Polymeric Drug Delivery of Antiepileptic Drugs to Neuronal Networks Cultured on Multielectrode Arrays

POLYMERIC DRUG DELIVERY OF ANTIEPILEPTIC DRUGS TO NEURONAL
NETWORKS CULTURED ON MULTIELECTRODE ARRAYS















By

ALEX JOSEPH CADOTTE


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2004

































Copyright 2004

by

Alex Joseph Cadotte






















This document is dedicated to my parents Alfred J. & Frances M. Cadotte.















ACKNOWLEDGMENTS

I would like to thank Dr. Thomas DeMarse for his guidance, patience, and

providing a wonderful atmosphere to work in. I would like to thank Dr. William Ditto

for his contributions to this project as well as his hard work in building the foundations to

what should prove to be a top notch Biomedical Engineering Department. I would like to

thank Dr. Chris Sackellares for his many helpful suggestions, references, and knowledge

in the field of epilepsy. I would like to thank Penny He for her help with cell

preparations and keeping the cell cultures alive which were necessary for this experiment.

I would also like to thank my parents, family, and friends for their continued support and

encouragement in my return to school for a master's degree.
















TABLE OF CONTENTS

page

ACKNOW LEDGM ENTS ........................................ iv

LIST OF FIGURES ........... ......... .............. ........... vii

ABSTRACT .....................x.............. ......................

CHAPTER

1 INTRODUCTION AND BACKGROUND ................................ ...............

B ackgrou n d ...................................... ...........................................
M ulti-electrode A rrays ....................................................... 1
Bursting ........................................................4
Epilepsy ...................................... ................... ............. ......... 5
Polymeric Antiepilepsy Drug Delivery ..........................................7
The H hypothesis ............... ................... ......... ..... ....... .. ..7
Polym eric D rug D delivery System ........................................................................8

2 DEVELOPMENT OF THE PHEMA MEA DRUG DELIVERY SYSTEM...........10

Poly Hydroxyethyl M ethacrylate Background .................................. ............... 10
Design, Fabrication, and Use of the PHEMA Drug Delivery Ring.......................... 12

3 M ATERIALS ................................................. ........ 16

Cell Culture............... .......................................... 16
Dissociated Neural Cultures ......................... ........ ...........16
Development of Culture Activity ....... ..................... ......... 17
H ardw are and Softw are ............................................................17
The Multi-electrode Array...................... ....................17
Culture System ........................................ ........ 18
Physiological Environm ent ........................................ ................. 19
D ata A acquisition System ........................................ ................... 20
Spike and Burst Detection Algorithms......................... ...............22

4 TESTING OF THE PHEMA MEA DRUG DELIVERY SYSTEM..........................24

Testing the PHEMA-MEA system with bicuculline .............................................24










Experimental Procedure................... .................. ........25
PHEMA Bicuculline Testing Results and Discussion............................... ....26
Conclusions............................ ........... .......... 30

5 ANTIEPILEPSY DRUG TESTING..............................................................32

A ntiepileptic D rugs............... ................ .. ........ ..................... .. 32
Phenytoin..................................................................................... 32
Ethosuximide............... ....... .............. .................33
Valproate ................................................33
Experimental Procedures............................ .............................34
Phase 1: Effects on Spontaneous Changes ........................................ .....35
Phase 2: Effect on Stimulated Activity ...................................... ........37

6 ANTIEPILEPSY DRUG RESULTS AND DISCUSSION...................................39

Phase 1 R results ....................................................... 39
Phenytoin...................................... .................. .............. ........39
Ethosuximide.................................................................. ... ........ 42
Valproate ................................................44
Summary Phase 1 ................................................... ........47
Phase 2 Results ................................................48
Discussion and Conclusions ..................................... ......... ........50
Discussion...................................... .................. ............. ........ 50
Conclusions ................................................52

APPENDIX

A BURST DETECTION ALGORITM .................. ......... ........53

B RASTER PLOTS FOR BICUCULLINE TREATMENTS................................56

C RASTER PLOTS FOR PHENYTOIN .................................................................59

D RASTER PLOTS FOR ETHOSUXIMDE .................... .......... .......62

E RASTER PLOTS FOR VALPROATE ................. ............ ...........65

LIST OF REFERENCES ......................................................... ................68

BIOGRAPHICAL SKETCH .................................................. ........72
















LIST OF FIGURES


Figure page

1-1 Two views of multi-electrode arrays...... ......... ..... ........ .................1

1-2 This is a raster plot showing neural activity from a dissociated neural culture on a
M EA .................................... ........... ............................... ........ 5

1-3 This is a representation of a seizure from an EEG from a human epilepsy patient ...6

2-1 The hydroxyethyl methacrylate monomer .........................................10

2-2 Teflon mold and new and used PHEM rings .................................. ...........13

3-1 This is a multi-electrode array with a white Teflon lid with the FEP Teflon
membrane attached............ .... ................ 18

3-2 The micro-incubator constructed to control the environment around the MEA-60
and the neural culture during experiments .................... ....................19

3-3 The MEA-60 from Multichannel systems with the 64 channel stimulator
attached..................... ... .. ........... ......... 20

3-4 Raw electrical activity from a neural culture on an MEA.....................................21

4-1 The chemical structure of bicuculline, a bioactive agent derived from Dicentra
cucullaria .................................................24

4-2 Bicuculline average burst duration.................................... .......... 27

4-3 Bicuculline mean interburst interval ............................. ...............27

4-4 Raster plots of spike activity for dish 6682 during baseline, placebo, and
bicuculline treatment spike activity............................ ........... 28

4-5 Dish 6685 with a standard concentration of 100mmol/L bicuculline ...................29

4-6 Dish 6685 with a concentration of 10mmol/L bicuculline...............................30

5-1 The chemical structure of phenytoin....................................... 32

5-2. The chemical structure of ethosuximide....................... ..................33

vii









5-3 The chemical structure of valproate ......... .... .................................... ............ 34

5-4 Figure 8 from (Keefer et al. 2001). This figure shows the responses of neural
cultures on MEAs to various pharmacological agents...................... ..............36

6-1 Interburst Interval for Phenytoin. .. .............. .............................39

6-2 Phenytoin Burst D uration................................................... 40

6-3 Dish 6686 Phenytoin Raster Plots.............. ....................... ..................41

6-4 Interburst Interval for Ethosuximide .................................... 42

6-5 Ethosuxim ide Burst Duration.................................................... 43

6-6 Dish 6686 Ethosuximide Raster Plots...................... ..................... 44

6-7 Interburst Interval for Valproate. ............................... ............... 45

6-8 V alproate Burst D uration ............................................... ...............46

6-9 Dish 6686 Valproate Raster Plots ................................ ............... 47

6-10 Dish 6686 Response to stimulation................................ ....... 49

6-11 Dish 1534 Response to stimulation................. ...............49

B-1 Raster plots of spike activity for dish 6685 during baseline, placebo, and
bicuculline treatment spike activity...... ...................... ........... 56

B-2 Raster plots of spike activity for dish 6688 during baseline, placebo, and
bicuculline treatment spike activity............................ ........... 57

B-3 Raster plots of spike activity for dish 7087 during baseline, placebo, and
bicuculline treatment spike activity............................ ........... 58

C-i Dish 6682 Phenytoin Raster Plots................................... 59

C-2 Dish 6692 Phenytoin Raster Plots................................... 60

C-3 Dish 7073 Phenytoin Raster Plots................................... 61

D-i Dish 6682 ethosuximide Raster Plots.............................. ............... 62

D-2 Dish 6692 ethosuximide Raster Plots.............................. ............... 63

D-3 Dish 7073 ethosuximide Raster Plots.................................................. 64

E-i Dish 6692 valproate Raster Plots ................................. ............... 65

viii









E-2 Dish 7073 valproate Raster Plots ................................. ............... 66

E-3 Dish 7076 valproate Raster Plots ................................. ............... 67
















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

POLYMERIC DRUG DELIVERY OF ANTIEPILEPTIC DRUGS TO NEURONAL
NETWORKS CULTURED ON MULTI-ELECTRODE ARRAYS

By

Alex Joseph Cadotte

August 2004

Chair: Thomas DeMarse
Major Department: Biomedical Engineering

Neurons in culture fire repetitive synchronized groups of action potentials called

bursts. These in-vitro bursts are theorized to be similar to in-vivo epilepsy. Three highly

effective anitepilepsy drugs (AEDs) phenytoin, ethosuximide, and valproate were

delivered to neural cultures on multi-electrode arrays (MEAs) to determine the extent to

which in-vivo epilepsy and in-vitro bursting are similar. A poly-hydroxyethyl-

methacrylate (PHEMA) drug delivery ring was developed to deliver these AEDs to the

MEAs. The PHEMA rings were found to have no effect on the native activity of neural

cultures and proved to be an effective in-vitro drug delivery device. Phenytoin, used for

grand mal seizures, was the most effective AED eliminating native and reduced

stimulated bursts. Ethosuximide, used for petit mal seizures, and valproate, used for

various forms of epilepsy, only slowed the bursting rate and did not prevent stimulated

bursts. These findings suggest that bursting in-vitro responds in a similar way to that of

in-vivo grand mal epilepsy to these AEDs.















CHAPTER 1
INTRODUCTION AND BACKGROUND

Background

Multi-electrode Arrays

Multi Electrode Array (MEA) technology is used for a variety of different cell

cultures. Basically, any cell that creates electrical impulses from ion channel activity can

be observed and stimulated on an MEA. These arrays consist of a large number of

electrodes embedded into the surface of a specialized culture dish. Figure 1-1 shows an

example of a 60 channel (electrode) MEA from Multichannel Systems. Each of




A) B'l)1
















Figure 1-1. Two views of multi-electrode arrays. A) Normal size view of a MEA
showing the electrode array inside a glass ring, which is filled with media.
The pads around the outside edge of the MEA are connected to individual
electrodes within the array at the center of the dish. The MEA is 5cm by 5cm
B) Magnified view of the center of the MEA showing the eight by eight grid
of electrodes with neural culture present. The point to point spacing of
electrodes in the grid is 200[tm.









the 60 electrodes can detect the electrical activity, depending on electrode spacing and

geometry, produced by action potentials from electrically active tissue growing nearby

(Claverol-Tinture and Pine 2002).

Two major neural tissue preparations used on MEAs are slice cultures (Egert et al.

