitude (panel A) and phase (panel B) of the frequency response curves of






- 66 -


two microelectrodes are shown. One was a movable electrode (STI--squares and

ST2--circles) and the other was a fixed electrode (SL--triangles). Two separate

determinations were made for the movable electrode. One measurement was made

near the surface of the air/perilymph interface and the other was made after

advancing 600 jpm to a position near the modiolus. When the pipet was first

placed in the fluid over scala tympani (STI--squares), it had a high-frequency

cutoff of 3.3 kHz with a 7-11 dB/octave rolloff, a total phase lag of 1.9 radians

(~ 1090) at 5 kHz, and a 10-90% rise time of 111 pisec (with a 9% overshoot). The

tip of the pipet apparently broke during the electrode advance (a good indicator

of this being a reduction in noise level), so the capacitance was recompensated

before the second determination was done. After recompensating (ST2--circles),

the high-frequency cutoff extended beyond 5 kHz. the phase lag at 5 kHz was

reduced to 0.6 radians (~ 290), and the rise time was reduced to 40 pUsec (with a

22% overshoot). The fixed micropipet (SL--triangles) was intentionally broken

back 50-100 pim to a tip diameter of 5-10 pm before it was placed in the fluid

over scala tympani. Thus, this electrode had a flat frequency response out to

5 kHz, a phase lag of 1.2 radians (~ 630) at 5 kHz, and a rise time of 49 jpsec

with only a 7% overshoot.

Since the calibration procedure just described was time-consuming,

requiring at least 10-15 minutes to complete, a faster method for setting the

capacitance compensation was employed that accomplished the same goal. A

continuous 1 kHz sinusoidal signal (1.5 mVrms) was superimposed on the animal

while the capacitance was compensated for each electrode to give a unity gain

response at this frequency. No corrections for the recording system were made to

the CM presented in this study. The CM magnitude should be accurate to within

2 dB out to 1.6 kHz, but may be off by +5 to -10 dB at 5kHz. A cumulative

phase lag is introduced by the recording system that is 0.1 radian at 50 Hz,

0.25-0.5 radian at 1.6 kHz, and 0.5-2.0 radians at 5 kHz.






- 67 -


The impedance of an electrode is composed of both a resistive and a

capacitative component. The resistive component is dominated by the size of

electrode tip and the resistivity of the electrolyte. The capacitative component is

distributed along the entire tip and shank of the electrode that is immersed in

fluid. Since the depth of an electrode varied from 20-600 um as it was being

advanced or -withdrawn, the effective capacitance of the electrode should vary

during a data collection track. This change in impedance could then lead to a

change in frequency response. To determine the effect of electrode depth on the

frequency response of the electrode, frequency response curves were determined

for an electrode at two positions in the cochlea (figure 3-7). The magnitude of

the frequency response curve of the electrode when it was deep in scala tympani

was 0.4-0.8 dB less than the frequency response at the surface of the fluid in the

0.05-3.0 kHz range, and was 1-2 dB less in the 3-5 kHz range. Only at the

highest frequencies were there any significant differences in the phase curves.

In addition, square-wave responses were averaged as the electrode was

withdrawn in 20 Am intervals to determine the depth range over which any

changes in the response occurred (panel C). Ten square-wave responses were

selected for display on the same plot to show the progression of the response.

The square-wave response did not change until the electrode was withdrawn to a

depth of about 400 jim. As the electrode was withdrawn further, both the

transient and steady-state part of the response grew larger, consistent with the

increase in the frequency response curves measured at the surface of the fluid

relative to the modiolar location.



Signal Averaging

At each stopping point along the electrode track one or more acoustic

stimuli were presented to the animal and the evoked responses picked up by the






- 68 -


fixed and movable electrodes were simultaneously averaged (usually 25 repetitions

at a presentation rate of 8-10/second). Five hundred and twelve points per

electrode signal were digitized at a rate that depended on stimulus frequency. A

digitization rate of 100 usec/point was used for 50 and 100 Hz, 50 Asec/point for

200 Hz, and 25 Asec/point for frequencies 300 Hz and above. The stimulus

duration was 40 msec (with a 2.5 msec rise/fall time unless stated otherwise)

except when a 100 usec digitization rate was used. In the latter case the stimulus

duration was increased to 60 msec so that the stimulus remained on throughout the

51.1 msec A/D conversion period.

In order to record both the baseline DC potential and evoked response,

custom-built synchronization circuitry was interfaced between the computer and

electronic switch so that the electrical signal to the speaker was gated on 2 msec

after the initiation of A/D conversion. The electronic switch was set so that the

electrical signal was gated at a positive zero crossing. By observing the gated

electrical signal with a very fast rise time, the positive zero crossing could be set

to within a few degrees of zero. The times of peak positivity of the electrical

signal, telec during the evoked potential waveforms to an I kHz stimulus were

computed as follows:



teiec (/) 2 + T + nT (3-10)



where n=0,1,2.... and T-1 The times of peak rarefaction at the tympanic

membrane, trar during the evoked potential waveforms were then computed with

the following formula:



trar (f) tlec (f) (3-11)
n A 21r






- 69 -


where 0(f) is peak rarefaction (in radians) relative to peak positivity of the

electric signal into the speaker (obtained from the acoustic calibration procedure).

In many of the evoked potential traces presented in this chapter, a solid vertical

line is plotted to indicate the time of peak rarefaction during one of the last few

cycles of the response.

The magnitude and phase of CM was determined from the evoked potential

waveforms by computing the discrete Fourier transform coefficients associated

with the frequency of the stimulating tone. The discrete Fourier transform was

computed over the steady-state portion of the evoked potential waveform,

beginning 3 msec after the onset of the stimulus (i.e., 5 msec into the trace) and

for a duration that included as many complete stimulus cycles as possible. Since

the cosine wave that was constructed to compute the discrete Fourier transform

was in phase with the oscillator, the CM phase values that were returned from

the transform algorithm were relative to the electrical input to the speaker. The

acoustical phase shift, 0(f), was then subtracted from the CM phase in order to

express the CM as peak positivity of CM relative to peak rarefaction at the

tympanic membrane.




