LUMINESCENCE RESONANCE ENERGY TRANSFER STUDIES OF …research.physics.illinois.edu/Publications/theses/copies/Posson.pdf · channels open when mechanical membrane stress or bending

LUMINESCENCE RESONANCE ENERGY TRANSFER STUDIES OF THE SHAKER K+ VOLTAGE-GATED ION CHANNEL

BY

DAVID JOHN POSSON

B.S., University of Cincinnati, 1997

DISSERTATION

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Physics

in the Graduate College of the University of Illinois at Urbana-Champaign, 2005

Urbana, Illinois

LUMINESCENCE RESONANCE ENERGY TRANSFER STUDIES OF THE SHAKER K+ VOLTAGE-GATED ION CHANNEL

David John Posson, Ph.D.

Department of Physics University of Illinois at Urbana-Champaign, 2005

Paul R. Selvin, Advisor

Members of the superfamily of voltage-gated ion channels are the molecular components

underlying electrical excitability in nerves and muscle. Voltage-gated channels allow the

selective flow of ions across the hydrophobic lipid bilayer of a cell, opening and closing

in response to changes in the voltage across the membrane. The Shaker K+ channel is a

standard model system for studying the structure function relationships in this important

class of ion channels. Voltage sensing is known to involve a highly charged segment of

the channel called S4. When a channel opens it moves some of these S4 charges across

the membrane electric field. The movement of four S4s, one from each identical subunit

of Shaker, is coupled to the “gate” which opens and closes the pore. Therefore, a central

question for understanding the functionality of these proteins is; how exactly does S4

move? Recently, the first crystal structure for a voltage-gated K+ channel was solved.

This structure, of the KvAP channel, led the authors to propose a new model of S4

movement. This new model, called the paddle model, hypothesized a large translational

motion of S4 across most of the lipid bilayer thickness, a view that has been very

controversial. In this study, we examine the conformational movements associated with

the S4 segment during voltage sensing. We use a technique called Lanthanide Resonance

Energy Transfer (LRET), which gives an accurate distance measurement between two

positions on the ion channel. We use two different configurations for LRET on Shaker.

The first measures distances between the four identical S4 amino-acid sites on the

homotetrameric K+ channel. The second measures distances between S4 sites and a

scorpion toxin bound to the symmetric axis of the channel just above the pore. These

LRET studies argue strongly against the paddle model of voltage sensing, and

demonstrate that the physical movements of the S4 segments of Shaker K+ channels are

quite small.

iii

Acknowledgements

So many people helped me out at every stage of grad school and in so many different

ways. I thank them all for their generosity.

• My parents and sisters have been my base of support for 30 years and counting.

• Paul Selvin taught me everything I know about LRET and how to do fluorescence

experiments. I thank him for passing on his techniques and skills and most of all

for believing in me during times when I was not the most ideal of grad students

and when I was just down right flakey.

• Francisco (Pancho) Bezanilla and Dorine Starace taught me so much about the

Shaker channel and about electrophysiology in general. Pancho’s knowledge is

inspirational and Dorine provided support both professional and personal.

Without them I could not have succeeded.

• Chris Miller was instrumental with his scorpion toxin preparations. Without his

help my Ph.D. would have turned out a bit more scrappy than it did. I thank him

most for his heady encouragement that buoyed me along through the final stages

of this work.

• I thank Pinghua Ge for his skill at preparing lanthanide chelates. Without his

synthetic abilities, none of my experiments would have been possible. I would

also like to thank Ming Xiao for determining the chelate’s quantum yield and Jeff

Reifenberger for measuring the anisotropy.

• Greg Snyder provided initial instruction in voltage clamping and oocyte work.

Both Greg and his wife, Tania, were wonderful colleagues in the Selvin lab and I

miss them.

iv

• Anne Gershenson taught me basic mutagenesis and provided friendship with a

needed dose of mentorship. For her crucial help, I am deeply grateful.

• Michel Bellini passed on his knowledge of Xenopus, including the surgical

procedure.

• Bob Gump and his staff at the Morrill hall animal facility took great care of our

frogs.

• Sara Chalifoux and Lisa Klodnicki performed numerous frog surgeries and oocyte

preps.

• Tatyana Lawrecki helped with surgeries and oocyte preps through the most

productive period these two years past. For her loyalty and the unwavering

quality of her efforts I am very grateful.

• I thank the Aldrich lab for providing the ILT-Shaker plasmid.

• Benoit Roux kindly combined the coordinates for two of his models, the Shaker

open state and the agitoxin-Shaker complex. I thank him for his general

enthusiasm for my experiments.

• Dane Sievers provided silicon wafers and greatly assisted with the oxide growth

for the silicon quenching project.

• Other members of the Selvin lab have provided friendship and support - Evan

Graves, Ahmet Yildiz, Comert Kural, Hyokeun Park, Hamza Balci, Sheyum

Syed, & Erdal Toprak. Jeff Reifenberger has been a good roommate, lab

colleague, and a great friend through some tough times.

v

• I have gratefully received personal support from Scott Stewart, Paul Melby,

Patrick Hentges, Trevor Vickey, & Mary Upton. Patrick showed me what

personal courage really is all about.

• I acknowledge the financial support for this work with NSF MCB99-84841,

Carver Foundation, and Cottrell funds of the research corp. CS0706

vi

Table of Contents

Chapters

1 Introduction..................................................................................... 1 1.1 Ion channels – the cartoon picture.............................................................1

1.2 Classical membrane topology of the Shaker K+ channel ..........................5

1.3 How does an ion channel handle its charge?............................................. 8

1.4 K+ channels – the atomic picture............................................................... 16

1.4.1 The KscA structure – a closed K+ channel ...................................... 17

1.4.2 The MthK structure – an open K+ channel ...................................... 18

1.4.3 The MscS structure – a slightly voltage sensitive, non-selective

mechanosensitive channel......................................................................... 20

1.4.4 The KvAP structure – a full-fledged, highly voltage-dependent

channel ...................................................................................................... 22

1.4.5 MacKinnon goes to Stockholm........................................................ 31

2 Electrophysiology of the Shaker K+ Channel.................................. 32 2.1 The Xenopus laevis oocyte expression system.......................................... 32

2.2 The two-electrode voltage clamp .............................................................. 33

2.3 Ionic currents ............................................................................................. 37

2.4 Gating currents .......................................................................................... 39

2.5 The ILT-Shaker phenotype ....................................................................... 45

3 Fluorescence Spectroscopy Methods .............................................. 47 3.1 Introduction to fluorescence...................................................................... 47

3.2 Resonance Energy Transfer (RET) theory ................................................ 50

3.3 Lanthanide Resonance Energy Transfer (LRET) ...................................... 52

3.4 LRET meets the voltage clamp ................................................................. 56

3.5 Instrumentation.......................................................................................... 65

vii

4 LRET Part I: The ILT-Shaker Channel .......................................... 66 4.1 What can measurements on the ILT channel tell us? ................................ 66

4.2 Initial LRET results on the ILT channel ................................................... 67

5 LRET Part II: Shaker with Scorpion Toxin.................................... 72 5.1 Ion channels are toxin receptors ................................................................ 73

5.2 LRET configuration using acceptor labeled toxin – putting the paddle

model to the test................................................................................................ 74

5.3 LRET results S3b, S3-S4 linker, and S4 ................................................... 78

5.4 Comparison to a model for Shaker............................................................ 81

5.5 Conclusions ............................................................................................... 88

5.6 Future experiments .................................................................................... 89

Appendices ............................................................................................ 92

A Molecular Biology........................................................................... 92 A.1 Shaker constructs....................................................................................... 92

A.2 Primers....................................................................................................... 96

A.3 Mutagenesis ............................................................................................... 100

A.4 mRNA synthesis ........................................................................................ 100

A.5 Toxin biochemistry.................................................................................... 102

B Animal Use Protocol ....................................................................... 103

C Xenopus Oocyte Preparation ........................................................... 111

D Electrophysiology Solutions ........................................................... 114

E Silicon Quenching Unbinding Bioassays........................................ 116

References.............................................................................................. 125

Vita......................................................................................................... 140

viii

Chapter 1

Introduction

1.1 Ion channels – the cartoon picture

Ion channels are protein pores that allow ions to flow across the otherwise

impermeant cell membrane. Most channels are highly selective for a single ion such as

Na+, K+, Ca2+, or Cl-. Well-conserved pore structures, called ‘selectivity filters’, confer

this specificity. Another essential structural element, called the ‘gate’, opens to permit

the flow of ions or closes to block the pore. Ion channels usually couple their gates to

some external influence such as ligand binding (‘ligand-gated’ channels), membrane

voltage (‘voltage-gated’ channels), or mechanical force (‘mechanosensitive’ channels).

Figure 1.1 shows cartoon representations of these ion channel types.

All animal, plant, and bacterial cells have membranes in order to keep inside

‘stuff’ separate from outside ‘stuff’. Ionic concentration gradients are maintained across

their plasma membranes and a negative voltage is present across the membrane when the

cell is at rest [1]. It is customary to express membrane voltage as Vintracellular – Vextracellular,

and let Vextracellular = 0. Animal cells typically have membrane voltages between –60 mV

and –100 mV at rest. Therefore, ionic currents across cell membranes depend on the

electrochemical gradients as well as the presence or absence of open ion channels. Table

1.1 gives the measured concentrations of the biologically relevant ions in mammalian

skeletal muscle. The Equilibrium Potential, the voltage at which the concentration

gradient exactly balances the electrical gradient, is given by equation 1.1. Zn is the

valence, [N]out is the extracellular concentration, and [N]in is the intracellular

concentration of ion n. Equation 1.1 is known as the Nernst equation.

in

out

nn

NN

FZRTE

][][ln= (1.1)

1

+ +

+ +

++ + + + ++

+ +

+ ++ + + + +

c

Closed Open

LL

L LaNeurotransmittergated

L

LL L

bSecondmessengergated

Voltage-gated

V = -100 mV V = 0 mV

dMechanicalForcegated

Figure 1.1 All ion channels have ‘gates’ (red doors) that open and close the channel. Different

types of channels open in response to different stimuli. a. Neurotransmitter-gated channels bind

diffusional ligands from the external solution in order to transfer electric excitation from one

neuron to another. b. Some ligand-gated channels open in response to intracellular second

messengers such as Ca2+. c. Voltage-gated channels have charged protein segments that move

in response to membrane voltage changes thereby opening the gate. d. Mechanosensitive

channels open when mechanical membrane stress or bending is applied.

Table 1.1 Free Ion Concentrations and Equilibrium Potentials for Mammalian Skeletal Muscle at

37ºC. Adapted from Hille, Ion Channels of Excitable Membranes Third Edition, pg. 17,

Sinauer Associates, 2001 [1].

Ion Extracellular Concentration (mM)

Intracellular Concentration (mM)

Equilibrium Potential (mV)

Na+ 145 12 +67 K+ 4 155 -98 Ca2+ 1.5 0.1 +129 Cl- 123 4.2 -90

2

Examining table 1.1 we note that the equilibrium potentials for K+ and Cl- are

close to the resting potential of –90 mV for the muscle cell. Na+ and Ca2+ on the other

hand have equilibrium potentials far from the resting potential. K+ and Cl- set and

stabilize the resting potential while Na+ and Ca2+ tend to drive or ‘excite’ the membrane

towards positive potentials. For this reason K+ and Cl- channels are fundamental to all

cells, with the K+ family displaying extraordinary diversity. Na+ channels are much less

diverse and are found more specifically in excitable cells.

Hille [1] broadly defines an excitable cell as any cell expressing voltage-gated

Na+ and Ca2+ channels. For the most part, we consider nerves and muscle as the

prototypical excitable cells. In these cells, we can easily appreciate the importance of the

large superfamily of voltage-gated ion channels. Although these channels have other

functions as well, it is perhaps clearest to introduce them as the molecular components

underlying the propagating ‘action potentials’ of neurons.

The action potential is the basic unit of electrical signaling in nerves and muscle.

Figure 1.2 shows an early recording of an action potential recorded from a giant squid

axon [2]. During the first half of the 20th century, Hodgkin, Huxley, and others set out to

understand the ionic basis of membrane excitation. For a detailed description of the

experimental history, see [1] and [3]. Armed with a newly invented instrument, the

‘voltage clamp’, Hodgkin and Huxley performed detailed studies of ionic conductance

changes in the giant squid axon [4-6]. These experiments made possible a quantitative

description of the action potential using an empirical kinetic description of the observed

membrane conductance changes [7]. At the time of the ‘HH model’, the molecular basis

of ionic flow across membranes was unknown. Today, we understand that ion

permeation occurs through ion channel proteins and describe the action potential with this

language, rather than referring merely to ionic conductance.

The action potential is simply a local, transient membrane voltage change towards

positive potentials. As a matter of vocabulary, we say the membrane is ‘depolarized’

whenever the membrane potential is more positive than the resting state. Therefore, an

action potential is a transient depolarization, even though the membrane voltage may

become positive. We say the membrane is ‘hyperpolarized’ whenever the membrane

potential is more negative than the resting state. In the nerve axon, the upstroke of the

3

action potential is generated by a sudden influx of Na+ ions through voltage-gated Na+

channels, which drives the membrane potential towards the Nernst equilibrium potential

for Na+ (Table 1.1, Fig. 1.2). The Na+ channels are designed to close after being open for

a short time, a process common to all voltage-gated channels, called ‘inactivation’. At

the same time, the membrane depolarization caused by sodium entry opens up voltage-

gated K+ channels allowing K+ to flow out of the cell, which drives the membrane

potential back to the resting state. Figure 1.2, right plots these ionic permeability changes

as a function of time. The action potential propagates along the nerve axon because the

local influx of depolarizing Na+ ions spread via electrodiffusion to neighboring

membrane regions and initiate an above-threshold depolarization to activate downstream

voltage-gated Na+ channels. In this manner, the process repeats down the length of the

nerve cell. The ability to propagate action potentials is ‘hard-wired’ or ‘programmed’

into the nerve axon by the voltage-gated channels present in the membrane.

Na+ rushes into the cell causingswift depolarization.

1ms

mV

K+ channels openre-polarizing the membrane with

outward K+ current.

Figure 1.2 Nerve and muscle cells pass transient electrical pulses along their membranes as

information. Left. An action potential recorded by Hodgkin and Huxley from a giant squid axon.

Right. A plot showing the time course of membrane permeability changes that produce the action

potential. Na+ rushes into the cell causing membrane depolarization followed by quick

inactivation of Na+ channels. The membrane depolarization then causes K+ channels to open

after a short delay, resulting in restoration of the resting potential.

Hodgkin and Huxley (H & H, Fig. 1.3) had no way of knowing what actually

constituted the ionic pores they were studying, however they logically hypothesized the

4

existence of channels that were gated with voltage. Furthermore, they suggested the

kinetic data implied the channels were controlled by several independent membrane-

bound ‘particles’ (they used four independent particles for their model [7]). These

particles should carry electrical charge in order to make their movements sensitive to

voltage. H & H pointed out that the motion of these charged gating-particles should also

produce a small detectable electrical current preceding ionic current. Twenty years later,

their prediction was validated with the first recording of ‘gating currents’ [8-10] (see

Chapter 2.4). In modern terms, the gating-particles are now called the ‘voltage-sensors’.

The charges on these sensors are called the ‘gating charges’.

Nobel Prize in Physiologyor Medicine, 1963"for their discoveries concerning the ionic mechanisms involved in excitation and inhibition in the peripheral and central portions of the nerve cell membrane"

Alan Lloyd Hodgkin Andrew Fielding Huxley

Figure 1.3 Hodgkin and Huxley shared the Nobel Prize in physiology or medicine with Sir John

Carew Eccles in 1963. Their groundbreaking electrophysiological work epitomized the peak of

‘classical biophysics’.

1.2 Classical membrane topology of the Shaker K+ channel In this work we study the conformational changes underlying voltage sensing in a

voltage-gated K+ channel from drosophila melanogaster (fruit fly) called Shaker. Shaker

was the first K+ channel to be cloned. The channel was identified and named because

mutations in this gene caused flies to shake their legs under anesthesia [11]. Similar

structural principles are assumed to underlie all voltage-gated channels because broad

sequence homology exists across the entire superfamily of channels. NaV and CaV

5

channels (modern nomenclature for voltage-gated Na+ and Ca2+ channels) have four

repeat domains I, II, III, and IV (Fig. 1.4a). KV channels (modern nomenclature for

voltage-gated K+ channels) have analogous sequence structures made from four identical

subunits – they are homotetramers (Fig. 1.4b) [12]. The proteins in Fig. 1.4 are called

channel α-subunits because they are the principal channel-forming subunits. Other

proteins called ‘auxiliary’ or ‘regulatory’ subunits can interact with the channel and

modulate various aspects of channel properties. Here we concern ourselves with KV

channels in the absence of auxiliary subunits. The KV α-subunit is made of 6

transmembrane helices, denoted S1-S6, and a pore-forming loop (P) that contains the K+

specific selectivity filter. Four α-subunits associate (tetramerization) in the membrane to

form a rotationally symmetric protein with a central ion conduction pore (Fig 1.5).

a NaV or CaV α-subunit

¼ of aKV channel(1 α-subunit)

bS1 S2 S3 S4 S5 P S6

Figure 1.4 Membrane topology for all voltage-gated channel α-subunits (principal channel-

forming subunits). a. The α-subunit for NaV and CaV channels consist of four repeat domains that

form the channel. b. The α-subunit for KV channels make up ¼ of a channel and tetramerization

6

of four KV α-subunits forms the channel. Each primary subunit of the KV channel has 6

transmembrane segments, S1-S6, and a pore-forming loop (P). S4 is called the ‘principle

voltage-sensor’ because it has intrinsic charge that constitutes most of the ‘gating charge’.

K+

K+

KV Channel tetramerFour identical subunits surround the K+ conduction pathway.

K+S5

S6S4

S3

S2S1

Figure 1.5 Four KV α-subunits form a rotationally symmetric homotetramer with the K+ conduction

pathway along the central axis. Cartoon channel viewed from above (left) with hypothetical S1-

S6 positions indicated for one subunit.

The channel topology of Fig. 1.4 is further divided into two major functional

parts. S5-S6 is called the ‘pore-domain’ because these segments make up both the pore

and the gate(s). Every K+ channel contains a pore domain homologous to S5-S6. S1-S4

7

is called the ‘voltage-sensor domain’ because it contains the structural elements

responsible for voltage-sensitive gating. S4 is called the ‘primary voltage-sensor’

because it is highly charged and interacts with the membrane electric field in order to

couple the gate (open probability) to voltage changes (cartoon Fig. 1.1c). The S4

segments are the charged ‘gating particles’ hypothesized by Hodgkin and Huxley. The

conformational changes of the voltage-sensor domain, particularly of the S4 segment, are

of great interest. In this study we use a spectroscopic technique called lanthanide

resonance energy transfer (LRET, see Chapters 3-5) to determine the physical

movements responsible for voltage sensing.

1.3 How does an ion channel handle its charge? The fundamental task of an ion channel is to provide a pathway for strongly

hydrophilic ions to cross the greasy, hydrophobic membrane barrier. Putting an ion into a

membrane (dielectric constant ε = 2) from water (ε = 80) costs about half the total

hydration energy of the ion, typically about 40-50 kcal/mol. It is expected that the

channel’s conduction pathway should be polar, so that the passageway will be

energetically favorable for charged ions. In fact, as early as the 1970’s, scientists

(including Clay Armstrong) had deduced the general topology of the interior conduction

pathway for K+ and Na+ channels. The principal gate was at the cytoplasmic side of the

pathway followed by an interior aqueous vestibule that can bind channel blockers

depending on whether the gate is open or closed [13]. Between the central vestibule and

the outside solution, the protein pathway narrows into a polar selectivity filter. These

features are shown in a cartoon from Hille, 1977 (Fig. 1.6) [14]. These early predictions

were elegantly validated by K+ channel X-ray crystallography (see Section 1.4).

8

Early cartoonrepresentationof the ion conductionpathway.Hille 1977.

Figure 1.6 The ion conduction pathway through an ion channel is hydrophilic. In between the

gate (inside face) and the selectivity filter (outside face) resides a wide aqueous cavity that can

contain fully hydrated ions. The selectivity filter is assumed to be very narrow and polar to

support high fidelity conduction of a particular ion.

We have established that a channel is voltage sensitive when its gate is coupled to

a charged voltage-sensor domain that moves across the membrane electric field. Since

the electric field falls across the membrane, it follows that the charged voltage-sensor

must be located in the membrane. Therefore, KV channels have to solve the problem of

moving charges across the membrane barrier twice; 1) the central ionic conduction

pathway that we have described above and 2) a ‘gating-charge conduction pathway’. X-

ray crystallography has provided exquisite molecular detail of the ionic conduction

pathway (see Section 1.4) however the structure and mechanism of the voltage-sensor

domain and gating charge motion are still under very active investigation.

The S4 segment has a very particular arrangement of positively charged residues

that is conserved in all voltage-sensitive channels. Arginines (R) and lysines (K) occur

every three residues along the S4 (Fig. 1.7a), which if folded into an α-helix would

produce a stripe of charge that slowly wraps around the helix (Fig. 1.7b). The positively

charged S4 segments are in their closed position when the membrane voltage is negative.

Membrane depolarization causes these voltage-sensor segments to move ‘outward’

causing a net motion of gating charge from the inside solution to the external solution.

9

For the Shaker KV channel, the total gating charge per channel is about 12-14 qe moving

across the membrane electric field [15-17] during channel opening. Through mutational

analysis, most of this gating charge for Shaker has been attributed to the first four

arginines on the S4 (R362, R365, R368, and R371), although one acidic site (E293) on

S2 also contributes to the charge movement [16,17].

a b

Figure 1.7 a. Amino acid sequences of S4 segments from various voltage-gated K+ channels with

positively charged residues highlighted. The amino acid numbers for Shaker charges are shown

on top. b. Positive charges every three amino acids results in a stripe of charge that slowly wraps

itself around an alpha helical secondary structure. Figures adapted from Isacoff, 2002 [18].

It is generally believed that the intrinsic charges on S4 must be kept isolated from

the low dielectric environment of the lipid membrane for energetic reasons. To

accomplish this, the structure of the protein was expected to bury the charged S4 face

against other protein segments, such as S2, that contain counter charges (acidic residues

and partial charges). This expectation turned out to be only partially correct, as many of

the basic residues have been shown to be in direct contact with either internal or external

water. Numerous studies on Shaker have established the aqueous accessibility of

cysteine substitutions along the S4 to thiol-reactive reagents (MTS reagents) applied from

either the inside or the outside solution [19-22]. Only a small fraction (~10 amino acids)

of S4 is not in contact with water and moving the channel from closed to open shifts

which residues are buried. These changes are shown in a topological cartoon (Fig. 1.8)

taken from Larsson et al., 1996 [19].

10

S4

S4

Figure 1.8 Topological changes in aqueous accessibility along the charged S4 segment moving

from closed (left) to open (right) states. Figure from Larsson et al., 1996 [19].

Accessibility results demonstrating a watery-environment for S4 motivated a

revised generation of voltage-sensor models which include watery invaginations that

penetrate the protein and put much of S4 in contact with the inside and outside solutions.

These watery crevices are imagined to form a ‘gating canal’ or ‘gating-pore’ through

which the S4 moves its gating charges. In general, three types of protein movement have

been commonly used to model voltage-gating: (1) S4 translates in the “up” direction,

perpendicular to the membrane, towards the external solution [19,23,24]. This type of

motion is implied in the topological diagram Fig. 1.8 and the first cartoon Fig. 1.1c. (2)

S4 rotates about its axis, moving charges from one aqueous crevice to another [25,26].

(3) Aqueous crevice reshaping [27-29]. These three types of protein rearrangements can

exist in any number of combinations. For instance, a model that includes both vertical

translation and rotation describes the motion of S4 as a ‘helical screw’ (Fig. 1.9). The

motivation for adding a twisting movement to a vertical translation of S4 is that the

counter-charges located on surrounding protein segments can then be stationary while

11

successive S4 charges pass by. Figure 1.10 shows a cartoon version of the crevice

reshaping model. Figure 1.11 shows the a purely rotational model for S4 proposed in

Cha et al., 1999 [25] (discussed at length in Chapter 3.5).

Voltage-DrivenHelicalScrew

+

++

++

++

-

+

++

+

++

+-

Figure 1.9 Helical Screw Model for S4 movement through an aqueous ‘gating canal’. Only the

gating canal is pictured. Upon channel opening, the S4 segment rotates while translating

outward into the external solution, carrying the gating charges outward. This cartoon is a

realization of the model presented in Gandhi and Isacoff, 2002 [18].

12

Voltage-DrivenCreviceShaping

+

++

++

++

+

++

++

++

--

No TranslationNo Rotation

Figure 1.10 Crevice reshaping model. Only the gating canal is pictured. S4 movement is not

required to move gating charge across the membrane field if the gating canal changes shape

during voltage-driven opening.

13

Figure 1.11 Pure rotation model for S4 movement. Top – channel closed state. Two subunits

of the entire channel are shown, with the other subunits removed for clarity. The S4 gating

charges are exposed to an internal crevice, outlined in blue. Bottom – open channel state. 180

degree rotation of S4 moves the gating charges from the internal crevice to the external crevice,

outlined in magenta. No transmembrane displacement of S4 is predicted in this model. Figure

taken from Cha et al., 1999 discussed in Chapter 3.5 [25].

In 2003, Roderick MacKinnon’s lab published a new model of voltage-sensing

based on their crystal structure of an archaebacterial voltage-gated channel, KvAP [30].

For a complete discussion of their data see below, Section 1.4.4. The new model was

14

called the “paddle model” because the principal voltage-sensor S4 was observed to form

a helix-turn-helix motif with the S3 (actually just part of S3, called S3b, Section 1.4), and

the structure looked paddle-like. The paddle model consists of two very striking features

that fly in the face of the generally accepted views shared among the conventional models

described above: (1) The S4 voltage-sensor was placed at the periphery of the protein

and in particular, the charges were allowed to make contact with the lipid membrane,

despite the energetic cost such exposure implies. (2) The S3b-S4 paddle structure was

hypothesized to undergo a large transmembrane displacement, generally from the bottom

of the membrane to the top (Fig. 1.12, taken from Jiang et al. 2003 [30]). As the paddle

was hypothesized to undergo such a large displacement (15-20 Å [31]) through the

membrane environment, it was described as a highly mobile hydrophobic cation.

Closed Channel Open ChannelPaddle Down Paddle Up

Figure 1.12 MacKinnon’s “paddle model” in cartoon form from Jiang et al. 2003 [30]. Left. In

the closed channel state, the paddles (S3b-S4 segments) are near the intracellular solution at the

periphery of the protein. Right. Membrane depolarization causes the paddles to move upward

towards the outside solution, transporting their charge across the membrane and pulling on the

K+ pathway gate.

15

In this dissertation, we examine the motion of the primary voltage-sensor S4 and

other segments. Experiments have tested both motions parallel to the membrane (Section

3.5 and Chapter 4) and perpendicular to the membrane (Chapter 5). We have established

that voltage-sensing segments undergo very small movements, and in particular the

vertical movement of sites on S4 is ~ 2 Å. Our work greatly constrains the type of

models that can be used to describe gating charge movement (Chapter 5.5, conclusions),

and in particular, we conclude the paddle model does not describe the true mechanism of

voltage sensing.

1.4 K+ channels – the atomic picture X-ray crystallography has greatly advanced the study of biological

macromolecules by resolving three-dimensional structures with atomic detail (2-3 Å

resolution). However, getting proteins to form an ordered crystal is not always

straightforward and is notoriously difficult for integral-membrane proteins such as ion

channels. The other stumbling block has been the need for milligrams of protein for

crystallization trials and so a basic molecular-biological problem requires resolution.

Great progress has been made in overcoming these problems, although much progress is

surely yet to come. Firstly, the genomic era has uncovered a startling fact: bacterial

organisms have ion channels of every important type, even those thought to be highly

specialized such as voltage-gated channels. Some of these bacterial channels can be

expressed at high levels (especially K+ channels) and purified using standard biochemical

procedures. Therefore, the protein quantity problem for crystallization trials is not

intractable. Secondly, the field has been cracked open by the heroic efforts of Roderick

MacKinnon, ion channel biophysicist-turned-crystallographer. MacKinnon has built his

laboratory on the sound principle of combining traditional structural-functional methods

– mutagenesis, electrophysiology, and the like – with structure determination. This

section summarizes in detail the relevant ion channel structures, paying special attention

to K+ channels.

16

1.4.1 The KcsA structure – a closed K+ channel. KcsA was the first ion channel

MacKinnon’s lab successfully crystallized [32]. It is a K+ channel from the bacterium

Streptomyces lividans [33] and is proton-gated. This ground-breaking structure at 3.2 Å

resolution showed how the K+ conduction pathway works, how the channel rejects Na+

but passes K+ at near diffusional limits. It was now possible to understand the pore

domain of potassium channels in great detail. As was shown in Fig. 1.6 above, classical

studies had outlined how the pore was thought to be shaped. The physical gate was

thought to reside near the intracellular side and the channel center should have an

aqueous vestibule that prepares ions to enter into close physical contact with a selectivity

filter. These predictions were beautifully shown by the atomic structure (Fig. 1.13a).