1998; Egert et al. 2002) and dissociated cultures (Tateno and Jimbo 1999). Slice cultures

are the preparation of a thin slice of tissue as a relatively intact representation of brain

structure. Dissociated cultures are a randomly laid down sample of neurons where

structural and connective tissue have been removed via digestion. Figure 1-1B shows an

example of a dissociated neural culture on a MEA. Neuronal cultures and hippocampal

slices are useful in the study of the electrical behavior of the brain, diseases such as

epilepsy, and the effect of drugs on the central nervous system. Much of the research

studying these areas has used single electrode patch systems where single neurons among

a small population can be studied in detail for several hours. MEA technology was

developed in the late 1970's (Thomas et al. 1972; Gross et al. 1977; Pine 1980) allowing

the researcher to study the activity of hundreds of neurons simultaneously.

However, there are many more electrically active cells in the human body and other

organisms that could be studied on an MEA. For example, cardiomyocytes differentiated

from murine embryonic stem cells can be cultured on an MEA (Igelmund et al. 1999).

Clusters of these cells form and begin to beat spontaneously and in synchrony depending

if pace maker cells are present within a cluster. Once the cell clusters are identified,

drugs can be tested in vitro to determine their effect cardiomyocytes activity. The ability

and mechanism of toxins that cause an arrhythmia and disturb the electrical coupling

between cells have been explored on a cellular level with cardiac cultures grown on a









MEA (Igelmund et al. 1999). Another example is sensory tissues such as photosensitive

cells. For example, retinal tissue cultured on MEAs (Meister et al. 1994) has been used

to understand how information is processed by the retina into neural signals that are sent

to the brain. Retinal tissue responds to photic stimulation in its natural state. When used

on an MEA recording is done with the ganglion cell layer. A retina is placed on the

MEA with the ganglion layer on the electrodes and optically stimulated and the MEA

records the neural response to these stimulations. Understanding the effects of multiple

locally applied signals to the intact retinal tissue on image production could help

determine mappings for future retinal prosthetics. In short, as long as the cells are

electrically active and/or sensitive to electronic potentials, the MEA can be a useful tool

for electrophysiology.

The major areas of study using neural culture are neurocomputing, behavioral

research, and pharmaceutical research. Companies such as Panasonic developed MEA

technology originally for use in neural hybrid computing. Neural computing involves

interfacing nonlinear neurons with linear computing to create a new type of computer that

could combine the number crunching power of silicon computers with the control and

learning capabilities of neural tissue. A great deal of work has been done to understand

and make sense of the intact brain and control it for use in computing (Freeman 1994;

Gross and Kowalski 1999; Garcia et al. 2002).

Behavioral research centers on the mechanisms and physiology of learning and

memory in neuronal networks. The work in this field involves investigating changes in

neural networks. Specifically long-term potentiation (LTP) and long-term depression

(LTD) both thought to be key to learning and memory in the brain (Tateno and Jimbo









1999; Shahaf and Marom 2001). Stimulation of the culture and feedback from the

culture are important to this area of study. Some researchers (DeMarse et al. 2001) have

closed this circuit into a control loop with a virtual animal called the Animat with a

neural culture brain on a MEA controlling a robotic body.

Pharmaceutical research using MEAs is a new area which has been dominated by

patch clamp recordings of single neurons. However, universities, institutes, and major

drug companies are beginning to use and this technology for pharmaceutical testing and

screening (Moorefield et al. 1999; Egert and Hammerle 2002).

Bursting

Neurons that are cultured on MEAs are spontaneously active, producing

stereotyped bursts of synchronous activity across the entire network (Jimbo et al. 2000;

Keefer et al. 2001). This synchronized activity is illustrated in the form of a raster plot

shown in Figure 1-2 in which activity (action potentials) on each channel (vertical axis) is

shown over time. Synchronized bursting appears as vertical groupings points where most

of the channels are active. Bursting in these cultures is semi-periodic occurring every 1 to

15 seconds and each burst often lasts from 100 ms to over one second.

During the process of mammalian development there is a period where the network

is primarily connected by excitatory synapses. Synchronous bursting behavior is

observed in this period or when the inhibitory systems are blocked in adult tissue (Keefer

et al. 2001). These bursts have been theorized to be important in the shaping of neural

circuits during development (Tosney and Landmesser 1985; Meister et al. 1991). The

mechanisms that initiate and terminate a burst are poorly understood. However, the

effect of various pharmacological agents on bursting rates and the interval between burst

has been studied (Keefer et al. 2001).














50






















on a MEA. Time is on the x-axis in seconds and channel number is on the y-
40



30
730 ---17= 7- IM-- ,-






10 7



0 10 20 30 40 50 60 70 80 90 100
times)
Figure 1-2. This is a raster plot showing neural activity from a dissociated neural culture
on a MEA. Time is on the x-axis in seconds and channel number is on the y-
axis. The blue dots represent individual action potentials. The vertical
groupings of action potentials are bursts.

Epilepsy

Interestingly, spontaneous bursting is reminiscent of patterns of activity found in

the intact (in-vivo) brain known as epileptic seizures. A seizure is abnormal electrical

activity in the brain that can cause abnormal motor movement, thoughts, and sensations

(Brenner 2000). Seizures are the observable symptoms of epilepsy, which can develop as

a result of many different origins. These origins include stroke, hypoxia, head trauma,

tumors, and many other sources that can alter the function of the brain (Marieb 1998).

Often seizures arise from a seizure focus, a group of neurons that elicits a synchronous

neuronal discharge, which may then spread from the focus into other regions of the brain.


























Figure 1-3. This is a representation of a seizure from an EEG from a human epilepsy
patient. The top two groups of electrodes are depth electrodes implanted into
the patients skull. The remaining electrodes are surface electrodes. The time
scale on the x-axis is Isecond per division. The seizure onset is at 5s and
produces synchronized large amplitude waves of neural activity.

Grand mal and petit mal seizures are two examples of generalized epileptic

seizures(Brenner 2000). Grand mal or tonic-clonic seizures are the most well known

types of seizures that involve a loss of consciousness and violent muscle contractions and

spasms. Petit mal or absence seizures are characterized by a brief loss of consciousness

where the subject will appear to be blank or absent with little to no signs of muscle

tension or spasms.

Normal and epileptic activity in the brain can be measured using an electro-

encephalogram (EEG). The EEG measures potential differences at multiple positions on

the scalp on a macro scale through the skull and surrounding tissues. Individual action

potentials cannot be resolved but more broad activity patterns at various frequencies can

be observed. Figure 1-3 shows a 15 second EEG recording from 25 electrodes located

across the surface of the scalp of a human patient (used with permission of the

Sackellares Lab).









Polymeric Antiepilepsy Drug Delivery

The Hypothesis

The onset of a seizure begins a few seconds before the appearance of large

amplitude waves and continues through the rest of the recording. Both in-vivo epilepsy

(Figure 1-3) and bursting (Figure 1-2) represent the onset of coordinated synchronous

neural activity and although there are large differences in the duration of each episode,

both could be based on a common pathology. In fact, some have hypothesized that the

bursting behavior observed in vitro may be epileptiform in nature in both hippocampal

slice (Gluckman et al. 1996) and dissociated neural cultures (Segal 1994; DeLorenzo et

al. 1998; Egert and Hammerle 2002). For example, Segal suggests that individual

neurons in micro-island culture experience epileptiform bursting similar to ictal or

interictal activity seen in the hippocampus and neocortex. Segal also references

hippocampal slice preparations that have been exposed to low calcium solutions as being

epileptic. DeLorenzo varies ion concentrations and adds various bioactive agents while

conducting intercellular recordings and calcium imaging in dissociated culture that results

in behavior described to be epileptogenic. Egert studies the genesis and spread of

epileptic activity in culture. Other experimenters use similar methods to quantify

bursting behavior through experiments but stop short of calling it epilepsy (Kamioka et

al. 1996; Watanabe et al. 1996; Gross and Kowalski 1999)

Unfortunately, other than these general comparisons, very few experiments have

been done to determine if the network activity on an MEA is epileptic or perhaps a part

of some other developmental processes such as an arrested developmental state (Meister

et al. 1991). This is an important question for those studying neural cultures in vitro









because it would mean that the tissue under study is in a pathological state and could

potentially limit any comparisons that might be drawn to normal tissue.

However, there are a number of current pharmacological treatments for epilepsy. If

epileptiform activity in-vivo and spontaneous bursting in culture is based on a common

pathology, then the prediction would be that the application of anti-epileptic drugs used

to prevent seizures in vivo should have similar effects in vitro. This thesis describes a

series of experiments to test three of the most common drugs used as anti-seizure

medication in-vitro. The drugs used will be phenytoin (used for grand mal seizures),

ethosuximide (used for petit mal seizures), and valproate (used for various seizure type

including petit mal and grand mal). Each drug will be delivered and tested in-vitro using

a novel polymeric delivery system to determine whether they are effective in abolishing

in-vitro bursting.

Polymeric Drug Delivery System

One of the challenges of conducting this work, however, is how to deliver these

drugs in such a way that these cultures can be studied continuously. For example, one of

the most common methods used in vitro is the perfusion system in which a solution is

pumped into one side of the MEA and simultaneously removed (via vacuum) on the

other. However, perfusion systems are prone to a variety of problems including the

possibility of infection, difficulty in setup, leaky tubing, open air chambers, etc.

Moreover, perfusion systems are somewhat expensive and not economically practical for

all drug delivery applications. As an alternative, a polymeric drug delivery device was

developed and tested that is placed in the media of the neural cell culture to dispense the

drugs using a concentration gradient. This monolithic device is made of poly

hydroxyethyl methacrylate (PHEMA), a hydrogel often used to make contact lenses






9


(Ratner 1996). The production costs of these small polymer devices are minimal. The

PHEMA ring was tested in the first phase of this project for toxicity and it's diffusion

capabilities were tested before testing anti-epilepsy drugs. The second phase of the

project will used these rings to deliver and test the efficacy of the antiepilepsy drugs.














CHAPTER 2
DEVELOPMENT OF THE PHEMA MEA DRUG DELIVERY SYSTEM

Poly Hydroxyethyl Methacrylate Background

Hydrogels are cross-linked polymeric structures that can become swollen with

water. These polymers are made up of monomers that by chemical reaction have been

polymerized into long chains connected in a macromolecular network, similar to a dense

spider web. Hydrophilic or water loving monomer is used so that the resulting polymer

has an affinity for water.





HO




Figure 2-1. The hydroxyethyl methacrylate monomer. The carbon-carbon double bond
in the top left opens up to form a vinyl linkage that forms the backbone of the
Poly-HEMA molecule.

Acrylic acids have been used to make polymers for biomedical applications such as

intraocular lenses since World War II (Ratner, 1996). As hydrogels for drug delivery

their uses have been extensively studied and documented(Ratner 1996; Hoffman 2002;

Nuttleman et al. 2002). Poly hydroxyethyl methacrylate (PHEMA) is an acrylate polymer

made from the HEMA monomer shown in Figure 2-1. It is a commonly used biomedical

hydrogel. PHEMA has been a notable material in this class of polymers due to the fact

that when fully swelled it has a similar density and water composition to that of living

tissue. PHEMA is biologically inert, nonionic (suitable for blood contact), resists









degradation, is permeable to metabolites, non-absorbable, can withstand autoclave

sterilization and can be molded to most any shape. Acrylates along with PHEMA are

also optically clear. As a biomedical polymer, these properties have been taken advantage

of for use most notably in contact lenses for the correction of vision (Ratner 1996).