Results



Scala Media Evoked Potential Waveforms

Evoked potentials from scala media of the second turn were recorded from

two animals. The approximate characteristic frequency (CF) of this location is

3 kHz. Data are presented in figures 3-8 and 3-9 from an animal (GPD7) with

initial CAP thresholds of 40-52 dB SPL in the 0.5-5.0 kHz range that did not

increase following otic-capsule surgery. Although thresholds deteriorated by

6-12 dB during the course of the experiment, the endocochlear .potential (EP)






- 70 -


remained steady at +88 mV until the animal died at the end of the experiment

(negative EP: -30 mV). Results from this experiment are shown in figure 3-8

over the 100-800 Hz range (80 dB SPL). Potentials from the fixed electrode that

remained outside the spiral ligament were recorded simultaneously and are shown

below the corresponding scala-media evoked potentials.

Note that the cochlear microphonic (CM) in scala media is about six to ten

time larger that the fixed electrode CM (the voltage calibrations are different) and

that the scala-media CM lags the CM recorded outside the spiral ligament (the

vertical solid line representing the time of peak rarefaction provides a vernier).

The details of the phase lag are given in figure 3-9 and show that the amount of

lag tends to increase with frequency. The rarefaction line also illustrates the

progressive phase lag with frequency of the CM as the peak of the traveling

wave moves from the extreme apex of the cochlea towards the recording location

in the second turn. Because of the progressive phase lag of the CM in scala

media with respect to the fixed electrode, the total phase lag that accumulates in

the 100-800 Hz range is larger in scala media (-150 at 100 Hz to +300 at 800 Hz:

a 1800 phase shift) than outside the spiral ligament (-125 at 100 Hz to +900 at

800 Hz: a 1450 phase shift). Phase-versus-frequency curves comparing scala

media with other electrode locations in the second turn are plotted in figure 3-20

(scala-media data from GPD7 shown with crosses and diamonds).



CM and Evoked Potential Waveforms in Scala Vestibuli

Evoked potentials from scala vestibuli of the second turn were recorded

from 7 guinea pigs. In the GPA7 experiment to be presented, initial CAP

thresholds varied from 40 to 55 dB SPL in the 0.5-5.0 kHz range, were raised

10-12 dB after making the otic-capsule exposure, but remained stable during the

two-hour data collection period. The EP was not encountered during this






- 71 -


experiment. Several examples of DC profiles obtained from this animal are shown

in chapter 2 (figure 2-11).

Intensity functions spanning 30 dB were obtained in this experiment over

the 50-1600 Hz frequency range. During a single electrode excursion, four

intensities of a pure tone stimulus were presented at each depth along the track

(20 Am intervals) while potentials were simultaneously averaged from the movable

and fixed electrodes. Figure 3-10 shows the magnitude (top panels) and phase

(bottom panels) of the CM determined from the evoked potentials by calculating

the pair of discrete Fourier transform coefficients associated with the frequency

of the stimulus. The data are normalized so that CM magnitude profiles across

the 30 dB intensity range can be compared on the same plot. The absolute CM

magnitudes at the normalization depths are plotted in the input-output functions

shown in figure 3-18 (scala-vestibuli data from GPA7 shown with squares and

circles).

The general shape of the magnitude profile does not change in the

50-800 Hz frequency range over the intensity range tested. The CM magnitude

increases by a factor of two over a 100 pm distance to a plateau value in scala

vestibuli that does not change by more than 5-10% in the last 200 Am of the

track. The 1600 Hz profiles are different in that the CM magnitude continues to

increase along the final 200 Aim length of the track. The CM data from the fixed

electrode outside the spiral ligament are shown below the corresponding scala-

vestibuli CM data. The fixed electrode data are normalized relative to the

magnitude of CM picked up by the fixed electrode when the movable electrode

was at the starting point of the track (i.e.. depth 0 pm). As the movable electrode

is advanced or withdrawn, the magnitude of the CM picked up by the fixed

electrode generally does not change by more than 5%.






- 72 -


The 400 Hz profiles illustrate what can occasionally happen when data are

collected along electrode excursions advancing through the spiral ligament and into

scala vestibuli. The abrupt increase in CM magnitude probably resulted from the

electrode bending and then suddenly penetrating through the spiral ligament.

Electrode bending could be observed on several occasions through the dissecting

microscope as the electrode was manually advanced. To prevent such artifacts

from appearing in the CM profiles, data were only collected while the electrode

was withdrawn in all experiments subsequent to GPA7.

The phase of CM relative to rarefaction is shown below the magnitude

data. The CM phase recorded by the movable or the fixed electrode does not

change by more than a few degrees as the movable electrode is advanced or

withdrawn, except at 1600 Hz where the phase of CM in scala vestibuli lags the

CM outside the spiral ligament by 10-200. The intensity-dependence of these

radial phase shifts of the CM in scala vestibuli is summarized in figure 3-11 for

GPA7 and one other scala-vestibuli experiment (GPA6). Radial CM phase shifts

larger than 50 are only seen at the lowest (50 Hz) and highest frequencies

(1600 Hz).

Also evident in figure 3-10 is a small CM phase shift with intensity

recorded by both the fixed and movable electrodes. This phase shift is

frequency-dependent such that at 50 Hz there is a small phase lag of about 100 as

intensity increases, whereas at 800 and 1600 Hz there is a much larger phase lead

of 20-50o with intensity. At intermediate frequencies, the phase first lags, then

switches to a phase lead, the switching intensity being 95 dB SPL at 100 Hz and

75 dB SPL at 200 and 400 Hz. These intensity-dependent phase shifts of the CM

in scala vestibuli are also illustrated in the input-output functions of figure 3-18

(squares and circles) and figure 3-19.