All potassium channels have a “signature sequence”, minimally GYG but very

often TVGYG. Mutations of these residues had resulted in decreased K+ selectivity so it

seemed likely that the signature sequence lined the selectivity filter [34,35].

MacKinnon’s laboratory solved a higher resolution structure (2 Å) of KcsA by

complexing the channel with monoclonal Fab antibody fragments [36]. At this

resolution, detailed protein chemistry and ordered water could be resolved. The structure

demonstrates unambiguously that K+ comes into close contact with the carbonyl oxygens

of the TVGYG sequence at the narrowest part along the permeation pathway (Fig. 1.13b).

The coordination shows that the protein provides surrogate oxygen atoms that mimic the

hydration shell for the K+ ion that is queued in the aqueous cavity preceding the

selectivity filter [37,38].

a b

17

Figure 1.13 Structure of the KcsA potassium channel. a. Two subunits are shown for clarity.

The C-terminal ends of the M2 helices (homologous to S6) form a bundle that mostly closes the

pore off to the inside solution. Therefore M2 (S6) helices form the gate. The central vestibule is

marked with a red asterisk. The selectivity filter is circled. (a. adapted from [39]) b. The selectivity

filter shown in detail (adapted from [36]). A fully hydrated ion resides in the aqueous vestibule

below and dehydrated ions (green spheres) reside in the filter itself, coordinated to backbone

oxygen atoms (red).

1.4.2 The MthK structure - an open K+ channel. The KcsA channel (above)

crystallized in the closed state so understanding the gating transition, how the channel

opens, could not be directly discerned. Since the M2 inner-helices (S6 helices for 6-

transmembrane channels) seemed to cross and close off the conduction pathway, there

must be a conformational change associated with these helices to open the channel.

MacKinnon’s laboratory cloned, characterized, and crystallized another K+ channel called

MthK from Methanobacterium thermoautotrophicum [40]. This channel is of the type

shown in cartoon Fig. 1.1b, it binds intracellular Ca2+ ligand to gate open. The

membrane-spanning domain has high sequence homology to KcsA and many other K+

channels. X-ray crystallographic analysis of this channel (including the intracellular Ca2+

binding domain) in the presence of calcium resulted in a structure (at 3.3 Å resolution)

that had a pore domain quite different from the KcsA. The M2 inner-helices of MthK

were not crossed into an excluding bundle, rather they were bent 30°, which appears to

open the conduction pathway (Fig. 1.14a). The bend in the M2 segment occurred at a

well-conserved glycine residue, called the ‘gating hinge’ [39]. Although bending at the

gating hinge was proposed to underlie the gating of all K+ channels, functional evidence

on the Shaker channel has suggested that in KV channels the pathway does not open

nearly as widely as MthK. It was suggested that a conserved P-X-P motif (7 amino acids

lower down from the gating hinge on M2) is responsible for opening up a narrower

passageway and that differences in the gating mechanism can explain the lower

conductance in these channels compared with their bacterial ancestors [41]. Furthermore,

in Shaker, a mutation (V478W) just three residues down from the P-X-P motif on S6

create a non-conductive mutant through the formation of a ‘hydrophobic seal’ [42].

Therefore, the MthK structure has provided powerful insight into the conformational

18

changes associated with K+ channel gates, though variation in gate structure between

different channels may exist.

a

b

c

Figure 1.14 Three comparisons between KcsA structure and MthK structure. a. Left shows the

MthK crystal structure that has been modified to take on the shape of the KcsA structure, i.e. the

M2 helices have been bent and twisted about the ‘gating hinge’. Right shows the unaltered MthK

crystal structure, an open K+ channel. Figure adapted from Jiang et al., 2002 [40]. b. The

structural information from KcsA and MthK are presented as the general mechanism of gating for

all K+ channels. c. Looking down the conduction pathway viewed from the inside of the cell. The

bundle crossing of KcsA almost completely closes off the permeation pathway (left). The MthK

structure shows a large opening (right) that allows K+ and larger pore-blockers access to the

aqueous vestibule. Figure taken from Webster et al., 2004 [41].

19

1.4.3 The MscS structure – a slightly voltage sensitive, non-selective

mechanosensitive channel. Doug Rees and colleagues crystallized the first

mechanosensitive ion channel (of large conductance), MscL from Mycobacterium

tuberculosis [43]. These channels are responsible for maintaining osmotic balance across

the bacterial cell membranes. In the event of a sudden drop in external osmolarity, water

begins to flow into the cell (down its concentration gradient) and the cell swells. The

flow of water can be reduced if the cell instead dumps ions into the external solution.

Therefore, the opening of mechanosensitive ion channels can keep the cell from

exploding in such situations. Mechanosensitivity is not what explicitly interests us here.

A channel called MscS is a mechosensitive channel (of small conductance) from

Escherichia coli that also displays a slight voltage-dependent gating [44].

The structure of MscS was solved by Doug Rees and colleagues at 3.9 Å

resolution [45]. This channel structure, though structurally distant from both MscL and

voltage-gated channels, offers a chance to understand mechanisms of gating. So how

does the voltage sensitivity work for MscS? Both voltage and tension are coupled to the

gate such that less depolarization is required as tension is increased and less tension is

required as depolarization is increased. The voltage dependence requires a voltage-

sensor that can create net movement of charge across the membrane field. The gating

charge for MscS has been estimated to be ~1.7 charges/channel (significantly lower than

KV channels, with ~13 charges/channel), possibly arising from the movement of two

arginine residues, R46 and R74, though this has not been determined conclusively. Since

MscS is constructed from 7 identical subunits, these proposed gating charge residues

actually represent 14 charges, therefore Bass et al. [45] note that a movement of ~2.4

angstroms/charge can account for the total gating charge (assuming a 20 Å hydrophobic

bilayer). The crystal structure surprisingly suggested these two arginines are likely to be

exposed to the bilayer, despite the energetic cost of placing charges in a membrane.

Molecular dynamics simulations have suggested that the charged residues interact

directly with the lipid polar head-groups and such interaction could explain the gating

mechanism’s voltage sensitivity [46]. Furthermore, new results using electron

paramagnetic resonance (EPR) probes have indicated the charged residues have low lipid

20

accessibility (Vásquez, Cortez, Perozo; Poster 1421, Biophys. Soc. Meeting 2005 [47])

for MscS reconstituted in a lipid bilayer. The mechanism for MscS gating originally

proposed by Bass et al. is shown in Fig. 1.15, however many more experiments will be

needed before this channel is fully understood.

Figure 1.15 Proposed mechanism for gating the MscS channel. Both tension and membrane

depolarization increase the open probability of this homoheptameric, nonselective ion channel.

Highlighted arginines are proposed to be the voltage-sensing residues. Figure taken from Bass

et al., 2002 [45].

It has been hypothesized [48] that the MscS structure suggests a shared voltage-

sensing mechanism with the ‘paddle model’ for KV channels (described in cartoon form,

section 1.3 above, see section 1.4.4 below). Both models have voltage-sensing structures

at the protein-lipid interface, allowing gating charges to contact the hydrophobic

membrane interior. In both cases the actual voltage sensor is a paddle-shaped helix-turn-

helix structure (for MscS, the paddle is pointed down, opposite to the KV paddle) with

arginine residues for gating charge. However, these channels have very little else in

common structurally. The KV paddle model proposes a large (15-20 Å) vertical

displacement of the voltage-sensor because KV channels have an extremely large total

21

gating charge that results in a very steep voltage dependence of channel opening [49].

MscS on the other hand is very weakly voltage-dependent and so a very small total gating

charge is required. Bass et al. [45] suggest a vertical gating charge displacement of only

2.5 Å may be required to account for the total gating charge movement. Furthermore,

evidence exists for both MscS and KV channels that the gating charges are not directly

exposed to the hydrophobic membrane bilayer [47,50]. Therefore, it seems there are

serious problems accepting paddle behavior as the mechanism for the straightforward

translocation of gating charges across a hydrophobic barrier.

1.4.4 The KvAP structure – a full-fledged, highly voltage-dependent

channel. The KcsA and MthK structures (above) were highly successful at detailing the

permeation pathway of K+ channels. The pore domain validated expectations and also

provided elegant insights into the energetics of highly selective, highly conductive

channels. However, how do channels operate in a voltage-dependent manner? MscS

provided clues perhaps, or at least teased us with clues, but the channel is very different

from the superfamily of voltage-gated channels. What we really need are structures of

full-fledged KV channels.

In a scientific tour de force, the MacKinnon lab diffracted X-rays off a crystal of

voltage-gated ion channels called KvAP [30]. Characterization of the

electrophysiological properties of this KV channel from the archaebacterial

hyperthermophile Aeropyrum pernix demonstrated that it was functionally similar to

eukaryotic channels like Shaker [51]. The KvAP voltage-dependence of opening occurs

at somewhat more negative voltages compared to Shaker and the sequence of KvAP has

very little linker regions between transmembrane segments (S1-S2 and S3-S4 linkers are

very large in Shaker). Otherwise, it is expected that sequence homology and functional

homology make these channels very similar in structure. MacKinnon and coworkers

used a channel from a hyperthermophile because they reasoned that the protein may be

exceptionally stable and favor crystallization. However, after many failed trials they

decided to raise antibodies against the voltage-sensor so that they could “hold” it in place

for crystal growth. Thus, they interpreted their difficulty in crystallization as evidence

that the voltage sensor is a highly mobile domain. The resulting monoclonal Fab

22

fragment-KvAP crystal structure (at 3.2 Å resolution) was in a very unexpected

conformation (Fig. 1.16a). The helices of both S3 and S4 were broken near the middle

and formed two separate helical segments. The parts of S3 were labeled S3a and S3b.

The two separate pieces of the traditional S4 segment were labeled S4 and S4-S5 linker

[30], though another study has defined the N-terminal piece S4a and the C-terminal piece

S4b [50]. It is unclear at this time whether these structural details extend exactly to other

KV channels like Shaker, there are likely differences since the prolines and glycines that

tend to break up α-helices occur at different places in the sequences. The Fab antibody

fragments bound to an epitope at the extracellular end of the S3b-S4 helices, and

apparently pulled it down towards the intracellular side (Fig. 1.16a). This led to a strange

artifact: the S4 and gating charges were all the way down near the inside, yet the pore

appeared to be in an open conformation. (This is contrary to the irrefutable

electrophysiological fact that channel opening must correspond to outward movement of

gating charge.) Furthermore, the S1-S4 helices were not packed into a tight structure but

appeared pulled apart (Fig. 1.16a), with S1, S2, S3b and S4 parallel to the membrane

rather than transmembrane as expected. The S3b-S4 structure was called the voltage-

sensing ‘paddle’ and is detailed in Fig. 1.16b.

a b

“S4-S5 linker”may be “S4b” The S3b-S4 “paddle”.

Figure 1.16 a. The KvAP crystal structure with two subunits removed for clarity. The pore

domain is white and S1-S4 are indicated in color. Part of S4 forms a continuous helix with the S5

of the pore domain and this was labeled by Jiang et al. as the S4-S5 linker [30], though if S4 is in

fact transmembrane as other models propose, then S4b may be more appropriate. Figure

23

adapted from Cuello et al., 2004 [50]. b. Detail of the S3b-S4 voltage-sensing “paddle” from the

full length crystal structure. This structure is hypothesized to act as a highly mobile hydrophobic

cation (the paddle), moving vertically through the lipid membrane. Figure adapted from Jiang et

al. 2003 [30].

Jiang et al. [30] published the crystallographic data shown in Fig. 1.16, aware that

distortions existed in this structure. Thus, directly interpreting the structure in terms of

the mechanism of voltage sensing was difficult if not impossible. More crystallographic

data and more experiments were needed. First, they crystallized an isolated voltage-

sensor domain S1-S4 (Figure 1.17a), and found that S3b and S4 were associated in the

same helix-turn-helix, paddle motif. (Any model of KV channels is likely to have S3 and

S4 next to each other in the membrane as S3-S4 linkers vary widely, from non-existent,

as in the KvAP channel, to very long, as in the Shaker channel. However, Shaker

functions normally even with the S3-S4 linker almost completely removed [52].)

Similarities in structure between the isolated voltage-sensor and the full-length channel

were interpreted as an indication that S1-S4 was not distorted very much in the full-

length structure. Therefore, they reasoned that the Fab fragments distorted the channel

by pulling segments down towards the intracellular space, but that the distortions were

not so severe that they could not make adjustments to the structure and then suggest a

mechanism for voltage sensing. After docking the isolated voltage-sensor structure to the

pore domain using the S2 segment as a docking reference guide, an altered voltage-

sensing domain structure was proposed (Fig. 1.17b). S3b-S4 was proposed to lie at the

channel periphery, in contact with the hydrophobic membrane, and voltage sensing is

based on a large charge-carrying traversal across the span of the membrane (the proposed

paddle model, Fig. 1.12).

24

a bThe voltage-sensingpaddle, S3b-S4

“S4-S5 linker”may be “S4b”

Isolated voltage-sensor. Docked.

Figure 1.17 a. The isolated voltage-sensor structure. Paddle segments are circled. The C-

terminal end of S4, called the S4-S5 linker in the full channel structure, is now part of a

continuous helix with the N-terminal side of S4. b. (Above) One subunit of the full-length channel

structure (S1 not shown for clarity). The isolated voltage-sensor is docked to the pore domain

(only 1 subunit shown) using the position of the S2 segment as a docking reference. Figures

adapted from Jiang et al., 2003 [30].

MacKinnon and coworkers sought to test experimentally the paddle model. A

bacterial toxin-channel called colicin forms a voltage-dependent channel with a charged

segment that traverses the membrane bilayer. Finkelstein and coworkers used biotin-

linkers attached to the translocating region to show that a biotin-binding protein called

avidin could attach to the linker from the internal solution or the external solution,

depending on the state of the channel. This type of experiment could be described as an

accessibility experiment using a fishing line that can determine how deep in the

membrane a protein site is located. MacKinnon applied this technique to the KvAP

paddle in order to test that the paddle is near the bottom of the membrane when the

channel is closed and translocates to the top during opening (Fig. 1.18) [31].

25

Figure 1.18 Experimental cartoon for testing KvAP paddle movement using avidin-biotin binding.

When the paddle is down, biotin linked sites are expected to bind internal avidin. When the

paddle is up, biotin linked sites are expected to bind external avidin. Binding from either side

should cause a reduction in K+ current. Figure taken from Jiang et al., 2003 [31].

Sites along S3b and S4 were labeled with the biotin linker fishing line, and

labeled KvAP was reconstituted into lipid bilayers so that the channels could be voltage-

clamped. A control current trace was recorded before the addition of avidin protein.

Next, avidin was added to either the inside or the outside solution and the resulting affect

on the current was recorded. For all sites on S3b and sites at the top of S4, current

inhibition occurred upon addition of avidin to the external solution. Membrane

depolarization significantly speeded up this inhibition. For two sites on S4, inhibition

occurred upon addition of avidin to both the internal solution (when the channel was

closed) and the external solution (when the paddle was open.) This behavior is attributed

to the biotin linker being dragged from the inside solution to the external solution upon

channel opening. For two sites lower down on S4, inhibition occurred upon addition of

avidin from the internal solution and never from the external solution. The complete data

set is shown in Fig. 1.19. This pattern of behavior was used to constrain how deeply

26

residues on the paddle could be located in the membrane, as a function of voltage. Such

constraints model the paddle’s ‘down’ configuration (Fig. 1.20a) and its ‘up’

configuration (Fig. 1.20b) and voltage sensing involves the ‘paddle’ moving between

down and up (Fig. 1.20c and 1.20d). This is the logic and experimental evidence put

forth by MacKinnon and colleagues in support of their paddle model.

Figure 1.19 Complete data set showing the pattern of current modification after addition of avidin

to the external solution (red traces) or internal solution (blue traces). Control currents are shown

in black, recorded before the addition of avidin to either side. Red sites required membrane

depolarization for fast inhibition from external avidin. Blue sites were only influenced by internal

avidin. Yellow sites showed inhibition when avidin was added to either the internal or external

side. Figure from Jiang et al., 2003 [31].

27

Figure 1.20 The derivation of a paddle model ‘cartoon’ from avidin accessibility data. a. The

biotin tether length puts accessible sites within 10 Å from the internal solution in the ‘down’

position. b. Accessibility data defines which sites are within 10 Å from the external solution in the

‘up’ position. c. Imagined closed state for the paddle model – this is a cartoon not a crystal

structure. d. Imagined open state for the paddle model – again this is not a structure. Figure

taken from Jiang et al., 2003 [31].

So is the paddle model a true representation of how voltage sensing works for KV

channels? Or is it just one big artifact? Or is KvAP just a very unusual channel, for

which the normal rules of structural homology sweetly flowing out of sequence and

functional homology just do not apply? The paddle model has been extremely

28

controversial, and many traditional data (usually obtained on the Shaker channel) have

not been easily understood in terms of the crystal structure. Contradictions between the

structure of Fig. 1.16 and traditional biophysical and biochemical data on functional

channels are numerous. The concept of the paddle model for voltage-sensing has also

been difficult to reconcile with many experimental results. Many of the issues have been

discussed in reviews [53-55]. For example, the channel N-terminus is known to be

intracellular (forming the T1 tetramerization domain in Shaker) [56] but in the structure

appears to be buried in the membrane. Cysteine-scanning mutagenesis studies examining

voltage-dependent accessibility to thiol-reactive reagents have shown that much of S3b

and the linker between S3b and S4 are accessible to external reagents independent of

voltage [19,21,57]. In the structure, these sites appear to be near the intracellular side of

the membrane. Recently, the S4 segment was shown to be very close to the pore domain

in the open state, as disulphide crosslinks are formed between pairs of cysteines

introduced in S4 and the pore domain, S5/S6 [57-59]. This indicates that the structure of

the open state is very different from Fig. 1.16. Interestingly, the residue on S4 that was

shown to crosslink with the top of the pore domain is just about the farthest residue away

from this crosslinking site in the structure of Fig. 2, and significantly, S1, S2, and S3b

would have to ‘get out of the way’ for this crosslink to occur.

A low-resolution KvAP structure (10.5 Å) from electron microscopic (EM) single

particle analysis (also from MacKinnon and coworkers) has also placed the voltage-

sensor paddle up against the pore domain (Fig. 1.21) [60]. This EM structure was not

subject to the crystal packing forces of the X-ray sample and suggests a very different

conformation. The location of the voltage-sensing paddle was determined because the

EM data included the Fab antibody fragments (easily identified in the low-resolution

structure) that bind to the paddle’s tip. Two possible docking orientations were possible

for S3b-S4 (both shown in Fig. 1.21). The second possible docking orientation (Fig.

1.21b and 1.21d) seems to agree best with the above-referenced S4 to pore domain

crosslinking data. Unfortunately, both X-ray structure determination and single-particle

EM analysis study channels outside of their native membrane environment. All of these

issues suggest that the full length KvAP crystal structure is very distorted, unlike the

conclusions of Jiang et al. [30].

29

Figure 1.21 Models of the KvAP channel using single particle electron microscopy analysis.

Crystallographic data for the pore-domain and the isolated paddle structure (S3b-S4) are docked

to the low 10.5 Å resolution density map (blue mesh c and d). Two docking orientations for the

paddle are possible. Orientation 1 is shown (a.) and from the top (c.). Orientation 2 is shown (b.)

and from the top (d.). Red ovals indicate density that is unaccounted for and is likely taken up by

segments S1, S2 and S3a.

I have covered the paddle model proposal in detail because our measurements

(Chapter 5) directly address the validity of this unconventional and surprising model.

Furthermore, the paddle model has been presented as a cartoon and is necessarily

ambiguous. Many of the data that run contrary to the paddle model are somewhat

indirect, and there is a problem with attributing predictions that are too specific to an

ambiguous model. We have tested the central hypothesis of the paddle model: a large

vertical translation of the voltage-sensing segments. We find that no such translation

occurs, and therefore conclude that the paddle model is not the general mechanism of

gating charge movement in voltage-gated ion channels.

30

1.4.5 MacKinnon goes to Stockholm. In 2003, Roderick MacKinnon shared the

Nobel Prize in Chemistry with Peter Agre. The structures discussed above, especially

KcsA, have pushed our understanding of ion channels forward at a tremendous rate. As

well as the structures discussed in detail above, MacKinnon crystallized a chloride

channel, demonstrating how anionic channels are structured [61]. Even though

contentions exist over the KvAP paddle model, research into KV channels has intensified

and progressed as a result. For a highly readable perspective on MacKinnon’s place in

the history of ion channel research, see Chris Miller’s essay, “Ion Channels go to

Stockholm – this time as proteins” [62].

Nobel Prize in Chemistry,2003"for structural and mechanisticstudies of ion channels”

Roderick MacKinnon

Figure 1.22 Roderick MacKinnon shared the Nobel Prize in Chemistry in 2003 with Peter Agre.

Agre pioneered the work on Aquaporins – water channels. MacKinnon propelled ion channel

research forward with structure after structure.

31

Chapter 2

Electrophysiology of the Shaker K+ Channel

2.1 The Xenopus laevis oocyte expression system

With the advent and development of cloning techniques in the 1970’s and 1980’s

it became possible to manipulate proteins such as Shaker and other ion channels in the

lab. In order to study the function of the gene products of these DNA clones, an

expression system was needed. By this time, the frog species Xenopus Laevis (Fig. 2.1,

left) had already been established as a model system for reproductive biology and

development, and the frogs were easy to handle and the extraction of egg cells

straightforward. Furthermore, Xenopus oocytes (unfertilized egg cells, Fig. 2.1, right)

were quite large (~1 mm diameter) and simple injection of mRNA encoding membrane

proteins resulted in protein expression [63,64]. The oocyte translates the mRNA,

incorporating the protein in a membrane vesicle, and then traffics it to the plasma

membrane (outer membrane of the cell). Thus, shortly after the first KV clone (Shaker)

was created [11], mRNA was injected into oocytes and a voltage dependent K+ current,

familiar from classical neuronal biophysics, appeared under voltage clamp. The study of

ion channels had entered the age of molecular biology. The further development of

molecular biology, namely mutagenesis and recombinant DNA manipulation, has

become the heart of basic ion channel biophysics. To this day, the standard procedure for

studying an ion channel follows the ‘mutate and measure’ strategy. Xenopus oocytes

continue to be a convenient system for the expression and study of channels of many

types. Voltage-gated channels are no exception, as these cells express huge amounts of

protein and are easy to record using voltage clamping instrumentation (see below). We

use the Xenopus system in our experiments - expressing ~1010 Shaker channels/cell!

32

1 mmXenopus laevis

Figure 2.1 Left. The African-clawed frog, Xenopus laevis laevis. Mature female frogs are about

10-15 cm long. Right. Isolated oocytes at two different magnifications. The dark side (the animal

pole) becomes the top of the frog and the light side (the vegetable pole) becomes the bottom.

2.2 The two-electrode voltage clamp In Chapter 1 we mentioned that Hodgkin and Huxley used a newly invented

instrument, called the voltage clamp, to propel forward their studies of the giant squid

axon. We also casually referred to channel currents, to K+ flowing through ion channels.

We referred to the phenomenon of gating currents, predicted by Hodgkin and Huxley

[65] and eventually measured by Armstrong and Bezanilla [9]. In this chapter we discuss

measurement of these currents using the Shaker K+ channel expressed in oocytes. First,

we describe the instrument used for taking these electrophysiological recordings.

The term voltage-clamping refers to the control of membrane voltage. In

particular, voltage clamps are used to hold the membrane voltage constant. This

innovation allows for the direct measurement of ionic currents across membranes without

contamination from capacitive current responses of the membrane (Cm*dVm/dt = 0, Fig.

2.2a). When an experiment requires a change in voltage, the instrument applies a fast

voltage step such that membrane capacitive currents occur as transients (Fig. 2.2b). To

33

demonstrate that the Xenopus oocyte membrane acts as a linear capacitor, we apply

voltage steps with a voltage clamp, integrate the measured current transient to obtain the

charge moved, and plot ∆Q vs. V. The slope shows the constant linear capacitance of the

oocyte membrane (Fig. 2.2c).

The membrane is a linear capacitor that is charged, Q = CmV

Cm- - - - --

+

- -

+ +++ ++

V

a

bc

∆Q = 31.045 + 0.31*V

C = 0.31µF

V

I

A voltage step command initiatesa transient charging current (I) of the

membrane capacity.

(V)

Figure 2.2 a. A biological membrane is like a capacitor. The lipid bilayer has a low dielectric

constant (ε=2) and the aqueous media on the external side and the internal side are conductors

34

(salt water). b. A sudden voltage step results in a transient charging/decharging current due to

the membrane capacitance. c. Xenopus oocytes used in our experiment have linear membrane

capacitance. Transients are generated with voltage clamp steps to various voltages from a

starting voltage of -100 mV. Integration of these transients results in capacitance charge vs.

voltage that is linear.

The current recordings shown in Fig. 2.2c are significant because they demonstrate

experimentally that depolarizing voltage steps elicit no currents other than the membrane

capacitance response. This is an important property of the Xenopus oocytes: they do not

exhibit endogenous expression of ion channels that interfere with the expression and

study of exogenous channels like Shaker. However, Chapter 1 mentioned that K+ and Cl-

channels were fundamental to all cells, and therefore it is likely that some native channels

are there. These endogenous channels are not likely to be voltage sensitive channels. For

example, an endogenous Ca2+ activated Cl- conductance has been described and studied

[66]. Also, the over-expression of ion channels has been shown to induce currents in the

oocyte that are unrelated to the exogenous channel expressed [67].

Voltage clamps come in several configurations, but the two principal techniques

are whole-cell clamping and patch clamping. Patch clamping is perhaps the central

modern technique that allows for the recording of small cells or a tiny excised patch of

cell membrane. The resolution of patch clamping extends down to the single ion channel

level [68]. Sakmann and Neher shared the Nobel Prize in Physiology or Medicine in

1991 “for their discoveries concerning the function of single ion channels in cells”. For

our studies (Chapter 4 and 5), we record a very large ensemble of ion channels, using a

standard whole-cell two electrode voltage clamp instrument.

The important elements of the two-electrode voltage clamp are shown in Fig. 2.3.

Two electrodes impale the oocyte. The first is a voltage measuring electrode (V1) that

simply determines the voltage inside the oocyte. The second electrode is the current

injection electrode (Vi) that is used to change the voltage inside the oocyte. The clamp

measures the current that crosses the membrane with an ammeter connected to ground.

Therefore, the instrument includes a measure of Vm, the membrane voltage, and a means

to change it (current injection). The last piece is a feedback amplifier, which compares

the measured Vm with the voltage command input by the experimenter and delivers the

35

appropriate current to the current injection electrode in order to null its inputs. In other

words, the amplifier clamps Vm to the value requested.

K+

K+

+++++++

+++++++

Inside Cell

OutsideCell

A

CommandVoltage Pulse

FeedbackAmp

VVoltageMeasuringElectrode V1

CurrentInjectionElectrode Vi

Figure 2.3 The two-electrode voltage clamp. Currents crossing the membrane such as K+

through ion channels and outward movement of voltage-sensing segments are recorded with an

ammeter (A) connected to ground. The feedback amplifier is the essential element that

compares the command voltage pulse to the measured voltage (V1) and supplies a current to Vi

(current injection electrode) in order to null the amplifier inputs.

The instrument we use (CA-1b high-performance oocyte clamp, Dagan) is in

reality more complicated than the schematic cartoon of Fig. 2.3, however the idea is the

36

same. Our clamp has the two electrodes, as shown, but the bath (external solution) is also

under active voltage clamp for faster, more accurate voltage control of Vm. Another

advanced element is a transient generator that allows the instrument to null out the

capacity transients shown in Fig. 2.2 (capacity compensation). This is required for

accurately measuring current responses that overlap the fast kinetics of the membrane

response and is otherwise useful for reducing this uninteresting, high magnitude current.

2.3 Ionic currents Shaker channels conduct K+ ionic current from the inside of Xenopus oocytes to

the outside. Thus, when channels open a positive current will result (cells have high [K+]

on the inside). Generally, ionic current is very large and so only very low channel

expression is required to detect K+ current across the membrane. All channels have more

states than simply closed and open, they have inactivation states. Inactivation refers to

the elimination of current upon extended depolarizations. It is clear that channels have

developed mechanisms to help safeguard the ionic gradients that they use for signaling.

Also, inactivation properties are just as important as the activation properties in defining

the shape and timing of action potentials in nerves.

For Shaker there are two types of inactivation. The simplest form is called fast

inactivation (or N-type inactivation). This type of inactivation is associated with a

peptide segment at the channel N-terminus that plugs the channel (from the inside) after

the activation gate opens [69]. This view of fast inactivation was called the ‘ball and

chain model’ because the peptide plug is tethered to the protein with a disordered peptide

that diffuses around according to polymer statistics [70-72]. This ‘inactivation ball’

could be cut off (destroyed with proteinase [70] or chopped off using molecular biology

[73,74]) and fast inactivation was eliminated. K+ currents persisted well after

inactivation usually turned the channels off. In all of our experiments we use a construct

called Shaker-IR, that has the inactivation ball chopped off with molecular biology (∆6-

46) [75]. The difference between Shaker K+ currents with and without the fast

inactivation ball is shown in Fig. 2.4a (Data taken from Hoshi et al., 1990 [73]).