Pharmaceutical applications for hydrogels including PHEMA have become a topic

of much investigation over the past fifteen years. The basic principle is to dissolve a

water-soluble drug in solution that is absorbed by the hydrogel. When placed into an

environment that lacks this agent a concentration gradient is formed and allows for

outward diffusion of the drug into the surrounding environment(Ratner 1996).

PHEMA is useful for use in delivering pharmaceutical agents to neural cultures due

to it's non-toxicity, inertness, optical clarity, and ability to absorb and release

pharmaceutical agents and metabolites. Ideally any material that is introduced into a

neural culture during an experiment should do nothing to disturb the native activity of the

culture. Ideally, only the agent that it carries should have an effect on the culture. Many

variables must be observed to insure no interactions occur when the PHEMA is placed in

the culture dish. The medium the PHEMA is soaked in must be the same temperature,

concentration, and gas content as the medium. Changes in the ionic concentrations of the

medium for example can change the behavior and firing rate of neurons (Hille 1992).

The PHEMA must be sterile, fully polymerized, clean, and carefully placed into the dish

to prevent contamination or agitation of the culture.

The initial testing of this material in neural culture was to simply test for acute and

chronic toxicity. Over the course of two weeks five control dishes were visually

compared to five dishes with sterile prepared PHEMA slabs added to the medium. In all









dishes there were no signs of acute or chronic toxicity. Even in cases where the PHEMA

slab was directly on top of the neural culture the cultures showed no signs of toxicity.

This case also demonstrates permeability similar to water of the PHEMA to dissolved

gases and metabolites in the medium.

Design, Fabrication, and Use of the PHEMA Drug Delivery Ring

Design and fabrication of the PHEMA ring for use in a neural culture on a MEA

was carried out to ensure repeatability, sterility, and material purity. The ring was

designed to be smaller in diameter, when swollen, than the inner diameter of the MEA

dish. The device is a ring rather than a disk to prevent the material from actually

touching the culture in the middle of the dish. This prevents abrasion of tissue from the

culture and prevents the PHEMA from influencing dissolved gas transport to and from

the culture.

The specific recipe for the PHEMA used in this experiment is adapted from the

laboratory section of Dr. Goldberg's Polymeric Biomaterials class (Goldberg 2002). The

major changes to this method include the use of a Teflon mold specific for creating rings

and the use of sterile procedures to ensure purity and sterility. To create the material

used in these experiments 4mL of ultra pure tissue culture water and 7.5 grams of 2 2'-

azobisisobutyronitrile (AIBN) initiator was added to 4mL HEMA monomer in a 50mL

test tube. The reaction mixture was blended by shaking and then pipetted as needed into

the sterile Teflon mold. The mold was then placed in an oven at 90C to 95C for 1 hour to

allow for polymerization. The smell of the monomer from the mold indicates that a

longer baking time is required. At lower temperatures the reaction will not occur and at

temperatures above 100C the PHEMA will develop bubbles in the solid that interfere

with the rings homogeneity.









After removal from the oven the PHEMA ring is ready to be removed from the

mold. If allowed to cool the PHEMA will become brittle and will be difficult to remove

from the mold. Cool the top of the PHEMA ring by filling the mold with sterile culture

water while the ring is hot and in a sterile environment. Immediately use a pick to

impale the ring and pull it from the mold. This should be easy if the ring did not cool for

too long. If the ring is brittle or cracks one solution is to try putting the ring back in the

oven with a little water to soften the ring. Once removed the ring was placed in a sterile

50mL test tube and sealed for later use. An example of a freshly molded ring, a used

ring, and their Teflon mold is shown in Figure 2-2. The rings can be kept in this sealed

sterile environment for an indefinite amount of time until hydrated for experiments.








B) C)









Figure 2-2. Teflon mold and new and used PHEM rings A) The Teflon mold that is used
to create PHEMA rings. B) A freshly created PHEMA ring removed from the
mold. Notice the optical clarity and rigidity of the ring. C) A used PHEMA
ring that has been soaked in a medium and drug solution and used to deliver
the drug to a neural culture.

When used in an experiment the PHEMA ring must be hydrated in fresh medium in

an incubator to allow carbon dioxide concentration and temperature of the ring to reach

equilibrium. Figure 2-2C shows a ring that has been soaked in medium as shown by its









red color. Sufficient time must be allowed for the equilibrium water content of the

PHEMA to be reached. It is recommended to prepare the rings with medium a few hours

before an experiment. Pharmaceutical and bioactive agents should be added at this time

to allow for the agent to soak into the PHEMA hydrogel. While placing the PHEMA

ring into an MEA culture dish care must be taken not to touch the neural culture and to

use sterilized forceps. The outer surface of the ring was rinsed with medium to remove

surplus liquid and to prevent delivery of excess pharmaceutical agent from the surface of

the material.

Several problems may emerge from incomplete polymerization of the HEMA (the

monomer is toxic) or non-optimal crosslink density of the polymer. Incomplete

polymerization could result from not baking the HEMA long enough to fully polymerize

the monomer. Preliminary testing has shown that baking at 90C results in a good

polymerization without damaging the resulting polymer. Crosslink density is controlled

directly by the amount of cross-linking agent added to the reaction mixture (Goldberg

2002). If the crosslink density is too high then the rate of diffusion into and out of the

PHEMA will decrease to the point where the PHEMA would be nonfunctional as a drug

delivery device.

The PHEMA rings were tested in-vitro for toxicity, effects on the activity of neural

culture, and basic functionality before they were used to deliver antiepilepsy drugs to

neural cultures. Chapter 3 describes the materials and equipment used to culture,

maintain, and record from neural cultures on MEAs. Understanding these principles and

the equipment used allows one to better understand the reasons for and the process of

testing PHEMA ring in-vitro, which will be discussed in Chapter 4. This testing will






15


insure that when the antiepilepsy drugs are delivered using the PHEMA ring, any effects

will be a result of the drugs and not the PHEMA alone.















CHAPTER 3
MATERIALS

Cell Culture

Dissociated Neural Cultures

Cortical tissue from E17 E18 Wistar rat embryos (Brain Bits Inc.) was digested

using the Papain Dissociation System (Worthington Biochemical Corporation) and

mechanically triturated separating the neuron's soma from the surrounding connective

tissue. The resulting suspension of cells consists of both neurons and glia. Immediately

after plating the neurons on an MEA they appear as small spheres with a dark center.

After dissociation, approximately 50,000 cells are placed in the center of a multi-

electrode array that has been coated with polyethyleneimine (PEI) (Lelong et al. 1992)

and laminin (Sigma) (Banker and Goslin 1998). The PEI creates a surface that is

hydrophilic and favorable to the growth of neurons. Laminin is a component of the

extracellular matrix of the neurons and serves as a building material for the neurons to

bind to the surface and grow axons and dendrites. Having the laminin present allows the

neurons to mature faster. If the surface treatment is not favorable the neurons may be

more attracted to each other rather than the surface. This will lead to clumping of the

neurons into small islands, which would limit the electrical contact and number of active

electrodes.

Medium is required to feed the neurons in an ex-vivo environment. It mirrors the

content of blood as far a proteins and nutrients. The neurons are cultured in medium

consisting of 90% Dulbecco's modified Eagle's medium (DMEM) (Gibco cat# 10569-

16









010) and 10% Equine Serum (HyClone cat# SH30074.03). The medium is allowed to

equilibrate to 5% carbon dioxide and 35C temperature in an incubator before use.

Medium is changed twice a week or as needed.

Development of Culture Activity

The neurons begin to connect to each other by regrowing their axons and dendrites

within the next few days after plating. Action potentials from these neurons can be

detected within this time. Over the next two weeks the embryonic cells begin to express

their ion channels. The excitatory N-methyl-D-aspartate (NMDA) ion channel system

matures quickly within the first 3 to 8 days postnatal. The excitatory systems are thought

to be responsible for epileptiform bursting behavior(Bugard and Hablitz 1993; Keefer et

al. 2001). The inhibitory gamma-amino butyric acid (GABA) ion channel system

develops and begins expression within the first 6 to 10 postnatal days (Luhmann and

Prince 1991). Inhibitory systems are notably absent for the first postnatal week. For

dissociated culture this translates into maturity being reached within the first four weeks.

Figure 1-1B shows an example of a mature neural culture. Experiments are typically

carried out with active mature cultures only past the 30-day post natal age to prevent

interference from the maturation process.

Hardware and Software

The Multi-electrode Array

The multi-electrode arrays used in the DeMarse lab are purchased from

Multichannel Systems (cf. Figure la and Ib). The electrodes are arranged in an eight by

eight grid with the corners missing giving a total of 60 electrodes. Each electrode records

the electrical activity produced by nearby neurons as they fire action potentials. These

electrodes can also be used to stimulate activity by applying a bi-phasic voltage pulse









(100 to 800 mV) through an electrode, which causes neurons near the electrode to fire.

Hence, these arrays can be used to both measure and stimulate the ongoing neural activity

of a small population of neurons or other electrically active tissue.




















Figure 3-1. This is a multi-electrode array with a white Teflon lid with the FEP Teflon
membrane attached. This effectively seals the culture from infectious agents
keeps moisture in the dish, and allows for gas transport across the membrane.

Culture System

Neural tissue is sensitive to infection by mold, yeast, and bacteria. Specialized lids

have been designed (Potter and DeMarse 2001) that are made of a Teflon ring that seals

around the base of the dish and at the top allows for a FEP Teflon film to be attached.

Figure 3-1 shows an example of an MEA with a Teflon lid with the FEP membrane

attached. This FEP covering allows for carbon dioxide and other gases to pass through

while preventing water and contagions from getting in or out. This effectively seals the

culture from infectious agents and retains water to prevent hyperosmolarity from the

evaporation of water in the medium. When used properly they can extend the lifetime of

a culture to as much as two years.































Figure 3-2. The micro-incubator constructed to control the environment around the
MEA-60 and the neural culture during experiments. The temperature,
humidity, and carbon dioxide concentrations are all controlled inside the
enclosure.

Physiological Environment

The environment around the MEA60 is temperature and gas compositionally

controlled. The MEA 60 is temperature controlled using the Multichannel Systems TC02

temperature controller. Air temperature and carbon dioxide content are controlled by

creation of a mini incubator around the MEA60 shown in Figure 3-2. In this case an

acrylic desiccators has been converted to house the MEA60 with an access door and

input/output ports for temperature controlled air, carbon dioxide, and electrical wiring.