- 73 -


Evoked potentials were selected at 80 pm intervals from the scala-vestibuli

experiment (GPA7) of figure 3-10 and are shown in figure 3-12. The vertical

solid line representing peak rarefaction at the tympanic membrane illustrates that

the CM in scala vestibuli (figure 3-12) is nearly in phase with the CM recorded

in scala media (figure 3-8). Phase-versus-frequency responses comparing scala

vestibuli with other electrode locations in the second turn are shown in

figure 3-20 (scala-vestibuli data from GPA7 shown with squares and circles).

Another feature of the evoked potentials in scala vestibuli in the 50-200 Hz

range is the presence of transient potentials during the negative phase of the first

cycle, predominantly negative-going, and 1.0-1.5 msec in duration. The transient

potentials do not change polarity with position of the electrode, nor do they

change appreciably in amplitude. These properties identify the transient

potentials as compound action potentials (CAPs), the result of synchronous action

potentials in a large number of auditory neurons. The time of peak transient

negativity is highlighted with vertical dotted lines in figure 3-12 (labeled APV).

This terminology for describing the CAPs is based on the displacement of the

basilar membrane as inferred from the CM (Bekesy, 1951b). Positivity in scala

media and scala vestibuli is associated with basilar membrane displacement

towards scala tympani, whereas negativity in scala media and scala vestibuli is

associated with basilar membrane displacement towards scala vestibuli. An AP,

can also be seen during the negative phase of the first cycle of the fixed

electrode waveform at 100 and 200 Hz in the scala-media experiment of

figure 3-8 (arrows). A small potential can even be seen in the 100 Hz scala-

media waveform at the same latency as the CAP on the fixed electrode.

Figure 3-13 shows evoked potentials recorded at locations deep in scala

vestibuli in response to tones over a 30 dB intensity range that were selected from

the same data collection tracks represented in figures 3-10 and 3-12. To highlight






- 74 -


the shape of the evoked potentials, each waveform has been plotted on a scale

that maintains the size of the evoked potentials relatively constant across intensity.

At the highest intensity presented (105 dB SPL) the 50 and 100 Hz CM appears

more square-wave-like than sinusoidal, presumably due to saturation of the

receptor cell transduction mechanism (Hudspeth and Corey, 1977). The only CAPs

that can be distinguished in the intensity range presented are above 85 dB SPL in

the 50-200 Hz range and at 104 dB SPL at 300 Hz.



CM and Evoked Potential Waveforms in Scala Tympani

Evoked potentials from scala tympani of the second turn were recorded

from 9 guinea pigs. In a given experiment, either 3 or 4 frequencies at

approximately the same intensity (5 experiments), or four intensities of the same

frequency (4 experiments) were presented at each depth along a radial data

collection track (usually 600 pm in length with 20 jim steps). In most experiments

data were collected for only a few frequencies (200, 400, and 800 Hz). However

in one experiment, GPA9, intensity functions were obtained at 50, 100, 200, 300,

400, 600, 800, and 1600 Hz. This animal had 5 iV CAP thresholds of 30-52 dB

SPL in the 0.5-5.0 kHz range at the beginning of the experiment, did not have

thresholds raised following the otic-capsule surgery, and had thresholds raised by

no more than 10 dB after all the data to be presented were collected (about

2 hours later). The initial advance of the micropipet encountered a healthy

+80 mV EP. Subsequent excursions were made along an electrode track that was

below the basilar membrane and along which no more large DC shifts were

encountered. Several examples of DC profiles recorded from this animal are

presented in chapter 2 (figures 2-5 and 2-6). All the data from this experiment

were collected while the electrode was withdrawn from scala tympani.






- 75 -


The magnitude and phase of the fundamental component of the scala-

tympani evoked potentials were calculated and the results presented in figure 3-14

in the same normalized fashion as the scala-vestibuli data of figure 3-10. The

absolute CM magnitudes at the normalization depths are also plotted in the input-

output functions shown in figures 3-18 and 3-19 (scala-tympani data from GPA9

shown with triangles and plus signs). The profiles of figure 3-14 demonstrate that

the magnitude and phase of CM change smoothly as a function of electrode depth

across the entire frequency and intensity range tested. The CM profiles for

400 Hz tones and lower all share a common pattern of change along the radial

electrode track. As the electrode is withdrawn, the CM magnitude gradually

decreases in the 560-400 pm depth range and then rapidly decreases to a

minimum value near the middle of the track (depth 320 Apm). A very large phase

shift (130-1800) and magnitude increase are then recorded as the electrode is

withdrawn past the location of the CM minimum. As the electrode approaches

the starting point outside the spiral ligament, the magnitude of the CM levels off

and the phase remains relatively constant. In contrast to the low-frequency CM

pattern, the profiles from 800 and 1600 Hz are characterized by relatively a large

CM deep in scala tympani that decreases and gradually goes through a 20-600

phase lag as the electrode is withdrawn (figure 3-14--continued).

At nearly all frequency/intensity combinations, the phase of the CM deep

in scala tympani leads the superficially-recorded CM. The exception to this rule

is at low intensities in the 200-400 Hz range where the CM deep in scala tympani

lags the superficially-recorded CM. These radial CM phase shifts in scala

tympani are summarized in figure 3-15 for the GPA9 experiment and for two

other experiments (GP94 and GP96). The vertical dotted line in each figure

identifies a break frequency that divides the scala-tympani phase shift plots into

two regions. At low frequencies (below 400-800 Hz) the relative phase of the CM






- 76 -


between the two recording locations is quite large, 90-1800. In the 40-60 dB SPL

range the CM deep in scala tympani lags the superficial CM. Above this

intensity range, the radial phase lag switches to a large phase lead. Above the

break frequency out to about 3 kHz, the CM phase difference between these two

recording locations is less than 600. Beyond 3 kHz, large radial phase shifts can

be seen in the GP96 experiment and also in the frequency response curve from

GPA9 shown in figure 3-21.

Intensity-dependent changes in the phase of the CM with respect to

rarefaction at the tympanic membrane are present at all frequencies in

figure 3-14. As intensity increases, the CM phase recorded at locations deep in

scala tympani (except at 50 Hz) becomes more negative relative to rarefaction. In

other words, there is an intensity-dependent CM phase lag, the amount of the

phase lag varying with frequency. At 800 and 1600 Hz there is a 350-450 phase

lag that can also be seen in the evoked potential intensity functions shown in

figure 3-17. The phase of the CM also lags with intensity at superficial locations,

except at 200 and 400-Hz where an intensity-dependent phase lead is observed.