37

The second and more complicated form of Shaker inactivation is called slow

inactivation (or C-type inactivation). This type of inactivation is more fundamental to the

channel and can not be eliminated by simply chopping off an otherwise inconsequential

part of the protein, like the fast inactivation peptide. In fact, the C-type inactivation gate

seems to be associated with the selectivity filter itself (narrowing of the filter perhaps)

[76]. Therefore, our Shaker-IR clones still inactivate. The timescale of slow inactivation

is on the order of seconds, roughly 100 times slower than fast inactivation (Fig. 2.4b). In

all of our experiments we depolarize and record for only 50 ms, a timescale for which

slow inactivation is completely negligible. Example ionic K+ currents through Shaker-IR

channels recorded with our setup are shown in Fig. 2.5.

a b

20 ms

Shaker B∆6-46 channel currents persist for many milliseconds

Addition of the fast-inactivation ball causesthe channels to get plugged

Slow inactivation turns ShakerOff over the course of seconds

Figure 2.4 a. Top traces show Shaker-IR currents that display no fast inactivation, the currents

persist over the course of 10s of milliseconds. Bottom traces show currents after the addition of

diffusing inactivation balls to the intracellular side of the membrane. Fast inactivation is apparent

from the sudden drop in current (data taken from [73]). b. Shaker-IR channels inactivate via slow

inactivation over the course of several seconds. Traces show the addition of the inactivation

peptide to the intracellular side of the membrane recovers fast inactivation. A mutant peptide has

little effect on inactivation (data taken from [77]).

38

Figure 2.5 Example K+ current recordings of Shaker-IR measured on the Selvin lab

electrophysiology rig. Oocytes were expressing low amounts of ion channels so that the current

is less than 50 µA. This form of presenting many recordings (current family) is common in

electrophysiology presentations. The test pulse was varied from -100 mV to +100 mV. The test

pulses are applied for 50 ms.

2.4 Gating currents In Chapter 1 we discussed models that describe movements of the highly charged

S4 voltage-sensor. This segment ultimately opens and closes the channel by moving

when the membrane electric field changes. Therefore, the conformational changes

associated with S4 are of particular interest in understanding the mechanism of voltage

sensing. Traditionally, S4 movements and voltage sensing were studied with

39

electrophysiology. The currents that arise from the movement of the S4 charges are

called ‘gating currents’. These microscopic currents are a direct measure of protein

conformational changes associated with the movement of S4. The magnitude of gating

current (approximately 100x smaller than ionic current [78]) is very small because it

arises from 13 elementary charges/channel crossing the field. Therefore, in order to

detect gating currents, ionic currents need to be eliminated.

Ionic currents are abolished by making mutations that eliminate conduction or by

blocking the channels with molecules or proteins that block the flow of K+. These

blocking methods need to leave the function of the voltage-sensing domains unaltered so

that S4 movement can still be detected. In this study we use several types of channel

block. In chapter 4, we use a common mutation, W434F [79], which makes Shaker non-

conducting, probably by closing the C-type inactivation gate of the selectivity filter [80].

In chapter 5 we block channels with scorpion toxins that bind to the external end of the

pore, clogging the mouth of the selectivity filter [17]. Both of these mechanisms block

K+ conduction but leave gating currents unaffected.

In Fig. 2.6 we repeat the experiment of Fig. 2.2c, except this time we do not

voltage clamp a ‘blank’ background oocyte, we clamp an oocyte expressing a huge

number (> 1010) of Shaker channels that are non-conducting via scorpion toxin block.

The current transients that result from depolarizing voltage steps are not as fast as in Fig.

2.2c and highly complex kinetics have become apparent (Fig. 2.6 left). The integral of

these recorded currents give a measure of the charge change across the membrane (which

is only due to membrane capacitance and the voltage-sensing domains of Shaker) as a

function of voltage (Fig. 2.6 right). Compare this plot with Fig. 2.2c. It is clear that a

nonlinear capacitance has appeared in addition to the normal linear capacitance of the

membrane. This nonlinearity is the charge movement of Shaker voltage-sensors.

40

2.00 ms

10.0 µA

Figure 2.6 Left. Transient currents recorded from voltage clamp steps from -100 mV to voltages

ranging from -150 mV to +50 mV, recorded on the Selvin lab electrophysiology rig. These

transients are a mixture of the membrane capacitance response and the movement of the

membrane-associated voltage-sensors (S4 segments). Currents show complex kinetics unlike

the background transients from non-expressing oocyte membranes (Fig. 2.2c). The protein in

this experiment was Shaker-IR blocked with 2 µM scorpion toxin (wild-type CTX). Right. The

integrated charge (∆Q) vs. test voltage (V) is now nonlinear (compare to Fig. 2.2c) because the

voltage sensor segments move gating charge across the membrane with very nonlinear voltage

dependence.

Unlike the recordings shown in Fig. 2.6 (left), normally we use membrane

capacitance compensation hardware to cancel the membrane response to voltage steps.

Therefore, the recordings will only contain the channel gating currents. In Fig. 2.7 (left)

we show gating currents recorded with a Shaker-IR/W434F non-conducting mutant using

our instrument. Gating currents recorded from Shaker-IR blocked with a scorpion toxin

are also shown (Fig. 2.7, right). The currents are transient because after a sudden step in

voltage, the ensemble of channels move to a new conformational equilibrium. While the

average conformational shift is occurring, there is a net movement of S4 charge across

the membrane that produces a current. After the new state is reached on average, no net

movement of S4 occurs and the current stops. At intermediate voltages, the S4s can be

41

moving between their closed and activated conformations, but the ensemble average

shows no net movement after the gating current transient is over.

Figure 2.7 Shaker gating currents recorded on the Selvin lab electrophysiology rig. Left. Shaker-IR/W434F, series of depolarizing pulses resulted in gating currents. Right. Shaker-IR

blocked with 2 µM external charybdotoxin, a pore blocking protein from scorpion venom.

The voltage dependence of S4 movement is usually expressed as the percentage

of total gating charge moved vs. voltage, called a normalized Q-V curve. This function is

easily calculated by the integration of gating currents (Fig. 2.7). This is exactly what we

have been showing in Fig. 2.2c and Fig. 2.6, however with the membrane capacity

response removed from our recordings we can now characterize the charge movement of

the voltage-sensors uncontaminated (Fig. 2.8). Typically, the Q-V curve is normalized

because the number of ion channels present in an experiment is not of fundamental

interest, however the total gating charge is a good way to quantify how many channels

you are managing to express. For our experiments we clamp ~1010 channels.

42

Figure 2.8 Normalized Shaker-IR Q-V curve obtained from the integration of clean gating

currents like those of Fig. 2.7 with no membrane capacity contaminations. In this experiment, K+

conduction was blocked using a scorpion toxin (2 µM CTX). By 0 mV all voltage sensors have

moved to their outward conformations. At hyperpolarized voltages, ~ -150mV, all voltage sensors

are in their resting (closed) conformations.

What is the relationship between channel opening and the voltage dependence of

S4 movement, the Q-V curve? This question goes back to the very beginning with the

Hodgkin and Huxley model. They assumed four independent gating particles moving

with identical kinetics opened a channel [65]. This model predicted the kinetics of

channel opening and predicted the existence of several closed channel states (1 gating

particle moved, 2 gating particles moved, etc.). However, they had no way of measuring

gating currents which give a direct measure of these otherwise ‘silent’ closed states.

Nowadays, scientists have access to detailed descriptions of macroscopic ionic currents,

single channel currents, and for voltage-gated channels, gating currents. Therefore, a

complete description of channel activation kinetics is possible if all of this information

can be understood in terms of a model. Several detailed models have attempted to

describe all the data for KV channels [81-83]. Although differences exist among these

various models, several essential features have become clear. KV channels have four

voltage sensors that move independently through at least two kinetic steps. All four

voltage sensors have to be in their fully outward position at the same time so that a

43

subsequent cooperative transition(s) can then open the channel. As a consequence, gating

currents precede opening and the Q-V curve is not identical to the open probability, as

there will not be significant activation of channels until the percentage of charge moved

is quite high. Fig. 2.9 plots the Shaker Q-V alongside the G-V (conductance curve or

open probability), data taken from Stefani et al., 1994 [84]. Although there is significant

overlap between Q-V and G-V, it is clear that a significant amount of the gating charge

can move in a voltage range where channels rarely open. These features of the gating

and activation voltage dependence are strikingly underscored in a mutant phenotype, the

ILT-Shaker, which we discuss in the next section.

Figure 2.9 Comparison of the Charge-voltage curve, Q-V, with the normalized conductance

curve (G-V or open probability) for Shaker. The Q-V curve measures the voltage sensor

transitions from the closed, inward position to the activated, outward positions. The plot has a

best-fit that shows the Q-V curve can not be described by a single boltzmann function. The G-V

describes the steep voltage dependence of channel opening. Data taken from Stefani et al.,

1994 [84].

44

2.5 The ILT-Shaker phenotype Aldrich and coworkers sought to understand how the S4 sequence affects the

voltage dependence of gating charge movement, channel opening, and activation kinetics.

Different KV channels behave quite differently in all of these respects, independent of the

total charge on S4. Therefore, it is expected that the uncharged residues of S4 influence

the basic activation pathway of the channels, probably by influencing the cooperative

transition(s) the channel undergoes immediately before opening. Aldrich’s approach was

to construct chimera channels, using the Shaker channel S1-S3 and S5-S6 but with an S4

spliced in from four very different donor channels. The resulting chimeras had properties

that were not predictable from the properties of the donor channels [85]. One channel,

the Shaker/Shaw-S4 chimera, had an activation curve (normalized conductance) that was

shifted to very positive voltages [86]. Evidently, the chimera greatly disrupted the

cooperative transition(s) required for opening. This change in Shaker activation was

shown to be mostly influenced by the single S4 amino acid change (I372L). However, a

triple Shaker-S4 mutant (V369I, I372L, S376T, called the ILT mutant) reproduced all of

the properties of the Shaw-S4 chimera channel. Study of the ILT-Shaker channel

demonstrated that the gating currents occur normally like wild-type Shaker, however the

channel is only opened by very strong depolarizations. Therefore, the Q-V curve of

voltage sensing and the G-V curve of channel opening are completely separated with

respect to voltage (Fig 2.10) [87]. The independent conformational transitions of the

voltage-sensors are unhindered, but the cooperative transition is isolated along the

voltage axis. The ILT-Shaker phenotype therefore, allows for the study of voltage

sensing transitions separately from the transition(s) of opening. In chapter 4, we use the

ILT-Shaker channels to study conformational changes associated with these separate

stages of the activation pathway.

45

Figure 2.10 The ILT-Shaker phenotype. Open symbols: The charge-voltage relation (Q-V) is

similar to the wild-type Shaker channel but with an approximately -40 mV shift (and a shallower

slope). Closed symbols: The normalized conductance (G-V) is very different from wild-type with

an approximately +110 mV shift (and a shallower slope). The Q-V and G-V no longer overlap.

There is a very slight amount of gating charge movement for the G-V voltage range (~5%,

[87,88]). Data taken from Ledwell et al., 1999 [87].

46

Chapter 3

Fluorescence Spectroscopy Methods

3.1 Introduction to fluorescence

Luminescence refers to the emission of light from an object as a result of excited

state relaxation. Fluorescence is a sub-class of luminescence that results from singlet to

single transitions (spin-preferred transitions) that occur rapidly. Organic dye molecules,

called fluorophores, exhibit excited state lifetimes on the order of 1-10 ns (fluorophores

used in this work are shown in Fig. 3.1). Other important properties of organic

fluorophores include; high extinction coefficients (efficient absorption of light), high

quantum yields, and broad emission spectra that are usually independent of the excitation

process [89]. Furthermore, the short excited state lifetimes plus the high quantum yields

make these molecules very bright and readily detectable as single molecules. The

lifetime is also much shorter than the time-scale of many biological processes and protein

motions, thus they act as instantaneous reporters when attached to biological

macromolecules.

Fluorescein-5-Maleimide(molecular probes)

Atto465-Maleimide(atto-tec)

Lucifer yellow-Iodoacetamide(molecular probes)

Bodipy Fl-Maleimide(molecular probes)

Figure 3.1 Organic dyes that are used in the experiments of this thesis. All of the molecules

have organic ring structures containing electronic states with fluorescent transitions. These

47

probes are all green and have a maleimide or iodoacetamide for attachment to the thiol groups of

protein cysteine residues.

The phenomenon of fluorescence is often explained in terms of the Jabłoński

diagram which illustrates the energy levels underlying fluorescence (Fig. 3.2, left).

Upon absorption of a photon of energy hν, the electrons are kicked from the ground state

S0 into an excited state, either S1 or S2. The electron then relaxes to the lowest

vibrational state of S1, a thermal equilibration process (internal conversion) that is

essentially instantaneous compared to the fluorescent excited state lifetime. The

extremely high density of closely spaced vibrational energy levels associated with S1 and

S2 explain the broad absorption spectra of fluorescent molecules (Fig. 3.2, right). Next,

fluorescence photons are emitted when the electron falls back to various vibrational

levels of the ground state. The distribution of emitted photon energies is explained by the

tremendous number of closely spaced vibrational states of S0 (see emission spectrum,

Fig. 3.2, right). Another excitation relaxation pathway occurs when the excited electron

undergoes a spin conversion to the triplet state T1 (intersystem crossing Fig. 3.2, left).

The transition from T1 to S0 is forbidden and so phosphorescence lifetime is on the order

of milliseconds, several orders of magnitude slower than fluorescence. Fluorophores

such as those used here (Fig. 3.1) exhibit no observable phosphorescence.

Figure 3.2 Left. Jabłoński diagram illustrating the transitions underlying fluorescence and

phosphorescence. Fluorescence occurs when an excited electron falls from the first excited

state S1 to the ground state S0. Phosphorescence occurs when the excited state undergoes

intersystem crossing to the triple state T1 and then falls to the ground state S0. Right. Absorption

48

and emission of a commonly used fluorophore, tetramethylrhodamine. The spectra are broad

because of the tremendous number of closely spaced vibrational energy levels of S0 and S1.

Applications of fluorescence to the study of biology are numerous, varying from

simple detection and localization studies to detailed structure-function studies exploiting

various spectroscopic properties. Ensemble techniques as well as single-molecule

techniques are now widely applied to a variety of biological systems [90-94]. The

attachment of fluorescent probes to biological molecules is accomplished conveniently

using thiol-reactive chemistry (amine reactive chemistry is also common). The

fluorophore is synthesized with a chemical group that reacts under appropriate conditions

with the cysteine thiols of proteins (maleimides and iodoacetamides in Fig. 3.1).

Cysteines are somewhat rare in proteins, and when they occur, they can often be removed

with mutagenesis without affecting protein function. Subsequently, cysteine mutations

made on a protein with a cys-lite ‘background’ define a unique labeling site for the

fluorescent probe. Cysteine-scanning mutagenesis refers to the production of many

protein mutants where the location of the introduced cysteine is systematically varied

along a protein sequence. The reaction of maleimide and iodoacetamide with a cysteine

thiol is illustrated in Fig. 3.3.

Figure 3.3 Common reactive groups that covalently attach to protein thiols. The protein is

represented by the R2 group (green) and the fluorescent molecule is represented by R1 (red).

Figures taken from the molecular probes handbook (www.probes.com).

49

3.2 Resonance Energy Transfer (RET) theory In our work, we use fluorescence as a spectroscopic ruler [95,96] to measure

intermolecular distances on the Shaker K+ channel (Chapters 4 and 5). The conventional

technique is called fluorescence resonance energy transfer (FRET) [97-99]. We apply a

modified version called lanthanide resonance energy transfer (LRET) [100-102]. Two

luminescent labels of different structures and spectra are attached to the protein of

interest. The shorter wavelength probe is called the donor and the longer wavelength

probe is called the acceptor. The donor probe is excited by a laser or other light source,

and can either emit photons or transfer its excitation energy to an acceptor. For efficient

energy transfer to occur the energy of donor emission transitions must overlap the energy

of acceptor excitation transitions. The energy transfer efficiency is used to calculate the

distance between donor and acceptor and distance changes can be measured during

protein conformational changes. Next, we discuss these techniques in detail before

introducing their application to the Shaker channel.

The original theoretical foundation for resonance energy transfer was worked out

by Förster [103]. Here we present a discussion adapted from Selvin, 1996 [104]. Energy

transfer occurs via the interaction of the electric dipole moments of the donor and

acceptor. The rate at which energy transfer occurs, ket, is given by Fermi’s golden rule: 2

*53

* ,))((3, ADR

RRR

ADk ADADet

rrrrrr⋅⋅

−⋅

∝µµµµ (3.1)

where D and A refer to the donor and acceptor, and µD and µA are their electric transition

dipole moments, R is the distance vector between the two dyes, and * denotes the excited

state.

The efficiency of energy transfer is given by:

et

ndndet

et

kkkk

kE+

=+

=1

1 (3.2)

where knd is the rate at which the donor de-excites through all non-distance dependent

pathways, such as internal vibrations. Equation 3.1 is used to find a distance from

50

equation 3.2. Combining all of the physical constants from knd and equation 3.1 into one

quantity, Ro, we are able to write the energy transfer in the common and useful form:

6)(1

1

oRR

E+

= . (3.3)

The sixth-power dependence on R comes from squaring the dipole-dipole energy in

Fermi’s golden rule. Because E falls so steeply with increasing R, energy transfer is most

sensitive to changes in distance when the distances are comparable to Ro. Ro is calculated

from the spectral properties of donor and acceptor:

6145 )1079.8( 2κ−−×= nJqR Do (in Ǻ) (3.4)

Where qD is the quantum yield of the donor, n is the refractive index of the surrounding

medium, κ2 is a factor that depends on the relative orientation of the donor and acceptor

dipoles, and J is the normalized spectral overlap integral given by:

∫

∫=λλ

λλλλε

df

dfJ

D

DA

)(

)()( 4

. (3.5)

Where εA(λ) is the molar extinction coefficient of the acceptor and fD(λ) is the emission

spectrum of the donor.

For a FRET experiment to be quantitatively accurate, Ro has to be accurately

determined. This can certainly be achieved, since J is computed from easily measurable

spectra, the quantum yield of the donor can be measured or approximated, and n is

usually equal to 1.33 for water. κ2 is usually the greatest source of error for energy

transfer measurements. In many instances however, the donor and acceptor are attached

to molecules with highly flexible attachments and the relative orientation between donor

and acceptor samples all possible angles. In this ideal case, where donor and acceptor are

completely unpolarized, κ2 = 2/3. For LRET, the donor probe is intrinsically unpolarized

and uncertainty in κ2 is greatly reduced (below).

51

3.3 Lanthanide Resonance Energy Transfer - LRET LRET is a modification and improvement on the widely-used technique of

fluorescence resonance energy transfer (FRET). These techniques utilize visible light

(roughly 500 nm wavelength), yet achieve (sub-) nanometer resolution. In both

techniques, a luminescent (fluorescent) probe called the donor, transfers energy via a

dipole-dipole interaction to a second structurally-different probe, called the acceptor

(Förster theory, above). FRET can measure distances between the probes over a range of

20-80 Å, and LRET extends this range out to beyond 100 Å [102]. This high spatial

resolution is possible, even with optical photons, because the amount of energy

transferred (E) is a strong function of distance between the donor and acceptor

fluorophores: E = 1/(1+ (R/Ro)6, where R

o is the distance at which half of the energy is

transferred and is generally 20-60 Å. By knowing Ro, which can be readily calculated or

experimentally determined, and measuring E, the distance between the probes can be

found. Labeling of probes to specific sites on biomolecules therefore enables the

distances between these sites to be measured. Energy transfer can be measured because it

reduces the donor’s intensity and excited-state lifetime (E = 1- Ida

/Id, = 1 - τ

da /τ

d), where

subscript refers to donor’s intensity or lifetime in the absence (Id, τ

d) and presence of

acceptor (Ida

, τda

), and also increases the acceptor’s emitted intensity (Fig. 3.11). The

donor's emission is at shorter wavelengths than the acceptor emission and hence they can

be independently measured (although spectral overlap can occur).

52

Figure 3.11 Energy transfer between donor and acceptor falls off with the sixth power of R (top).

As donor and acceptor get closer, donor fluorescence becomes dimmer, acceptor fluorescence

increases, and donor lifetime is reduced (bottom).

In LRET, the donor is a luminescent lanthanide atom encased in a small chelate

(Fig 3.12a), and the acceptor is a conventional (organic) fluorophore (see Fig. 3.1).

FRET uses conventional organic-based donors and acceptors. While relying on the same

fundamental dipole-dipole mechanism, LRET has many technical advantages over FRET,

including greater: distance accuracy and range; ability to resolve multiple D-A distances

(donor populations); ability to isolate signal from proteins labeled with both donor and

acceptor, even in the presence of proteins labeled only with donor or only with acceptor;

and less sensitivity of energy transfer to orientation of dyes (which is often unknown).

The fundamental advantages of LRET arise because the donor emission is long-

lived (Fig 3.12b; millisecond lifetime compared to nanosecond lifetime of acceptor or

conventional dyes), sharply-spiked emission (Fig. 3.12c; peaks of a few nanometer

width), has a high quantum yield [105], and is unpolarized [106]. (The chelate’s atomic

structure has also been determined [107]).

53

c

ba

Figure 3.12 a. Structure of Tb-DTPA-cs124-emph (maleimide-chelate used in our experiments,

top). Crystal structure of Eu-DTPA-cs124 (dimensions 12.8 Å x 8.1 Å x 8.3 Å) [107]. b. Luminescence lifetime data for Tb-DTPA-cs124 and Eu-DTPA-cs124. c. Tb-DTPA-cs124

spectrum is sharply spiked and acceptor fluorescence (at 520nm) is measured with no donor

contamination.

• An order of magnitude greater accuracy in distance-determination is achieved with

LRET because the energy transfer process is dominated by the distance between the

donor and acceptor, and their relative orientations play only a minor role in determining

energy transfer efficiency. (A worst case scenario is 12% uncertainty in distance

determination due to orientation effect.) This advantage results from the fact that the

terbium donor emission is unpolarized [106]. This contrasts with FRET where the errors

due to orientation effects can be unbounded. We have shown that angstrom changes due

to protein conformational changes can readily be measured with LRET [25,108,109].

• A 100-fold improvement in signal to background (S/B) is achieved with LRET.

Specifically, energy transfer can be measured with essentially no contaminating

background, in stark-contrast to FRET. By temporal and spectral discrimination, donor

emission and acceptor emission – both intensity and lifetime — can independently be

54

measured. This leads to dramatically improved signal to background compared to FRET.

Specifically, in LRET the acceptor emission due only to energy transfer — called

sensitized emission — can be measured with no background. Contaminating background

in FRET when trying to measure energy transfer via an increase in acceptor fluorescence

arises from two sources: direct excitation of the acceptor by the excitation light, and

donor emission at wavelengths where one looks for acceptor emission. In LRET both

sources are eliminated. For example, by choosing an acceptor such as fluorescein and

looking around 520 nm, donor emission is dark (Fig. 3.5c). By using pulsed excitation

and collecting light at 520 nm only after a few tens of microseconds, all the direct

acceptor emission (which has nanosecond lifetime) has decayed away.

• Samples that contain donor-only or acceptor-only can be spectrally and temporally

discriminated against. Often when labeling proteins, particularly in living cells, one gets

an unknown distribution of donor-donor, donor-acceptor, acceptor-acceptor mixture. In

FRET this makes distance-determination difficult. In LRET, sensitized emission from

acceptor arises only from donor-acceptor labeled complex (see preceding paragraph).

Energy transfer of this D-A labeled complex can then be determined by comparing the

lifetime of sensitized emission (τad

), which decays with micro- to millisecond lifetime,

with the donor-only lifetime (τd): E = 1- τ

ad /τ

d. This ability to measure energy transfer

even in complex labeling mixtures is essential for the LRET studies on ion channels

presented below [25].

We have published a number of papers on LRET (partially reviewed in [102,104])

showing its advantage in model systems such as DNA oligomers [110,111], the ability to

measure distance changes of an angstrom reliably even on large protein complexes such

as actomyosin [108,109,112], and of most relevance to the studies in this dissertation, in

ion channels in living cells [25]. Other workers have now successfully used the

technique on DNA-protein complexes [113-115], actomyosin [116,117], protein-protein

interactions in cells [118], and detection of binding of many different biomolecules [119-

121].

55

3.4 LRET meets the voltage clamp

In 1999, two groups reported resonance energy transfer results on the Shaker KV

channel. Isacoff and coworkers used conventional FRET [26], however nonspecific

labeling and other technical limitations of FRET seemed to make the measurements

inaccurate. The above technical advantages we have reviewed for LRET however,

enabled Cha, Snyder, Selvin, & Bezanilla to measure accurate distances within the

Shaker potassium channel in oocytes using LRET [25]. Here we review this application

of LRET in detail because it is a direct blueprint for new LRET results presented in

Chapter 4. Chapter 5 then presents further LRET results taken with a new donor-

acceptor labeling configuration in order to obtain complimentary structural information.

Methodology: Site-specific labeling was obtained by substituting a cysteine for a

particular residue (Fig. 3.13a) via site-directed mutagenesis and attaching cysteine-

reactive terbium donor (Fig. 3.12a) or acceptor (fluorescein maleimide, Fig. 3.1), to the

four identical subunits of the channel (Fig. 3.13b). We refer to this arrangement as the

S4-donor to S4 acceptor version of the experiment, because it measures distances from

one subunit to the other, distances that are parallel to the membrane (Fig. 3.13c).

Intersubunit distances were obtained by measuring the time constant of sensitized

emission of the acceptor, i.e., fluorescence of the acceptor after receiving energy from the

donor, and comparing this to the time constant of the donor attached to the same site in

the absence of the acceptor. Results for residues S346C, S351C, S352C, N353C are

reported here: they are in the S3-S4 linker, (Fig. 3.13a), accessible to labeling from the

outside of the cell, and near S4, the region of primary interest. As a control, results at

F425C are also reported here (see below).

56

LRET

X

Donor Acceptor

a b

c

Figure 3.13 a. Membrane topology of Shaker with grey circles showing sites studied with LRET

in Cha et al, 1999. Charged residues are indicated on S4 and S2. b. Labeling of homotetramers

that have 1 cysteine per subunit results in a square geometry with two distances, RSC (subunits

contiguous) and RSA (subunits across). Labeling is done with a 4:1 mixture of Tb-donor:Fl-

acceptor such that most channels have only 1 or no acceptor. c. Cartoon of LRET experiment on

the Shaker-IR channel in the S4-donor to S4-acceptor configuration. Distances are measured

parallel to the membrane.

Although labeling leads to a heterogeneous population of channels with different

numbers of donors and acceptors, associated problems are greatly minimized for three

reasons. 1) Channels that are labeled with four donors or four acceptors have no donor-

57

acceptor pair generating acceptor sensitized emission, i.e., they do not contribute to the

signal. 2) Assuming four-fold symmetry of the channel, as demonstrated by the crystal

structures of the closely related potassium channels KcsA and KvAP [30,32], there are

only two possible inter-subunit distances: the distance between residues on neighboring

subunits (RSC

) and the distance between residues across the pore (RSA

), which are related

by the Pythagorean theorem (Fig. 3.13b). 3) The labeling is done with excess donor (4

donors for every 1 acceptor) so that there is typically only one acceptor per channel,

which can readily accept energy independently from multiple donors.

To avoid problems with slow-inactivation of the channel, yet ensure that the

position of the channel residues were at a steady-state, the oocytes were voltage clamped

at a resting potential of –90 mV, brought up to the desired test voltage for 50 msec, and

then the laser fired and LRET data collected (Fig. 3.14b). The process was then repeated

approximately once per second until sufficient signal to noise was achieved.

a b

Figure 3.14 a. Oocytes are voltage clamped on an inverted microscope so that fluorescence can

be collected from below. Both voltage clamping and LRET are acquired by timed pulses (voltage

pulses and laser pulses, respectively) and therefore the techniques are nicely integrated. (Note:

The data from Cha et al., 1999 [25] used a cut-open oocyte voltage clamp different from the two-

electrode clamp pictured, however the concept is exactly the same. Selvin lab is equipped with a

two-electrode clamp, which we use in our experiments, Chapter 4 and 5.) b. Voltage steps were

applied from -90 mV to test potentials between -120 mV and 50 mV. The laser is fired 50 ms

58

after the initiation of voltage pulses so that channels have plenty of time to reach conformational

equilibrium.

Results: The intersubunit distances at several sites were measured by comparing

the time constants of acceptor sensitized emission, and donor emission without acceptor:

R = Ro (τad

/(τd-τ

ad)

1/6, where Ro was determined to be 45 Å [98]. (Note: a corrected

calculation shows that the Ro for Terbium to Fluorescein is 43 Å, using the correct

quantum yield for the Tb-donor [105].) Donor lifetime in the absence of acceptor was

independent of voltage, indicating no significant change in the environment of the caged

terbium (data not shown). Similarly, the intensity, polarization (which was minimal),

spectra and labeling efficiency of the acceptor were unchanged with voltage, indicating

that neither the acceptor nor donor are likely moving with respect to the protein as a

function of voltage. Thus changes in distance between labels likely indicate changes in

the underlying protein conformation.