This setup is wrapped in conductive foil to reduce heat loss. The temperature and carbon

dioxide controllers were made in the DeMarse lab specifically for this application.


























Figure 3-3. The MEA-60 from Multichannel systems with the 64 channel stimulator
attached. A MEA with it's Teflon lid is in the apparatus.

Data Acquisition System

The connection to the MEA dish is provided by the MEA60 amplifier system also

made by Multichannel Systems. This allows for a cable connection to the MC_Card,

which is a 64-channel analog to digital (A/D) card for data acquisition that fits into a PCI

card slot in a PC. Stimulation is provided to the dish by a 64-channel stimulation board

that attaches to the top of the MEA60. This stimulation board was developed and built in

house by Dr. Tom DeMarse and Vihn Trinh. These components are shown in Figure 3-3.

Raw electrical activity is recorded at 25kHz per channel. Since, a typical action potential

is approximately Ims long this system provides more that adequate resolution to measure

the wave form. The MEA60 system consists of three distinct parts. An amplifier to

amplify neural signals whose amplitude typically ranges from 20 [ V to over 100 [ V.

The data is digitized using an in host data acquisition card (PCI interface) capable of bus


I '









mastered transfer to main system memory at 3.3 MB/sec. The host computer can then

perform analyses (e.g. spike detection, artifact suppression) that are required.

I j dI F -j IT.. .


Figure 3-4. Raw electrical activity from a neural culture on an MEA. The large
deviations seen in the individual boxes represent action potentials occurring
near the corresponding electrode on the MEA. This picture shows the raw
activity of the culture during a burst. For each window voltage is represented
on the y-axis and time (approximately 200ms) on the x-axis.

As mentioned previously, raw electrical activity is sampled at 25KHz on each

channel and digitized (12 bit). For each sample the 64 available channels are sampled in

hardware order around the outside of the MEA dish instead of any geometric pattern. A

map of the hardware order to the geometric channel locations is necessary to determine

the location of a channel in the MEA dish as the data is recorded in hardware order.

Culture electrical activity is then viewable and is processed in real time on the computer

monitor. An example of this activity is shown in Figure 3-4. The raw data can then be









streamed in real-time (low latency) to other machines via a dedicated TCP/IP 1GB

ethernet for further processing, analysis, or stimulation.

Spike and Burst Detection Algorithms

Raw data from the MEA is available to the computer at 25kHz for 64 channels.

For the most part the bulk of this data is not spike information. A raw recording of this

magnitude could take up 3.3 megabytes of storage per second. This file size becomes

unwieldy when 20 minutes of data are recorded, 3.5 gigabytes in this case. To prevent

such large file sizes and to eliminate data that is not relevant to spiking events, only the

spike events were recorded. The MEABench (written by Daniel Wagenaar) software is a

set of command line and GUI utilities to process data from the MEA60. This software is

used in these experiments to visualize, process, and record data. Each recorded spike

event contains information about the time (at peak), channel, height, width, threshold

used, and one ms of context raw data around the peak time. Spikes are detected in real

time via a moving voltage threshold based on 5 standard deviations of the noise.

This spike file is later processed offline to extract information about any bursts that

may have occurred. The burst detection algorithm, which was developed in MATLAB,

is shown in Appendix A. This algorithm uses a leaky integrator to determine when

temporally related spiking events constitute the beginnings of a burst. This algorithm is

an adaptation of a method described by Gross (Stenger and McKenna 1994). Gross'

algorithm uses the difference between two leaky integrators to determine the threshold

for entering into and exiting the bursting condition. The end of a burst is characterized

by a period of inactivity with no spiking similar to the refractory period for individual

neurons. Instead of using the same leaky integrator to exit the bursting condition this

algorithm searches for this period of inactivity to establish the end of the burst. The burst






23


start and stop time are recorded. This algorithm also calculates the burst duration and the

interburst interval (the time between bursts) along with statistical information such as

standard deviation for these measures. These measures will be critical for measuring

changes in culture activity in the following chapters where the PHEMA ring and

eventually antiepileptic drugs will be tested in-vitro.














CHAPTER 4
TESTING OF THE PHEMA MEA DRUG DELIVERY SYSTEM

Testing the PHEMA-MEA system with bicuculline

A series of experiments was conducted to determine the effect of the poly HEMA

ring on the activity of neurons cultured on a MEA and the effectiveness of the drug

delivery device in general. It was first important to determine that the poly HEMA ring

alone does not have an effect on the normal electrical activity of the neural culture. Test

of the ability of the ring to delivering bioactive agents was conducted using bicuculline,

which produces a well known stereotypic response in neural culture.











I -*---"

Figure 4-1. The chemical structure of bicuculline, a bioactive agent derived from
Dicentra cucullaria. Bicuculline is a pro-convulsive agent that evokes
rhythmic bursting from neural cultures.

Bicuculline shown in Figure 4-1 is a phthalide isoquinoline a bioactive agent

derived from Dicentra cucullaria a flowering plant common in North America. Manske

first isolated it in 1932. Bicuculline is a specific competitive agonist of GABA (Razet et

al. 2000). When applied to neurons ex-vivo stereotypic rhythmic bursting behavior

results (Stenger and McKenna 1994). Bicuculline, when applied to neural culture on an









MEA, has been shown by Keefer to result in increased burst durations (Keefer et al.

2001). Direct injection of bicuculline into the motor cortex of monkeys produces

rhythmic dystonia which is characterized by involuntary muscle contractions that cause

involuntary movement and twisting of body parts (Guehl et al. 2000).

Experimental Procedure

First, ten-minute baseline recordings of neural activity from four MEAs were

conducted without HEMA rings. This baseline data will be compared to the neural

activity in the presence of a HEMA ring with and without bicuculline. The primary

measures of neural activity were interburst interval (IBI) and burst duration (BD). IBI is

a measure of the time between bursts while BD is a measure of the duration of a burst.

The maximum intensity of the burst (determined by maximum average firing rate during

the burst) will also helpful in characterizing bursting behavior.

Following baseline experiments, a placebo poly HEMA ring soaked in medium and

equilibrated in a 5% C02 incubator, was placed in each of the MEAs and activity will be

measured for 10 minutes. A placebo was conducted to insure that the PHEMA with fresh

media does not have an effect on either burst or firing rates.

Data from the placebo will be compared to cultures in which the PHEMA ring has

been soaked in 100[tmol/L bicuculline. Bicuculline removes inhibitory mechanisms from

the neurons and should create a highly stereotypic periodic rapid burst (Moorefield et al.

1999). Bicuculline induced burst activity was recorded and compared to control (native)

and placebo activity. The MEAs were then rinsed with media to remove the bicuculline

and placed in the incubator for later use.









PHEMA Bicuculline Testing Results and Discussion

Application of the placebo ring produced no apparent effect on burst activity

relative to baseline recordings. In contrast, rings with bicuculline showed substantial

increases in burst duration and interburst interval.

Figure 4-2 shows the burst durations for the bicuculline treatment for four dishes in

baseline, placebo, and bicuculline conditions. The placebo and baseline conditions do

not show appreciable differences in mean or variation suggesting that the PHEMA has

little effect on burst duration. The mean burst duration increases greatly when comparing

bicuculline treatment to baseline, while the variation stays relatively the same.

Figure 4-3 shows the interburst interval for bicuculline treatment for the same four

dishes in baseline, placebo, and bicuculline conditions. The placebo and baseline

conditions do not show any significant changes in the mean or variation of interburst

interval suggesting that PHEMA has little effect on interburst interval. The mean and

variation of interburst interval increase when comparing bicuculline treatment to

baseline.

Figure 4-4 shows a raster plot with the neural activity for culture 6682 during the

ten-minute baseline, application of the placebo ring, and application of a ring with

bicuculline. The bicuculline raster plots for the remaining cultures are in Appendix B.

The changes in burst duration and interburst interval are plainly visible when comparing

baseline and treatment runs while baseline and placebo runs look very similar.











3.500


0.500


0.000


1+


TI


6682


6688


6685


MEA Dish


Figure 4-2. Bicuculline average burst duration. Mean burst duration during baseline,
placebo, and bicuculline treatments.


7087 6682


6688


6685


MEA Dish


Figure 4-3. Bicuculline mean interburst interval. Mean interburst interval in seconds (y-
axis) during baseline, placebo, and bicuculline treatments for four MEAs (x-
axis).












dish 6682
60



20
0 I


0 100 200 300 400 500 600
times)
60

CO

0





bicu -culline i activi Time is on th x i secondsa
0
0 100 200 300 400 500 600
times)











Figure 4-4. Raster plots of spike activity for dish 6682 during baseline, placebo, and












for dish 6688 but the mean increases as it did in the other cultures. This difference can be


seen graphically for interburst interval in figure 4-2 and BD in figure 4-3. The effect of


the bicuculline on the burst duration is consistent with those of Keefer (Keefer et al.


2001) in that the burst duration significantly increased. The IBI grew longer where


Keefer shows no major change in burst rate with bicuculline. Keefer measures the burst


rate rather than interburst interval, which are inversely related.
rate rather than interburst interval, which are inversely related.





























Figure 4-5. Dish 6685 with a standard concentration of 100mmol/L bicuculline. A) The
full 600 seconds of treatment B) The first 100 seconds of treatment. These
plots shows the almost immediate effects of bicuculline on the culture at this
concentration.

Application of bicuculline showed an effect within the first ten seconds at this high

concentration such as in Figure 4-5 A & B. One might expect to see a gradual change in

burst characteristics as the drug diffuses out of the ring. However, the effect in this

experiment at this concentration occurred within the first 10 seconds. The relatively high

concentrations of bicuculline used show no such release rate shown by a slow activity

change but rather a fast abrupt change. Two cultures were tested at a much lower

concentration, 10mmol/L bicuculline, to assess whether concentration has an impact.

Figure 4-6 shows the effect of a PHEMA ring soaked in 10mmol/L bicuculline. The IBI

gradually increases until around time = 40s where the major effects of BCC can be seen

with increased BD and IBI. This shows that the BCC is released slowly from the

PHEMA ring. Both the interburst interval and burst duration during the first 40 seconds

following the application of the drug gradually increase. However at 40 seconds there is










an abrupt increase in BD and IBI. This effect would be expected due to the first order

release rate from the monolithic PHEMA (Ratner 1996).















.. Bi-







Figure 4-6. Dish 6685 with a concentration of 10mmol/L bicuculline. A) The full 600
seconds of treatment B) The first 100 seconds of treatment.