Evoked potential waveforms were selected at 80 pm intervals from the

scala-tympani experiment (GPA9) of figure 3-14 and are shown in figure 3-16.

The vertical solid line indicating the time of peak rarefaction at the tympanic

membrane provides a convenient landmark to visualize the radial phase shifts of

the CM, especially at 600 Hz and above where the phase shifts are small. The

rarefaction line also demonstrates that the superficially-recorded CM outside the

spiral ligament bordering scala tympani is nearly in phase (250) with the CM in

scala vestibuli, except at 600 and 800 Hz where the scala-vestibuli CM lags the

superficial scala-tympani CM by 1350 and 860, respectively. On the other hand,

if the phase of the CM deep in scala tympani is compared to the phase of the

CM in scala vestibuli, the phase difference is 1800 300 over the 50-800 Hz range.





- 77 -


These details are more graphically presented in the phase-versus-frequency curves

of figures 3-20 and 3-22 (scala-tympani data from GPA9 shown with triangles

and plus signs).

Another feature of the evoked potentials in scala tympani is the presence

of transient potentials that have characteristics similar to the CAPs observed in

scala vestibuli. Two classes of negative-going transients are distinguished in

evoked potentials recorded from scala tympani: (1) those synchronous with the

positive phase of CM recorded deep in scala tympani (AP,--vertical dotted lines in

figure 3-16) and (2) those synchronous with the negative phase of CM (APt--

vertical dashed lines in figure 3-16). Since the CM in scala tympani is

approximately 1800 out of phase with the CM in scala vestibuli, this nomenclature

is consistent with that introduced for the CAPs in scala vestibuli, i.e., AP, is

synchronous with displacement of the basilar membrane towards scala vestibuli,

whereas APt is synchronous with basilar membrane displacement towards scala

tympani. As was observed in scala vestibuli, the CAPs in scala tympani do not

change polarity with position of the electrode, nor do they change appreciably in

amplitude.

An AP, is clearly present in the evoked potential in response to 50-600 Hz

85 dB SPL tones. In the 50 and 100 Hz waveforms, AP, only appears during the

first CM cycle. Although AP, is present in every cycle in the 200-600 Hz range,

the potential adapts with each successive cycle. At 800 Hz only a slight

deflection during the rising positivity is present and at 1600 Hz there is very

little distortion due to CAPs, resulting in a nearly sinusoidal CM waveform. At

85 dB SPL. APt is present in the first half cycle at 50 Hz and 100 Hz, in the

second cycle at 200 Hz, and in every cycle at 300 Hz.

Figure 3-17 shows evoked potentials recorded at locations deep in scala

tympani in response to tones over a 30 dB intensity range. Note that most of






- 78 -


these intensity functions cover a 10 dB lower intensity range than the scala-

vestibuli data presented in figure 3-13. To highlight the shape of the evoked

potentials, each waveform has been plotted on a scale that maintains the size of

the evoked potentials relatively constant across intensity.

The threshold, latency, and shape of the AP, and APt potentials are a

complex function of frequency and intensity. The threshold of APV is lowest in

the 200-400 Hz range, being present at the lowest intensity tested (65 dB SPL), but

does not consistently appear until 75 dB SPL at 50 and 100 Hz. At 600 and

800 Hz, the distortion of the positive phase of the CM associated with AP, at

these frequencies does not become evident until about 85 dB SPL. although it is

possible to see AP, at 65 dB SPL in superficially-recorded 600 Hz waveforms

where the CM is relatively small (not shown).

The latency of AP, elicited during the first stimulus cycle decreases with

intensity at all frequencies. At 300 and 400 Hz where APy is present in every

cycle, the latency shift also occurs in these subsequent cycles. The net effect of

this intensity-dependent latency shift and CM phase lag in the 200-400 Hz range

is to move the position of AP, from the falling phase of the CM at 65 dB SPL to

the peak of the positive phase at 95 dB SPL. Another feature that is especially

evident in the 200-400 Hz waveforms is a change in shape of AP., the negative

peak becoming less sharp at high intensities. At 300 Hz, the duration of the

transient negativity also increases at high intensities.

An AP, is only present in the 50-300 Hz range. It can be seen at 75 and

85 dB SPL in the 50 Hz waveforms, at 65 and 75 dB SPL at 100 Hz, and at 85

and 95 dB SPL at 200 and 300 Hz. Even though the distortion of the negative

wave of the CM at 400 and 600 Hz is similar to that seen at 200 and 300 Hz, the

distortion is not large and sharp enough to be reliably distinguished from the CM.





- 79 -


CM Input-Output Functions

Input-output functions of the CM recorded from several locations in the

second turn are shown in figure 3-18 (magnitude--top panels, phase--bottom

panels). The cochlear cross-section of figure 3-1 illustrates the approximate

recording location associated with each of the symbols used in this figure. Four

input-output functions are shown for each frequency. Two are from the GPA7

scala-vestibuli experiment and two are from the GPA9 scala-tympani experiment.

In each experiment data were selected at two locations for display: a location

deep in scala tympani or scala vestibuli (squares and triangles, respectively) and a

location outside the spiral ligament bordering scala tympani or scala vestibuli

(circles and plus signs, respectively). Also included in the 100-800 Hz plots are

the 80 dB SPL CM recordings from scala media (crosses) and from the fixed

electrode outside scala media (diamonds).