The sensitized emission for all sites displayed two time constants, reflecting two

donor-acceptor distances that showed a Pythagorean relationship (Fig. 3.15a shows

representative data for probes at S346C). This indicates that the technique is measuring

distances between contiguous subunits and across the pore simultaneously. For example,

at S346C, distances of 28 Å and 41 Å were measured, where the Pythagorean relationship

predicts a distance of 40 Å for RSA given that the measured distance is 28 Å between

contiguous subunits. Further verification of the technique was achieved by measuring

distances at F425C, where a homologous residue is present in the crystal structure of the

KcsA bacterial analog [32]. A distance of 30 Å was measured across the pore, in

excellent agreement with the 29 Å distance obtained between α-carbons for the

homologous residue from the crystal structure.

59

a b

Figure 3.15 a. Representative data from S346C, a residue at the top of the S3-S4 linker on

Shaker. Two time-constants were obtained from LRET sensitized emission and calculated

distances followed the expected Pythagorean relationship. b. Sensitized emission at S346C

displayed an obvious voltage-dependence corresponding to a voltage dependent conformational

change.

Next, voltage-dependent movements near the voltage-sensing regions were

measured by determining inter-subunit distances as a function of voltage. Residue

S346C demonstrated a robust change in the time constant of acceptor sensitized emission

(Fig. 3.15b). This change in time constant reflected a voltage-dependent movement of

~3.2 Å between S346C residues on contiguous subunits, or ~4.5 Å across the pore, as the

channel moves into the open state (Fig. 3.16). These results were the first measurements

of actual distance changes around the voltage sensor as the channel goes from the closed

to the open state. Furthermore, the voltage dependence of the physical movement for

S346C correlated very well with the gating charge movement for the same channels (Fig.

3.16). In other words, the movement of this particular residue, S346C, is well correlated

to the movement of all the charged residues moving in the channel, the latter creating the

gating current, which is fundamental to the gating of the channel.

60

Figure 3.16 Open symbols: RSC vs. voltage for S346C (at the top of the S3-S4 linker in Shaker).

The data shows a 3.2 Å change with the sites getting farther away from each other as the

channel opens. Closed symbols: Q-V for the labeled S346C channels. Gating currents were

recorded simultaneously with LRET. Charge movement and changes in LRET were strongly

correlated.

Voltage-dependent energy transfer changes were also detected at sites S351C,

S352C, and N353C - three successive residues near the S4 segment (Fig. 3.13a).

Surprisingly, the changes are different for each site (Fig. 3.17a). The donor and acceptor

attached to S351C residues move ~1 Å further apart as the channel opens. In contrast,

site S352C shows no significant change in distance, and site N353C moves ~1 Å closer

together as the voltage is depolarized. Note that while these changes are small, they are

highly reproducible (see error bars in figure). It is also possible that the energy transfer

changes at these sites are caused by a reorientation of the acceptor without a

corresponding translation. (Energy transfer in LRET is weakly dependent on the

acceptor orientation with respect to the radius vector joining the donor and acceptor.)

These changes in distance were argued to demonstrate a rotation of this protein region,

although it is possible that a simple tilting or some other small rearrangement could

produce these changes. If these residues were undergoing a large translation (as opposed

to a rotation) across the membrane, which is the simplest and most common hypothesis

for the movement of S4, we would expect all residues to change distances in the same

direction, which is not found. A physical model consistent with the data is shown in Fig.

3.17b. In this scenario, this region of the protein is portrayed as an α-helix, and a 180°

61

rotation moves one residue further from the pore (S351C; black), one residue closer

(N353C; red), and one residue maintains the same distance (S352C; blue).

a b

closed open

Figure 3.17 a. Three residues on the S3-S4 linker near S4, the primary voltage sensor, moved a

small amount, with RSA for S351C becoming 1 Å greater, RSA for S352C remaining the same, and

RSA for N353C becoming 1 Å less upon depolarization. b. Such a pattern in distance changes is

thought to be consistent with a rotation model for the protein conformational change. S351C is

represented by black dots. S352C is represented by blue dots. N353C is represented by red

dots.

A rotation of the S4 region with no apparent outward translation is a surprising

result since it is known that the open and closed state of the channel differ in energy

arising from the movement of charge through the electric field across the membrane

(from inside to outside potential). Indeed, several older models have postulated a

translation of S4 as much as 16 Å across the membrane [19,24] and the recent

controversial KvAP paddle model predicts a vertical movement around 15-20 Å [31].

However, not only does the data at 351-353 suggest a rotation in the S4 (or more

precisely the S3-S4 region), it was argued the sigmoidal shape of the voltage-dependent

movements of S346C (Fig. 3.16), S531C, S352C, and N353C are inconsistent with a

significant translation. To understand this, consider a model where S4 undergoes a

significant translation (Fig. 3.18). Note that there are four S4 segments per channel, and

62

to go from the open to closed state all S4 segments must move. At negative potentials,

the channel is closed and all S4s are down (“closed” state in Fig. 3.18). At intermediate

potentials (“active” state), some S4s are up and some are down, although the channel

remains closed. At more positive potentials, all S4s are up and the channel is open. This

picture has the S4 voltage-sensors moving independently of one other, which has been

demonstrating convincingly by several different lines of evidence [87,88,122]. With a

transmembrane movement of 16 Å, the inter-subunit distance versus voltage would

demonstrate a bell-shaped voltage dependence with a peak change in distance of

approximately 2.2 Å at intermediate potentials (Fig. 3.18). Smaller translations would

still be bell-shaped with simply a lower peak change. Since the actual voltage-dependent

movement is sigmoidal, and not bell-shaped, it is unlikely that the S4 segment undergoes

a significant transmembrane movement with voltage.

This argument however, is substantially indirect and relies on the S4 helices to be

slowly moving (between closed and open states at intermediate potentials) on the time

scale of the LRET measurement. If LRET measurements were instantaneous then the

distribution of S4s would be exactly like the central cartoon of Fig. 3.11, however since

LRET measurements take longer, usually several hundred microseconds, the S4 helices

will be moving during the measurement. Generally, the S4s go between closed and open

on the order of milliseconds, therefore the LRET measurement is significantly faster.

However this argument against vertical translation is subtle and indirect, and the

experimental data could not quantify a small vertical translation if it exists. In Chapter 5

we show data that overcomes this limitation of the S4-donor to S4-acceptor version of the

experiment. In order to test the MacKinnon paddle model, we use LRET to rule out a

significant vertical translation of S4 conclusively.

63

Figure 3.18 A model for S4 movement that includes a significant vertical displacement

perpendicular to the membrane. At intermediate potentials (“active”) some S4s will be displaced

to their outward positions while other S4s will be in their ‘down’ state. Therefore the measured

distance will be greatest at intermediate potentials while a mixture of S4 states is occurring. The

closed and open states show the same distance because all S4s are in the same conformational

state. This bell-shaped change in distance between S4s was not observed for any site near or on

S4 in Cha et al., 1999 [25].

How can a rotation of S4 carry 13 electronic charges per channel across the

electric field – from inside potential to outside potential? Cha et al., 1999 [25] suggested

a model that is consistent with the LRET data and chemical accessibility studies (Fig.

1.11). In both closed and open states of the channel, residues in the S4 segment reside in

crevices for which protons have deeper accessibility than cysteine-reactive reagents, as

shown by histidine and cysteine-scanning accessibility studies [19,21,123-125]. As the

channel goes from the closed to open state, the S4 segment rotates, moving the key

charged residues – R362, R365, R368, and R371 – from one crevice connected to the

intracellular potential to another crevice connected to the extracellular potential. The

water-filled crevices focus the electric field across a relatively thin hydrophobic region,

permitting a small conformational change such as a rotation to move the charge across

64

the field. This rotation of S4 also changes the chemical accessibility of these residues

from the intracellular to the extracellular solution. No S4 translation required.

In summary, these results show that the technique of LRET, combined with site-

directed fluorescent labeling, has the power to study atomic-scale structural changes in

the Shaker channel in vivo. Initial results on the Shaker potassium channel suggested a

rotational model of the voltage-sensing S4 region with the motion of the S3-S4 linker

domain strongly correlated to gating currents arising from S4 charge movement [25].

3.5 Instrumentation The Selvin lab voltage-clamp is the CA-1B oocyte clamp in two electrode mode

(Dagan). Electrophysiology solutions are listed in Appendix D. Recordings were filtered

at 20 kHz and digitized with an A/D conversion card (Innovative Integration). Voltage

command pulses were produced using a D/A (Innovative Integration). The

electrophysiology apparatus was controlled and data was collected and analyzed using in-

house software from the Bezanilla laboratory.

The optical setup consisted of an Olympus inverted IX-70 microscope with a 40x

quartz objective (numerical aperture 0.8, Partec). The lanthanide was excited with a

pulsed 337 nm nitrogen laser source (Oriel), reflected by a 400DCLP dichroic (Chroma).

Donor and acceptor fluorescence were separated using a Q505lp beam splitter and

collected simultaneously with D490/10 and HQ520/20 filters, respectively (Chroma).

Fluorescence was detected with two water-cooled R943-02 photomultiplier tubes

(Hamamatsu) operated at -1760 V. Prompt fluorescence was rejected using an electronic

gate (Products for Research) with a dead-time of 70 µs. The detector current was

converted to voltage (106 V/A, Hamamatsu), filtered at 50 kHz (8-pole Bessel filter,

Dagan), and digitized with an A/D (National Instruments).

65

Chapter 4

LRET Part I: The ILT-Shaker Channel

4.1 What can measurements on the ILT channel tell us?

Cha et al. measured changes in intersubunit distances on Shaker using the S4-

donor to S4-acceptor LRET configuration (Chapter 3.5). A summary of all the results

obtained in that study are presented here in Fig. 4.1. Even though the change in distance

measured for S346C (at the top of the S3-S4 linker) displayed a strong correlation with

gating charge movement (Fig. 3.16), it is possible that the movement is in fact correlated

with the opening steps of the channel. Alternatively, a part of the distance change may

correlate with gating charge movement and the remaining distance change may correlate

with the final opening steps of the channel. This can not be determined by studying wild-

type Shaker due to the high overlap of the charge-voltage relation (Q-V) and conductance

curve (G-V, see Fig. 2.10). Applying LRET to the ILT-Shaker mutant we can now

determine the physical movements associated with the movement of gating charge

separately from movements associated with the cooperative steps of opening. To first

order, we can take the Cha et al. results (Fig. 4.1) as indicating the total movement at a

particular site and we can measure in the ILT channel what voltage range displays this

movement. Several sites from Cha et al. showed no movement so we naturally base our

study on the sites that did show a movement; S346C, S351C, and N353C. However, it

will also be important to study sites that didn’t show a change in the former study

because movements opposite each other along the activation pathway may become

exposed whereas they simply masked each other in the wild-type channel.

66

Figure 4.1 Summary of all the distance measurements obtained from Cha et al., 1999 [25]. For

the location of these sites on the Shaker membrane topology see Fig. 3.13a. The measurements

demonstrate a movement for three sites; S346C, S351C, and N353C. The columns RSC and RSA

were determined from two exponential fits and R’SA was determined by taking RSC x 1.414. We

examine 346 and 351 with LRET using the ILT-Shaker channel to determine where these

changes occur along the activation pathway.

4.2 Initial LRET results on the ILT channel The methodology used to perform LRET measurements on the ILT-Shaker were

identical to those used by Cha et al. and were described in detail in Section 3.5. The

constructs used are ILT-Shaker-IR/W434F. The W434F mutant prevents K+ conduction

and allows us to measure gating currents simultaneously with LRET. However, at the

very positive potentials (0 mV to +200 mV) contamination K+ currents are recorded

because W434F is not strictly nonconducting [80] - strong depolarizations push ions

through. Therefore, measurement of the small amount of gating charge that is moved

during the isolated channel opening steps (Fig. 2.10) could not be repeated in our

measurements. Slight modification was made to the voltage pulsing protocol. Test

pulses from -150 mV to -25 mV (in 25 mV increments) were initiated from a prepulse

potential of -100 mV in order to study the voltage range for charge movement (Q-V).

For the measurements across the activation voltage range (G-V), test pulses from 0 mV to

67

+200 or +225 mV were initiated from a prepulse potential of 0 mV. No slow-type

inactivation was detected using a prepulse to 0 mV and in fact, the ILT mutations seem to

inhibit slow inactivation along with the cooperative opening transition(s). This result was

tested by measuring whether gating charge was immobilized by prolonged

depolarizations to 0 mV, a property of slow-type inactivation [72]. If slow inactivation

occurs a ‘hysteresis’ appears in the charge-voltage relation because greater

hyperpolarizations are required to return the gating charges to their resting state (inward,

down state). This result was qualitatively described in the initial characterization of ILT

[87] and quantitatively presented in another study [122].

Fig. 4.2 plots the distance between contiguous subunits Rsc vs. voltage obtained

for the ILT-Shaker-S346C mutant (at the top of the S3-S4 linker) from the shorter of two

lifetime components from LRET (raw data not shown). The normalized gating charge vs.

voltage curve (Q-V curve) is also plotted to show how well the LRET voltage

dependence correlates with the traditional electrical measurement of voltage-sensor

movement. Note how well Fig. 4.2 reproduces the distances and distance change

published for the site S346C in the wild-type Shaker channel in Fig. 3.16 (the change is a

bit greater, about 4.3 Å). The other important result of Fig. 4.2 is that the physical

movement of residue 346 correlates with the charge movement, Q-V, and very little

movement (< 0.5 Å) is seen for this site in the voltage range of channel opening (between

0 and 225mV, see the G-V in Fig. 2.10). Two sets of data are plotted in 4.2. The blue

data were obtained from cells that were labeled with a ratio of 4 donors to every 1

acceptor (data is the average of 7 cells, error bars are the standard error of the mean). To

check that the labeling conditions actually produce data representative of channels with

only 1 acceptor probe and that signals from channels with 2 or more acceptors are

negligible, we measured LRET on cells labeled with 9 donors to every 1 acceptor. This

labeling ratio produces less overall signal, but gives even greater weight to the channels

of interest (labeled with only 1 acceptor). The red data shows the distances vs. voltage

for these cells and clearly the two labeling conditions give the same distance vs. voltage

result (data is the average of 7 cells, error bars are the standard error of the mean).

68

Figure 4.2 LRET measurements for ILT-Shaker-S346C. Movement of this site correlated

strongly with the Q-V and very little distance change (< 0.5 Å) is observed between 0 mV and

+200 mV (channel activation).

In Fig. 4.3 we present data for the ILT-Shaker-S351C mutant. The distance

between contiguous S351C residues, RSC, is again calculated from the short lifetime

component of sensitized acceptor emission data (blue data, average of 5 cells). Oocyte

cells were labeled with 4 donors to every 1 acceptor and data were fit to 2 exponentials.

This residue was previously measured to have an RSC of about 28 Å with very little

movement (< 1 Å) [25]. Here we show again that the distance between contiguous

S351C sites is about 28 Å and we detect an overall movement of about 1 Å (again

somewhat greater than that detected by Cha et al.) that mostly correlates with the ILT-

69

Shaker charge movement, the Q-V curve. The movement of this site between 0 and

225mV, the voltage range of channel opening, is small, ~0.5 Å or less.

Figure 4.3 LRET measurements for ILT-Shaker-S351C. Most of the movement of this site is

correlated strongly with the Q-V and a small distance change (~0.5 Å) is observed between 0 mV

and +200 mV (channel activation).

These data for S346C and S351C demonstrate unambiguously that almost all of

the physical movement of these sites is directly coupled to the movement of gating

charge across the membrane. Perhaps this is expected, since the voltage-sensing

arginines are near the extracellular side of the membrane (that is, in a transmembrane

model, see Chapter 5) rather far from the actual activation gates (the S6 helical bundle).

It could be that motion of segments near the outside of the channel will correlate with

gating charge movement, which in turn allows gating conformational changes that occur

70

deeper in the channel (physically closer to the activation gates themselves). The ILT

mutation that mostly disrupts channel opening (I372L) is 21 amino acids farther down on

S4 compared to our labeling site S351C.

Another site was tested on the Shaker wild type channel, L361C, which is on the

S4 itself right before the first charged arginine (data not shown). No movement was

found for this site, which is consistent with the finding by Cha et al. that V363C, the site

right after the first charged arginine does not move. It will be necessary to test these sites

with the ILT channel background to make sure that no movement occurs. Therefore, it

appears that we can not detect movements of S4 itself using this S4-donor to S4-acceptor

arrangement of LRET. This suggests that if S4 moves at all, it is likely a purely vertical

movement. We address the issue of S4 vertical movement with direct measurements in

Chapter 5. Whether other ILT channel sites, on S4 or elsewhere such as S1-S3, show

movements that correlate with the actual opening of the channel is the next important

question we will address. We will explore other sites and look for movements that are

related distinctly to charge-movement or channel opening. This project is ongoing. We

have put it aside temporarily because the paddle model inspired a new measurement that

could address the issue of vertical motion for voltage-sensing segments. We now turn to

this project (Chapter 5) that has been completed.

71

Chapter 5

LRET Part II: Shaker with Scorpion Toxin

As soon as MacKinnon presented his new paddle model for gating charge

movement [31] based on the first X-ray structure of a KV channel, KvAP (see Chapter 1)

[30], many labs sought to disprove the paddle concept with new experiments. Various

experiments included a new accessibility study [57], crosslinking studies between the

pore domain and S4 [57-59], and a clever study that examined which sites on S4 are

capable of carrying charge across the electric field [126]. These studies were presented

as contradicting the paddle mechanism of voltage sensing. However, these studies tested

predictions of the new model that were at least one step removed from the basic idea of

the paddle mechanism. In our lab, we sought to directly test the central feature of the

model, a large vertical translation of the S3b and S4 segments. Others have not been able

to put this basic feature under scrutiny, and it is just the kind of thing that energy transfer

experiments are good for. Testing the vertical translation of S3b-S4 is of central

importance in evaluating the validity of the paddle mechanism, as the model’s other

unusual feature, the location of S4 at the lipid interface, has been shown to be

experimentally plausible [50,127]. Here we show that LRET demonstrates no large

perpendicular movement of the voltage sensing segments, in stark contrast to the idea of

a voltage-sensing paddle. The results are consistent with S3 and S4 segments oriented in

a transmembrane fashion for all voltages. The small conformational changes that we

have been able to detect using LRET have demonstrated that the voltage sensor does not

need to move very much in order to carry its huge gating charge across the membrane

field. This fact is consistent with a model of a highly focused electric field.

72

5.1 Ion channels are toxin receptors Puffer fish, though a delicacy in Japan, can be lethal if not prepared properly.

Japanese scientists attempting to understand the paralytic substance secreted by puffer

fish discovered the first toxin - ion channel interaction [128,129]. They discovered that

tetrodotoxin (TTX) blocked sodium currents in axons when only nanomolar

concentrations were applied. TTX is now known to be a potent and selective blocker of

the Na channel pore [130]. Homologous block of KV channels is achieved with a family

of scorpion toxins called α-K-toxins [131]. These toxins are typified by charybdotoxin

(CTX) and agitoxin (AgTX) from the scorpion, Leiurus quinquestriatus (Fig. 5.1a)

[132,133]. These two toxins are 37 and 38 amino acid peptides, respectively, with 3

disulfide bonds stabilizing their structure (6 conserved cysteines). NMR structures have

been determined (Fig 5.1b) [134,135]. Both toxins have a conserved lysine, K27, that

interacts directly with the K+ channel pore (Fig. 5.1c) [136-138].

Leiurus quinquestriatus

a b

c

Figure 5.1 a. Scorpion venom contains many small peptide blockers of Na+, Ca2+, and K+

channels. The first potent blockers for KV channels were isolated from the scorpion Leiurus

quinquestriatus. b. NMR structures for the toxins used in our study, Agitoxin-2 (AgTX) and

73

Charybdotoxin (CTX). Highlighed are residues mutated to cysteine for fluorophore attachment,

AgTX-D20 and CTX-R19. Also highlighted, K27, the critical lysine residue that interacts directly

with the selectivity filter. c. Molecular dynamics model for AgTX bound to the Shaker channel.

Figure taken from Eriksson and Roux, 2002 [138].

Therefore, these small proteins bind to the top of the channel on the central axis. Before

the days of K+ channel crystal structures several mutagenesis studies detailed the

interaction between toxin and channel [139-142]. Models have docked the toxins onto

channels starting from experimental constraints and using molecular dynamics [138,143].

5.2 LRET configuration using acceptor labeled toxin – putting

the paddle model to the test LRET experiments have been presented (Chapter 3.5 and Chapter 4) that were

performed in an S4-donor to S4-acceptor configuration. Here we present data using an

S4-donor to toxin-acceptor configuration. In this new approach, all four Shaker subunits

are labeled with the lanthanide luminescent donor probe, and the acceptor dye is attached

to a scorpion toxin. The toxin is either charybdotoxin (CTX-R19C [144]) or agitoxin-2

(AgTX2-D20C [145]). Procedures for labeling these toxins have been described by our

collaborator Chris Miller [144]. The toxin mutations for labeling (Fig. 5.2b) are on the

‘top’ side of the toxin, whereas the ‘bottom’ side forms the interaction surface with

Shaker. We have shown that labeling these mutant toxins with organic dyes has only a

small effect on the binding affinity to the channels by qualitatively testing the toxin off-

rate (data not shown). We put in the mutations F425G and K427D in our Shaker

construct, which greatly increases the affinity of wild-type CTX with Shaker (Ki ≈ 1.5

pM [139,140]). We have shown that these mutations do not interfere with the strong

binding of agitoxin-2 with Shaker (data not shown). Both of these toxins bind very

strongly and specifically to the external side of the Shaker channel pore (see Fig. 5.1c),

displaying a residency time of many tens of minutes [140,144]. With this geometry the

acceptor is located on a stationary reference point at the very top of the channel and near

the central axis. Therefore, if voltage-sensing segments are undergoing large translations

74

across the membrane, as the KvAP paddle model predicts, this LRET experiment will

detect it. Fig. 5.2 illustrates the geometry, assuming a very conservative version of the

paddle model with the paddle segments undergoing a 15 Å vertical translation, LRET

between S4-donors and a toxin-acceptor show a change in distance of 10 Å. It should be

noted that these values come from conservative structural estimates, and in fact the

paddle model has been presented with even more extreme numbers [31].

Figure 5.2 Cartoon representation of the paddle model. LRET measures distances from donor

labelled sites (blue circles) on the S3b-S4 paddle (structure taken from the isolated voltage-

sensor [30]) to the toxin-acceptor. The voltage-sensing arginines are shown in red. The energy

transfer acceptors (green circles) are attached to the top of the channel with a scorpion toxin. The

paddle model predicts a change in distance, Dc - Do, of 10 Å, estimated by a conservative

geometric calculation assuming a 15 Å vertical translation (red arrows).

The experiments are performed as follows: Xenopus oocyte cells are injected with

mRNA (20 ng) for the Shaker channel (for example N353C with the F425G, K427D

background) such that they over-express channels a few days post-injection. The cells

are labeled with thiol-reactive Tb chelate (the donor species) such that the N353C

75

residues become labeled with donor. (Other cysteines on endogenous proteins in the

oocyte membrane will also become Tb-labeled.) Then the cell is voltage clamped in a

bath solution containing a high-blocking concentration of fluorescent toxin (100nM

toxin). The toxins bind to the Shaker channels in a matter of seconds yielding donor-

acceptor pairs close enough for energy transfer.

There are four important advantages with this configuration. 1) All Shaker

channels give the signal of interest (no mixed populations from stoichiometric labeling as

in the S4-donor to S4-acceptor LRET) with the maximum possible number of donors

(greater signal). 2) The toxin introduces the acceptor probe very specifically (less

background). 3) The acceptor probe location is a stationary and well-defined point,

constrained to a high degree of certainty from toxin docking models [138] (see Fig. 5.1c).

4) It is extremely unlikely that the acceptor will move or change its orientation as a

function of voltage, since the probe is not directly attached to the Shaker channels. This

removes any doubt regarding environmental changes or dye orientation factors for the

acceptor.

In general, background LRET signals arise from two sources: 1) Non-specific

binding of toxin such as binding to endogenous cation channels that may have cysteines

that label with Tb-donor. 2) Diffusional LRET from unbound toxin in the cell bath

solution diffusing close to Tb-donor molecules on the cell surface. Diffusional LRET is

not significant below µM concentrations of acceptor dye [105], and in any case, can be

tested by varying the amount of acceptor-labeling. To approximate the background

signals, control oocyte cells are injected with cysteine-lite Shaker mRNA and allowed to

express similar amounts of channels as the Shaker-cysteine containing cells. The control

cells are labeled with donor side-by-side with the Shaker-cysteine cells. LRET acceptor

signals are measured from both Shaker-cysteine cells and Shaker-cysteine-lite cells in the

same acceptor-toxin bath solution. Here we show that the signal to background achieved

is ~100 when performing measurements in the presence of 100nM Bodipy-Fl-toxin. That

is, the intensity of LRET sensitized emission (acceptor fluorescence) on Shaker-cysteine

expressing cells is about 100 times brighter than the LRET signals on the Shaker-

cysteine-lite expressing oocytes. This is a several-fold improvement over experiments

76

with the S4-donor to S4-acceptor arrangement (Chapter 4) and more optimization may be

possible.

For LRET, ionic currents of Shaker expressed in Xenopus oocytes were blocked

with 100 nM fluorescent charybdotoxin (CTX) or agitoxin-2 (AgTX) such that almost all

channels were blocked and residual unblocked current was limited to 10-30 µA to

minimize voltage-clamping errors [146] (Fig. 5.3a). The charge-voltage relations (Q-V)

for donor-labelled channels were measured separately from cells blocked with a

saturating level of unlabelled toxin (Fig. 5.3b).

337 nm laser shot40ms

a b

Figure 5.3 a. Current recordings during LRET (data for L361C shown). Shaker was blocked

using 100 nM Bodipy-Fl CTX, which produced ~99% block. Residual K+ currents are apparent

and make the measurement of gating currents simultaneous with LRET impossible. The laser

fires 40 ms after the initiation of the voltage step to ensure the Shaker channels have reached

conformational equilibrium. b. Gating currents (data for E333C shown) are recorded with

saturating toxin block (2 µM wild type CTX).

Distance calculations. The lifetime of acceptor sensitized emission was used to

calculate energy transfer efficiency using the relation E = 1 – τAD/τD, where τD is the

lifetime of the donor in the absence of acceptor. τD was measured on channel sites in the

absence of acceptors. On a few sites τD displayed a slight voltage dependence (S346C

and S351C < 10%, E335C < 5%) and these changes were included in the analysis

77

(distances changed < 1 Å). Sensitized emission data were fit to two exponentials using

four parameters; A1, τ1, A2, τ2. Multiple time constants indicate that the acceptor

molecule is not an equal distance from all four donors labeled to the voltage-sensing

domains. An average lifetime was calculated normalizing the sensitized emission

lifetimes by the rate of energy transfer to obtain a ‘population average’ [147];

)/()/()/()/(

2211

222111

kAkAkAkA

++

=ττ

τ where Dn

nkττ11

−= . Distances from τ1, τ2, τ were

calculated by finding E (above) and using R = Ro(E-1 – 1)1/6 where Ro is the characteristic

distance of 50% energy transfer. Most data were taken using Bodipy Fl-maleimide

acceptors (Molecular Probes) for which Ro = 39 Å. Other data were taken using Atto465-

maleimide (Atto-Tec), Ro = 27 Å, and Lucifer Yellow-iodoacetamide (Molecular

Probes), Ro = 23 Å.

5.3 LRET results S3b, S3-S4 linker, and S4 Acceptor sensitized emission data from E333C on S3b and background controls

are shown in Fig. 5.4, left. LRET signals were fit well to two time constants that reflect

the asymmetry of the toxin-acceptor position with respect to the central axis of the

channel (see Fig. 5.6). Distances from both time constants were calculated as well as a

population-weighted average distance vs. voltage (see methods above). These distances

are shown for E333C on S3b and L361C on S4 (Fig. 5.4, right). These results clearly

show that sites homologous to the KvAP voltage-sensor paddle move less than 1 Å with

respect to the toxin when going from the closed to the open channel positions. If the S4

segment moves in a purely vertical direction, a change in LRET distance of 1 Å

corresponds to a 2 Å vertical displacement, as estimated by a conservative geometric

calculation similar to that shown in Fig. 5.2.