Conclusions

One dish was killed immediately due to a rough edge on the PHEMA ring that

accidentally removed the culture from the dish surface. This was corrected by increasing

the inner diameter of the PHEMA ring to help prevent contact between the PHEMA and

the neural culture. When the ring was not removed properly the edges of the culture can

be disturbed and leading to peeling of the culture from the edges. This peeling will

eventually remove the culture from the surface.

Overall the PHEMA ring performed as anticipated as a drug delivery device. The

bicuculline was absorbed into the polymer matrix and released over time. There was no

evidence of infection or contamination due to the use of the rings. In the few cases that a

failure occurred the cause was primarily poorly made rings that physically abraded and






31


damaged cultures. This was remedied by making the inner diameter of the rings larger.

When manufactured correctly in a sterile environment the PHEMA drug delivery ring

proves a viable cost effective alternative to a perfusion system for drug delivery. These

rings were used for all future drug delivery in this project.















CHAPTER 5
ANTIEPILEPSY DRUG TESTING

Antiepileptic Drugs

If bursting originates from the same mechanisms antiepilepsy drugs should have a

similar effect on in-vitro bursting as they do in-vivo epilepsy. Three antiseizure drugs

were chosen for testing. They were: phenytoin, ethosuximide, and valproate. Each was

delivered to the cultures using the previously tested PHEMA drug delivery device. Each

of these drugs has been in documented use for treatment of epilepsy for many years

(Brenner 2000; Levy et al. 2002).



C'










Figure 5-1. The chemical structure of phenytoin. Phenytoin is a white powder that
dissolves in a basic aqueous solution. As a antiepilepsy drug it is mainly used
for grand mal seizures.

Phenytoin

Phenytoin shown in Figure 5-1 is a hydantoin derivative used to treat partial

seizures and generalized tonic-clonic seizures (Brenner 2000). Merritt and Putnam first

used it in the treatment of epilepsy in 1938 (Levy et al. 2002). It was the first

anticonvulsant that does not result in sedation of the patient. Its major method of action









is blockage of voltage sensitive sodium channels in a neuron's cellular membrane. The

spread of electrical discharge from a seizure focus to the surrounding tissue is though to

be reduced by this effect, as the neuron is no longer as voltage sensitive. The therapeutic

serum concentration is 10 20 ug/mL (Brenner 2000).



0


,/ NH





Figure 5-2. The chemical structure of ethosuximide. Ethosuximide is a nonionic white
powder that dissolves in ethyl alcohol. As a antiepilepsy drug it is mainly
used for petit mal (absence) seizures.

Ethosuximide

Ethosuximide shown in Figure 5-2 is a succinimide derivate that is used primarily

in the treatment of generalized absence or petit mal seizures (Brenner 2000). Its primary

effect is thought to be on the T type calcium channels that regulate the entrance of

calcium into the neuron. It has many other effects and has proven to be difficult to isolate

its anticonvulsant actions experimentally. It is known not to have any effect on voltage

sensitive sodium channels like phenytoin and other antiepileptic drugs (Levy et al. 2002).

The therapeutic serum concentration is 40 -100 ug/mL (Brenner 2000).

Valproate

Valproate shown in Figure 5-3 is a branched chain fatty acid with a relatively low

molecular weight. Its mechanism of action is wide and is effective on many types of

seizures including both generalized and focal types (Brenner 2000). It has been shown to









block not only voltage sensitive sodium channels but also increases GABA production

and slows degradation of GABA. No single effect of valproate can be claimed to be it's

primary mechanism of action as the efficacy of its antiepileptic inhibition is broad(Levy

et al. 2002). Valproate has also been shown to prevent rapid firing of action potentials of

CNS neurons in culture(Levy et al. 2002). The therapeutic serum concentration is 50 -

100 ug/mL (Brenner 2000).







.-I



Figure 5-3. The chemical structure of valproate. Valproate is a clear crystalline solid that
dissolves easily in aqueous solution. As an antiepilepsy drug its uses are wide
including both petit mal and grand mal seizures.

The three drugs were chosen because they represent a good spectrum of common

antiepileptic drugs. Phenytoin is primarily for grand mal seizures, but has also been

found to increase absence seizure activity. In contrast, ethosuximide is mainly used to

treat absence seizures while valproate can be used in the treatment of a wide variety

seizure types. Due to their differing methods of action there would be potential

differences in their effects. For example, if bursts in-vitro resemble grand mal seizures

then phenytoin should be effective in attenuating or even removing bursts while

ethosuximide would not.

Experimental Procedures

A PHEMA drug delivery ring was prepared, as discussed previously, prior to each

experiment. After polymerization the PHEMA ring was soaked overnight in a media and









drug solution to thoroughly hydrate and saturate the PHEMA. This was conducted in the

incubator with a FEP Teflon lid on the container to allow equilibration of C02, 02,

temperature and pH to match that of the media in the MEA dishes.

Phase 1: Effects on Spontaneous Changes

Phase one of the procedure tests the change in interburst interval and burst duration

when the cultures are exposed to the antiepilepsy drugs. The MEA cultures were

individually removed from the incubator and placed in the test rig. The test rig acts like

an incubator maintaining a similar temperature and carbon dioxide concentration as that

of the incubator. Four MEA cultures were used for experiments for each drug. Baseline

recordings were conducted for all dishes pretreatment and post-wash. The PHEMA ring

was removed after each treatment and the MEA was rinsed with media in a series of five

washes to remove the residual concentrations of the drugs. A typical experimental session

began with 10 minute baseline spike recordings. This baseline data was compared to the

treatment and return to baseline recordings. The treatment for each dish begins after

baseline recordings when a drug laced PHEMA ring was placed into the MEA. Treatment

recordings will be 20 minutes duration. The dish was then washed five times with

medium after removal of the PHEMA ring. Return to baseline recordings repeat the same

ten-minute recordings to match the baseline.











o 1 native 2 BCC
A 1 *3- STR A 4 SCH50119
A to 0E5-ChTX A6-APA
'* *7 NBQX (NMDA only) 8 3 mM CaC2
*AA **9-4.7mMCaCI2 10-6.2 mM CaCI2
2 2 Ef 11-8mMCaCI2 12 8.5mMKCI
CC



1L, ^3 4 0 5




0 B o

0 10 20 30 40 50 60
burst rate (bpm)

0.9

0.8 N
NMDA-only
S0.7
3 mM CaC*12

0.6 4.7 mM CaCI

05 6.2 mM CaC12C

0.4
8 mM CaCI12 5
0.3 a a Tm W 10 mM-K

0.2
8.5 mM KCI
0.1

0 '^ --- ....... .- .. ..-- -- --- --
0 5 10 15 20 25 30 35 40
burst rate (bpm)
Figure 5-4. Figure 8 from (Keefer et al. 2001). This figure shows the responses of neural
cultures on MEAs to various pharmacological agents. Figure B is a close up
view of the bottom left of Figure B. Prior experiments with BCC matches
results shown above. The effect of antiepilepsy drugs should fall in the
bottom left quadrant of B if they effectively attenuate bursting in-vitro.









Differences in bursting duration and interburst interval was used to compare

treatment to baseline native activity (Jimbo et al. 2000; Keefer et al. 2001). Standard

deviations were calculated for each subject per condition to insure that any differences in

the normalized data sets are statistically significant. The change in firing rates and

bursting rates during treatment were compared to determine if a relationship to the

release kinetics from the monolithic PHEMA drug delivery device can be detected

(Ratner 1996). Keefer (Keefer et al. 2001) has previously tested various compounds on a

neural culture on a MEA shown in figure 5-4. The results for bicuculline were matched

in the PHEMA ring testing. If the antiepilepsy drugs have an effect on bursting like the

effect they have on seizures, reduced burst rates and burst durations should appear as

points clustered near the origin.

Phase 2: Effect on Stimulated Activity

Phase two of the experiment tested the response to stimulation of a neural culture in

the presence of each drug. Response to stimulation was quantified by the average firing

rate during a burst. Burst duration and magnitude derived from average firing rates for

each drug post stimulation was compared to baseline burst responses. Electrical

stimulation of a healthy neural culture produces a stereotypical burst response. Jimbo

(Jimbo et al. 2000) describes two phases of a burst, an early phase and a late phase. The

early phase is very stereotypical and characterized by synchronous spiking over many

channels while spiking seen during the late phase is more random. The metric that Jimbo

used to quantify spiking during these phases is the average firing rate. Jimbo varied

treatments by increased ion concentrations and showed that the early phase did not

change. The late phase changed in response to the same treatments.









Phase 2 consisted of recordings on two dishes with baseline recordings of 20

stimulations thirty seconds apart on a channel with active neurons. This was followed by

20 minutes of probing with stimulations every 30 seconds. The drug was then rinsed and

a second baseline was recorded. The next drug was then applied to the culture, probed,

and the procedure was repeated for the third and final drug. Spikes were recorded 20ms

before each probe and for 800ms afterward. Occasions where spikes occurred before the

probe indicate that the probe delivery had occurred during a burst rather than initiating

one. Hence, these occasions were excluded from the analysis.

The anti epilepsy drugs should, from researching their effect on the ion channels of

the neurons, depress the activity of the culture in general (Hille 1992; Kapur and

Macdonald 1997; Brenner 2000). The firing rate for the culture should decrease

somewhat and a large increase in interburst interval should be seen. If bursting is

epileptiform a very large reduction or complete suppression of bursting should be seen

with the proper concentration of drug. The proper concentration could be an issue if high

enough concentrations are not reached in the media to even cause an effect. This could

be due to problems with the poly HEMA ring as well as problems with the chosen

concentration. All concentrations (Brenner 2000; Levy et al. 2002) are 10% of

pharmacological blood serum concentrations, which generally corresponds to the

concentrations that surround neurons in-vivo due to the blood-brain barrier.















CHAPTER 6
ANTIEPILEPSY DRUG RESULTS AND DISCUSSION

Phase 1 Results

Phenytoin

Figure 6-1 shows interburst interval measured in seconds on the y-axis and the x-

axis shows baseline, treatment, and return to baseline results for each MEA. The

interburst interval significantly increased for each culture after application of phenytoin

relative to baseline measurements. Phenytoin resulted in the eventual complete cessations

of bursting over the course of the 20-minute recording, which translates into an infinite

interburst interval in each case. This is why each bar is at the ceiling of figure for

treatment conditions. Return to baseline results show no difference from baseline

measurements suggesting that the washes completely removed phenytoin from the MEA

dishes with no apparent carryover effects.








2 I








Figure 6-1. Interburst Interval for Phenytoin. Baseline, Treatment, and Return to
Baseline results shown for IBI in seconds (on the y-axis) for each of the four
dishes (shown on the x-axis) used with phenytoin. Note that IBI approaches
infinity as bursting ceased in each dish.