A comparison of the input-output functions obtained from the two

endpoints of each track helps to explain the intensity dependence of the

normalized CM magnitude profiles in scala tympani (figure 3-14) and the intensity

independence of the magnitude profiles in scala vestibuli (figure 3-10). The pair

of CM magnitude input-output functions from the two locations in the scala-

vestibuli experiment are parallel to one another over the frequency and intensity

range tested, a result that is consistent with the normalized CM magnitude

functions that superimpose in figure 3-10. On the other hand the input-output

functions from scala tympani are not parallel to each other in the 100-800 Hz

range. This implies that the ratio of the CM recorded outside the spiral ligament

bordering scala tympani relative to the CM deep in scala tympani changes as a

function of intensity. Figure 3-14 shows that this ratio varies from 0.1-0.8

depending on both frequency and intensity. At 50 and 100 Hz, the CM ratio

decreases as intensity is increased. This is explained by noting that the CM





- 80 -


input-output function recorded by the SL electrode has a slope less than unity

while the CM input-output function deep in scala tympani is still in the linear

range. At 200 and 400 Hz, this is also true in the 65-80 dB SPL range. Above

this intensity, the ratio begins to increase and at 800 and 1600 Hz the ratio

increases with intensity over the entire intensity range studied. The reason for

the increase in the ratio of the superficial CM relative to the CM deep in scala

tympani is that once the CM reaches the limit of linearity, the slope of the input-

output function decreases with intensity and that a greater decrease with intensity

occurs deep in scala tympani than at the superficial location.

The CM is largest in scala media over the entire frequency range tested

(figure 3-18--top panels). Below 400 Hz, the CM within scala vestibuli and

outside the spiral ligament bordering scala vestibuli is larger than the CM in scala

tympani. The smallest CM is recorded from the location outside the spiral

ligament bordering scala tympani, near the location where the CM is at a

minimum. At some frequency in the 800-1600 Hz range, the relative magnitude

of the CM in scala vestibuli versus scala tympani changes so that at 1600 Hz the

CM in scala tympani exceeds that in scala vestibuli. The plots of CM phase

versus intensity (figure 3-18--bottom panels) demonstrate that the microphonics in

scala media, scala vestibuli, and outside the spiral ligament bordering scala

tympani are all nearly in phase up to 400 Hz.

Input-output functions were selected at locations deep in scala tympani and

scala vestibuli (near the end of the track where the CM is largest) and are shown

in figure 3-19. The CM magnitude input-output functions (top panels) are linear

up to a certain sound pressure level. Above the limit of linearity the slope is

less than unity and decreases with intensity, although the slope remains positive

even at the highest intensities tested (100-115 dB SPL). Figure 3-19 illustrates the

frequency dependence of the limit of linearity of the CM in the second turn.






- 81 -


This limit is about 95 dB SPL at 50 Hz, decreases to 85 dB SPL at 200 Hz.

decreases to 75 dB SPL at 600 and 800 Hz, and at 1600 Hz the limit is lower that

65 dB SPL. The limit of linearity is about 5-10 dB lower in scala vestibuli. The

intensity-dependent phase shifts of the CM that were described in figures 3-10

and 3-14 are also evident in figure 3-19 (bottom panels). In scala tympani the

CM phase progressively lags with intensity, except at 50 Hz where a small phase

lead is observed. The phase lags are larger at higher frequencies resulting in a

larger phase accumulation versus frequency at higher intensities. In scala

vestibuli the total phase accumulation is actually reduced at the highest intensities

due to a significant phase lead with intensity at 1600 Hz and the absence of any

phase shifts at lower frequencies.



CM Frequency Response Curves

The data presented in the previous figures are recast in figure 3-20 to

show the frequency response of the CM recorded from six locations in the second

turn at a constant 80 dB SPL. Low-frequency CM magnitude responses curves

(below 300 Hz) recorded deep in scala tympani and in scala vestibuli grow at a

rate of 4-6 dB/octave. Since 80 dB SPL is within the limit of linearity in this

very low-frequency range, the same low-frequency slope would be expected at

lower intensities. All of the phase-versus-frequency curves have shallow slopes

up to 300 Hz, but above this frequency large phase lags accumulate. Note that in

the scala-media phase curves (figure 3-20. crosses and diamonds) and the scala-

tympani curve (triangles) there are small phase leads from 100 Hz to 200 Hz.

This phase lead was also present in both of the frequency response curves that

were measured with the lock-in analyzer (figure 3-21).

An inflection in the CM magnitude-versus-frequency curve recorded by the

electrode outside the spiral ligament bordering scala tympani is evident in





- 82 -


figure 3-20. Figure 3-21 shows that a notch was present in a 70 dB SPL

frequency response curve obtained with the lock-in analyzer from the same

animal (GPA9) at the spiral ligament location (SL(ST)--plus signs) and also from

animal GP94 at 60 dB SPL. The inflection or notch in the magnitude function is

associated with a large phase shift as the stimulating tone is swept through this

frequency region.

A closer look at the relative phase of the CM recorded from different

locations in the second turn is shown in figure 3-22. Phase values are plotted in

degrees relative to the phase of the CM deep in scala tympani. At frequencies

400 Hz and below, the relative phase values range from 135-1800 lagging and

from 1550-180 leading the CM deep in scala tympani. At 800 Hz, a near 1800

phase shift is maintained across the cochlear partition (i.e., scala-media phase re

scala-tympani phase--crosses). The large phase shift that is present along radial

tracks in scala tympani at low frequencies is replaced by a gradual 50 radial

phase shift in scala tympani (SL(ST)--plus signs) and a vertical CM phase shift

along the lateral wall. between scala tympani (SL(ST)--plus signs) and scala

vestibuli (SL(SV)--circles).



Modulation of the Standing Current

Of great interest is the relation between the standing currents presented in

chapter 2 and the spatial variations of CM presented in this chapter. Radial

tone-evoked potential gradients were calculated from the data collected along

scala-tympani and scala-vestibuli tracks by computing the first spatial derivative

of the field potentials, the same calculations that were used to compute click-

evoked radial potential gradients (Brownell et al., 1983). The acoustically-evoked

potential gradients represent a modulation of the radial standing current (computed

from an averaged DC potential profile). Standing currents in scala vestibuli (top)






- 83 -


and scala tympani (bottom) and their modulation with a 200 Hz stimulus are

shown in figure 3-23. The spatial profile of the modulation of the radial

potential gradient is frozen at two points in time, the two points chosen being

maximal displacement of the cochlear partition toward scala tympani (triangles)

and the other maximal displacement of the cochlear partition toward scala

vestibuli (circles).