78

-150 -100 -50 0 5027282930313233343536

-150 -100 -50 0 5027

28

29

30

31

32

33

34

35

0 500 1000 15000.01

0.1

1

10

L361

C D

ista

nce

(ang

stro

ms)

Voltage (mV)

Voltage (mV)

E333

C D

ista

nce

(ang

stro

ms)

LRET on E333C

Acceptor -150mVτAD = 160 µs, 509 µs

Acceptor +50mV

Donor only τD = 1600 µs

Average control data (n=6)

Time (µs)

Fluo

resc

ence

Figure 5.4 LRET raw data and distance calculations. Acceptor sensitized emission at two

extreme voltages are shown (left) for the E333C mutant near the top of the paddle. The time

constants displayed a small voltage dependence corresponding to a small movement 0.8 Å away

from the toxin (top-right). The distances calculated from the two lifetime components and the

weighted-average (methods above) are shown. L361C showed voltage dependent movement of

0.8 Å towards the toxin (bottom-right). Error bars for the average distance represent standard

error of the mean (n = 13 for E333C, n = 8 for L361C).

Small but unambiguous voltage-dependent movements were seen at many sites

(Fig. 5.5) with S3 moving ~1 Å away from the toxin, S4 moving ~1 Å towards the toxin,

and the sites in the linker moving up to 2.5 Å in a manner consistent with a change in

linker tilt [25]. We note that S3b and S4 move in opposite directions, instead of

translating together as a rigid unit. For three sites, N353C, E335C, and L361C, both

AgTX and CTX gave similar calculated distances. Two sites on the S3-S4 linker were

79

studied using two different acceptors, CTX-Lucifer Yellow and CTX-Atto465, useful for

measuring distances as short as 15 Å. For S346C, the calculated distances differed by

only 2.5 Å, which may be attributed to differences in dye size and linker lengths. For

S351C the distances obtained using CTX-Atto465 vs. CTX-Bodipy Fl differed by 5 Å,

but the gating-induced change in distance was unaffected by the choice of acceptor.

Thus, the absolute distances are slightly uncertain, but the changes in distance are very

reproducible. These small changes refute the most central feature of the paddle model:

substantial physical movement of gating charge transverse to the membrane plane.

-150 -100 -50 0 50 10030

31

32

33

34

35

36

-150 -100 -50 0 5030

31

32

33

34

35

36

-150 -100 -50 0 5015

20

25

30

35S4

Aver

age

dist

ance

s (a

ngst

rom

s) S

4

Voltage (mV)

R365C BodipyFl-CTX∆D = -0.8 angstromsn = 7

L361C BodipyFl-AgTX∆D = -0.7 angstromsn = 8

E333C BodipyFl-CTX∆D = 0.8 angstromsn = 13

S346C Atto465-CTX∆D = 0.5 angstromsn = 6

L361C BodipyFl-CTX∆D = -0.7 angstromsn = 8

S3b

E335C BodipyFl-CTX∆D = 0.7 angstromsn = 3

E335C BodipyFl-AgTX∆D = 1.1 angstromsn = 5

V330C BodipyFl-CTX∆D = 0.1 angstromsn = 11

Aver

age

dist

ance

s (a

ngst

rom

s) S

3b

Voltage (mV)

S346C lucifer yellow-CTX∆D = 0.7 angstromsn = 6

S351C Atto465-CTX∆D = -2.3 angstromsn = 17

S351C BodipyFl-CTX∆D = -2.4 angstromsn = 16

S352C BodipyFl-CTX∆D = -1.7 angstromsn = 5

N353C BodipyFl-CTX∆D = -2.0 angstromsn = 5

N353C BodipyFl-AgTX∆D = -1.4 angstromsn = 4

S3-S4 linker

Ave

rage

dis

tanc

es (a

ngst

rom

s) S

3-S

4 lin

ker

Voltage (mV)

Figure 5.5 Average distances for many Shaker sites. The S4 and S3b sites are homologous to

sites on the KvAP voltage-sensing paddle. The distances for S4 change just 0.8 Å, consistent

with an approximately 2 Å vertical translation. S3b moves in the direction opposite to S4, moving

just 0.8 Å away from the toxin. Sites in the S3-S4 linker are clearly closer to the toxin than the

transmembrane segments, as expected, and move no more than a few angstroms.

80

We are confident that this LRET technique estimates distances faithfully, on proteins in

general and K+ channels in particular. Beyond the technique’s agreement with known

structures in soluble proteins [102], distances measured here agree well with independent

estimates of distances from the Shaker voltage-sensors to the pore using tethered

tetraethylammonium blockers [148]. For example, Q348C, D349C, and K350C were

found to be 17-18 Å from the pore in the open state, very similar to our measurements of

17-19 Å and 21-25 Å for S3-S4 linker residues S346C and S351C, respectively.

Likewise, E334C and E335C were found to be ~30 Å from the pore, close to our

measurements for these same residues at the end of S3b, 32-34 Å. Although tethered

blocker data and LRET measure distances to two different points near the central pore,

the close agreement between the approaches demonstrate their power for constraining

structural distances on the Shaker channel. Furthermore, tethered blockers measure

distances only for the open state whereas LRET has the advantage of probing both open

and closed states.

5.4 Comparison to a model for Shaker LRET measures absolute distances with less systematic error than traditional

energy transfer techniques [99,102], and can therefore be used to evaluate and constrain

structural models. Recently, a structural model was proposed for the Shaker open state

based on a combination of experimental data and molecular dynamics [54,59]. This

model was supplemented with a computationally docked agitoxin [138] so that

theoretical distances from the toxin labeling site to sites on the Shaker voltage sensor

could be compared directly with our LRET measurements (Fig. 5.6). The model predicts

four theoretical distances and we have used simulations to test how well LRET

experiments can measure the average distance for situations of such geometric

complexity (Table 5.1). The simulations reproduce average distances in close agreement

to model values, with the exception of S351 using CTX-Atto465 where the small Ro

caused an underestimation. The LRET experimental results for two sites on S4

demonstrate very close agreement between model and data. The model prediction for

81

S3b was unique in that it predicted a shorter distance (~ 4 Å) than the distance measured

experimentally. However, the LRET measurements may systematically underestimate

distances slightly because the position of the probes can wobble around their linker

attachment points, weighting the measurement towards the distance of closest approach.

The distance values obtained with LRET are thus consistent with a general structural

view that S3 and S4 segments are transmembrane segments at all voltages.

L361C-TbL361C-Tb

L361C-Tb

L361C-Tb

AgTX-20C-Fl

Shaker

Agitoxin2

40.3 Å

42.7 Å

28.4 Å

31.7 Å

Figure 5.6 A model of Shaker with docked agitoxin predicts four distances for each LRET

experiment. Distances for L361C on S4 are shown measured from alpha-carbons (right). The

coordinates (left) provide an opportunity to compare our measurements with a picture of Shaker

that has S3 and S4 placed against the pore domain.

Coordinates for the Shaker open-state model with docked agitoxin [59,138]

provide predictions for four different distances between the agitoxin-D20 alpha carbon

and the four alpha carbons of selected sites on the voltage-sensors (Fig. 5.6). Assuming

these four distances, LRET signals were simulated assuming a bi-exponential donor with

dominant component, 75% at 1600 µs, and a minor component, 25% at 300 µs. The

minor component adds a systematic error that slightly underestimates distances (< 5%,

see below). The multiple components can be well described by fitting to two

exponentials (Supplementary Data) as was the experimental data. These calculations

82

show how the complicated geometry of the model can be reduced to distance estimations

in close agreement with actual LRET measurements (Table 5.1).

Table 5.1 Summary of Distance Comparisons to a Shaker model (Fig. 5.6) Donor site Acceptor used Avg. model D (Å)* Avg. simulation

D (Å)** LRET

Avg. D (Å) (S4) L361 BodipyFl-CTX 35.8 33.4 33.3

(S4) R365 BodipyFl-CTX 39.8 35.5 33.2

(S3b) V330 BodipyFl-CTX 30.9 30.2 34.8

(S3-S4 Linker) S351 BodipyFl-CTX 30.2 30.5 25.5

(S3-S4 Linker) S351 Atto465-CTX 30.2 25.0 21.0

*The mean of four distances measured from the Cα of D20 on a docked AgTX to the Cα of

specified sites on each subunit (Figure 4).

**See calculations below.

The D20 alpha-carbon on a docked agitoxin was used to measure four distances to

the alpha carbons of S351, S352, N353, V330, L361, and R365. Assuming a perfect

donor species of 1600 µs lifetime and using the BodipyFl Ro = 39 Å, 4 time constants

and amplitudes can be calculated. Our donor is not perfectly single exponential, but has a

minor species (25%) with lifetime of 300 µs. For this species Ro = 29.5 Å with BodipyFl.

The minor donor species yields 4 more components to the LRET simulation, though they

change the resulting analysis by less than 5% typically (below).

For L361C, the calculated signals for the dominant donor species were: 42.36

exp(-t/205.7) + 21.88 exp(-t/355.5) + 5.13 exp(-t/879.0) + 3.63 exp(-t/1011.8) and for the

minor species were: 14.11 exp(-t/132.2) + 7.29 exp(-t/181.2) + 1.71 exp(-t/260.1) + 1.21

exp(-t/270.5) These signals, even though there are 8 components, can be faithfully

described with 2 exponentials (four parameters) just as the real LRET data was analyzed.

Distance calculations assume that there was just a single donor species of 1600 µs.

83

Therefore, two distances are obtained and an average distance is calculated just as the

experimental data.

1

10

100

0 500 1000 1500

361C simulation data

361C with bodipyFL major LRET componentsTOTAL simulation 361C with BodipyFL

time (microsec)

y = m1*exp(-m0/m2)+m3*exp(-m...ErrorValue

0.05615557.316m1 0.23005227.93m2

0.06543915.424m3 1.7506733.01m4

NA0.054732ChisqNA1R

y = m1*exp(-m0/m2)+m3*exp(-m...ErrorValue

0.1318276.434m1 0.40889198.45m2 0.1605619.85m3

2.9093652.66m4 NA0.45126ChisqNA1R

Figure 5.7 Simulatated LRET signals assuming the four Donor-Acceptor distances from

Laine/Roux model [59]. Black curve represents signals assuming a perfect single exponential

donor species with a lifetime of 1600 µs. Blue curve represents signals assuming a bi-

exponential donor with 75% 1600 µs and 25% 300 µs. These signals can be fit to a four

parameter, two exponential fit, as the experimental data. Results are shown right.

The following tables compare distances obtained from simulations that follow the above

procedure with the measured LRET distances.

84

L361C/BodipyFl-CTX: Model distances: 28.4, 31.7, 40.3, 42.7. Model Davg = 35.8 Å

Major τ components (µs): 205.7, 355.5, 879.0, 1011. τAVG = 613.0 DAVG = 36.0 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

28.9 37.9 34.5

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

28.2 36.7 33.4

Actual LRET measurements (open state) 27.3 34.5 33.3

R365C/BodipyFl-CTX: Model distances: 33.5, 35.8, 44.0, 45.8. Model Davg = 39.8 Å

Major τ components (µs): 457.9, 600.8, 1076, 1158. τAVG = 823.0 DAVG = 39.4 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

31.3 36.5 36.4

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

27.4 35.8 35.5

Actual LRET measurements (open state) 26.5 35.3 33.2

85

V330C/BodipyFl-CTX: Model distances: 22.6, 26.6, 35.8, 38.4. Model Davg = 30.9

Major τ components (µs): 58.98, 146.3, 596.4, 762.8. τAVG = 391.1 DAVG = 32.3 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

24.0 34.5 31.1

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

23.8 33.9 30.2

Actual LRET measurements (open state) 28.0 35.6 34.8

S351C/BodipyFl-CTX: Model distances: 19.1, 28.3, 33.7, 39.6. Model Davg = 30.2

Major τ components (µs): 22.05, 202.7, 471.5, 838.4. τAVG = 383.7 DAVG = 32.2 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

22.4 32.3 31.2

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

22.5 31.8 30.5

Actual LRET measurements (open state) 22.3 27.7 25.5

86

S351C/Atto465-CTX: Model distances: 19.1, 28.3, 33.7, 39.6. Model Davg = 30.2

Major τ components (µs): 180.2, 909.6, 1266, 1455. τAVG = 952.65 DAVG = 28.8 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

19.1 30.0 26.6

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

18.8 28.5 25.0

Actual LRET measurements (open state) 16.6 22.7 21.0

S352C/BodipyFl-CTX: Model distances: 16.0, 25.1, 30.4, 36.0. Model Davg = 26.8

Major τ components (µs): 7.51, 105.2, 291.7, 610.9. τAVG = 253.8 DAVG = 29.5 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

25.4 33.1 30.6

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

25.2 32.8 30.0

Actual LRET measurements (open state) 22.7 28.5 26.5

87

S353C/BodipyFl-CTX: Model distances: 17.7, 24.3, 31.9, 36.1. Model Davg = 27.5

Major τ components (µs): 13.7, 89.3, 369.3, 615.9. τAVG = 272.0 DAVG = 29.9 Å

D1 (Å) D2 (Å) Davg (Å)

Simulated LRET distances – 4 major components (2 exp fit analysis)

24.2 33.5 31.1

Simulated LRET distances – 4 major components + 4 minor components (2exp fit)

24.1 33.0 30.3

Actual LRET measurements (open state) 23.8 30.6 28.5

Here we have shown that model coordinates for Shaker sites overall show a

disposition consistent with the distances measured using LRET. In the model, the S4

segment is based on numerous experimental constraints, whereas the placement of the

S1-S3 helices is much more uncertain. However, the view of voltage-sensing segments

situated in a transmembrane orientation peripheral to the pore-domain is generally

consistent with our measurements and inconsistent with the paddle model.

5.5 Conclusions The small vertical S4 movements presented here supplement the even smaller

lateral movements between voltage-sensors previously obtained [25] and indicate that the

conformational changes that underlie gating charge movement are subtle rather than

substantial. The paddle model could be altered to account for the present data obtained

with toxin by allowing the paddle segments to swing laterally outward while undergoing

vertical movement such that distances to the toxin remain constant. However, this kind

of movement would be flatly inconsistent with the small lateral displacements observed

in previous LRET measurements using the S4-donor to S4-acceptor arrangement [25].

The small physical movements of voltage-sensing segments suggest that the membrane

88

electric field must be focused over a very tight region of the voltage-sensing region, as if

aqueous crevices penetrate the protein and thereby shape the field profile [125,149].

Small S4 movements relative to these crevices and voltage-induced changes in crevice

shape can produce the large gating charge that must traverse the field to account for the

steep voltage dependence of voltage-gated channels.

In summary, our understanding of how voltage-gated channels move their gating

charges across the membrane electric field has been locked in a struggle between two

broad models. The paddle model predicts that the voltage-sensing segments move a large

distance across the thickness of the membrane, ~15 to 20 Å. The focused-field model

predicts that the electric field does not take a simple form across the voltage-sensing

domains, but rather is shaped by aqueous crevices that focus the field strongly across a

narrow hydrophobic region. Gating charges can move across this highly focused field

with very small conformational changes. Our measurements have demonstrated that the

large conformational changes required for the paddle mechanism do not occur.

Therefore, we believe that the protein has transmembrane segments that shape the electric

field allowing gating charge motion to be coupled to a small conformational change.

5.6 Future experiments As a matter of control, we will repeat some of the LRET experiments using toxin

but with the configuration reversed. We have made a charbydotoxin labeled with Tb-

chelate. Therefore, we can attach acceptors to the voltage-sensor sites and measure

LRET between 1 donor and 4 acceptors. In this configuration, we can test acceptors that

have very short linkers and therefore the position of the probe will be more rigorously

coupled to the site of interest. This will address potential concerns that the Tb-chelate is

a rather large probe with a quite long linker and therefore may not faithfully reflect the

movements of the labeling site.

Questions remain about how voltage-gated channels structure their voltage-

sensing domains. If the focused field theory is correct, how S1-S4 is structured to

accomplish the shaping of the electric field is a fundamental question. Crystallography

may help address this question but it seems likely that the distortions in KvAP structure

89

were more extensive than MacKinnon and coworkers believe. Perhaps new structures

will tell a different story.

Interesting structural discrepancies have cropped up between studies of eukaryotic

KV channels like Shaker and studies performed on bacterial channels. EPR on KvAP

[50] and LRET on a bacterial sodium channel [150,151] (Bezanilla, personal

communication) have indicated S1 is very close to the pore domain with S1-S3 probably

packed around it. Both studies used channels that were not voltage-clamped and without

membrane polarization and were therefore in a deeply inactivated conformation. These

structural conclusions seem inconsistent with data from eukaryotic KV channels. LRET

on Shaker (Cha et al. [25]) suggested the S4 is the closest segment to the pore domain

with S1-S3 likely surrounding it. Other data are consistent with such a picture, tethered

blockers [148] and S4-pore crosslinking data [57-59]. Scanning mutagenesis on

eukaryotic KV channels suggest that S1 has a lipid accessible face [152,153] in contrast to

the EPR finding for KvAP. The Laine-Roux model for the Shaker open state (Fig. 5.6

[59]) examined in this work was based on these data for Shaker. Perhaps these

discrepancies are reflected in the conformational changes associated with slow

inactivation. Possibly, the bacterial channels are different and the focused field

mechanism can be used with different arrangements of S1-S4.

We will perform more S4-donor to S4-acceptor LRET measurements in order to

better constrain all of the segments, S1-S4. Cha et al. [25] did not take data on S1 or S3.

Furthermore, the inactivated state was not studied. Therefore we will measure voltage-

dependent distances and then measure the distance for the inactivated state. The

mechanism of gating charge immobilization [72] in the inactivated conformation is

waiting to be discovered.

Other labs have made linked dimers of Shaker subunits. Two dimers come

together to form a full tetrameric channel in the membrane. With these constructs, a

single cysteine can be put into the dimmer. Therefore, experiments in the S4-donor to

S4-acceptor arrangement can be done with only one pair of cysteines and only one donor-

acceptor distance. This eliminated the complications due to channel populations from

stoichiometric labeling and donor populations (RSC vs RSA). This geometric

90

simplification will help us build more confident constraints on the structural packing of

S1-S4.

We will pursue more S4-donor to toxin-acceptor experiments, expanding our

study to include S1 and S2. The ILT mutant can also be studied using the toxin-acceptor

arrangement. Many structural constraints can be obtained and perhaps a new generation

of molecular dynamics modeling will use our constraints.

91

Appendix A

Molecular Biology

A.1 Shaker constructs Our Shaker constructs originate from the Bezanilla lab. The channel is the Shaker

H4IR (H4 [154,155] inactivation removed [73]). Improvements in plasmid design for

obtaining high channel expression in oocytes were made by Dorine Starace (personal

communication) and described in Shih et al. [156].

Low expression constructs for an ILT-Shaker (Shaker B? which only differs from

H4 in two residues near C-terminus) channel were obtained from the Aldrich lab. A

section of this clone, between the BsiWI and SpeI restriction sites, were subcloned into

the high expression vector from Starace/Bezanilla, resulting in a high expression ILT-

Shaker construct. We have these plasmids both with and without the nonconducting

W434F mutation [79].

Shaker H4IR/W434F (marked ZH4IR/W434F-PBSTA or PBGLT): Oocyte clone: Kozak consensus. seq + H4IR CDS inserted into the BglII site of pBSTA vector. Mutation W434F (wt is TGG) T7 promotor: nt#635-653 ShH4IR translation start site: nt(#759) ShH4IR translation stop site: nt(#2604) Cloning sites: nt#749(BglII site destroyed), #2618(BglII site maintained). Xen. beta globin 5'UT: #703-748 Xen. beta globin 3'UT: #2619-2821 Zh4ir-Pbt.Seq Length: 5062 August 1, 1996 12:11 Type: N Check: 7071 .. 1 GGAAATTGTA AACGTTAATA TTTTGTTAAA ATTCGCGTTA AATTTTTGTT 51 AAATCAGCTC ATTTTTTAAC CAATAGGCCG AAATCGGCAA AATCCCTTAT 101 AAATCAAAAG AATAGACCGA GATAGGGTTG AGTGTTGTTC CAGTTTGGAA 151 CAAGAGTCCA CTATTAAAGA ACGTGGACTC CAACGTCAAA GGGCGAAAAA 201 CCGTCTATCA GGGCGATGGC CCACTACGTG AACCATCACC CTAATCAAGT 251 TTTTTGGGGT CGAGGTGCCG TAAAGCACTA AATCGGAACC CTAAAGGGAG 301 CCCCCGATTT AGAGCTTGAC GGGGAAAGCC GGCGAACGTG GCGAGAAAGG

92

351 AAGGGAAGAA AGCGAAAGGA GCGGGCGCTA GGGCGCTGGC AAGTGTAGCG 401 GTCACGCTGC GCGTAACCAC CACACCCGCC GCGCTTAATG CGCCGCTACA 451 GGGCGCGTCG CGCCATTCGC CATTCAGGCT GCGCAACTGT TGGGAAGGGC 501 GATGAATTCA TCGGTGCGGG CCTCTTCGCT ATTACGCCAG CTGGCGAAAG 551 GGGGATGTGC TGCAAGGCGA TTAAGTTGGG TAACGCCAGG GTTTTCCCAG 601 TCACGACGTT GTAAAACGAC GGCCAGTGAA TTGTAATACG ACTCACTATA 651 GGGCGAATTG GGTACCGGGC CCCCCCTCGA GGTCGACGGT ATCGATAAGC 701 TTGCTTGTTC TTTTTGCAGA AGCTCAGAAT AAACGCTCAA CTTTGGCAga 751 tcgccaccAT GGCCGCCGTT GCCCTGCGGG AGCAGCAGCT CCAGCGCAAC 801 TCCCTCGATG GTTACGGGTC TTTGCCCAAA TTGAGCAGTC AAGACGAAGA 851 AGGGGGGGCT GGTCATGGCT TTGGTGGCGG ACCGCAACAC TTTGAACCCA 901 TTCCTCACGA TCATGATTTC TGCGAAAGAG TCGTTATAAA TGTAAGCGGA 951 TTAAGGTTTG AGACACAACT ACGTACGTTA AATCAATTCC CGGACACGCT 1001 GCTTGGGGAT CCAGCTCGGA GATTACGGTA CTTTGACCCG CTTAGAAATG 1051 AATATTTTTT TGACCGTAGT CGACCGAGCT TCGATGCGAT TTTATACTAT 1101 TATCAGAGTG GTGGCCGACT ACGGAGACCG GTCAATGTCC CTTTAGACGT 1151 ATTTAGTGAA GAAATAAAAT TTTATGAATT AGGTGATCAA GCAATTAATA 1201 AATTCAGAGA GGATGAAGGC TTTATTAAAG AGGAAGAAAG ACCATTACCG 1251 GATAATGAGA AACAGAGAAA AGTCTGGCTG CTCTTCGAGT ATCCAGAAAG 1301 TTCGCAAGCC GCCAGAGTTG TAGCCATAAT TAGTGTATTT GTTATATTGC 1351 TATCAATTGT TATATTTTGT CTAGAAACAT TACCCGAATT TAAGCATTAC 1401 AAGGTGTTCA ATACAACAAC AAATGGCACA AAAATCGAGG AAGACGAGGT 1451 GCCTGACATC ACAGATCCTT TCTTCCTTAT AGAAACGTTA TGCATTATTT 1501 GGTTTACATT TGAACTAACT GTCAGGTTCC TCGCATGTCC GAACAAATTA 1551 AATTTCTGCA GGGATGTCAT GAATGTTATC GACATAATCG CCATCATTCC 1601 GTACTTTATA ACACTAGCGA CTGTCGTTGC CGAAGAGGAG GATACGTTAA 1651 ATCTTCCAAA AGCGCCAGTC AGTCCACAGG ACAAGTCATC GAATCAGGCT 1701 ATGTCCTTGG CAATATTACG AGTGATACGA TTAGTTCGAG TATTTCGAAT 1751 ATTTAAGTTA TCTAGGCATT CGAAGGGTTT ACAGATCTTA GGACGAACTC 1801 TGAAAGCCTC AATGCGGGAA TTAGGTTTAC TTATATTTTT CTTATTTATA 1851 GGCGTCGTAC TCTTCTCATC GGCGGTTTAT TTTGCGGAAG CTGGAAGCGA 1901 AAATTCCTTC TTCAAGTCCA TACCCGATGC ATTTTtcTGG GCGGTGGTTA 1951 CCATGACCAC CGTTGGATAT GGTGACATGA CACCCGTCGG CGTTTGGGGC 2001 AAGATTGTGG GATCACTTTG TGCCATTGCT GGCGTGCTGA CCATCGCACT 2051 GCCGGTGCCG GTCATCGTCA GCAATTTCAA CTACTTCTAT CACCGCGAAA 2101 CGGATCAGGA GGAGATGCAG AGCCAGAACT TTAATCACGT TACTAGTTGT 2151 CCATATTTGC CCGGGACATT AGTAGGTCAA CACATGAAGA AATCATCATT 2201 GTCTGAGTCC TCATCGGATA TGATGGATTT GGACGATGGT GTCGAGTCCA 2251 CGCCGGGATT GACAGAAACA CATCCTGGAC GCAGTGCGGT GGCTCCATTT 2301 TTGGGAGCCC AGCAGCAGCA GCAACAACCG GTAGCATCCT CACTGTCGAT 2351 GTCGATCGAC AAACAACTGC AGCACCCACT GCAGCAGCTG ACGCAGACGC 2401 AACTGTACCA ACAGCAGCAA CAGCAGCAGC AGCAGCAGCA AAACGGCTTC 2451 AAGCAGCAGC AGCAACAGAC GCAGCAGCAG CTGCAACAGC AACAGTCCCA 2501 CACAATAAAC GCAAGTGCAG CAGCGGCGAC GAGCGGCAGC GGCAGTAGCG 2551 GTCTCACCAT GAGGCACAAT AATGCCCTGG CCGTTAGTAT CGAGACCGAC 2601 GTTTGACTAC TGgtngcaGA TCTGGTTACG TTACCACTAA ACCAGCCTCA 2651 AGAACACCCG AATGGAGTCT CTAAGCTACA TAATACCAAC TTACACTTTA 2701 CAAAATGTTG TCCCCCAAAA TGTAGCCATT CGTATCTGCT CCTAATAAAA 2751 AGAAAGTTTC TTCACATTCT AAAAAAAAAA AAAAAAAAAA AAAAAAAAAA 2801 AAACCCCCCC CCCCCCCCCC CTGCAGCCCC TAGAGCGGCC GCCACCGCGG 2851 TGGAGCTCCA GCTTTTGTTC CCTTTAGTGA GGGTTAATTC CGAGCTTGGC 2901 GTAATCATGG TCATAGCTGT TTCCTGTGTG AAATTGTTAT CCGCTCACAA 2951 TTCCACACAA CATACGAGCC GGAAGCATAA AGTGTAAAGC CTGGGGTGCC 3001 TAATGAGTGA GCTAACTCAC ATTAATTGCG TTGCGCTCAC TGCCCGCTTT

93

3051 CCAGTCGGGA AACCTGTCGT GCCAGCTGCA TTAATGAATC GGCCAACGCG 3101 CGGGGAGAGG CGGTTTGCGT ATTGGGCGCT CTTCCGCTTC CTCGCTCACT 3151 GACTCGCTGC GCTCGGTCGT TCGGCTGCGG CGAGCGGTAT CAGCTCACTC 3201 AAAGGCGGTA ATACGGTTAT CCACAGAATC AGGGGATAAC GCAGGAAAGA 3251 ACATGTGAGC AAAAGGCCAG CAAAAGGCCA GGAACCGTAA AAAGGCCGCG 3301 TTGCTGGCGT TTTTCCATAG GCTCCGCCCC CCTGACGAGC ATCACAAAAA 3351 TCGACGCTCA AGTCAGAGGT GGCGAAACCC GACAGGACTA TAAAGATACC 3401 AGGCGTTTCC CCCTGGAAGC TCCCTCGTGC GCTCTCCTGT TCCGACCCTG 3451 CCGCTTACCG GATACCTGTC CGCCTTTCTC CCTTCGGGAA GCGTGGCGCT 3501 TTCTCATAGC TCACGCTGTA GGTATCTCAG TTCGGTGTAG GTCGTTCGCT 3551 CCAAGCTGGG CTGTGTGCAC GAACCCCCCG TTCAGCCCGA CCGCTGCGCC 3601 TTATCCGGTA ACTATCGTCT TGAGTCCAAC CCGGTAAGAC ACGACTTATC 3651 GCCACTGGCA GCAGCCACTG GTAACAGGAT TAGCAGAGCG AGGTATGTAG 3701 GCGGTGCTAC AGAGTTCTTG AAGTGGTGGC CTAACTACGG CTACACTAGA 3751 AGGACAGTAT TTGGTATCTG CGCTCTGCTG AAGCCAGTTA CCTTCGGAAA 3801 AAGAGTTGGT AGCTCTTGAT CCGGCAAACA AACCACCGCT GGTAGCGGTG 3851 GTTTTTTTGT TTGCAAGCAG CAGATTACGC GCAGAAAAAA AGGATCTCAA 3901 GAAGATCCTT TGATCTTTTC TACGGGGTCT GACGCTCAGT GGAACGAAAA 3951 CTCACGTTAA GGGATTTTGG TCATGAGATT ATCAAAAAGG ATCTTCACCT 4001 AGATCCTTTT AAATTAAAAA TGAAGTTTTA AATCAATCTA AAGTATATAT 4051 GAGTAAACTT GGTCTGACAG TTACCAATGC TTAATCAGTG AGGCACCTAT 4101 CTCAGCGATC TGTCTATTTC GTTCATCCAT AGTTGCCTGA CTCCCCGTCG 4151 TGTAGATAAC TACGATACGG GAGGGCTTAC CATCTGGCCC CAGTGCTGCA 4201 ATGATACCGC GAGACCCACG CTCACCGGCT CCAGATTTAT CAGCAATAAA 4251 CCAGCCAGCC GGAAGGGCCG AGCGCAGAAG TGGTCCTGCA ACTTTATCCG 4301 CCTCCATCCA GTCTATTAAT TGTTGCCGGG AAGCTAGAGT AAGTAGTTCG 4351 CCAGTTAATA GTTTGCGCAA CGTTGTTGCC ATTGCTACAG GCATCGTGGT 4401 GTCACGCTCG TCGTTTGGTA TGGCTTCATT CAGCTCCGGT TCCCAACGAT 4451 CAAGGCGAGT TACATGATCC CCCATGTTGT GCAAAAAAGC GGTTAGCTCC 4501 TTCGGTCCTC CGATCGTTGT CAGAAGTAAG TTGGCCGCAG TGTTATCACT 4551 CATGGTTATG GCAGCACTGC ATAATTCTCT TACTGTCATG CCATCCGTAA 4601 GATGCTTTTC TGTGACTGGT GAGTACTCAA CCAAGTCATT CTGAGAATAG 4651 TGTATGCGGC GACCGAGTTG CTCTTGCCCG GCGTCAATAC GGGATAATAC 4701 CGCGCCACAT AGCAGAACTT TAAAAGTGCT CATCATTGGA AAACGTTCTT 4751 CGGGGCGAAA ACTCTCAAGG ATCTTACCGC TGTTGAGATC CAGTTCGATG 4801 TAACCCACTC GTGCACCCAA CTGATCTTCA GCATCTTTTA CTTTCACCAG 4851 CGTTTCTGGG TGAGCAAAAA CAGGAAGGCA AAATGCCGCA AAAAAGGGAA 4901 TAAGGGCGAC ACGGAAATGT TGAATACTCA TACTCTTCCT TTTTCAATAT 4951 TATTGAAGCA TTTATCAGGG TTATTGTCTC ATGAGCGGAT ACATATTTGA 5001 ATGTATTTAG AAAAATAAAC AAATAGGGGT TCCGCGCACA TTTCCCCGAA 5051 AAGTGCCACC TG

ILT-Shaker H4IR/W434F (marked (ILT)ZH4IR/W434F-PBSTA or PBGLT): GGAAATTGTAAACGTTAATATTTTGTTAAAATTCGCGTTAAATTTTTGTTAAATCAGCTCATTTTTTAACCAATAGGCCGAAATCGGCAAAATCCCTTATAAATCAAAAGAATAGACCGAGATAGGGTTGAGTGTTGTTCCAGTTTGGAACAAGAGTCCACTATTAAAGAACGTGGACTCCAACGTCAAAGGGCGAAAAACCGTCTATCAGGGCGATGGCCCACTACGTGAACCATCACCCTAATCAAGTTTTTTGGGGTCGAGGTGCCGTAAAGCACTAAATCGGAACCCTAAAGGGAGCCCCCGATTTAGAGCTTGACGGGGAAAGCCGGCGAACGTGGCGAGAAAGGAAGGGAAGAAAGCGAAAGGAGCGGGCGCTAGGGCGCTGGCAAGTGTAGCGGTCACGCTGCGCGTAACCACCACACCCGCCGCGCTTAATGCGCCGCTACAGGGCGCGTCGCGCCATTCGCCATTCAGGCTGCGCAACTGTTGGGAAGGGCGATGAATTCATCGGTGCGGGCCTCTTCGCTATTACGCCAGCTGGCGAAAGGGGGATGTGCT

94

GCAAGGCGATTAAGTTGGGTAACGCCAGGGTTTTCCCAGTCACGACGTTGTAAAACGACGGCCAGTGAATTGTAATACGACTCACTATAGGGCGAATTGGGTACCGGGCCCCCCCTCGAGGTCGACGGTATCGATAAGCTTGCTTGTTCTTTTTGCAGAAGCTCAGAATAAACGCTCAACTTTGGCAgatcgccaccATGGCCGCCGTTGCCCTGCGGGAGCAGCAGCTCCAGCGCAACTCCCTCGATGGTTACGGGTCTTTGCCCAAATTGAGCAGTCAAGACGAAGAAGGGGGGGCTGGTCATGGCTTTGGTGGCGGACCGCAACACTTTGAACCCATTCCTCACGATCATGATTTCTGCGAAAGAGTCGTTATAAATGTAAGCGGATTAAGGTTTGAGACACAACTACGTACGTTAAATCAATTCCCGGACACGCTGCTTGGGGATCCAGCTCGGAGATTACGGTACTTTGACCCGCTTAGAAATGAATATTTTTTTGACCGTAGTCGACCGAGCTTCGATGCGATTTTATACTATTATCAGAGTGGTGGCCGACTACGGAGACCGGTCAATGTCCCTTTAGACGTATTTAGTGAAGAAATAAAATTTTATGAATTAGGTGATCAAGCAATTAATAAATTCAGAGAGGATGAAGGCTTTATTAAAGAGGAAGAAAGACCATTACCGGATAATGAGAAACAGAGAAAAGTCTGGCTGCTCTTCGAGTATCCAGAAAGTTCGCAAGCCGCCAGAGTTGTAGCCATAATTAGTGTATTTGTTATATTGCTATCAATTGTTATATTTTGTCTAGAAACATTACCCGAATTTAAGCATTACAAGGTGTTCAATACAACAACAAATGGCACAAAAATCGAGGAAGACGAGGTGCCTGACATCACAGATCCTTTCTTCCTTATAGAAACGTTATGtATTATTTGGTTTACATTTGAACTAACTGTCAGGTTCCTCGCATGTCCGAACAAATTAAATTTCTGCAGGGATGTCATGAATGTTATCGACATAATCGCCATCATTCCGTACTTTATAACACTAGCGACTGTCGTTGCCGAAGAGGAGGATACGTTAAATCTTCCAAAAGCGCCAGTCAGTCCACAGGACAAGTCATCGAATCAGGCTATGTCCTTGGCAATATTACGAGTGATACGATTAGTTCGAaTcTTTCGAcTgTTTAAGTTAaCgAGGCATTCGAAaGGccTACAaATaTTAGGACGAACTCTGAAAGCCTCAATGCGGGAgcTcGGTTTACTTATATTTTTCTTATTTATAGGCGTCGTACTCTTCTCATCGGCGGTTTATTTTGCGGAAGCTGGAAGCGAAAATTCCTTCTTCAAGTCCATACCCGATGCATTTTtcTGGGCGGTcGTTACCATGACCACCGTTGGATATGGTGACATGACACCCGTCGGCGTTTGGGGCAAGATTGTGGGATCACTTTGTGCCATTGCTGGCGTGCTGACCATCGCACTGCCGGTGCCGGTCATCGTCAGCAATTTCAACTACTTCTATCACCGCGAAACGGATCAGGAGGAGATGCAGAGCCAGAACTTTAATCACGTTACTAGTTGTCCATATTTGCCCGGGACATTAGTAGGTCAACACATGAAGAAATCATCATTGTCTGAGTCCTCATCGGATATGATGGATTTGGACGATGGTGTCGAGTCCACGCCGGGATTGACAGAAACACATCCTGGACGCAGTGCGGTGGCTCCATTTTTGGGAGCCCAGCAGCAGCAGCAACAACCGGTAGCATCCTCACTGTCGATGTCGATCGACAAACAACTGCAGCACCCACTGCAGCAGCTGACGCAGACGCAACTGTACCAACAGCAGCAACAGCAGCAGCAGCAGCAGCAAAACGGCTTCAAGCAGCAGCAGCAACAGACGCAGCAGCAGCTGCAACAGCAACAGTCCCACACAATAAACGCAAGTGCAGCAGCGGCGACGAGCGGCAGCGGCAGTAGCGGTCTCACCATGAGGCACAATAATGCCCTGGCCGTTAGTATCGAGACCGACGTTTGACTACTGgtngcaGATCTGGTTACGTTACCACTAAACCAGCCTCAAGAACACCCGAATGGAGTCTCTAAGCTACATAATACCAACTTACACTTTACAAAATGTTGTCCCCCAAAATGTAGCCATTCGTATCTGCTCCTAATAAAAAGAAAGTTTCTTCACATTCTAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAACCCCCCCCCCCCCCCCCCTGCAGCCCCTAGAGCGGCCGCCACCGCGGTGGAGCTCCAGCTTTTGTTCCCTTTAGTGAGGGTTAATTCCGAGCTTGGCGTAATCATGGTCATAGCTGTTTCCTGTGTGAAATTGTTATCCGCTCACAATTCCACACAACATACGAGCCGGAAGCATAAAGTGTAAAGCCTGGGGTGCCTAATGAGTGAGCTAACTCACATTAATTGCGTTGCGCTCACTGCCCGCTTTCCAGTCGGGAAACCTGTCGTGCCAGCTGCATTAATGAATCGGCCAACGCGCGGGGAGAGGCGGTTTGCGTATTGGGCGCTCTTCCGCTTCCTCGCTCACTGACTCGCTGCGCTCGGTCGTTCGGCTGCGGCGAGCGGTATCAGCTCACTCAAAGGCGGTAATACGGTTATCCACAGAATCAGGGGATAACGCAGGAAAGAACATGTGAGCAAAAGGCCAGCAAAAGGCCAGGAACCGTAAAAAGGCCGCGTTGCTGGCGTTTTTCCATAGGCTCCGCCCCCCTGACGAGCATCACAAAAATCGACGCTCAAGTCAGAGGTGGCGAAACCCGACAGGACTATAAAGATACCAGGCGTTTCCCCCTGGAAGCTCCCTCGTGCGCTCTCCTGTTCCGACCCTGCCGCTTACCGGATACCTGTCCGCCTTTCTCCCTTCGGGAAGCGTGGCGCTTTCTCATAGCTCACGCTGTAGGTATCTCAGTTCGGTGTAGGTCGTTCGCTCCAAGCTGGGCTGTGTGCACGAACCCCCCGTTCAGCCCGACCGCTGCGCCTTATCCGGTAACTATCGTCTTGAGTCCAACCCGGTAAGACACGACTTATCGCCACTGGCAGCAGCCACTGGTAACAGGATTAGCAGAGCGAGGTATGTAGGCGGTGCTACAGAGTTCTTGAAGTGGTGGCCTAACTACGGCTACACTAGAAGGACAGTATTTGGTATCTGCGCTCTGCTGAAGCCAGTTACCTTCGGAAAAAGAGTTGGTAGCTCTTGATCCGGCAAACAAACCACCGCTGGTAGCGGTGGTTTTTTTGTTTGCAAGCAGCAGATTACGCGCAGAAAAAAAGGATCTCAAGAAGATCCTTTGATCTTTTCTACGGGGTCTGACGCTCAGTGGAACGAAAACTCACGTTAAGGGATTTTGGTCATGAGATTATCAAAAAGGATCTTCACCTAGATCCTTTTAAATTAAAAATGAAGTTTTAAATCAATCTAAAGTATATATGAGTAAACTTGGTCTGACAGTTACCAATGCTTAATCAGTGAG

95

GCACCTATCTCAGCGATCTGTCTATTTCGTTCATCCATAGTTGCCTGACTCCCCGTCGTGTAGATAACTACGATACGGGAGGGCTTACCATCTGGCCCCAGTGCTGCAATGATACCGCGAGACCCACGCTCACCGGCTCCAGATTTATCAGCAATAAACCAGCCAGCCGGAAGGGCCGAGCGCAGAAGTGGTCCTGCAACTTTATCCGCCTCCATCCAGTCTATTAATTGTTGCCGGGAAGCTAGAGTAAGTAGTTCGCCAGTTAATAGTTTGCGCAACGTTGTTGCCATTGCTACAGGCATCGTGGTGTCACGCTCGTCGTTTGGTATGGCTTCATTCAGCTCCGGTTCCCAACGATCAAGGCGAGTTACATGATCCCCCATGTTGTGCAAAAAAGCGGTTAGCTCCTTCGGTCCTCCGATCGTTGTCAGAAGTAAGTTGGCCGCAGTGTTATCACTCATGGTTATGGCAGCACTGCATAATTCTCTTACTGTCATGCCATCCGTAAGATGCTTTTCTGTGACTGGTGAGTACTCAACCAAGTCATTCTGAGAATAGTGTATGCGGCGACCGAGTTGCTCTTGCCCGGCGTCAATACGGGATAATACCGCGCCACATAGCAGAACTTTAAAAGTGCTCATCATTGGAAAACGTTCTTCGGGGCGAAAACTCTCAAGGATCTTACCGCTGTTGAGATCCAGTTCGATGTAACCCACTCGTGCACCCAACTGATCTTCAGCATCTTTTACTTTCACCAGCGTTTCTGGGTGAGCAAAAACAGGAAGGCAAAATGCCGCAAAAAAGGGAATAAGGGCGACACGGAAATGTTGAATACTCATACTCTTCCTTTTTCAATATTATTGAAGCATTTATCAGGGTTATTGTCTCATGAGCGGATACATATTTGAATGTATTTAGAAAAATAAACAAATAGGGGTTCCGCGCACATTTCCCCGAAAAGTGCCACCTG

A.2 Primers The Following are the primers used for site-directed mutagenesis of Shaker. The

numbers in parenthesis correspond to the high expression vector we used from the

Bezanilla lab. Underlined letters in the sequences are the altered bases. All DNAs were

bought from Integrated DNA Technologies, Inc. Sequencing primers were purified with

standard desalting and mutagenic primers were had HPLC purification.

Cysteine mutations: Primers for Y323C, TAC (1602-1604) to TGC: Forward Primer Name: for_Y323C

5’ CGA CAT AAT CGC CAT CAT TCC GTG CTT TAT AAC ACT AGC G 3’

Reverse Primer Name: rev_Y323C

5’ CGC TAG TGT TAT AAA GCA CGG AAT GAT GGC GAT TAT GTC G 3’ Primers for T326C, ACA (1611-1613) to TGC: Forward Primer Name: for_T326C

5’ CCA TCA TTC CGT ACT TTA TAT GCC TAG CGA CTG TCG TTG CCG 3’

Reverse Primer Name: rev_T326C

5’ CGG CAA CGA CAG TCG CTA GGC ATA TAA AGT ACG GAA TGA TGG 3’ Primers for T329C, ACT (1620-1622) to TGC:

96

Forward Primer Name: for_T329C

5’ CGT ACT TTA TAA CAC TAG CGT GCG TCG TTG CCG AAG AGG AGG 3’

Reverse Primer Name: rev_T329C

5’ CCT CCT CTT CGG CAA CGA CGC ACG CTA GTG TTA TAA AGT ACG 3’ Primers for V330C, GTC (1623-1625) to TGC: Forward Primer Name: for_V330C

5’ CGT ACT TTA TAA CAC TAG CGA CTT GCG TTG CCG AAG AGG AGG 3’

Reverse Primer Name: rev_V330C

5’ CCT CCT CTT CGG CAA CGC AAG TCG CTA GTG TTA TAA AGT ACG 3’ Primers for V331C, GTT (1626-1628) to TGT: Forward Primer Name: for_V331C

5’ CGT ACT TTA TAA CAC TAG CGA CTG TCT GTG CCG AAG AGG AGG 3’

Reverse Primer Name: rev_V331C

5’ CCT CCT CTT CGG CAC AGA CAG TCG CTA GTG TTA TAA AGT ACG 3’ Primers for E333C, GAA (1632-1634) to TGC: Forward Primer Name: for_E333C

5’ GCG ACT GTC GTT GCC TGC GAG GAG GAT ACG TTA AAT CTT CC 3’

Reverse Primer Name: rev_E333C 5’ GGA AGA TTT AAC GTA TCC TCC TCG CAG GCA ACG ACA GTC GC 3’ Primers for E335C, GAG (1638-1640) to TGC: Forward Primer Name: for_E335C

5’ CGA CTG TCG TTG CCG AAG AGT GCG ATA CGT TAA ATC TTC C 3’

Reverse Primer Name: rev_E335C

5’ GGA AGA TTT AAC GTA TCG CAC TCT TCG GCA ACG ACA GTC G 3’ Primers for S346C, AGT (1671-1673) to TGT: Forward Primer Name: for_S346C

5’ GCG CCA GTC TGT CCA CAG GAC AAG TCA TC 3’ Reverse Primer Name: rev_S346C

5’ GAT GAC TTG TCC TGT GGA CAG ACT GGC GC 3’ Primers for S351C, TCA (1686-1688) to TGC:

97

Forward Primer Name: For_S351C

5’ CCA CAG GAC AAG TGC TCG AAT CAG GCT ATG 3’ Reverse Primer Name: Rev_S351C

5’ CAT AGC CTG ATT CGA GCA CTT GTC CTG TGG 3’ Primers for S352C, TCG (1689-1691) to TGC: Forward Primer Name: For_S352C

5’ CAG GAC AAG TCA TGC AAT CAG GCT ATG TC 3’ Reverse Primer Name: Rev_S352C

5’ GAC ATA GCC TGA TTG CAT GAC TTG TCC TG 3’ Primers for N353C, AAT (1692-1694) to TGT: Forward Primer Name: For_N353C

5’ GAC AAG TCA TCG TGT CAG GCT ATG TCC TTG 3’ Reverse Primer Name: Rev_N353C 5’ CAA GGA CAT AGC CTG ACA CGA TGA CTT GTC 3’ Primers for M356C, ATG (1701-1703) to TGC: Forward Primer Name: For_M356C

5’ GGA CAA GTC ATC GAA TCA GGC TTG CTC CTT GGC AAT ATT ACG 3’ Reverse Primer Name: Rev_M356C

5’ CGT AAT ATT GCC AAG GAG CAA GCC TGA TTC GAT GAC TTG TCC 3’ Primers for A359C, GCA (1710-1712) to TGC: Forward Primer Name: For_A359C

5’ CGA ATC AGG CTA TGT CCT TGT GCA TAT TAC GAG TGA TAC GAT TAG 3’ Reverse Primer Name: Rev_A359C 5’ CTA ATC GTA TCA CTC GTA ATA TGC ACA AGG ACA TAG CCT GAT TCG 3’ Primers for L361C, TTA (1716-1718) to TGC: Forward Primer Name: For_L361C

5’ GCT ATG TCC TTG GCA ATA TGC CGA GTG ATA CGA TTA GTT CG 3’ Reverse Primer Name: Rev_L361C

5’ CGA ACT AAT CGT ATC ACT CGG CAT ATT GCC AAG GAC ATA GC 3’ Primers for V363C, GTG (1722-1724) to TGT: Forward Primer Name: For_V363C

5’ CTT GGC AAT ATT ACG ATG TAT ACG ATT AGT TCG 3’ Reverse Primer Name: Rev_V363C

5’ CGA ACT AAT CGT ATA CAT CGT AAT ATT GCC AAG 3’ Primers for I364C, ATA (1725-1727) to TGC:

98

Forward Primer Name: For_I364C

5’ CCT TGG CAA TAT TAC GAG TGT GCC GAT TAG TTC GAG TAT TTC G 3’ Reverse Primer Name: Rev_I364C

5’ CGA AAT ACT CGA ACT AAT CGG CAC ACT CGT AAT ATT GCC AAG G 3’

Primers for R365C, CGA (1728-1730) to TGC: Forward Primer Name: For_R365C

5’ CCT TGG CAA TAT TAC GAG TGA TAT GCT TAG TTC GAG TAT TTC G 3’ Reverse Primer Name: Rev_R365C

5’ CGA AAT ACT CGA ACT AAG CAT ATC ACT CGT AAT ATT GCC AAG G 3’

Background mutations:

Primers for W434F, TGG (1935-1937) to TTC (Makes Shaker non-conducting [79]): Forward Primer Name: FOR_W434F

5’ CAT ACC CGA TGC ATT TTT CTG GGC GGT GGT TAC 3’ Reverse Primer Name: REV_W434F

5’ GTA ACC ACC GCC CAG AAA AAT GCA TCG GGT ATG 3’

Primers for V478W, GTC (2067-2069) to TGG (Makes Shaker non-conducting [42]): Forward Primer Name: FOR_V478W

5’ GCC GGT GCC GGT CAT CTG GAG CAA TTT CAA CTA CTT CTA TC 3’ Reverse Primer Name: REV_V478W

5’ GAT AGA AGT AGT TGA AAT TGC TCC AGA TGA CCG GCA CCG GC 3’

Primers for F425G and K437D, TTC (1908-1910) to GGC and AAG (1914-1916) to GAC (Enhances Charybdotoxin binding 100,000-fold over wild-type Shaker [140]): Forward Primer Name: 425G427D_for

5’ GCT GGA AGC GAA AAT TCC GGC TTC GAC TCC ATA CCC GAT GC 3’ Reverse Primer Name: 425G427D_rev

5’ GCA TCG GGT ATG GAG TCG AAG CCG GAA TTT TCG CTT CCA GC 3’

Sequencing primers:

Sh_1069r: 5’ GCA TCG AAG CTC GGT CGA C 3’

Sh_1698f: 5’ GCT ATG TCC TTG GCA ATA TTA CG 3’

Sh_2209f: 5’ CCT CAT CGG ATA TGA TGG ATT TGG 3’

99

Sh_1791r: 5’ GCA TTG AGG CTT TCA GAG TTC GTC C 3’

ILT_677: 5’ GCA ACA CTT TGA ACC CAT TCC TCA CGA TCA TG 3’

ILT 1082: 5’ GTA TCC AGA AAG TTC GCA AGC 3’

ILTFOR_1252: 5’ GAA GAC GAG GTG CCT GAC AT 3’

ILT_1252: 5’ ATG TCA GGC ACC TCG TCT TC 3’

ILT_1619r: 5’ ACC GAG CTC CCG CAT TGA GGC TTT CAG 3’

ILT_1647: 5’ GTC GTA CTC TTC TCA TCG G 3’

ILTFOR_2191: 5’ CGC AAC TGT ACC AAC AGC AGC 3’

ILT_2191: 5’ GCT GCT GTT GGT ACA GTT GCG 3’

A.3 Mutagenesis Site-directed mutagenesis was performed using QuikChange (Stratagene). The

entire Shaker gene sequence was checked at the UIUC Keck center sequencing facility to

verify that random errors had not occurred.

Plasmids were propagated with XL1-Blue E.coli (Stratagene). 4 mL cultures of

transformed XL1-Blues were grown up for 18 hours in terrific broth and DNA was

purified using a standard miniprep kit (Qiagen).

All restriction enzymes were from New England Biolabs.

A.4 mRNA synthesis I used the Ambion “mMESSAGE mMACHINE High Yield Capped RNA

Transcription Kit.” Calalog # 1344. Now a “T7-ultra kit” has been developed and this

should be used to increase usable mRNA. See Ambion manual and also there website:

www.ambion.com.

100

Template preparation: For mutants based on the Bezanilla high-expression Shaker construct, ZH4IR/W434F in

pBSTA(PBGLT) vector, I linearize with NotI from New England Biolabs. Here is the

digestion assembly (50µL):

20 µL plasmid DNA (5-8 µg)

5 µL NEB buffer 3

1.5 µL NotI (30 units)

23 µL sterile H2O

0.5 µL 100X BSA (provided with enzyme)

Incubate for 2.5 hours at 37°C. It is important to digest to completion, however you need

to avoid significant star activity.

Precipitate the linearized DNA, Add:

2.5 µL 0.5 M EDTA (1/20 volume)

5 µL 3 M sodium acetate (1/10 volume)

100 µL 100% ethanol (2 volumes)

Chill at -20°C for at least 10 min. and centrifuge at maximum speed for 15 min. Remove

the supernatant, re-spin for a few seconds, and then remove residual fluid with a very

fine-tipped pipet. Resuspend in RNase-free H2O at a concentration of 0.5-1 µg/µL (I

used 10 µL). Measure the concentration with absorbance at 260nm (I made 80X

dilutions and remember that for DNA 1 o.d. is 50 ng/µL). From this concentration

calculate how much volume to add to the transcription reaction.

mRNA synthesis follows the Ambion kit procedures. The resulting mRNA was purified

using Lithium Chloride precipitation. I typically obtain about 30 µg or mRNA. 40 µg is

possible. Dilute and aliquot the mRNA as required and store aliquots at -80 °C. Avoid

large numbers of freezing and thawing.

101

A.5 Toxin biochemistry Expression, fluorescent labeling, and purification of scorpion toxins were

performed as previously described in the laboratory of Christopher Miller (Brandeis)

[144]. Recombinant wild-type charybdotoxin and Agitoxin-2 were obtained from

Alomone labs.

102

Appendix B

Animal Use Protocol

Below are substantial excerpts from the IACUC approved animal use protocol, #04123,

outlining procedures used with Xenopus. The protocol was approved on 5-19-04.

Project Overview. Using lay person terms, provide a description of this project and its intended benefits. If this is a teaching activity, describe the specific educational goals that will be met through the proposed use of animals.

We study shape changes in ion channels. These channels are proteins in the cell membrane which help the cell communicate to the outside world and also control the flow of ions (sodium, potassium) in the cell interior. We are studying two classes of ion channels, one which is opened or closed by the voltage across the cell membrane, and the other by the presence of a small molecule. These channels are involved in nerve transmission and are also major drug targets. For example, the benzodiazapines (e.g. valium), are small molecules that bind to one of the channels we study (the GABA receptor), and affect its operation, thereby giving it pharmaceological affects. In order to study these channels, we produce (express) them in the oocyte and study them in this environment. Such types of studies are standard in the literature.

Rationale for Involving Animal Subjects. Provide the rationale for using animals and for the appropriateness of the species to be used. The rationale should include a description of non-animal alternatives that have been considered and explain why non-animal models are not adequate substitutes for the proposed use of animals. If the proposed activity is a field study of wild animals in a natural setting please enter “N/A”.

Ion channel proteins are membrane proteins that must be expressed in a cell membrane in order to be functional. The only possible ways to study membrane proteins in such a membrane are to utilize live stem cells (oocytes), cultured mammalian cells, or an artificial bilayer membrane. As described below, the Shaker potassium ion channel has never been successfully expressed in mammalian cells, and although some work has been done with artificial bilayer membranes, these cannot be studied with electrophysiological techniques such as those necessary for our experiment, and the rate of diffusion of the channel in such bilayers would be too great to allow studies of rotational behavior of the type we wish to probe. Further, our collaborators, led by Dr. Francisco Benzanilla at UCLA, have been using Xenopus oocytes for almost 20 years to study the properties of ion channels, and in particular have made great progress in understanding the properties of the Shaker potassium ion

103

channel. Their past successes in this field have demonstrated that the Xenopus oocyte is an ideal system for studying potassium channels.

Rationale for Numbers of Animals. Describe how the numbers of animals requested in item 1 were determined to be appropriate. Whenever possible, the number of animals requested should be justified statistically . The IACUC recognizes that it is not always possible to predict at the initiation of field studies the number of animals to be encountered or even the species to be encountered.

An ovarian biopsy is performed in the morning and single oocytes are isolated and prepared for mRNA injection that evening or the following day. Our experiments require 3-4 days for the oocyte to produce the quantities of ion channel proteins necessary for our work. Oocytes are rarely healthy enough for our experiments past day 4 post-biopsy. Therefore, 1 ovarian biopsy gives us the ability to perform about 2 experiments (about 100 injected oocytes.) Initially 3 years ago we tested the protocol in which we kept Xenopus alive for a total of 6 surgeries with 3 month recovery periods. However we found that after the first surgery/recovery period the oocytes obtained from subsequent surgeries were much reduced in quality and were not adequate for our demanding application. Following our collaborators’ (Dr. Francisco Bezanilla, Dr. Miles Akabas) advice we began using a protocol where we operated once on a frog, kept it alive, then later that week (generally 2-3 days) operated again, this time terminating the frog. We have found this optimizes the number of usable oocytes obtained from each frog. Specifically, we can perform around 4 experiments from oocytes obtained from a single frog. By using 1 frog/week we can make steady scientific progress on our ion channel studies. Sometimes oocytes obtained from certain frogs are not healthy and robust enough to support the high protein expression our experiments require. In fact, certain times of the year and occasionally entire shipments from Nasco yield poor oocytes. During these periods we like to increase the rate of experiments by using 1.5-2 frogs/week. This is necessary to maintain steady progress when the cell quality drops.

We are working on experiments which clearly elucidate the conformational changes which open and close voltage-gated ion channels. Although it is not known exactly how many more experiments we will perform for our current studies before we publish, current progress suggests it is very likely that we will write at least 2 papers in the next six months. These papers will constitute David Posson's Ph.D. dissertation. As outlined in the previous paragraph, we make steady progress towards our project goals by using on average 1 frog/week. Our approach to studying ion channels is powerful and unique. In particular, we have recently developed a new system which is promising to open up an entirely new view for how these proteins work. For this reason, it is possible a new graduate student in our lab will begin working in this area and continue these new experiments. Therefore, we seek approval for using the number of Xenopus that we require for a continous supply of oocytes for the three year term of this protocol.

With these considerations in mind, I calculated the maximum number of animals that we would possibly use during the 3-year protocol duration using 1.5 frogs/week for

104

156 weeks for a total of 234 frogs. It is expected that we will mostly use 1 frog/week.

Description of Animal Use. Provide a complete description of the proposed use of the animals. Include descriptions of: non-standard housing systems; use of capture/restraint devices; methods of marking; volumes and frequency of collections of bodily fluids; names, dosages, and routes of administration of test compounds and other materials administered; animal manipulations; and similar details. Details of surgical procedures must be provided in section 10.

The frogs will be used for oocyte collection. Oocyte collection requires ovarian biopsy, described in section 10 below.

Surgery. (For additional guidance refer to: http://www.dar.uiuc.edu/Policies/iacuc_surgery.htm) If the project involves surgery, complete the following: a. Will animals be permitted to regain consciousness/recover following surgery?

Yes

No

b. Will animals be subjected to more that one major operative procedure following which they will regain consciousness or recover?

Yes

No If yes, provide the scientific rationale for this.

As mentioned above, animals are allowed to recover from ovarian biopsy #1. However, within 2-5 days a second biopsy procedure is performed and the animal is immediately euthenized. We have found this maximizes the quantity of eggs usable for our experiments while minimizing possible discomfort induced in the animal.

c. List the anesthetic(s), including dosage(s), frequency of dosing, and route(s) of administration that will be used, and describe how you will monitor the depth/quality of anesthesia to ensure it is adequate.

Tricaine methane sulfonate (MS222) is the anesthetic used. It is dissolved in deionized water at a dose of 1.5 g/L. The solution is then buffered using NaHCO3 to a neutral pH. A fresh working solution of MS222 is prepared before each surgery. The adult female to be anesthesized is directly placed into a small plastic bowl containing the MS222 solution. Depth of anesthesia can be monitored through loss of the righting response and loss of response to painful stimuli. Respiratory movement will slow and then cease as anesthesia deepens, but alternate routes of gas exchange in amphibians (skin and buccopharyngeal cavity) can suffice for short procedures.

105

Once the female is anesthetized, it is removed from the solution. Anesthesia will persist for 10-20 minutes, sufficient time for ovarian biopsy.

.d. Describe the preoperative care of the animal, e.g. withholding of feed and water, etc.

.e. Describe the methods employed to minimize microbial contamination of the surgical site. Include brief descriptions of the preparation of the animal, surgeon, and instruments.

The frog surgical area is a portion of the room that can be easily sanitized. The surface is kept clean, and free from overhanging objects and chemicals. The area is wiped down with a 10% bleach solution before and after surgery. Instruments are sterilized (autoclave) before surgery. A sterile wrap is provided as an aseptic surface on which to place instruments during the procedure. Great care is also given to disturb as little as possible the powerful antibacterial mucus layer over the skin, which will prevent very efficiently any bacterial infection. The surgic al area on the frog’s skin will be sterilized using 10% povidone-iodine solution before incision. This procedure has been recommended in the literature (Elsner et al. Comp Med 2000 Apr;50(2):206-11) for preventing contamination of the ovarian biopsy material to microorganisms which have been shown to exist on the frog’s skin. Elsner et al. show convincingly that such contamination can greatly degrade oocyte quality and therefore undermine experiments. Povidone-iodine is well tolerated by the frogs and will further protect the animal from potential internal infections resulting from the surgical procedure.

f. Describe the surgical procedure. Include descriptions of methods and materials for ligatures and wound closure.

The female frog is placed belly up on a paper towel covered ice bucket. The paper towel is kept moist to prevent skin damage. Using forceps, the skin is grabbed on one side of the body (approximately 3/8 in. above the junction between the legs and the body), and an incision of 1/2 inches is made with a pair of fine scissors. The same procedure is then repeated on the muscle layer immediately under the skin. Using fine forceps, the ovary is located by gently probing around. Usually, the ovary is easily identified as a "bag of oocytes" located immediately under the muscle incision. However, sometimes a very careful and gentle probing is necessary to visualize the ovary. In that case, great care is given not to damage any organ. Part of the ovary is pulled out through the incision and a small fragment is separated using very fine scissors. The remaining part of the ovary is left alone, and will heal and regenerate on its own, naturally. The isolated fragment is placed directly into an appropriate saline buffer. The muscle closure is done with 3-4 stitches, then the skin closure is done similarly but with nonabsorbable sutures. g. Describe the post-surgical care. Include information regarding the use of pain-

relieving drugs, monitoring of animals for normal recovery from anesthesia and wound healing, and provision of supportive care, such as supplemental heat and fluid or antibiotic therapy.

106

The frog is rinsed in fresh dechlorinated water for recovery, and placed alone in a small tank. The water is kept shallow during recovery from anethesia, as frogs can drown before complete consciousness can be restored. After the frog regains consciousness, more dechlorinated water is added to the tank. The tank has a continuous flow system for water exchange. No food is provided for the next few days while the frog awaits the second, terminal biopsy procedure 2-5 days later. Euthenasia follows the second procedure. **Amendment 3-3-03 from protocol #01168 We request a waiver of the first 3 15-minute post-surgical observations as outlined on the surgical/post-surgical documentation form. Our lab now has 3 years experience performing these procedures on Xenopus and our frogs rarely regain consciousness before 45 minutes. If an animal were to receive an accidental overdose of anaesthetic (which has never happened) we do not have any procedure to save the animal. Furthermore, we do not have procedures for taking meaningful scientific measurements relative to anaesthetic recovery, e.g. pulse or respiratory rate. The frogs recover from anesthesia in a flat-bottomed tupperware container that has a tight lid. The lid has multiple holes for air exchange. The tupperware has very shallow water for moisture. Generally, it takes about an hour for the frog to wakeup enough to begin moving around. However, the frogs are always very sluggish and inactive even after an hour of recovery. Mostly they don’t even appear to be awake unless you squirt some water on them, gently touch them, or begin to turn them over. We have never observed a frog even approach tipping the tupperware over even when fully active before anesthesia. Therefore it seems extremely unlikely that a very sluggish frog that is not at all active will compromise the stability of its container.

We will check on the frog no later than 60 minutes and verify that it is awake enough to return to a recovery tank full of water. If the frog does not exhibit the righting response when flipped in the water it may drown. It is possible that the frog will recover enough before the full hour, however the tupperware containing the frog contains enough moisture and air exchange for the frog to remain safely for 1 hour. We feel that the greatest danger to the frog is returning it to the deep water too early.

Reminder: Documentation of the surgical procedure and post-surgical care is required and is the responsibility of the principal investigator. Copies of the surgical/post-surgical records must be provided to the unit animal care supervisors and readily available to the Division of Animal Resources veterinary staff, the IACUC and federal regulatory officials.

11. Clinical Outcomes. Are any clinical signs and/or lesions expected in the animals as a result of the procedures (e.g., tumors, surgical wounds, weight loss, behavioral abnormalities, fever, illness, etc.) whether as an integral part of the planned procedures or as an incidental side effect?

Yes

No If yes, please describe.

107

Between first and second surgeries the frogs have an approximately 1/2" surgical wound. The wound is sutured and the animal is euthenized before the wound has a chance to heal. No other clinical outcomes are expected to result from the first ovarian biopsy. It is difficult to assess the discomfort level for these animals. However, generally frogs that remain at the bottom of the water tank are recovering normally. Activity such as swimming away from your hand or a net is a good sign that the animal is doing ok.

If transgenic or knockout animals will be used, describe expected clinical abnormalities.

12. Minimizing Pain and Distress. Describe the procedures or methods designed to assure that discomfort and pain to animals will be limited to that which is unavoidable in the conduct of this research, and that analgesic, anesthetic, and tranquiliz ing drugs will be used where indicated and appropriate to minimize discomfort and pain to animals. Include specific objective criteria or end-points that you will use to ensure that animals that would otherwise experience severe or chronic pain or distress that cannot be relieved will be euthanized.

During surgery, frogs are anesthesized to prevent any pain. Great care is made to adequately suture the surgical wound so that wound closure will not be comprimised during the 2-5 day post-surgical survival period. Frogs are observed daily after surgery to determine if they are in severe or chronic pain. Euthenasia is performed within a few days of surgical procedures as described.

13. Duplication of Activity. Does the animal use described in this protocol unnecessarily duplicate previous research or teaching?

Yes

No

14. Disposition of Animals/Carcasses at the End of the Project. a. euthanasia

Animals are first euthanized with an excess dose of MS222 (4g/l) for 1 hour. The MS222 is a sodium bicarbonate buffered solution made with deionized water. The MS222 solution is placed in a tupperware bowl and the frog is completely submersed for 1 hour. Following this, the animals are quickly frozen to -20°C (cryoeuthanasia) . Ref: JAVMA, Jan 15, 1993, Vol 202(2) 243-4

c. For animals that will be euthanized, the carcasses will be disposed of by:

incineration

108

15. Alternatives to Painful Procedures. Will animals used in this project be subjected to procedures that will cause more than slight or momentary discomfort, distress, deprivation or prolonged restraint? (Note: A painful procedure is defined as any procedure that would reasonably be expected to cause more than slight or momentary pain and/or distress in a human being to which that procedure is applied. This includes painful procedures, such as survival and terminal surgeries, that are performed under anesthesia.)

Yes

No

If yes, federal regulations require that you provide below a written narrative description of the methods and sources used to determine that alternatives to these procedures were not available. Alternatives may include methods that use non-animal systems or less sentient animal species to partially or fully replace animals (for example, the use of an in vitro or insect model to replace a mammalian model), methods that reduce the number of animals to the minimum required to obtain scientifically valid data, and methods that refine animal use by lessening or eliminating pain or distress and, thereby, enhancing animal well-being. The performance of a database search is an effective and efficient method for demonstrating compliance with the requirement to consider alternatives to painful/distressful procedures. However, in some circumstances (as in highly specialized fields of study), conferences, colloquia, subject expert consultants, or other sources may provide relevant and up-to-date information regarding alternatives in lieu of, or in addition to, a database search. In these cases, sufficient documentation, such as the consultant's name and qualifications and the date and content of the consult, should be provided to demonstrate the expert's knowledge of the availability of alternatives in the specific field of study. When a database search is the primary means of meeting this requirement, the narrative must, at a minimum, include: the names of the databases searched, the date of the search, the period covered by the search, and the key words and/or search strategy used. If alternatives to painful or distressful procedures exist, but were not chosen, explain the reasons for not using the alternatives. For more information regarding alternatives searches, please see the Institutional Animal Care and Use Committee Webpage at http://www.dar.uiuc.edu/iacuc.htm.

The use of Xenopus oocytes for membrane protein studies is extremely common in the literature. Searches for alternatives showed many examples of Xenopus oocytes used for this purpose, but did not find any alternatives. Specific searches that were performed were: "xenopus AND oocyte AND alternative AND surgery" and "xenopus AND anesthetic AND alternative". The search was performed on Medline, Biological Abstracts, and the Science Citation Index. Another search was performed with the keywords "shaker AND mammalian AND cell" to determine if cultured mammalian cells could be substituted for oocytes. Many studies used cultured mammalian cells to investigate "Shaker related proteins" (i.e. proteins that were similar in function to Shaker but were native to mammalian cells), but no studies

109

were found in which Shaker was expressed in mammalian cells. A third search was performed using the keywords "bilayer AND lipid AND membrane AND shaker" to determine if it was feasible to perform these studies on Shaker expressed in an artificial lipid bilayer. Studies in which this type of system were used did not express complete proteins, and none did any voltage clamping. Voltage clamping is necessary to study the motion of the protein during depolarization.

110

Appendix C

Xenopus Oocyte Preparation

1x Solutions (mM):

Ca2+ free OR2 ND96*

NaCl 82.5 96

KCl 2.5 2

CaCl2 0 1.8

MgCl2 1 1

HEPES 5 5

pH (with NaOH) 7 7.6

Osmolarity (mOsm/kg H2O) 160-180 180-200

*1x ND96 also contains 2.5 mM Na-pyruvate and is often supplemented with 50 µg/mL

gentamycin antibiotic and DTT as needed.

For ND96, 20x stock solutions were prepared at pH 7.85, sterile filtered, and diluted to

1x in sterile Baxter irrigation water. 20x stock of Ca2+ free OR2 had pH 7.5 and treated

similarly.

Dishes used: Falcon polystyrene tissue culture dishes.

353001 (35 x 10 mm)

353002 (60 x 15 mm)

1. Xenopus ovarian lobe samples are removed from frog and placed in Ca2+ free

OR2.

2. Transfer a selected sample of ovary to a fresh dish of Ca2+ free OR2. Tease apart

lobes with ethanol washed forceps such that eggs lay in dish as a ‘monolayer’.

111

There should be no clumps, but the oocytes should be loosely connected with

connective tissue.

3. Weigh out ~15 mg of Collagenase Type 1A (Sigma C-9891, ~260 units/mg) in a

50 mL corning tube. Dissolve in ~20 mL of Ca2+ free OR2.

4. Add teased oocytes to collagenase solution. First pour out as much of the dirty

teasing solution as possible and therefore add as little volume as possible to the

collagenase solution.

5. Tape the tube horizontally onto the cell incubator/shaker. Keep the lid open so

that the temperature does not rise above room temperature and shake at 240 rpm.

This produces vigorous shaking.

6. Shake for 45-70 minutes. You want many oocytes to be free and the follicular

layer should begin to pop off the cells. Determine the minimum time that works

and stick with it. Time required varies from frog to frog, enzyme batch to enzyme

batch, and extent of teasing performed. Prepare a large excess of eggs, as you can

then check the progress of defolliculation as it proceeds and in the end select only

the very best oocytes.

7. Rinse the oocytes free of collagenase by adding and pouring out 7 volumes of

Ca2+ free OR2.

8. Empty oocytes into a 60 x 15 mm dish filled with 5:1 mixture of Ca2+ free

OR2:ND96. (That is 4 parts Ca2+ free OR2 plus 1 part ND96. Use ND96 with

pyruvate and gentamycin.) Pipette selected oocytes into a fresh 60 x 15 mm dish

containing 5:1. Incubate in 18 degree incubator for ~3 hours.

9. Pipette selected oocytes into a 60 x 15 mm dish filled with 3:1 mixture of Ca2+

free OR2:ND96. (That is 2 parts Ca2+ free OR2 plus 1 part ND96. Use ND96

with pyruvate and gentamycin.) Incubate in 18 degree incubator for 30-60

minutes.

10. Pipette selected oocytes into 1x ND96 with pyruvate and gentamycin. Change the

solution everyday, washing the oocytes copiously. Wait at least 1 hour typically

before mRNA injection. Wash all injected oocytes 1 or 2 times a day, throwing

away any dead ones that have serious problems or milky looking places. Wash

oocytes with 4 to 5 dish volumes of ND96.

112

Xenopus Oocyte Labeling Preblocking

Oocytes were often preblocked with β-MPA, in order to block background cysteines.

After cells were injected with mRNA they are placed in a 12 degrees incubator so that

Shaker is expressed but not trafficked to the outer membrane. I often put the cells in 500

µM DTT at this stage, but I have come to believe that it is too high. Therefore, I

recommend having 100 µM or maybe less, like 25 µM DTT at this stage. Then on day 2

or 3, put the cells in 100 µM DTT (if they have only been in low DTT) for 30 minutes.

Next, wash away DTT completely. Place the oocytes into depolarizing solution + 500

µM β-MPA (β-maleimidopropionic acid, Sigma M-9154) for 1 hour. Next, wash away

MPA completely. Place oocytes in low DTT (0 or 25 µM) at 18 degrees for 24-30 hours.

Proceed with labeling.

Fluorophore labeling

Prior to labeling, put the cells in depolarizing solution plus 100 µM DTT for 30 minutes.

Next, wash away DTT completely. Place oocytes in 80 µM Tb-chelate-maleimide for 30

minutes. This is for toxin LRET experiments. OR, place oocytes in 80 µM Tb-chelate-

maleimide plus 20 µM fluorescein-maleimide for 30 minutes. This is for S4-donor to S4-

acceptor experiments.

These high concentrations of dye facilitate labeling of more buried cysteine sites on

Shaker. However, for sites that are very accessible, out on the linker for example, you

can use 0 DTT after preblocking, then put oocytes into depolarizing solution plus 50 µM

DTT for 30 minutes, wash, and label with only 4 µM Tb-chelate-maleimide for

30minutes. Wash. This type of protocol labels background cysteines much less. You

also get less labeling of Shaker, but better overall specificity (for sites that are quite

accessible). It may be possible to optimize these protocols, however this is what I settled

on after fiddling. Raw specificity (labeled Shaker compared with background labeled

sites) for many sites seems basicly impossible.

Always label control oocytes so that signal to background can be evaluated.

113

Appendix D

Electrophysiology Solutions

Bridges

Agar bridges for the voltage clamp are 3% agar dissolved in 1 M NaMES (NaOH

neutralized to pH 7 with MES) with 10 mM HEPES. The bridges are stored in 1 M

NaMES with 10 mM HEPES pH 7.

Electrodes

V1, the voltage measurement electrode, is filled with 2.7 M NaMES with 10 mM NaCl.

Vi, the current injection electrode, is filled with 2.3 M K-citrate with 10 mM NaCl.

Electrode resistances should be between 0.2 and 1 MΩ.

3 M KCl can be used in electrodes, however the electrode glass has to have a narrow

taper such that the electrode resistance is above 200 kΩ. K-citrate is better at passing

higher currents and won’t clog as easily as KCl electrodes.

Recording Solution

Experimental solution for the bath is (mM):

120 N-methylglucamine

2 CaMES

10 HEPES

pH 7.8 with MES (high for using with fluorescein)

114

Depolarizing Solution

High K+ solutions depolarize membranes (mM):

120 KOH

2 CaCl2

1 MgCl2

10 HEPES

pH 7-7.4 with MES (Need to adjust pH for salts to completely dissolve.)

Initially add 3 mL MES before using pH meter.

115

Appendix E Silicon Quenching Unbinding Bioassays My initial project involved the development of a new technology, a bioassay

based on high efficient fluorescence quenching via proximity to a silicon substrate

acceptor. The mechanism of energy transfer (quenching) involves in part a Förster-like

mechanism, although the acceptor is no longer a point dipole but a continuum half-space.

Here we present initial measurements obtained on model systems that demonstrate the

technique. We have produced silicon substrates with a “bio-friendly” adhesion surface

for the attachment of virtually any molecule of interest. We have successfully attached

short pieces of DNA for monitoring in real time enzymatic cleavage and release of DNA

fragments by the digestive enzymes, EcoRI and DdeI. Experiments are performed with a

simple optical setup consisting of a microscope and a PMT. Nanomolar sensitivity and a

reasonable dynamic range have been obtained and much room for experimental

optimization still remains. The assay requires only one molecule to be fluorescently

labeled and there is no general requirement on the size of the biological macromolecules.

This is important because there are many critical biological systems where molecular

complexes are huge, for example in activation and regulation of DNA transcription.

Experiments on silicon wafers are also ideal for miniaturization and automation, with

perhaps even integration of excitation, detection, and data collection equipment directly

on the wafer. The possibility also exists for the study of different semiconductors or

amorphous silicon as the quenching substrate. The simple setup used to take

measurements is shown in Fig. 3.4.

116

Microscope Objective

Hg Lamp filtered togive blue excitation.

PMT

Reaction chambercontaining Fluoresceinglowing green.

Si

Mirror reflects blueexcitation and passesgreen emission.

Figure 3.4 Simple optical setup for detecting signals from a silicon substrate-based fluorescent

bioassay. A microscope objective collects fluorescence from an experimental chamber and the

intensity is read by a photomultiplier tube (PMT). Excitation is achieved via a Hg lamp.

For a model system, to determine the distance scale (and hence size of

biomolecules) over which our technique will work, we followed Nakache et al. [157].

Silicon wafers were coated with Langmuir-Blodgett lipid monolayers which were doped

with the fluorescent dye NBD (Fig. 3.5a). The lipid monolayers consisted of 98% egg

PC (Phosphatidylcholine, Avanti Polar Lipids #830051) and 2% NBD-C6-HPC

(Molecular Probes, N-3786). The distances between the fluorescent layers and the silicon

crystal were controlled by thermally growing SiO2 into the wafer either 360 Å or 1000 Å

(Figure 2). Wafers with the native ~30 Å oxide layer were also tested resulting in three

fluorescent intensity measurements (Fig. 3.5b). The wafers we used were n-type (111)

and were doped with phosphorous for a resistivity of 0.2-1.0 Ω cm (Crysteco inc.).

Intensity measurements throughout this paper were all taken with an Olympus IX-70

inverted epifluorescence microscope fitted with a PMT from Electron Tubes inc. (Fig.

3.4). Our result illustrates the extremely long range of fluorescence quenching.

Assuming the 1000 Å sample represents zero quenching, the signal at 360 Å is quenched

by 92% and the fluorophores on the native 30 Å oxide are quenched by >99%. Thus,

117

fluorescently labeled macromolecular complexes hundreds of angstroms in size will be

quenched when bound to the silicon substrate.

Oxide Thickness (Å)

0

5 1 0 5

1 1 0 6

1 .5 1 0 6

2 1 0 6

2 .5 1 0 6

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0

c o u nt s /s ec ab o v e b a c k g ro u n d

92% Quenched

>99% Quenched

Assume 0%Quenching at1000 Å.

1000 Å SiO2

Does NOT Quench

Silicon

360 Å SiO2

SiliconSilicon

a

b

Figure 3.5 a. Lipid monolayers with 2% fluorescent lipid content were deposited onto three

silicon wafer samples with various oxide layer thicknesses. b. The fluorescence intensity was

measured and long distance quenching over hundreds of angstroms was observed.

We tested the silicon substrate unbinding assay using digestion of short DNA

molecules by restriction endonucleases. These enzymes bind non-specifically to the

DNA helix and diffuse along its length. When the endonuclease comes upon a specific

sequence typically 6 base pairs long, called the recognition sequence, it binds tightly and

catalyzes a cleavage break of both DNA strands. As a model bioassay, we have

successfully monitored in real time the enzymatic cleavage of 20mer DNA by the EcoRI

restriction enzyme. The 3’ end of the DNA is biotinylated for attachment to silicon

wafers which are coated with streptavidin. The 5’ end (free end) has a fluorescein label

118

and the central sequence is the EcoRI recognition sequence, 5’-G↓AATTC-3’. Addition

of EcoRI cleaves the DNA, allowing the fluorescent probe to diffuse away from the

surface and become highly fluorescent (Fig. 3.6a). For a complete picture of the DNA

immobilization procedure see Fig. 3.6b.

30 Å SiO2

SiliconEnergyTransfer

EcoRIRecognitionSite

BeforeCleavage After

Cleavage

a b

BiotinylatedDNA

BSA-biotin30 Å SiO2

Silicon

Streptavidin

Fluorescein

Figure 3.6 a. Schematic cartoon for DNA cleavage bioassay. After cleavage, a fluorescently

labeled DNA fragment diffuses far enough away from the substrate to fluoresce unquenched. b. The DNA immobilization is accomplished using standard BSA-biotin, avidin adsorption.

For the immobilization, BSA-biotin was non-covalently adsorbed onto the silicon

wafer for a “bio-friendly” adhesion layer (Fig. 3.6b). Streptavidin can bind up to four

biotins very tightly and serves as an efficient crosslinker between the biotinylated surface

and any biotinylated molecule, in this case DNA. Fluorescent intensity measurements

were taken in real time with several enzyme concentrations. As expected, the reaction

proceeds faster with greater amounts of enzyme (Fig. 3.7).

119

0

200

400

600

800

1000

0 40 80 120 160

T im e (m inutes)ADDENZYME

Figure 3.7 DNA digestion bioassays with EcoRI. More enzyme causes the digestion reaction to

proceed faster as expected.

We performed a control experiment to measure what percentage of immobilized

DNA is cut in the assay. Ideally, 100% of the substrate is accessible for enzymatic

cleavage. To test this, we performed the reaction on our non-quenching wafers which

have 1000 Å silicon dioxide and after washing measured the fluorescent intensity

remaining on the wafer. The ratio of this intensity to the initial intensity of the substrate

preparation gives a measure of how much DNA is remaining on the surface after the

reaction has finished. We found that 90% of our immobilized DNA was accessible to

EcoRI (Figure 3.8).

120

1000 Å SiO2Control Wafer

Does NOT Quench

Silicon Silicon

1000 Å SiO2Control Wafer

Does NOT Quench

2 Hour DigestAnd WashBefore Cleavage

Figure 3.8 Schematic diagram of control experiment. Intensity of the initial DNA preparation is

measured and compared with the intensity after 2 hour digestion reaction with EcoRI and

washing. Reactions with 53 units and 33 units of enzyme were tested. 90% of the immobilized

DNA is cleaved off the surface.

Digestion with restriction enzymes can in a few special instances be used to

understand something important about the sequence or mutations in the sequence.

Sickle-Cell Anemia is a heritable disease which results in polymerization of hemoglobin

and subsequent elongation of red blood cells into a shape resembling a sickle. The

polymerization of the hemoglobin protein is the result of a single point mutation in the β-

globin gene [158]. The most common mutation causing this genetic disease occurs in a

DNA sequence containing the 5 base pair recognition sequence of the endonuclease

DdeI. The wild type (non-mutant) β-globin DNA is cleaved by DdeI at the sequence 5’-

C↓TCAG-3’. In sickle-cell anemia, the thymine (T) has been replaced with an adenine

(A), 5’-CACAG-3’, and DdeI does not cut the DNA. We have used our restriction

enzyme bioassay to distinguish mutant β-globin DNA from wild type DNA. Fluorescein

labeled 30mer DNAs were immobilized on our silicon wafers. The DNA contained 15

base pairs of the β-globin gene, either wild type or sickle cell mutant (Fig. 3.9). Addition

of DdeI cleaved the wild type DNA and produced a large spectroscopic signal as the

fluorescein labeled DNA fragments diffused away from the quenching silicon. The

121

sickle-cell DNA was not cleaved and no detectable spectroscopic signal resulted from the

addition of the enzyme (Fig. 3.10). As an internal control, the 30mer DNA also

contained an EcoRI recognition sequence (Fig. 3.9) so that the DNA could be cleaved

independent of the β-globin gene segment. This was convenient for proving the

successful immobilization of a large quantity of DNA in the sickle-cell mutant

preparation. After DdeI shows no cleavage of DNA and thus no rise in signal, EcoRI was

added to cleave the DNA and thus prove that the mutation does in fact prevent cleavage

by DdeI.

EcoRI

3’-GCTCGAGCTTAAGCAGAG GAGTCCTCAAGT-5’5’-CGAGCTCGAATTCGTCT C CTCAGGAGTTCA-3’

DdeI

3’-GCTCGAGCTTAAGCAGAG GTGTAC

C CTCAAGT-5’5’-CGAGCTCGAATTCGTCT C C AGGAGTTCA-3’

EcoRI DdeIDdeIDoes NotCut!

Sickle Cell Mutant

Wild Type

Mutation

Figure 3.9 Sequence and restriction map for the oligos used in the detection bioassay for sickle

cell mutation in β-globin DNA.

122

0

100

200

300

400

500

600

700

0 20 40 60 80 100 120 140 160

Time (minutes)

33 Unitsof DdeI

33 Units of EcoRI

Wash here

103

Figure 3.10 Discrimination of sickle-cell mutant DNA from wild type DNA by restriction digest

assay with DdeI.

This initial project was never completed. In fact, it essentially worked, however

there were reproducibility problems that made it difficult to achieve any kind of

quantitative measurements of unbinding. Other experiments were done using peptides

cut by the tobacco etch virus protease, a highly sequence specific protease. Quite a bit of

effort was put into improving sample preparation. Polyethylene glycol was used to make

the surface resistant to non-specific binding of protein. Covalent forms of attachment

were investigated in order to make the immobilization cleaner. Progress was little at the

expense of great effort, so when the chance came to begin a project on ion channels,

that’s what I did. In fact, working in an area that has focused biological questions turned

123

out to be much more enjoyable for me. Therefore, I completed my graduate work with

ion channel studies.

124

References

1. Hille, B. Ion channels of excitable membranes (Sinauer, Sunderland, Mass.,

2001).

2. Hodgkin, A. L. & Huxley, A. F. Resting and action potentials in single nerve fibres. J. Physiol. 104, 176-95 (1945).

3. Aidley, D. J. The Physiology of Excitable Cells (Cambridge University Press, Cambridge, 1998).

4. Hodgkin, A. L. & Huxley, A. F. Currents carried by sodium and potassium ions through the membrane of the giant axon of Loligo. J. Physiol. 116, 449-72 (1952a).

5. Hodgkin, A. L. & Huxley, A. F. The components of membrane conductance in the giant axon of Loligo. J. Physiol. 116, 473-96 (1952b).

6. Hodgkin, A. L. & Huxley, A. F. The dual effect of membrane potential on sodium conductance in the giant axon of Loligo. J. Physiol. 116, 497-506 (1952c).

7. Hodgkin, A. L. & Huxley, A. F. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117, 500-44 (1952d).

8. Schneider, M. F. & Chandler, W. K. Voltage dependent charge movement of skeletal muscle: a possible step in excitation-contraction coupling. Nature 242, 244-6 (1973).

9. Armstrong, C. M. & Bezanilla, F. Currents related to movement of the gating particles of the sodium channels. Nature 242, 459-461 (1973).

10. Keynes, R. D. & Rojas, E. Kinetic and steady-ste properties of the charged system controlling sodium conductance in the squid giant axon. J. Physiol. (Camb.) 239, 393-434 (1974).

125

11. Tempel, B. L., Papazian, D. M., Schwarz, T. L., Jan, Y. L. & Jan, L. Y. Sequence of a probable potassium channel component encoded at Shaker locus of Drosophila. Science 237, 770-5 (1987).

12. MacKinnon, R. Determination of the subunit stiochiometry of a voltage-activated potassium channel. Nature 350, 232-5 (1991).

13. Armstrong, C. M. Interaction of tetraethylammonium ion derivatives with the potassium channels of giant axons. J. Gen. Physiol. 58, 413-437 (1971).

14. Hille, B. Ionic basis of resting and action potentials. Handbook of Physiology. The Nervous System I (eds. Brookhart, J. M., Mountcastle, V. B., Kandel, E. R. & Geiger, S. R.) (American Physiological Society, Washington, D.C., 1977).

15. Schoppa, N. E., McCormack, K., Tanouye, M. A. & Sigworth, F. J. The Size of Gating Charge in Wild-Type and Mutant Shaker Potassium Channels. Science 255, 1712-1715 (1992).

16. Seoh, S.-A., Papazian, D. M., Sigg, D. & Bezanilla, F. Voltage-sensing residues in the S2 and S4 segments of the Shaker K+ channel. Neuron 16, 1159-1167 (1996).

17. Aggarwal, S. K. & MacKinnon, R. Contribution of the S4 segment to gating charge in the Shaker K+ channel. Neuron 16, 1169-1177 (1996).

18. Gandhi, C. S. & Isacoff, E. Y. Molecular models of voltage sensing. J Gen Physiol 120, 455-63 (2002).

19. Larsson, H. P., Baker, O. S., Dhillon, D. S. & Isacoff, E. Y. Transmembrane movement of the shaker K+ channel S4. Neuron 16, 387-97 (1996).

20. Yusaf, S. P., Wray, D. & Sivaprasadarao, A. Measurement of the movement of the S4 segment during the activation of a voltage-gated potassium channel. Pflugers Arch 433, 91-7 (1996).

21. Baker, O. S., Larsson, H. P., Mannuzzu, L. M. & Isacoff, E. Y. Three transmembrane conformations and sequence-dependent displacement of the S4 domain in shaker K+ channel gating. Neuron 20, 1283-94 (1998).

126

22. Wang, M. H., Yusaf, S. P., Elliott, D. J., Wray, D. & Sivaprasadarao, A. Effect of cysteine substitutions on the topology of the S4 segment of the Shaker potassium channel: implications for molecular models of gating. J Physiol 521 Pt 2, 315-26 (1999).

23. Sigworth, F. J. Voltage gating of ion channels. Q Rev Biophys 27, 1-40 (1994).

24. Durell, S. R. & Guy, H. R. Atomic scale structure and functional models of voltage-gated potassium channels. Biophys. J. 62, 238-247 (1992).

25. Cha, A., Snyder, G. E., Selvin, P. R. & Bezanilla, F. Atomic scale movement of the voltage sensing region in a potassium channel measured via spectroscopy. Nature 402, 809-813 (1999).

26. Glauner, K. S., Mannuzzu, L. M., Gandhi, C. S. & Isacoff, E. Y. Spectroscopic mapping of voltage sensor movement in the Shaker potassium channel. Nature 402, 813-7 (1999).

27. Horn, R. How S4 segments move charge. Let me count the ways. J Gen Physiol 123, 1-4 (2004).

28. Vemana, S., Pandey, S. & Larsson, H. P. S4 movement in a mammalian HCN channel. J Gen Physiol 123, 21-32 (2004).

29. Nguyen, T. P. & Horn, R. Movement and crevices around a sodium channel S3 segment. J Gen Physiol 120, 419-36 (2002).

30. Jiang, Y. et al. X-ray structure of a voltage-dependent K+ channel. Nature 423, 33-41 (2003).

31. Jiang, Y., Ruta, V., Chen, J., Lee, A. & MacKinnon, R. The principle of gating charge movement in a voltage-dependent K+ channel. Nature 423, 42-8 (2003).

32. Doyle, D. A. et al. The Structure of the Potassium Channel: Molecular Basis of K+ Conduction and Selectivity. Science 280, 69-77 (1998).

33. Schrempf, H. et al. A prokaryotic potassium ion channel with two predicted transmembrane segments from Streptomyces lividans. Embo J 14, 5170-8 (1995).

127

34. Heginbotham, L., Abramson, T. & MacKinnon, R. A functional connection between the pores of distantly related ion channels as revealed by mutant K+ channels. Science 258, 1152-5 (1992).

35. Heginbotham, L., Lu, Z., Abramson, T. & MacKinnon, R. Mutations in the K+ channel signature sequence. Biophys J 66, 1061-7 (1994).

36. Zhou, Y., Morais-Cabral, J. H., Kaufman, A. & MacKinnon, R. Chemistry of ion coordination and hydration revealed by a K+ channel-Fab complex at 2.0 A resolution. Nature 414, 43-48 (2001).

37. Morais-Cabral, J. H., Zhou, Y. & MacKinnon, R. Energetic optimization of ion conduction rate by the K+ selectivity filter. Nature 414, 37-42 (2001).

38. Zhou, Y. & MacKinnon, R. The occupancy of ions in the K+ selectivity filter: charge balance and coupling of ion binding to a protein conformational change underlie high conduction rates. J Mol Biol 333, 965-75 (2003).

39. Jiang, Y. et al. The open pore conformation of potassium channels. Nature 417, 523-6. (2002).

40. Jiang, Y. et al. Crystal structure and mechanism of a calcium-gated potassium channel. Nature 417, 515-22 (2002).

41. Webster, S. M., Del Camino, D., Dekker, J. P. & Yellen, G. Intracellular gate opening in Shaker K+ channels defined by high-affinity metal bridges. Nature 428, 864-8 (2004).

42. Kitaguchi, T., Sukhareva, M. & Swartz, K. J. Stabilizing the closed S6 gate in the Shaker Kv channel through modification of a hydrophobic seal. J Gen Physiol 124, 319-32 (2004).

43. Chang, G., Spencer, R. H., Lee, A. T., Barclay, M. T. & Rees, D. C. Structure of the MscL homolog from Mycobacterium tuberculosis: a gated mechanosensitive ion channel. Science 282, 2220-6 (1998).

44. Martinac, B., Buechner, M., Delcour, A. H., Adler, J. & Kung, C. Pressure-sensitive ion channel in Escherichia coli. Proc Natl Acad Sci U S A 84, 2297-301 (1987).

128

45. Bass, R. B., Strop, P., Barclay, M. & Rees, D. C. Crystal structure of Escherichia coli MscS, a voltage-modulated and mechanosensitive channel. Science 298, 1582-7 (2002).

46. Sotomayor, M. & Schulten, K. Molecular dynamics study of gating in the mechanosensitive channel of small conductance MscS. Biophys J 87, 3050-65 (2004).

47. Vasquez, V., Cortes, D. M. & Perozo, E. Structural Dynamics of MscS Transmembrane Segments in the Closed State. A Site-directed Spin-labeling Analysis. Biophysical Society Annual Meeting Poster 1421 (2005).

48. Bass, R. B. et al. The structures of BtuCD and MscS and their implications for transporter and channel function. FEBS Lett 555, 111-5 (2003).

49. Bezanilla, F. The voltage sensor in voltage-dependent ion channels. Physiol Rev 80, 555-92 (2000).

50. Cuello, L. G., Cortes, D. M. & Perozo, E. Molecular architecture of the KvAP voltage-dependent K+ channel in a lipid bilayer. Science 306, 491-5 (2004).

51. Ruta, V., Jiang, Y., Lee, A., Chen, J. & MacKinnon, R. Functional analysis of an archaebacterial voltage-dependent K+ channel. Nature 422, 180-5 (2003).

52. Gonzalez, C., Rosenman, E., Bezanilla, F., Alvarez, O. & Latorre, R. Periodic perturbations in Shaker K+ channel gating kinetics by deletions in the S3-S4 linker. Proc Natl Acad Sci U S A 98, 9617-23 (2001).

53. Cohen, B. E., Grabe, M. & Jan, L. Y. Answers and questions from the KvAP structures. Neuron 39, 395-400 (2003).

54. Laine, M., Papazian, D. M. & Roux, B. Critical assessment of a proposed model of Shaker. FEBS Lett 564, 257-63 (2004).

55. Ahern, C. A. & Horn, R. Stirring up controversy with a voltage sensor paddle. Trends Neurosci 27, 303-7 (2004).

129

56. Shen, N. V., Chen, X., Boyer, M. M. & Pfaffinger, P. J. Deletion analysis of K+ channel assembly. Neuron 11, 67-76 (1993).

57. Gandhi, C. S., Clark, E., Loots, E., Pralle, A. & Isacoff, E. Y. The orientation and molecular movement of a k(+) channel voltage-sensing domain. Neuron 40, 515-25 (2003).

58. Broomand, A., Mannikko, R., Larsson, H. P. & Elinder, F. Molecular movement of the voltage sensor in a K channel. J Gen Physiol 122, 741-8 (2003).

59. Laine, M. et al. Atomic proximity between S4 segment and pore domain in Shaker potassium channels. Neuron 39, 467-81 (2003).

60. Jiang, Q. X., Wang, D. N. & MacKinnon, R. Electron microscopic analysis of KvAP voltage-dependent K+ channels in an open conformation. Nature 430, 806-10 (2004).

61. Dutzler, R., Campbell, E. B., Cadene, M., Chait, B. T. & MacKinnon, R. X-ray structure of a ClC chloride channel at 3.0 A reveals the molecular basis of anion selectivity. Nature 415, 287-94. (2002).

62. Miller, C. Ion channels go to Stockholm--this time as proteins. Neuron 40, 1049-51 (2003).

63. Gurdon, J. B., Lane, C. D., Woodland, H. R. & Marbaix, G. Use of frog eggs and oocytes for the study of messenger RNA and its translation in living cells. Nature 233, 177-82 (1971).

64. Noda, M. et al. Expression of functional sodium channels from cloned cDNA. Nature 322, 826-8 (1986).

65. Hodgkin, A. L. & Huxley, A. F. A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 117, 500-44 (1952).

66. Weber, W. M. Endogenous ion channels in oocytes of xenopus laevis: recent developments. J Membr Biol 170, 1-12 (1999).

130

67. Kuruma, A., Hirayama, Y. & Hartzell, H. C. A hyperpolarization- and acid-activated nonselective cation current in Xenopus oocytes. Am J Physiol Cell Physiol 279, C1401-13 (2000).

68. Neher, E. & Sakmann, B. Single-channel currents recorded from membrane of denervated frog muscle fibres. Nature 260, 799-802 (1976).

69. Zhou, M., Morais-Cabral, J. H., Mann, S. & MacKinnon, R. Potassium channel receptor site for the inactivation gate and quaternary amine inhibitors. Nature 411, 657-61 (2001).

70. Armstrong, C. M., Bezanilla, F. & Rojas, E. Destruction of sodium conductance inactivation in squid axons perfused with pronase. J Gen Physiol 62, 375-91 (1973).

71. Bezanilla, F. & Armstrong, C. M. Inactivation of the sodium channel. I. Sodium current experiments. J Gen Physiol 70, 549-66 (1977).

72. Armstrong, C. M. & Bezanilla, F. Inactivation of the sodium channel. II. Gating current experiments. J Gen Physiol 70, 567-90 (1977).

73. Hoshi, T., Zagotta, W. N. & Aldrich, R. W. Biophysical and molecular mechanisms of Shaker potassium channel inactivation. Science 250, 533-8 (1990).

74. Zagotta, W. N., Hoshi, T. & Aldrich, R. W. Restoration of inactivation in mutants of Shaker potassium channels by a peptide derived from ShB. Science 250, 568-71 (1990).

75. Hoshi, T., Zagotta, W. N. & Aldrich, R. W. Two types of inactivation in Shaker K+ channels: effects of alterations in the carboxy-terminal region. Neuron 7, 547-56 (1991).

76. Yellen, G. The moving parts of voltage-gated ion channels. Q Rev Biophys 31, 239-95 (1998).

77. Encinar, J. A. et al. Probing the channel-bound shaker B inactivating peptide by stereoisomeric substitution at a strategic tyrosine residue. Biochemistry 42, 8879-84 (2003).

131

78. Bezanilla, F. & Stefani, E. Gating currents. Methods Enzymol 293, 331-52 (1998).

79. Perozo, E., MacKinnon, R., Bezanilla, F. & Stefani, E. Gating Currents from a Nonconducting Mutant Reveal Open-closed Conformations In Shaker K+ Channels. Neuron 11, 353-358 (1993).

80. Yang, Y., Yan, Y. & Sigworth, F. J. How does the W434F mutation block current in Shaker potassium channels? J Gen Physiol 109, 779-89 (1997).

81. Bezanilla, F., Perozo, E. & Stefani, E. Gating of Shaker K+ channels: II. The components of gating currents and a model of channel activation. Biophys. J. 66, 1011-1021 (1994).

82. Zagotta, W. N., Hoshi, T. & Aldrich, R. W. Shaker potassium channel gating. III: Evaluation of kinetic models for activation. J Gen Physiol 103, 321-62 (1994).

83. Schoppa, N. E. & Sigworth, F. J. Activation of Shaker potassium channels. III. An activation gating model for wild-type and V2 mutant channels. J Gen Physiol 111, 313-42 (1998).

84. Stefani, E., Toro, L., Perozo, E. & Bezanilla, F. Gating of Shaker K+ channels: I. Ionic and gating currents. Biophys J 66, 996-1010 (1994).

85. Smith-Maxwell, C. J., Ledwell, J. L. & Aldrich, R. W. Role of the S4 in cooperativity of voltage-dependent potassium channel activation. J Gen Physiol 111, 399-420 (1998).

86. Smith-Maxwell, C. J., Ledwell, J. L. & Aldrich, R. W. Uncharged S4 residues and cooperativity in voltage-dependent potassium channel activation. J Gen Physiol 111, 421-39 (1998).

87. Ledwell, J. L. & Aldrich, R. W. Mutations in the S4 region isolate the final voltage-dependent cooperative step in potassium channel activation. J Gen Physiol 113, 389-414 (1999).

88. Horn, R., Ding, S. & Gruber, H. J. Immobilizing the moving parts of voltage-gated ion channels. J Gen Physiol 116, 461-76. (2000).

132

89. Lakowicz, J. R. Principles of Fluorescence (Kluwer Academic, New York, 1999).

90. Clegg, R. M. in Methods in Enzymology (eds. Lilley, D. M. J. & Dahlberg, J., E.) 353-388 (Academic Press, N.Y., 1992).

91. Wouters, F. S., Verveer, P. J. & Bastiaens, P. I. Imaging biochemistry inside cells. Trends Cell Biol 11, 203-11 (2001).

92. Weiss, S. Fluorescence Spectroscopy of Single Biomolecules. Science 283, 1676-1683 (1999).

93. Weiss, S. Measuring conformational dynamics of biomolecules by single molecule fluorescence spectroscopy. Nat Struct Biol 7, 724-9 (2000).

94. Ha, T. Single-molecule fluorescence resonance energy transfer. Methods 25, 78-86. (2001).

95. Stryer, L. & Haugland, R. P. Energy Transfer: A Spectroscopic Ruler. Proc. Natl. Acad. Sci., USA 58, 719-726 (1967).

96. Stryer, L. Fluorescence Energy Transfer as a Spectroscopic Ruler. Annual Review of Biochemistry 47, 819-846 (1978).

97. Clegg, R. M. Fluorescence Resonance Energy Transfer. Current Opinions in Biotechnology 6, 103-110 (1995).

98. Selvin, P. R. in Methods in Enzymology (ed. Sauer, K.) 300-334 (Academic Press, Orlando, 1995).

99. Selvin, P. R. The Renaissance in Fluorescence Resonance Energy Transfer. Nature Structural Biology 7, 730-734 (2000).

100. Selvin, P. R., Rana, T. M. & Hearst, J. E. Luminescence resonance energy transfer. J. Amer. Chem. Soc. 116, 6029-6030 (1994).

101. Heyduk, E. & Heyduk, T. Thiol-reactive luminescent Europium chelates: luminescence probes for resonance energy transfer distance measurements in biomolecules. Anal. Biochem. 248, 216-227 (1997).

133

102. Selvin, P. R. Principles and Biophysical Applications of Luminescent Lanthanide Probes. Annual Review of Biophysics and Biomolecular Structure 31, 275-302 (2002).

103. Förster, T. Experimental and Theoretical Investigation of the Intermolecular Transfer of Electronic Excitation Energy. Z. Naturforsch A 4, 321-327 (1949).

104. Selvin, P. R. Lanthanide-based resonance energy transfer. IEEE J. of Selected Topics in Quantum Electronics: Lasers in Biology 2, 1077-1087 (1996).

105. Xiao, M. & Selvin, P. R. Quantum Yields of Luminescent Lanthanide Chelates and Far-Red Dyes Measured by Resonance Energy Transfer. J. Am. Chem. Soc. 123, 7067-7073 (2001).

106. Reifenberger, J. G., Snyder, G. E., Baym, G. & Selvin, P. R. Emission polarization of europium and terbium chelates. Journal of Physical Chemistry B 107, 12862-12873 (2003).

107. Selvin, P. R., Jancarik, J., Li, M. & Hung, L.-W. Crystal structure and spectroscopic characterization of a luminescent europium chelate. Inorganic Chemistry 35, 700-705 (1996).

108. Burmeister-Getz, E., Cooke, R. & Selvin, P. R. Luminescence resonance energy transfer measurements in myosin. Biophys. J. 75, 2451-2458 (1998).

109. Xiao, M. et al. Conformational changes between the active-site and regulatory light chain of myosin as determined by luminescence resonance energy transfer: The effect of nucleotides and actin. Proc. Nat'l. Acad. Sci., USA 95, 15309-15314 (1998).

110. Selvin, P. R. & Hearst, J. E. Luminescence energy transfer using a terbium chelate: Improvements on fluorescence energy transfer. Proc. Natl. Acad. Sci, USA 91, 10024-10028 (1994).

111. Selvin, P. R. & Hearst, J. E. (University of California, Berkeley, 1994).

112. Xiao, M. et al. An actin-dependent conformational change in myosin. Nat Struct Biol 10, 402-8 (2003).

134

113. Heyduk, E., Heyduk, T., Claus, P. & Wisniewski, J. R. Conformational changes of DNA induced by binding of chironomus high mobility group protein 1a (cHMG1a). J. Biol. Chem. 272, 19763-19770 (1997).

114. Heyduk, E. & Heyduk, T. Architecture of a complex between the sigma70 subunit of Escherichia coli RNA polymerase and the nontemplate strand oligonucleotide. Luminescence resonance energy transfer study. J Biol Chem 274, 3315-22 (1999).

115. Callaci, S., Heyduk, E. & Heyduk, T. Core RNA polymerase from E. coli induces a major change in the domain arrangement of the sigma 70 subunit. Mol Cell 3, 229-238 (1999).

116. Xu, J. & Root, D. D. Domain motion between the regulatory light chain and the nucleotide site in skeletal myosin. J Struct Biol 123, 150-61 (1998).

117. Lidke, D. S. & Thomas, D. D. Coordination of the two heads of myosin during muscle contraction. Proc Natl Acad Sci U S A 99, 14801-6 (2002).

118. Root, D. D. In situ molecular association of dystrophin with actin revealed by sensitized emission immuno-resonance energy transfer. Proc. Natl. Acad. Sci., USA 94, 5685-5690 (1997).

119. Mathis, G. Rare earth cryptates and homogeneous fluoroimmunoassays with human sera. Clinical Chem. 39, 1953-1959 (1993).

120. Mathis, G. Probing molecular interactions with homogeneous techniques based on rare earth cryptates and fluorescence energy transfer. Clinical Chem. 41, 1391-1397 (1995).

121. Vazquez-Ibar, J. L., Weinglass, A. B. & Kaback, H. R. Engineering a terbium-binding site into an integral membrane protein for luminescence energy transfer. Proc Natl Acad Sci U S A 99, 3487-3492. (2002).

122. Pathak, M., Kurtz, L., Tombola, F. & Isacoff, E. The cooperative voltage sensor motion that gates a potassium channel. J Gen Physiol 125, 57-69 (2005).

123. Starace, D., Stefani, E. & Bezanilla, F. Voltage-dependent proton transport by the voltage sensor of the Shaker K+ channel. Neuron 19, 1319-1327 (1997).

135

124. Starace, D. M. & Bezanilla, F. Histidine scanning mutagenesis of basic residues of the S4 segment of the shaker k+ channel. J Gen Physiol 117, 469-90 (2001).

125. Starace, D. M. & Bezanilla, F. A proton pore in a potassium channel voltage sensor reveals a focused electric field. Nature 427, 548-53 (2004).

126. Ahern, C. A. & Horn, R. Specificity of charge-carrying residues in the voltage sensor of potassium channels. J Gen Physiol 123, 205-16 (2004).

127. Hessa, T., White, S. H. & von Heijne, G. Membrane Insertion of a Potassium-Channel Voltage Sensor. Science (2005).

128. Narahashi, T., Moore, J. W. & Scott, W. R. Tetrodotoxin blockage of sodium conductance increase in lobster giant axons. J. Gen. Physiol. 47, 965-974 (1964).

129. Nakamura, Y., Nakajima, S. & Grundfest, H. The action of tetrodotoxin on electrogenic components of squid giant axons. J Gen Physiol 48, 985-996 (1965).

130. Kao, C. Y. Structure-activity relations of tetrodotoxin, saxitoxin, and analogues. Ann N Y Acad Sci 479, 52-67 (1986).

131. Miller, C. The charybdotoxin family of K+ channel-blocking peptides. Neuron 15, 5-10 (1995).

132. Miller, C., Moczydlowski, E., Latorre, R. & Phillips, M. Charybdotoxin, a protein inhibitor of single Ca2+-activated K+ channels from mammalian skeletal muscle. Nature 313, 316-8 (1985).

133. Garcia, M. L., Garcia-Calvo, M., Hidalgo, P., Lee, A. & MacKinnon, R. Purification and characterization of three inhibitors of voltage-dependent K+ channels from Leiurus quinquestriatus var. hebraeus venom. Biochemistry 33, 6834-9 (1994).

134. Bontems, F., Gilquin, B., Roumestand, C., Menez, A. & Toma, F. Analysis of side-chain organization on a refined model of charybdotoxin: structural and functional implications. Biochemistry 31, 7756-64 (1992).

136

135. Krezel, A. M., Kasibhatla, C., Hidalgo, P., MacKinnon, R. & Wagner, G. Solution structure of the potassium channel inhibitor agitoxin 2: caliper for probing channel geometry. Protein Sci 4, 1478-89 (1995).

136. MacKinnon, R. & Miller, C. Mechanism of charybdotoxin block of the high-conductance, Ca2+-activated K+ channel. J Gen Physiol 91, 335-49 (1988).

137. Goldstein, S. A. & Miller, C. Mechanism of charybdotoxin block of a voltage-gated K+ channel. Biophys J 65, 1613-9 (1993).

138. Eriksson, M. A. & Roux, B. Modeling the structure of agitoxin in complex with the Shaker K+ channel: a computational approach based on experimental distance restraints extracted from thermodynamic mutant cycles. Biophys J 83, 2595-609 (2002).

139. Goldstein, S. A. & Miller, C. A point mutation in a Shaker K+ channel changes its charybdotoxin binding site from low to high affinity. Biophys J 62, 5-7 (1992).

140. Goldstein, S. A., Pheasant, D. J. & Miller, C. The charybdotoxin receptor of a Shaker K+ channel: peptide and channel residues mediating molecular recognition. Neuron 12, 1377-88 (1994).

141. Swartz, K. J. & MacKinnon, R. Mapping the receptor site for hanatoxin, a gating modifier of voltage-dependent K+ channels. Neuron 18, 675-82 (1997).

142. Gross, A. & MacKinnon, R. Agitoxin footprinting the shaker potassium channel pore. Neuron 16, 399-406 (1996).

143. Rauer, H. et al. Structure-guided transformation of charybdotoxin yields an analog that selectively targets Ca(2+)-activated over voltage-gated K(+) channels. J Biol Chem 275, 1201-8 (2000).

144. Shimony, E., Sun, T., Kolmakova-Partensky, L. & Miller, C. Engineering a uniquely reactive thiol into a cysteine-rich peptide. Protein Eng 7, 503-7 (1994).

145. MacKinnon, R., Cohen, S. L., Kuo, A., Lee, A. & Chait, B. T. Structural Conservation in Prokaryotic and Eurkaryotic Potassium Channels. Science 280, 106-109 (1998).

137

146. Baumgartner, W., Islas, L. & Sigworth, F. J. Two-microelectrode voltage clamp of Xenopus oocytes: voltage errors and compensation for local current flow. Biophys J 77, 1980-91 (1999).

147. Heyduk, T. & Heyduk, E. Luminescence energy transfer with lanthanide chelates: interpretation of sensitized acceptor decay amplitudes. Anal Biochem 289, 60-7. (2001).

148. Blaustein, R. O., Cole, P. A., Williams, C. & Miller, C. Tethered blockers as molecular 'tape measures' for a voltage-gated K+ channel. Nat Struct Biol 7, 309-11 (2000).

149. Asamoah, O. K., Wuskell, J. P., Loew, L. M. & Bezanilla, F. A fluorometric approach to local electric field measurements in a voltage-gated ion channel. Neuron 37, 85-97 (2003).

150. Ren, D. et al. A prokaryotic voltage-gated sodium channel. Science 294, 2372-5. (2001).

151. Koishi, R. et al. A superfamily of voltage-gated sodium channels in bacteria. J Biol Chem 279, 9532-8 (2004).

152. Li-Smerin, Y., Hackos, D. H. & Swartz, K. J. alpha-helical structural elements within the voltage-sensing domains of a K(+) channel. J Gen Physiol 115, 33-50 (2000).

153. Hong, K. H. & Miller, C. The lipid-protein interface of a Shaker K(+) channel. J Gen Physiol 115, 51-8 (2000).

154. Schwarz, T. L., Tempel, B. L., Papazian, D. M., Jan, Y. N. & Jan, L. Y. Multiple potassium-channel components are produced by alternative splicing at the Shaker locus in Drosophila. Nature 331, 137-42 (1988).

155. Kamb, A., Tseng-Crank, J. & Tanouye, M. A. Multiple products of the Drosophila Shaker gene may contribute to potassium channel diversity. Neuron 1, 421-30 (1988).

138

156. Shih, T. M., Smith, R. D., Toro, L. & Goldin, A. L. High-level expression and detection of ion channels in Xenopus oocytes. Methods Enzymol 293, 529-56 (1998).

157. Nakache, M. Heterogeneity of membrane phospholipid mobility in endothelial cells depends on cell substrate", , vol 317 sept 5, 1985 by et al. Nature 317, 75-7 (1985).

158. Caskey, C. T. Disease diagnosis by recombinant DNA methods. Science 236, 1223-9 (1987).

139

Vita

David John Posson was born on March 17th, 1975 in Dayton, Ohio. In 1996, he received

the Arts and Sciences Alumni Scholarship of the University of Cincinnati and in 1997

graduated with a B.S. in Physics. During his years as an undergraduate physics student,

David worked with Professor Leigh Smith on several projects including a mechanical

chaos experiment now used as a lab demonstration and an autocorrelator instrument used

to measure the pulse width of a fast Ti-Sapphire laser.

During graduate school, David spent a summer in Jim Wolfe’s lab measuring surface

acoustic waves excited on a regularly patterned substrate. However, interest in the

biological sciences soon led David to Professor Paul Selvin’s laboratory where he

completed several spectroscopic studies of the Shaker K+ voltage-gated ion channel in

collaboration with Professors Francisco Bezanilla and Christopher Miller.

His poster presentations include:

1. Posson, D. J. & Selvin, P. R. Binding/unbinding bioassays utilizing long range fluorescence quenching by silicon substrates. Biophysical Journal 78, 254a-254a (2000).

2. Posson, D. J., Starace, D. M., Bezanilla, F. & Selvin, P. R. Conformational

changes associated with gating and activation in the ILT mutant of the Shaker potassium channel measured by lanthanide-based resonance energy transfer. Biophysical Journal 82, 232a-232a (2002).

3. Posson, D. J., Miller, C, Bezanilla, F. & Selvin, P.R. Luminescence Resonance Energy Transfer Studies of the Shaker K+ Voltage-Sensor using Fluorescent Scorpion-Toxin Acceptor Probes. Biophysical Society Annual Meeting (2005).

He has submitted a paper:

1. Posson, D.J., Ge, P., Miller, C., Bezanilla, F. & Selvin, P.R. Angstrom scale vertical movement of a K+ channel voltage-sensor measured with luminescence energy transfer. Nature, 2005. submitted.

140