39







40


Figure 6-2 shows the burst durations for phenytoin in seconds on the y-axis and


baseline, treatment, and return to baseline organized by dish on the x-axis. The quantities


of bursts for phenytoin were limited due to the fact that bursting stops about five minutes


into the twenty-minute recording for all dishes. Hence the burst durations reported here


reflect the culture behavior during these first five minutes rather than the last fifteen


minutes where there were no bursts. However, for the bursts that did occur there was


little difference from the baseline.


2.50 -


2.00 4


1.50 4-


1.00 4


0.50 4


0.00


6692


* Baseline
*Treatment
OR to Baseline


6682 7073 6686
MEA


Figure 6-2. Phenytoin Burst Duration. Baseline, Treatment, and Return to Baseline
results for BD shown in seconds (on the y-axis) for each of the four dishes
(shown on the x-axis) used with phenytoin. Note that no significant
difference is seen between baseline, treatment, or return to baseline for all of
the dishes.







41


Figure 6-3 shows an example of raster plots for dish 6686 (one of the four dishes


used for phenytoin). The raster plots for the remaining three dishes can be found in


Appendix C. Figure 6-3 is made up of four 600-second raster plots The first raster plot


represents the baseline recording. The second and third plots represent the first and


second half respectively of the continuous treatment run with phenytoin. The fourth plot


represents the return to baseline run. Notice the smooth transition in the second plot,


beginning at 100s and ending near 400s, between the insertion of the phenytoin laced


PHEMA ring at time Os and the drugged condition as the phenytoin is released from the


PHEMA ring.




Phenytoin dish 6686
60


20
0
S0 100 200 300 400 500 600





0 100 200 300 400 500 600
E Co
S40 -
5 20 -
20
o00 700 800 900 1000 1100 1200

40
20

cc 0 100 200 300 400 500 600
times)

Figure 6-3. Dish 6686 Phenytoin Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording.







42



Ethosuximide

Figure 6-4 shows the interburst interval for baseline, treatment, and return to


baseline for each culture during application of ethosuximide. This figure shows a


difference between baseline and treatment means and their corresponding variances for


interburst interval. Only one culture, MEA 7073, resulted in cessations of bursting that


translates into an infinite interburst interval. The remaining three cultures show and


increased interburst interval that is maintained at a steady state for the remainder of the


treatment run. Return to baseline results show no difference from baseline measurements


suggesting that the washes completely removed ethosuximide from the MEA dishes with


no apparent carryover effects.


50.00

45.00

40.00

35.00

30.00


T


M Baseline
) 25.00 -UTreatment
0 6OR to Baseline
20.00

15.00

10.00

5.00

0.00
6692 6682 7073 6686
MEA


Figure 6-4. Interburst Interval for Ethosuximide. Baseline, Treatment, and Return to
Baseline results shown for IBI in seconds (on the y-axis) for each of the four
dishes (shown on the x-axis) used with ethosuximide. Note that IBI
approaches infinity as bursting ceased in culture 7073 and culture 6692's
mean IBI was 128 seconds.








43



Figure 6-5 shows the burst durations for ethosuximide. There was little difference


in burst duration between baseline, treatment, and return to baseline. Even for culture


7073 the burst durations never changed even though bursting ceased completely for this


culture.




0.90


0.80


0.70


0.60


0.50
0.50 M Baseline
W ETreatment
BOR to Baseline
0.40


0.30


0.20


0.10


0.00
6692 6682 7073 6686
MEA



Figure 6-5. Ethosuximide Burst Duration. Baseline, Treatment, and Return to Baseline
results for BD shown in seconds (on the y-axis) for each of the four dishes
(shown on the x-axis) used with phenytoin. Note that no difference is seen
between Baseline, Treatment, or Return to Baseline for all of the dishes.


Figure 6-6 shows the raster plots for dish 6686, one of the four dishes used for


ethosuximide. The other three cutures can be found in Appendix D. Figure 6-6 is made


up of four 600-second raster plots similar to those for phenytoin. The first raster plot


represents the baseline recording. The second and third plots represent the first and


second half respectively of the continuous treatment run with ethosuximide. The fourth


plot represents the return to baseline run. Application of ethosuximide at Os eventually







44


produces an abrupt change in burst activity near 325 seconds relative to native activity

and the drugged condition as the ethosuximide is released from the PHEMA ring.



Ethosuximide dish 6686




0
C 0 100 200 300 400 500 600



E
E 0 100 200 300 400 500 600

4C0
5 20

600 700 800 900 1000 1100 1200

CCO


0 100 200 300 400 500 600
times)

Figure 6-6. Dish 6686 Ethosuximide Raster Plots. Each plot has channel number on the
y-axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording.

Valproate

Figure 6-7 shows interburst interval during baseline, treatment, and return to

baseline for valproate. This figure shows a large difference between baseline and

treatment means and variances for interburst interval. Only one culture, MEA 6692,

resulted in cessations of bursting that translates into an infinite interburst interval. The

remaining three cultures show an increased interburst interval that is maintained at a

steady state for the remainder of the treatment run. Similar to both phenytoin and











ethosuximide there was no difference between baseline and return to baseline and hence,

no evidence of carryover effects.


50.00

45.00

40.00

35.00

30.00

25.00

20.00

15.00

10.00

5.00

0.00


7073 6686


Figure 6-7. Interburst Interval for Valproate. Baseline, Treatment, and Return to Baseline
results shown for IBI in seconds (on the y-axis) for each of the four dishes
(shown on the x-axis) used with valproate. Note that IBI approaches infinity
as bursting ceased in culture 6692 and culture 7073's mean IBI was 120
seconds.

Figure 6-8 shows the correspondent burst durations for valproate. This graph does

not show any difference between baseline, treatment, and return to baseline means and

variances. There is no significant change in the burst duration when valproate is added to

a MEA culture. Even for culture 6692, those bursts that did occur were not different

from that of baseline burst durations.


i


* Baseline
*Treatment
OR to Baseline


6692


7076
MEA








46



1.20




1.00




0.80



3 H Baseline
S0.60 ETreatment
OR to Baseline



0.40 T




0.20




0.00
6692 7076 7073 6686
MEA




Figure 6-8. Valproate Burst Duration. Baseline, Treatment, and Return to Baseline
results for BD shown in seconds (on the y-axis) for each of the four dishes
(shown on the x-axis) used with valproate. Note that no difference is seen
between Baseline, Treatment, or Return to Baseline for all of the dishes.

Figure 6-9 shows the raster plots for dish 6686 (Appendix E contains the remaining


three dishes). Notice the smooth transition between insertion of the PHEMA ring at time


zero where activity resembles baseline activity and the drugged condition as the valproate


was released from the PHEMA ring.







47



Valproate dish 6686




0
c 0 100 200 300 400 500 600


0 100 200 300 400 500 600



E CO
20 -
So
E 60



0






0 100 200 300 400 500 600
times)

Figure 6-9. Dish 6686 Valproate Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording.

Summary Phase 1

Each drug resulted in an increase in the interburst interval over time. Phenytoin


eventually silenced all bursting in all four dishes resulting in a very large and infinite


(bursting ceases) interburst interval that translates to a zero bursting rate. Valproate and

ethosuximide slowed the bursting rates of three of four dishes in each case while

silencing the fourth dish. Burst duration was not effected by any of the three drugs. The

mean burst duration moved very little in any dish with any drug and the standard

deviations of the data heavily overlapped suggesting that there is no distinguishable

difference in the burst duration between baseline and drugged states.









Phase 2 Results

Stimulation of drugged dishes paralleled the results of Phase 1 in that phenytoin

had the largest effect on the cultures. Valproate and ethosuximide did little to prevent

bursting after an electrical stimulation. Figures 6-10 and 6-11 show that phenytoin

treatments reduce or eliminate the burst response to stimulation seen in baseline

recordings in-vitro.

The 20 stimulations events for each condition were recorded into a single file for

each condition. Each stimulation event starts with a 500mV 200us biphasic pulse. The

average firing rate for each run is determined by creating a histogram of all spikes

following the stimulation. The spikes that make up the resulting burst are sorted into ims

bins. The algorithm that does this sorting is shown in Appendix A. The resulting bins

are graphed in chronological order to show the average firing rate of the resulting burst

post-stimulation.

Figure 6-10 shows the response to stimulation for dish 6686. Phenytoin, shown in

the second graph had a noticeable effect on the average firing rate of the culture during

the initial early phase where a small number of action potentials can be seen. Phenytoin

almost completely eliminates the late phase bursting response. Both valproate and

ethosuximide show a small but consistent increase in average firing rate compared to the

baseline conditions over the first 30ms for this culture.

Figure 6-11 shows the response to stimulation for dish 1534. This was the first time

drugs were applied to this dish. The early phase for each drug and all baselines were

nearly identical. Valproate and phenytoin both show little deviation from the baseline

conditions in the late phase. The results are similar to dish 6686 in that only phenytoin










49




has a large noticeable effect on response to stimulation. However, dish 1534s average



firing rate for phenytoin during the late phase was only reduced not eliminated.


Dish 6686 Response to Stimulation


30 -

L 20 -

10 -

0
0


V -~y


time(ms)


30 -

S20 -

10 -

0
0


baseline 1
ethosuximide






151


baseline 2
- phenytoin
basln


time(ms)


30

S20



0
0


time(ms)


- baseline 3
valproate






150


100


Figure 6-10. Dish 6686 Response to stimulation. The y-axis of this figure is a count of

the average number of action potentials detected for each time bin over the 20

stimulation events.


150


Dish 1534 Response to Stimulation


baseline 1
phenytoin


baseline 3
ethosuximide





150


Figure 6-11. Dish 1534 Response to stimulation. The y-axis of this figure is a count of

the average number of action potentials detected for each time. This graph

shows the average shape of a burst from 20 stimulation events.


30 -

20 -

10 -
O0
0




20 -

10

0
0
40

30 -

20 -2

10


time(ms)


50


time(ms)


100


- - - - - -









The first 15ms after stimulation differ greatly for the two dishes. They both show

large spikes synchronized activity within the first 5ms. It is possible that some of this

initial activity is artifact from the stimulation. A dish-to-dish comparison shows

enormous differences during the early phase. Only phenytoin indicates any major

differences between baseline and treatment for dish 1534. After the initial 6-10ms

phenytoin demonstrates less than 40% of the activity of the baseline conditions. Dish

6686 shows differences in each case between baseline and treatment conditions. For

phenytoin there are fewer action potentials and only a few channels in close proximity to

the stimulating electrode that show and signs of action potentials. When drugs are present

for both valproate and ethosuximide the early phase demonstrates more action potentials.

Discussion and Conclusions

Discussion

If in-vitro bursting is similar to in-vivo epilepsy antiepilepsy the drugs used for

treating in-vivo epilepsy should have similar effects on in-vitro bursting. When a

properly diagnosed patient takes phenytoin for grand mal (tonic-clonic) epilepsy the

patient can expect to have a reduction in grand mal seizure activity (Levy et al. 2002).

Phenytoin also increases the voltage threshold for seizures caused by electrical

stimulation in animals and humans (electroshock therapy). Phenytoin has been shown to

be ineffective in the treatment of petit mal (absence) seizures. Phenytoin silences in-vitro

bursting within the first 6-7 minutes in these experiments and the response to stimulation

for phenytoin was similar to in-vivo. That is, bursting in response to stimulation is either

largely reduced or eliminated with phenytoin. The main mechanism of action for

phenytoin is the blockage of voltage sensitive sodium channels that are important for

initiating action potentials. When these channels are blocked the likelihood of a neuron









responding to electrical stimulation from electrodes or a nearby neuron is reduced (Hille

1992). It is logical that blockage of these channels would reduce epileptic seizures and

bursting in-vitro.

Ethosuximide is a successful drug for the treatment of absence seizures but fairly

ineffective in the prevention of grand mal seizures (Levy et al. 2002). Similarly the

experiment in-vitro indicates ethosuximide is the least effective of the three drugs tested

in terms of its ability to stop or slow the bursting. It does cause what appears to be a

sudden change in the burst rate for all treatment runs. This phase change results in a

slowed bursting rate but does not lead to a cessation of burst behavior. Ethosuximide

does little to nothing to prevent bursts in response to electrical stimulation in-vitro. The

application of ethosuximide may have actually increased the average firing rate of

neurons in the early phase of bursting in one dish. Its mechanisms of action, which are

poorly understood, do little to make it effective as an anticonvulsant in-vivo. In-vitro it is

not able to eliminate bursting or prevent bursting in response to stimulation.

Valproate is a drug with many actions within the nervous system that make it an

effective antiepilepsy drug for both grand mal and petit mal seizures(Levy et al. 2002).

In-vitro valproate shows similarities to phenytoin in that a smooth change in bursting rate

occurs over a long period of time. The data seems to indicate the steady state may not

have been reached in the experimental window for treatment conditions. It is possible

that longer exposure to valproate may cause bursting to cease in-vitro. However,

valproate is similar to ethosuximide in that it did not fully stop bursting in most dishes

and also did little to prevent bursting responses from electric stimulation. As one might









expect the results for valproate are mixed as exactly how valproate works to prevent

many forms of seizures are not fully understood.

Conclusions

Overall the experiments suggest that phenytoin was the most effective drug in the

prevention of bursting by neural cultures. Valproate was the next most effective drug

with ethosuximide having the least effect on the cessation of bursting. All three drugs

increased the interburst interval resulting in a slowed the bursting rate, and phenytoin

completely eliminates bursting. However, none of the drugs had an effect on burst

duration.

It stands to reason that if in-vivo epilepsy and in-vitro bursting share common

behaviors that treatment of bursting with antiepilepsy drugs should determine whether

they might share similar origins. In-vitro bursting does indeed show a similar response to

antiepileptic drugs as in-vivo epileptic seizures. It also stands to reason that the cultures

may respond better to drugs meant for a specific type of epilepsy. In fact this experiment

shows that in-vitro bursting responds better to one type of epilepsy treatment than others.

The effects of phenytoin on the neural cultures bursting behavior suggest that in-vitro

bursting may be most similar to in-vivo grand mal (tonic-clonic) seizures. However, since

the drugs tested only treat symptoms of seizures it is still possible that the causes of

seizure in-vitro and in-vivo may be very different.

This evidence does suggest that further investigation is warranted. The presence of

these effects is important not only for its relevance to similarities between the in-vivo and

in-vitro systems, but it would also open the possibility of the study of epilepsy in an in-

vitro system that permits detailed (multi-channel) long term recordings on a platform

where experimental manipulation is more easily accomplished.
















APPENDIX A
BURST DETECTION ALGORITM

Burstdet.m this is the main burst detection function it only requires data in the form of a
spikefile from MEABench.


function [bst,burstdata,globalrate] = burstdet(data)
time = 0.00004;
leaky = zeros(3,60);
burst = 0;


A = data.time;
B = data.channel;
N = length(A);
J= 1;
globalrate = A 0;
burstdata = A 0;
decay = 0.9999;

for I= 1:N
ifB(I) <= 59
%if A(I) >= 180

time = A(I) leaky(l,(B(I)+1));
lastspike = max(leaky(l,(B(I)+1)));
leaky(2,:) = leaky(2,:) (decayA(25000* timeme));
leaky(2,(B(I)+1)) = leaky(2,(B(I)+1)) + 1;
globalrate(I) = sum(leaky(2,:))/60;
leaky(l,(B(I)+1))= A(I);
%end
end
end
C= median(globalrate(1:I));
D = std(globalrate(1:I));
threshold = C + 0.25*D; %0.25

for I= 1:N
ifB(I) <= 59
%if A(I) >= 180










time = A(I) leaky(l,(B(I)+1));

leaky(2,(B(I)+1)) = leaky(2,(B(I)+1)) (decay^(25000* timeme)) + 1;
%globalrate(I) = sum(leaky(2,:))/60;

if burst == 1
lastspike = max(leaky(l,(B(I)+1)));
Z =I-1;
if(A(I) A(Z)) >= 0.05 %0.05
if(A(Z) bst(J,1)) >= 0.10
burst = 0;
burstdata(I-1) = 0;
bst(J,2)= A(Z);
bst(J,3)= bst(J,2) bst(J,1);
J =J+1;
end
end
end
if burst == 0
%if A(I) >= 180
if globalrate(I) > threshold
burst = 1;
bst(J,1) = A(I);
end
%end
end
burstdata(I)=burst;
leaky(l,(B(I)+1))= A(I);
%end
end
end
figure
subplot(3,1,1);plot(data.time,burstdata); axis([l 600 0 1.5]);
subplot(3,1,2);plot(data.time,globalrate); axis([l 600 0 2]);
subplot(3,1,3);plot(data.time,data. channel,'.','markersize',1);axis([1 600 0 60]);


burstrate.m this file calls the burstdet function to create files with burst data and also
creates files with statistical data on the bursts.

function [rundata,runstats] = burstdet(datal,data2,data3)

[burstsl,burstdatal,globalrate 1] = burstdet(datal);
[bursts2,burstdata2,globalrate2] = burstdet(data2);
[bursts3,burstdata3,globalrate3] = burstdet(data3);






55


rundata=
struct('burstsl', [bursts 1l],'burstdatal', [burstdatal ],'globalratel ',[globalratel ],'bursts2', burss
ts2],'burstdata2',[burstdata2],'globalrate2', [globalrate2],'bursts3', [bursts3],'burstdata3', [bur
stdata3],'globalrate3', [globalrate3]);
runstats = struct('rates',[length(burstsl) length(bursts2)
length(bursts3)],'meanburstduration',[mean(burstsl(:,3)) mean(bursts2(:,3))
mean(bursts3(:,3))],'stdburstduration',[std(burstsl(:,3)) std(bursts2(:,3)) std(bursts3(:,3))])


burst.m this file ties everything together when three sets of data (baseline, treatment,
return to baseline are used.

[rundata,runstats] = burstrate(datal,data2,data3)



















APPENDIX B
RASTER PLOTS FOR BICUCULLINE TREATMENTS




dish 6685


0 100 200 300 400 500 600
times)








o 100 200 300 400 500 soc
0 -- -----




20

0 100 200 300 400 500 600
times)


40





0 100 200 300 400 500 600
times)



Figure B-1. Raster plots of spike activity for dish 6685 during baseline, placebo, and
bicuculline treatment spike activity. Time is on the x-axis in seconds and
channel in hardware order is on the y-axis each point represents an action
potential from a nearby neuron.








57




dish 6688
so




d 20


0 100 200 300 400 500 600
times)
60.

S40-



0

0 100 200 300 400 500 600
times)
60 .- -i- -- i ; ; -; i ; i -.. -




20

0 1
0 100 200 300 400 500 600
times)


Figure B-2. Raster plots of spike activity for dish 6688 during baseline, placebo, and
bicuculline treatment spike activity. Time is on the x-axis in seconds and
channel in hardware order is on the y-axis each point represents an action
potential from a nearby neuron.













dish 7087


S40 -

m 20


0
0 100 200 300 400 500 609
times)


S40 -




0
0 100 200 300 400 500 60'
times)
60
C,

u 40
CO 0 0 _0 .0 _0 ..


300
times)


Figure B-3. Raster plots of spike activity for dish 7087 during baseline, placebo, and
bicuculline treatment spike activity. Time is on the x-axis in seconds and
channel in hardware order is on the y-axis each point represents an action
potential from a nearby neuron.



















APPENDIX C
RASTER PLOTS FOR PHENYTOIN






Phenytoin dish 6682


20 -

0
0 100 200 300 400 500 6C
60



20

0 : :
0 100 200 300 400 500 6C
CO *---------------------- ----


100 200 300 400 500
times)


C
0
E 40 -
N
c
ci2
(M





4C
t 20
oI
l- 600

n


0 0
(1 lo


Figure C-1. Dish 6682 Phenytoin Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording.


E
rC



E

Gi







60



Phenytoin dish 6692


100 200


100 200


300 400 500


300
times)


nn00 500


Figure C-2. Dish 6692 Phenytoin Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording


n00








61










Phenytoin dish 7073


c40-

20 -

0
0 100 200 300 400 500 600
E 60



20
i o
0 100 200 300 400 500 600
E 60

S40 -

S20 -

0
I- 600 700 800 900 1000 1100 1200

60

2CO


0 100 200 300 400 500 600
times)


Figure C-3. Dish 7073 Phenytoin Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording



















APPENDIX D
RASTER PLOTS FOR ETHOSUXIMDE


Ethosuximide dish 6682


100 200


300 400 500


E


0
0 eo-
E 60
o


20 -








0 0
l 600





^ O1


300


400


500


100 200 300 400 500
times)


Figure D-1. Dish 6682 ethosuximide Raster Plots. Each plot has channel number on the
y-axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording













Ethosuximide dish 6692


300


60 -

c 40

20

0
0 -
S60 -
C


E

00

E6
'r


Gi 0 o
FE 600
C
E 40
N
C
1 20


l- 600
2 so



r20
"u

Sn
CC 0~


300
times)


400


500


400 500


Figure D-2. Dish 6692 ethosuximide Raster Plots. Each plot has channel number on the
y-axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording


100 200


600








64














Phenytoin dish 7073


40

20 -

0
0 100 200 300 400 500 600




2 20 -
E so0




i 0 100 200 300 400 500 600
E 60

S40 -.

S20 -

r0
g 600 700 800 900 1000 1100 1200
j 60

40-

20
| 0
0 100 200 300 400 500 600
times)



Figure D-3. Dish 7073 ethosuximide Raster Plots. Each plot has channel number on the
y-axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording



















APPENDIX E
RASTER PLOTS FOR VALPROATE


Valproate dish 6692


60

- 40

20

0

E 60

do

61







E
600







60
0



S20 -

r 0C


300
times)


400 500


Figure E-1. Dish 6692 valproate Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording


100 200













Valproate dish 7073


-
60

c 40

20
& ^


(V


200


c 0
E 60 -
C
*a 40

E


20
4 20

20
E 60k



E
c 20





0
S600


S40


300


400


500


600


300
times)


Figure E-2. Dish 7073 valproate Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording








67




Valproate dish 7076


CO

c ,C

20

0
0 -



E
F so

.40
61

E o

20
'a d0 .





E
M
.4C
CiO-
ci
0


2 o
ri 40-




c
i 20





" 0
l 600


Figure E-3. Dish 7076 valproate Raster Plots. Each plot has channel number on the y-
axis and time on the x-axis. Each point on a given channel at a given time
represents an action potential from a nearby neuron. The first raster plot
represents the 10-minute baseline recording. The second and third represent
the continuous 20-minute treatment recording. The fourth represents the
return to baseline recording


200 300
times)















LIST OF REFERENCES


Banker, Goslin (1998) Culturing Nerve Cells. MIT Press: Cambridge, Massachusetts.

Bledi Y, Domb AJ, Linial M (2000) Culturing Neuronal Cells on Surfaces Coated by a
Novel Polyethyleneimine-Based Polymer. Brain Research Protocols 5:282-289.

Brenner GM (2000) Pharmacology. WB Saunders Company: Philadelphia, PA.

Bugard EC, Hablitz JJ (1993) Developmental Changes in NMDA and Non-NMDA
Receptor-Mediated Synaptic Potentials in Rat Neocortex. Journal of
Neurophysiology 69:230-240.

Claverol-Tinture E, Pine J (2002) Extracellular Potentials in Low-density Dissociated
Neuronal Culture. Journal of Neuroscience Methods 117:13-21.

DeLorenzo RJ, Pal S, Sombati S (1998) Prolonged Activation of the N-
methyl-D-aspartate Receptor Ca2+ Transduction Pathway Causes Spontaneous
Recurrent Epileptiform Discharges in Hippocampal Neurons in Culture.
Proceedings of the National Academy of Science USA 95:14482-14487.

DeMarse TB, Wagenaar DA, Blau AW, Potter SM (2001) The Neurally Controlled
Animat: Biological Brains Acting with Simulated Bodies. Autonomous Robots
11:305-310.

Egert U, Hammerle H (2002) Application of the Microelectrode-Array (MEA)
Technology in Pharmaceutical Drug Research. Baselt JP and Gerlach G (editors)
Sensoren im Fokus neuer Anwendugen Web Univeritatsverlag, Dresden: 51-54.

Egert U, Heck D, Aertsen A (2002) Two-dimensional Monitoring of Spiking Networks in
Acute Brain Slices. Experimental Brain Research 142:268-274.

Egert U, Schlosshauer B, Fennrich S, Nisch W, Fejtl M, Knott T, Muller T, Hammerle H
(1998) A Novel Organotypic Long-term Culture of the Rat Hippocampus on
Substrate-integrated Multielectrode Arrays. Brain Research Protocols 2:229-242.

Freeman WJ (1994) Neural Networks and Chaos. Journal of Theoretical Biology 171:13-
18.

Garcia PS, Calabrese RL, DeWeerth SP, Ditto WL (2002) Simple Arithmetic with Firing
Rate Encoding in Leech Neurons: Simulation and Experiment (Pre-print).









Available on the WWW at
http://wwwneuroengineeringcom/publications/ 111180pdf 7/6/04

Gluckman BJ, Neel EJ, Netoff TI, Ditto WL, Spano ML, Schiff SJ (1996) Electric Field
Suppression of Epileptiform Activity in Hippocampal Slices. Journal of
Neurophysiology 96:4202-4205.

Goldberg ea (2002) Polymeric Biomaterials EMA 6581 Laboratory Handout, Fall 2002.
Department of Materials Science and Engineering, University of Florida.

Gross GW, Kowalski JM (1999) Origins of Activity Patterns in Self-Organizing
Neuronal Networks in Vitro. Journal of Intelligent Material Systems and
Structures 10:558-564.

Gross GW, Rieske E, Kreutzberg GW, Meyer A (1977) A New Fixed-array
Multielectrode System Designed for Long-term Monitoring of Extracellular
Single Unit Neuronal Activity In-vitro. Neuroscience Letters 6:101-105.

Guehl D, Burbaud P, Boraud T, Bioulac B (2000) Bicuculline Injections into the Rastral
and Caudal Motor Thalamus of the Monkey Induce Different Types of Dystonia.
European Journal of Neuroscience 12:1033-1037.

Hille B (1992) Ionic Channels of Excitable Membranes. Sinauer Associates Inc.:
Sunderland, MA

Hoffman AS (2002) Hydrogels for Biomedical Applications. Advanced Drug Delivery
Reviews 43:3-12.

Igelmund P, Fleischmann BK, Fisher IR, Soest J, Gryshchenko 0, Bohm-Pinger MM,
Sauer H, Liu Q, Jurgen H (1999) Action Potential Propagation Failures in Long-
term Recordings from Embryonic Stem Cell-derived Cardiomyocytes in Tissue
Culture. European Journal of Physiology 437:669-679.

Jimbo Y, Kawana A, Parodi P, Torre V (2000) The Dynamics of Neuronal Culture of
Dissociated Cortical Neurons of Neonatal Rats. Biological Cybernetics 83:1-20.

Kamioka H, Maeda E, Jimbo Y, Robinson HPC, Kawana A (1996) Spontaneous Periodic
Synchronized Bursting During Formation of Mature Patterns of Connections in
Cortical Neurons. Neuroscience Letters 206:109-112.

Kapur J, Macdonald RL (1997) Rapid Seizure-Induced Reduction of Benzodiazepine and
Zn2+ Sensitivity of Hippocampal Dentate Granule Cell GABAa Receptors. The
Journal of Neuroscience 17:7532-7540.









Keefer EW, Gramowski A, Gross GW (2001) NMDA Receptor-Dependent Periodic
Oscillations in Cultured Spinal Cord Networks. Journal of Neurophysiology
86:3030-3042.

Lelong IH, Petegneif V, Rebel G (1992) Neural Cells Mature Faster of Polyetyleneimine
Coated Plates Than on Polylysine Coated Plates. Journal of Neuroscience
Research 32:562-568.

Levy RH, Mattson RH, Meldrum BS, Perucca E (2002) Antiepileptic Drugs, 5 Edition.
Lippincot, Williams, & Wilkins: Philadelphia, PA.

Luhmann HJ, Prince DA (1991) Postnatal Maturation of the GABAergic System in Rat
Neocortex. Journal of Neurophysiology 65:247-263.

Marieb E (1998) Human Anatomy & Physiology, 4 Edition. Benjamin/Cummings
Science Publishing: Menlo Park, CA.

Meister M, Pine J, Bayor DA (1994) Multi-neuronal Signals from the Retina: Acquisition
and Analysis. Journal of Neuroscience Methods 51:95-106.

Meister M, Wong ROL, Baylor DA, Shatz CJ (1991) Synchronous Bursts of Action
Potentials in Ganglion Cells of the Developing Mammalian Retina. Science
252:939-943.

Moorefield SI, Keefer EW, Chapman KD, Gross GW (1999) Drug Evaluations Using
Neuronal Networks Cultured on Microelectrode Arrays. Biosensors &
Bioelectronics 15:383-396.

Nuttleman CR, Henry SM, Anseth KS (2002) Synthesis and Characterization of
Photocrosslinkable, Degradable Poly(vinyl alcohol) -Based
Tissue Engineering Scaffolds. Biomaterials 23:3617-3626.

Pine J (1980) Recording Action Potentials from Cultured Neurons with Extracellular
Microcircuit Electrodes. Journal of Neuroscience Methods: 19-31.

Potter SM, DeMarse TB (2001) A New Approach to Neural Cell Culture for Long Term
Studies. Journal of Neuroscience Methods 110:17-24.

Ratner (1996) Biomaterials Science: An Introduction to Materials in Medicine. Academic
Press: San Diego, CA.

Razet R, Thomet U, Furtmuller R, Jursky F, Sigel E, Sieghart W, Dodd RH (2000) Use of
Bicuculline, a GABA Agaonist, as a Template for the Development of a New
Class of Ligands Showing Positive Allosteric Modulation of the GABAa
Receptor. Bioorganic & Medicinal Chemistry Letters 10:2579-2583.









Segal MM (1994) Endogenous Bursts Underlie Seizurelike Activity in Solitary
Excitatory Hippocampal Neurons in Microcultures. Journal of Neurophysiology
72:1874-1884.

Shahaf G, Marom S (2001) Learning in Networks of Cortical Neurons. The Journal of
Neuroscience 21:8782-8788.

Stenger DA, McKenna TM (1994) Enabling Technologies for Cultured Neural Networks.
Academic Press: San Diego, CA.

Tateno T, Jimbo Y (1999) Activity-dependent Enhancement in the Reliability of
Correlated Spike Timings in Cultured Cortical Neurons. Biological Cybernetics
80:45-55.

Thomas CA, Springer PA, Loeb GE, Berwald-Netter Y, Okun LM (1972) A Miniature
Microelectrode Array to Monitor the Bioelectric Activity of Cultured Cells. Exp
Cell Res 74:61-66.

Tosney KW, Landmesser LT (1985) Specificity of Early Motorneuron Growth Cone
Outgrowth in the Chick Embryo. Journal of Neuroscience 5:2336-2344.

Watanabe S, Jimbo Y, Kamioka H, Kirino Y, Kawana A (1996) Development of Low
Magnesium-induced Spontaneous Synchronized Bursting and GABAergic
Modulation in Cultured Rat Neocortical Neurons. Neuroscience Letters 210.















BIOGRAPHICAL SKETCH

Alex Cadotte received his B.S. in chemical engineering from Virginia Tech in

1999. He worked in industry from 1999 to 2002 at companies such as General Electric

and Corning Optical Fiber in Wilmington, NC. Alex moved to Gainesville, FL, in the

Fall of 2002 to pursue a Master of Science at the University of Florida in biomedical

engineering specializing in tissue and neural engineering. This thesis is a summary of the

work done by Alex under the mentorship of Dr. Thomas DeMarse in this field of study.