The figure illustrates the reciprocal nature of the modulation of potential

gradients in scala tympani and scala vestibuli. During scala-vestibuli displacement

of the cochlear partition (circles), the modulation of the potential gradient is in the

positive direction in scala vestibuli (i.e., toward the modiolus), whereas in scala

tympani the modulation is in the negative direction (i.e., toward the spiral

ligament). During displacement of the cochlear partition toward scala tympani

(triangles), the direction of potential gradient modulation is reversed relative to

scala-vestibuli displacement: the potential gradient is modulated in the negative

direction in scala vestibuli (i.e., toward the spiral ligament), whereas in scala

tympani the modulation is in the positive direction (i.e., toward the modiolus).

A qualitative difference between radial potential gradient modulation in

scala tympani and scala vestibuli is the relation between the location of the

standing current peak and the place where maximal modulation of the radial

current occurs. In the scala-vestibuli experiment shown, the peak of the radial

standing current is at depth 240 /pm (squares). During peak cochlear partition

displacement toward scala vestibuli, the peak of the standing current through the

scala-vestibuli pathway is reduced by a factor of two (circles). During peak

cochlear partition displacement toward scala tympani, the peak of the radial

standing current through the scala-vestibuli pathway is augmented by a factor of

two (circles). Thus, at least with acoustic stimuli of moderate intensity

(80 dB SPL) the actual direction of radial current flow does not change in scala

vestibuli.






- 84 -


In contrast, the peak of the radial standing current in scala tympani

(depth 160 Am) is not significantly modulated during acoustic stimulation. Peak

modulation of the radial potential gradient is located about 200 pm modiolar to the

standing current peak, a location where there is little or no radial component to

the standing current. Thus, at this location in scala tympani (depth 360 Am) the

current flow actually changes direction during the different phases of the

sinusoidal stimulus. During scala-vestibuli displacement a current is directed

toward the spiral ligament, whereas during scala-tympani displacement a current

is directed toward the modiolus.




Discussion



Comparison with Previous CM Studies

The results presented in this chapter provide the first detailed

characterization of tone-evoked field potentials in the cochlea. By recording at

finely-spaced intervals along radial tracks, it is shown that the cochlear

microphonic (CM) waveform changes smoothly in the perilymphatic spaces of the

cochlea and within the spiral ligament. The changes in magnitude and phase of

the CM observed along radial tracks in scala tympani and scala vestibuli provide

part of the missing link between the large differences in the magnitude and phase

of the CM in scala tympani versus scala vestibuli that have been characterized

with differential electrodes (Tasaki et al., 1952; Dallos, 1973). The CM should

also change along a vertical path through the spiral ligament to account for the

difference in CM magnitude that is present between superficial locations in scala

tympani and scala vestibuli (a factor of about three).





- 85 -


The low-frequency pattern of CM in the second turn of the guinea pig can

be summmarized by the pair of schematic diagrams in figure 3-24. The

modulation of the potential field is frozen at two points in time (left--CM positive

in scala media; right--CM negative in scala media) and the magnitude and polarity

of the CM is shown by size of the plus and minus signs. The figure illustrates

that the CM is of the same polarity in scala media, scala vestibuli, and throughout

most or all of the spiral ligament. Two large phase shifts are evident: (1) across

the reticular lamina (i.e., the boundary formed by the apical membranes of the

cells of the organ of Corti) and (2) near the border of the spiral ligament in scala

tympani. The phase shift across the reticular lamina has been recognized since

the days of Tasaki et al. (1954). What is quite surprising is that the low-

frequency phase reversal of the CM in scala tympani of the second turn has not

been described previously despite many years of differential electrode recording.

Dallos et al. (1971) recorded the phase difference of the CM between their

standard scala-tympani and scala-vestibuli differential electrodes. In addition to a

normal low-frequency phase difference of 1800, they found that in some animals

the CM recorded by their differential electrodes was in phase at low frequencies.

They interpreted this in-phase pattern as the result of electrical crosstalk between

second-turn scala vestibuli and third-turn scala tympani. In light of the large

radial phase shifts of the CM that are present in scala tympani, the dichotomy of

differential electrode phase responses most likely results from scala-tympani

electrode placements that are sometimes on the modiolar side of the phase reversal

and at other times lateral to the phase reversal.

The theory of the differential electrode technique is based on the

assumption that the source of the CM at a particular point in time can be

represented by a dipole (Dallos, 1969). An infinite array of such dipoles

longitudinally arranged in a homogeneous conductor would generate a virtual






- 86 -


ground plane midway between the dipole sources. However, the insulating

membranes of scala media and the presumed inhomogeneity of the spiral ligament

with respect to perilymph will distort this idealized model of a virtual ground

plane. Instead, what might be predicted along the lateral wall is a two-

dimensional virtual ground surface, locally orthogonal to the flow of microphonic

current and located somewhere along the lateral wall between scala tympani and

scala vestibuli. The presence of a large phase shift and minimal CM magnitude

along tracks in scala tympani suggest that the electrode track has cut through such

a two-dimensional surface at this location.

A similar argument could be made for a two-dimensional virtual ground

surface in the modiolus. However, the largest changes in CM occur near the

lateral wall and not near the modiolus, arguing that the dominant pathway for

current flow is through the spiral ligament. A functional consequence of a

dominant flow of current through the lateral wall is that a relatively large CM

can be generated at modiolar locations in the extracellular spaces outside the hair

cells and auditory nerve dendrites. Such an influence of the CM on inner hair

cell receptor potentials has previously been used to explain an apparent

discrepancy in the low-frequency excitatory phases of inner hair cells and the

neurons that innervate the cochlear base (Russell and Sellick, 1983; Ruggero and

Rich, 1983).



Relation of Microphonic Current to Standing Current

Arguments are presented in the discussion of chapter 2 (Radial DC

Gradients in Scala Tympani and Scala Vestibuli) that identify the peak of the DC

potential gradient as most likely being associated with that part of the electrode

track within the spiral ligament. Since the maximum modulation of the standing

current in scala vestibuli occurs at the location of the DC gradient peak






- 87 -


(figure 3-23), the peak of the radial microphonic current is also probably located

in the. spiral ligament bordering scala vestibuli.

Anecdotal evidence supports the presence of large potential gradients in the

spiral ligament. Micropipets were observed through the dissecting microscope as

they were advanced into scala vestibuli. Sometimes the electrode would bend as

it made contact with the spiral ligament and then straightened out after

penetrating through. A consequence of such bending and electrode slippage

during a data collection excursion is that the tip of the pipet may remain at the

same depth for several 20 pim advances of the microdrive spindle and then

suddenly advance 40-100 fIm on a subsequent depth increment. If radial potential

gradients exist within the spiral ligament, then such irregularities in the travel of

the electrode through the spiral ligament would give rise to abrupt changes in the

measured DC potential and CM profiles. Discontinuities were found at lateral

locations of the CM magnitude profiles along several forward data collection

tracks (e.g., figure 3-10--400 Hz) that were usually correlated with abrupt DC

shifts as well. Thus, these data are consistent with and support the interpretation

that the DC and CM potential gradient peaks are located within the spiral

ligament bordering scala vestibuli.

The assignment of the region of the CM magnitude increase to the spiral

ligament may also explain the small gradual phase shifts of the CM along scala-

vestibuli electrode tracks at the higher frequencies used in this study

(figure 3-10--800 and 1600 Hz). The CM phase shifts occur at the same location

as the CM magnitude increase, have little dependence on stimulus intensity, and

the amount of the phase shift increases with stimulus frequency. One explanation

of these gradual phase shifts is that the conductivity of the spiral ligament has a

significant reactive component. The membranes of the cells that are embedded in

the extracellular fiber matrix of the spiral ligament could be the anatomical

substrate for such a reactance.






- 88 -


Important conclusions can be drawn by comparing the CM phase profiles

in scala tympani and scala vestibuli. In light of the radial CM phase shifts that

are observed in scala vestibuli and the interpretation that they are generated by a

spiral-ligament reactance, a similar argument can be entertained for the radial CM

phase shifts in scala tympani. Since cells are distributed throughout all of the

spiral ligament, the effects of a reactive component of the spiral ligament should

be similar in both scala tympani and scala vestibuli. However, in contrast to the

scala-vestibuli CM phase shifts, the gradual CM phase shifts in scala tympani

above 600 Hz are intensity-dependent and inversely proportional to stimulus

frequency (figure 3-14, 800 and 1600 Hz). The scala-tympani CM phase shifts

also occur over a several hundred micron distance and are much larger than the

phase shifts in scala vestibuli (600 versus 250). Thus, while a spiral-ligament

reactance may contribute to small phase shifts in the lateral part of the scala-

tympani tracks, the major part of the gradual phase shifts in scala tympani

probably result from a different mechanism.

The peak of the microphonic current in scala tympani is located about

200 um modiolar to the standing current peak (figure 3-23). Since the spiral

ligament is only 100 pm wide, the DC and microphonic currents cannot both be

located within the spiral ligament bordering scala tympani. Thus, if the

assignment of the DC gradient peak to the spiral ligament is correct, then the

peak of the radial microphonic current is located within scala tympani just below

or lateral to the outer hair cells.

The presence of microphonic currents near the lateral wall in. scala

vestibuli argues that at least some of the current that is modulated by the hair

cells is shunted through the scala-vestibuli pathway. A likely pathway for

current between the hair cells and scala vestibuli is through scala tympani and

the spiral ligament. However, because of the spatial separation of the radial DC





- 89 -


and CM current peaks in scala tympani, very little radial current is modulated in

the region of the DC gradient peak, i.e., the presumed location of the spiral

ligament. This result implies that (1) either the net current through the scala-

tympani pathway does not change during acoustic stimulation or (2) that both the

magnitude and direction of the current through the spiral ligament bordering scala

tympani change such that the radial component of the current remains unchanged.

If the first possibility is correct, then alternative pathways must be considered for

current flow from scala tympani to the lateral wall of scala vestibuli.

The Henson's cells of the organ of Corti have been shown to be electrically

coupled by gap junctions (lurato et al., 1976; Gulley and Reese, 1976). Perhaps

they are just one cell type of an electrical network that provides a low-impedance

transcellular pathway for current flow from the spaces of Nuel (the large

extracellular spaces around the outer hair cells) to a region of the spiral ligament

near the stria vascularis. Such an intracellular pathway would be shielded from

the extracellular currents measured in this study and could provide part of the

pathway for the modulation of current that is shunted from the hair cells to scala

vestibuli.

If the microphonic currents in scala tympani do not make their way to

scala vestibuli, then where do they go? One possibility is that an alternative

leakage pathway of the endolymph/perilymph barrier is present lateral to the

organ of Corti and modiolar to the spiral ligament, placing it in the external

sulcus. During acoustic stimulation, there could be a reciprocal modulation of

current through the hair cells and this leakage pathway, i.e., when current is

increased through the hair cells, current through the leakage pathway is reduced

and vice versa. Such a local pathway of reciprocal currents could maintain the

net current through scala tympani relatively constant during acoustic stimulation.





- 90 -


Other evidence can also be brought forth in support of a leakage current

through the external sulcus. Subpeaks were sometimes measured in the DC

gradient profiles on the modiolar side of the major DC gradient peak. The

subpeaks may correspond to a leakage current from the region of the external

sulcus spreading out into scala tympani. Such a leakage current could explain the

presence of 5-10 mV positive DC shifts that were stable over a 20-50 pm range

along tracks in several scala-tympani experiments. On subsequent excursions

along the same track, the small positive DC shifts were replaced with the

endocochlear potential (EP) (probably because the micropipet traversed a slightly

different path through the spiral ligament) that was recorded only over a

40-60 pm range. The CM phase changed very little upon entry into scala media

and then shifted about 180 when the DC returned to the near-zero level after

exiting scala media. Such a short track through scala media with these DC and

CM characteristics could only result from the electrode passing into scala media in

the region of the spiral prominence (the area bordering scala media below the

stria vascularis), passing through the external sulcus in scala media, and exiting

into scala tympani. Thus, the electrode excursions with the 5-10 mV positive

shifts preceding the excursions that penetrated scala media likely passed just

below the external sulcus.



Reciprocal Modulation of Ionic Currents

Also shown in figure 3-24 is the displacement of the cochlear partition as

inferred from CM (Bekesy, 1951b; left--towards scala tympani; right--towards

scala vestibuli). During displacements towards scala tympani (left panel), the CM

is positive in scala media and the relatively large CM in scala vestibuli drives

current from scala vestibuli into the spiral ligament. A potential gradient is

oriented in the opposite direction in scala tympani where the movement of





- 91 -


positive charge is directed into scala tympani from the spiral ligament. The

simplest interpretation for the reduced current through the scala-tympani pathway

during cochlear partition displacement towards scala tympani is that the

transduction current through the outer hair cells is transiently reduced. Based on

the polarization of stereocilia in the cochlea and the excitatory direction of hair

bundle displacement (Hudspeth and Corey, 1977), the outer hair cell stereocilia

should be displaced towards the modiolus during cochlear partition displacement

towards scala tympani. Furthermore, the closure of transduction channels will

result in an increase of the net resistance of the endolymph/perilymph barrier. If

the stria vascularis generates a relatively constant current, then the EP will

increase concomitant with the resistance increase and drive more current through

the leakage pathways out of scala media. Thus, a leakage current powered by

the EP normally flows through Reissner's membrane and possibly the cells of the

spiral limbus in silence. During acoustic stimulation this leakage current is

indirectly modulated by the relatively large CM in scala media.

The opposite phase of stimulation is illustrated on the right (i.e., CM

negative in scala media and cochlear partition displaced towards scala vestibuli).

The direction of radial current modulation is reversed during peak negativity of

CM in scala media. During this phase of stimulation the transduction channels

are in an open state, leading to an increased current through the hair cells and

through the scala-tympani pathway. The decrease in resistance of the

endolymph/perilymph barrier transiently reduces the EP and less current is driven

through Reissner's membrane and scala vestibuli.



Comparison with Previous CAP Recordings

The evoked potential recordings presented in this chapter demonstrate that

multiple compound action potentials (CAPs) can be recorded from intracochlear






- 92 -


electrodes in the second turn of the guinea-pig cochlea. The AP, recorded in

both scala tympani and scala vestibuli of the second turn is almost certainly the

CAP originally identified by Derbyshire and Davis (1935) and more recently

labeled API by Ruggero et al. (1986). The distinguishing features of this CAP

are its relatively low threshold and activation associated with basilar membrane

velocity towards scala vestibuli. Based on an analysis of the preferred firing

times of single auditory units, Ruggero et al. (1986) proposed that this CAP is

generated by neurons in the apical regions of the cochlea. The intracellular

recordings of Dallos (1985) are also consistent with this view. For low-frequency

stimuli, apical inner hair cell depolarization leads outer hair cell depolarization by

900, implying that apical inner hair cells respond to basilar membrane velocity

towards scala vestibuli.

Ruggero et al. (1986) also recorded two other CAPs, AP2 and AP3, that

rode on the negative phase of the CM recorded from the chinchilla round

window. These authors argued that AP2 was the near-threshold response of

neurons innervating the cochlear base and that AP3 was the high-intensity

response of basal neurons (based on their observation of peak-splitting and level-

dependent phase shifts). Whether APt corresponds to either AP2 or AP3 of

Ruggero et al. (1986), or to neither, remains to be determined. It is interesting,

however, that APt is not present in the evoked potentials recorded by the scala-

vestibuli electrode, even at 105 dB SPL. Perhaps the relatively large low-

frequency CM in scala vestibuli masks the small APt. On the other hand, if APt

is generated in the basal coil, then the scala-tympani electrode may be able to

pick up this remote potential since it is closer to the presumed site of activation.

The possibility of such selectivity of intracochlear electrodes for recording

CAPs has not been discussed in the literature. Auditory nerve CAPs are almost

exclusively recorded from the round window. The current view is that the round






- 93 -


window location picks up action potentials from all cochlear fibers about equally

(Kiang et al., 1976) and is therefore capable of giving a faithful indication of

auditory thresholds across the full frequency spectrum. The rationale behind this

view is that the action potential generated by each neuron as it passes through the

modiolus of the basal turn on its way to the brain, results in a small (< 1 pV)

potential that can be recorded at the round window. The CAP response is

merely the sum of all such action potentials and models based on these

assumptions work well in reproducing CAP responses at the round window (Dolan

et al., 1983). In contrast, the action potentials from fibers that innervate the basal

turn might not be expected to spread to the second, third, and fourth turns.

Thus, it is possible that intracochlear electrodes in the upper turns preferentially

pick up CAPs from fibers originating at apical locations. In support of this

hypothesis, Davis et al. (1950) have remarked that CAPs recorded from the third

turn or apex are larger than round window CAPs for low-frequency stimuli,

resulting in CAP thresholds about 10 dB lower than a round-window electrode

below 1.5 kHz.



Capacitative Currents through the Walls of Scala Media

One explanation for the frequency-dependent phase lag of the CM in scala

media with respect to the CM in the spiral ligament is that as stimulus frequency

is increased, the modulation of current through the hair cells results in a

modulation of leakage current not only through the DC leakage pathways

(figure 2-15), but through capacitative pathways represented by the membranes of

cells that comprise the endolypmh/perilymph barrier. According to the relation

Q=CV, charge density Q is proportional to the voltage gradient V across a

membrane, the constant of proportionality being the specific capacitance C of the

membrane. By differentiating this relation with respect to time, one obtains




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ECZRVBPKC_ZUI2PX INGEST_TIME 2012-02-20T22:16:40Z PACKAGE AA00009082_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES