LIBRARY

Michigan State
i Unlverslty

 

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

MTE DUE MTE DUE MTE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1M C’JCIRCJDIDDIIpfi-p.“

 

IMAGING THE MOLECULAR DIMENSIONS AND OLIGOMERIZATION OF
PROTEIN MOLECULES AT THE SOLID/LIQUID INTERFACE BY SURFACE

ORIENTED MOLECULAR SIZING (SOMS) MICROSCOPY

Mark Joseph Waner

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

Doctor of Philosophy

1998

ABSTRACT

IMAGING THE MOLECULAR DIMENSIONS AND OLIGOMERIZATION OF
PROTEIN MOLECULES AT THE SOLID/LIQUID INTERFACE BY SURFACE

ORIENTED MOLECULAR SIZING (SOMS) MICROSCOPY

Mark Joseph Waner

The structure and behavior of proteins at the solid/liquid interface is of great
scientific interest. It has application both to fundamental biochemical understanding, as
well as to biotechnological purposes. Interfaces play a critical role in many physiological
processes. The mechanism of protein adsorption to surfaces is not very well understood.
The current model put forth in much of the literature assumes a two step model. In the
first step of this model the protein collides with the surface and adsorbs if its energy is
sufficient to overcome the free energy of desorption of surface adsorbed solvent. The
second step is often assumed to involve significant conformational change of the
secondary and tertiary structure of the protein or enzyme, akin to denaturation. This
unfolding of the protein would tend to indicate that loss of fimction would occur
concomitantly, but studies have found very little loss in activity upon adsorption for a

number of different protein systems.

The recent development of the atomic force microscope (AF M) offers another
tool for the examination of protein structure at liquid/solid interfaces. For atomically flat
crystals the AF M has been used to determine atomic positions to < l A resolution. In the
case of samples with topographic features larger than atoms, the probe tip of the AF M
‘convolutes’ with the size and shape of surface features. This has hindered the use of
AF M for molecular level structural determination of proteins at the liquid/solid interface.

The work presented in this dissertation covers the development of the surface
oriented molecular sizing (SOMS) technique which makes use of the angstrom height
resolution of the AF M and a physically based mathematical framework for the analysis of
the height distribution of adsorbed protein molecules. The surface adsorption and
orientation (SAO) model is developed using statistical thermodynamics to model the
expected height distributions for molecules adsorbed on a surface.

The SOMS technique will be shown to be viable through studies of ferritin and
concanavalin A (Con A) at the water/mica interface. Using this technique we are able to
determine both the three-dimensional size and the oligomerization state of the adsorbed
molecules. This technique will then be utilized for the examination of denaturation of
Con A at the interface, by a number of mechanisms. Further, the structural and
orientational changes in Con A as a function of pH will also be presented. The final
chapter of this dissertation will present an extension of these studies to the deposition and

structure of Con A thin films on mica.

DEDICATION

To my parents, James and Carol, for all their love, support and encouragement. I

wouldn’t have made it through this without you.

iv

ACKNOWLEDGEMENTS

I have had the great pleasure to work with and have support from many people
during my graduate work at M.S.U. First, however, I must express my gratitude to those
who helped me get here in the first place. The encouragement of my parents and
undergraduate advisors, Dr. Nick Baumgartner and Dr. David Ewing, was integral to my
decision to pursue the PhD. Here I would also like to mention Dr. George Leroi, for his
advice and support, which began with my first visit to the department as a prospective
graduate student in 1991.

I would like to thank Dr. Marcos Dantus for his insight, support and time. He has
supported not only my development as a researcher, but also as an educator. Our
collaborator, Dr. Melvin Schindler has been of considerable help in the development of
the science that is presented in this dissertation and has provided valuable advice and
critical insight. The initial work in our group on this project began with Martha Gilchrist,
who helped greatly in my learning of the field. I have also had considerable assistance
and support from Dr. James Geiger. He has provided insightful reviews of this
manuscript and our Journal of Physical Chemistry publication, as well as allowing me the
use of instrumentation and equipment in his laboratory and his biochemical expertise.

I would also like to acknowledge the other professors in the department who have
been both excellent teachers and mentors. I would especially like to thank Dr. Paul
Hunter, who has been a great help in my development as a teacher. I would also like to

thank Dr. Katharine C. Hunt, who has been a role model both as a top-notch teacher and

researcher. I would also like to thank Dr. Gerald T. Babcock for his support during my
studies, and his excellent example as a teacher and researcher.

The strong support staff in the department of chemistry have also played a key
role during my stay here. The glass, machine and electronics shops have provided first
class expertise. The business office and the secretarial staff have always done an
outstanding job in making sure that I could get my job done on time, and their efforts are
greatly appreciated.

Last, but certainly not least, I would like to thank my future wife, Kristina Henson
for her love, support and understanding as I finished my research and wrote this

dissertation.

vi

TABLE OF CONTENTS

Page

List of Tables x
List of Figures xi
List of Abbreviations xv
Introduction 1
Chapter 1: Perspective on the field of study

1.1 Proteins at interfaces 2

1.2 Atomic force microscopy (AF M) 5

1.3 Proposal of SOMS microscopy 19

1 .4 References 23
Chapter 2: Conceptual and Theoretical Framework

2.1 Introduction to SOMS technique 26

2.2 SAO model 27

2.3 Other considerations 36

2.4 References 40
Chapter 3: Experimental methods

3.1 Materials and instrumentation 41

3.2 Sample preparation 43

3.3 AF M imaging 46

3.4 Data Analysis 48

vii

Page

3.5 References 56
Chapter 4: Application of SOMS
4.1 Analysis of gold and ferritin at the liquid/solid interface
4.1.1 Measurement of colloidal gold at the
water/mica interface 57
4.1.2 SOMS analysis of ferritin at the
water/mica interface 64
4.2 Analysis of Con A at the liquid/solid interface
4.2.1 Background information for Con A 69
4.2.2 SOMS analysis of Con A at the
water/mica interface 72
4.2.3 Analysis of Con A denaturation at the
water/mica interface 85
4.2.4 Effect of pH on the structure and
adsorption of Con A molecules at the
water mica interface 90
4.3 Discussion and conclusions 100
4.4 References 1 08
Chapter 5: Study of Con A Orientation at the Liquid/Solid Interface: From Isolated
Molecules to Thin Films
5.1 Introduction to Con A film formation 110
5.2 Experimental and theoretical details 114

viii

5.3 Results
5.3.1 Growth and orientation of Con A films
from pH=7.3 solutions
5.3.2 Growth and orientation of Con A films
at acidic pH
5.3.3 Kinetics of the deposition process at
physiological and acidic pH
5.3.4 Morphology of Con A thin films at
the water/mica interface
5.4 Discussion
5.5 References
Appendix 1: Re-print of publication
Appendix 2: Computer program source code
Appendix 2.1 Source code for Mathematica
calculation of Chapter 2
Appendix 2.2 Source code for Image Analysis software

in Visual Basic 5.0

Page

119

130

140

143

147

152

153

163

164

170

Table I

Table II

Table III

force.

Table IV

Table V

Table VI

Table VII

LIST OF TABLES
Some common techniques for the determination of protein structure.
SOMS data obtained for Con A at the water/mica interface.

SOMS data for Con A at the water/mica interface as a function of imaging

SOMS data for succinyl Con A at the water mica interface.
SOMS data for SDS denatured Con A at the water/mica interface.
SOMS data for SDS denatured Con A at the water/mica interface.

SOMS data for Con A at the water/mica interface deposited from solutions

at two different pH values.

Table VIII

Free energy of orientation values for Con A at the water/mica interface

deposited from solutions at two different pH values.

Table IX

A) SOMS data for Con A at the water/mica interface, as deposited from a

pH=1O solution. B) Free energy of orientation values for Con A at the water/mica

interface deposited from a solution at pH=10.

Table X

Table XI

Table XII

Table XIII

Summary of literature results for Con A thin films deposited on surfaces.
Data for Con A films deposited in HEPES buffer (pH=7.3).
Data for Con A films deposited in Acetate buffer (pH=4.6).

RMS and Average roughness for Con A thin films.

LIST OF FIGURES
Figure 1 Schematic representation of the standard model for protein adsorption ath
the liquid/solid interface.
Figure 2 Scale drawing of Park Scientific Microlever type A cantilever.
Figure 3 Schematic force vs. distance curve for AF M in air.
Figure 4 Schematic diagram of A) the atomic force microscope (AF M) and B) PZT
scanner tube.
Figure 5 Error mode AF M image of the mica lattice. The raw image is shown in
A), while B) is the same image after application of a Fourier filter.
Figure 6 Lateral tip/sample ‘convolution’ in AF M.
Figure 7A Image of tobacco mosaic virus on mica illustrating the effect of lateral
convolution of the AF M probe tip.
Figure 7B & C Effect of tip convolution problem when the probe tip is B) non-
spherical or when there is a C) double tip.
Figure 8 Scale model of AF M probe tip scanning ellipsoidal particles adsorbed to
mica in three different orientations.
Figure 9A Rendering of Con A dimer circumscribed by a parallelepiped of
dimensions a, b, c.
Figure 9B Schematic of AFM probe tip measuring dimension a as the height (side be
adsorbed to substrate surface).
Figure 10 Height distribution predicted by SAO model, for given dimensions.
Figure 11 Experimental data modeled from an ideal height distribution.

Figure 12 Log-log plot of variance versus number of measurements.

xi

Figure 13 AFM height measurement discrepancy due to digitization.

Figure 14 AF M images of Con A molecules on mica, shown in standard and three-
dimensional renderings.

Figure 15A Example of image requiring first order flattening.

Figure ISB Example of image requiring second order flattening.

Figure 16 Height distribution measured by AFM for three different sizes of colloidal
gold on mica.

Figure 17 A) Compressibility of 5 nm colloidal gold on mica as a filnction of force.
B) AF M image from this data set.

Figure 18 A) Compressibility as a function of force for 5 nm colloidal gold using a

new probe tip. B) AFM image from this data set.

Figure 19 Height distribution measured for ferritin molecules at the water/mica
interface.
Figure 20 Plot of compressibility as a function of force for ferritin molecules

adsorbed at the water/mica interface.
Figure 21 Ribbon rendering of the Con A dimer, based on XRD structure of Shoham

et al. [BNL Protein DataBank, 1CN1].

Figure 22 Height distribution measured for Con A molecules at the water/mica
interface.

Figure 23 Height distribution measured for Succinyl Con A molecules at the
water/mica interface.

Figure 24 Height distribution measured for SDS treated Con A molecules at the

water/mica interface.

xii

Figure 25 Height distribution measured for Con A molecules treated with 6 M urea,
prior to dilution and adsorption at the water/mica interface.

Figure 26 Positively charged amino acids in Con A at two pH values. Positive
residues are shaded.

Figure 27 Height distributions measured for Con A molecules deposited from
solutions at A) pH=7.3 and B) pH=4.6.

Figure 28 Model illustrating the AFM lateral resolution expected for measurement of
thin films on mica.

Figure 29 1 min. deposition of Con A in HEPES on mica. A) AF M image B) Height
histogram

Figure 30 Portion of image from Figure 29, rendered as a contour plot. Approximate
lateral resolution is represented by a square drawn to scale of image.

Figure 31 5 min. deposition of Con A in HEPES on mica. A) AF M image B) Height
histogram

Figure 32 10 min. deposition of Con A in HEPES on mica. A) AF M image B)
Height histogram

Figure 33 60 min. deposition of Con A in HEPES on mica. A) AF M image B)
Height histogram

Figure 34 1 min. deposition of Con A in Acetate on mica. A) AF M image B) Height
histogram

Figure 35 5 min. deposition of Con A in Acetate on mica. A) AF M image B) Height
histogram

Figure 36 10 min. deposition of Con A in Acetate on mica. A) AF M image B)
Height histogram

Figure 37 60 min. deposition of Con A in Acetate on mica. A) AFM image B)
Height histogram

Figure 38 First order kinetics plot of Con A data at two different pH values.

xiii

Figure 39 Line profiles taken from AF M images of Con A thin films, deposited at A)
pH=7.3 and B) pH=4.6.

Figure 40 Possible orientations of Con A molecules at the water/mica interface, as a
function of time: A) early growth, B) intermediate growth, C) continuous film (as
discussed in the text).

xiv

AF M
Con A
HEPES
HPLC
NMR
NSOM
PSPD
PZT
QCM
SANS
SAXS
SAO
SEM
SFA
SOMS
SPM
STM
TEM

LIST OF ABBREVIATIONS

atomic force microscope or microscopy
Concanavalin A

N-2 hydroxyethylpiperazine N’-2 ethanesulfonic acid
high-performance liquid chromatography

nuclear magnetic resonance

near-field scanning optical microscope or microscopy
position sensitive photodiode

lead zirconium titanate

quartz crystal microbalance

small angle neutron scattering

small angle x-ray scattering

surface adsorption and orientation model

scanning electron microscope or microscopy

surface force apparatus

surface oriented molecular sizing

scanned probe microscope or microscopy

scanning tunneling microscope or microscopy
transmission electron microscope

x-ray diffraction

XV

INTRODUCTION

In this dissertation I will discuss the work I have done over the past two and a half
years, under the direction of Dr. Marcos Dantus. My interest has been focused on the
better understanding of the physical characteristics of protein molecules at the liquid/solid
interface. A large part of my work has been the development of a new technique for the
direct observation of tertiary structure and oligomerization state of protein molecules at

the solid/liquid interface using the atomic force microscope (AFM).

The first chapter of this work is devoted to a description of the current
understanding of protein behavior at liquid/solid interfaces, and some of the questions
that remain to be adequately understood or investigated. I will then discuss the use of the
AFM as a tool for investigating biomolecular structures on surfaces and its limitations for
accurate lateral dimensions of biomolecules. Chapter two will lay out the theoretical
framework for the surface oriented molecular sizing (SOMS) technique and describe how
this approach is used to overcome one of the shortcomings of the AF M, and the other
related scanned probe microscopes (SPM). The experimental methods and materials will
be described in chapter three, along with an explanation of our data analysis procedure.
Chapter four will give the experimental results for SOMS measurements of ferritin and
concanavalin A (Con A). The system which has been the focus of most of my work is
Con A, and most of chapter four is devoted to better understanding its tertiary and

quaternary structure at the water/mica interface as it adsorbs molecule by molecule. The
1

final chapter contains results which use the AF M, and the understanding gained by the
results of chapter four, to examine the structure of Con A as it adsorbs to form monolayer

films.

Chapter 1 : Perspective on the Field of Study

1.1 Proteins at interfaces

The behavior of proteins at the solid/liquid interface has been of great scientific
interest, as it has application to fundamental biochemical understanding, as well as to
more commercial purposes. Interfaces play a critical role in many physiological
processes. Many processes involving proteins occur at or near interfaces, such as
between cell membranes and the cytoplasm. For example, the cell adhesive protein
fibrinogen is regarded as causing thrombosis when it adsorbs to many biomaterials [1].
On the more practical side is the interest in gaining understanding of protein structure and
function at interfaces for the development of biological assays, biocompatible materials
for implants and surgical instruments and even materials which might reduce contact lens
fouling [2-4]. Recently there have been ACS symposia devoted specifically to the studies

of proteins at interfaces [2,4].

The adsorption of a protein molecule from solution onto a solid surface is a
complex process [5,6]. Current literature tends to favor a two step model for the

adsorption process, illustrated schematically in Figure 1 [4]. The first step involves the
2

 

1) Collision of protein with surface and displacement of surface adsorbed solvent

. ' 1" - r v, .
. _ . ‘ .
.I 7" .to- .4! ‘D‘

2) Conformational change of protein to maximize electrostatic & Van der Waals
interactions with the surface.

‘@ *&

Figure 1: Schematic representation of the standard model for protein adsorption

at the liquid/ solid interface.
L J

     

    

 

 

collision of the protein, which is translating by Brownian motion, with the surface. If the
energetic benefit of the removal of surface adsorbed water is greater than the thermal
energy of the protein molecule, then it remains on the surface, otherwise it returns to
solution [6]. Once the protein is adsorbed, the second step is thought to involve a
conformational change that maximizes electrostatic interactions, local interactions
between electron donor moieties in the protein with electron acceptors on the surface and
maximization of van der Waals forces [7]. It is often assumed that this conformational
change involves fairly substantial denaturation. There is now some experimental
evidence that ‘compact’ globular proteins will conserve their native structure when
adsorbed, and that only some ‘soft/flexible’ proteins suffer some degree of denaturation
upon adsorption, especially on hydrophobic surfaces [8]. The second step, therefore,

3

involves primarily an orientation process which is thermodynamically driven and yields

proteins in the most energetically favorable orientation.

Much of the work in this field has been focused primarily on three areas of study:
a) kinetics of the adsorption process, b) monolayer and 2D crystal growth and c)
structural transitions of proteins. The bulk of the studies, up to this time, have
investigated the kinetics and thermodynamics of the adsorption process and the formation
of monolayer films. The study of isolated, individual protein molecules at interfaces is

even more limited.

While there is much interest in the structure of biomolecules at interfaces,
publications in this area are relatively few. Much of the reason for this is that the number
of techniques available to quantitatively measure structure and structural changes of
individual protein molecules at interfaces is very limited. Table I lists the tools most
commonly used for the determination of protein structure today, as well as information
about the limits of each technique. Of the techniques listed, only scanning and
transmission electron microscopy (SEM and TEM, respectively), small angle x-ray and
neutron scattering (SAXS and SANS, respectively), and AFM are readily utilized for the
study of protein structure at interfaces. X-ray diffraction (XRD), Electrophoresis, HPLC
and MS are applicable for protein structure determination in pure form or in solution.
Although NMR spectroscopy primarily has been used for low molecular weight proteins
in solution, new developments which increase the intensity of observed NMR signals

may extend its use to the study of proteins at interfaces [9,10].
4

Table I: Some common techniques for the determination of protein structure.

 

 

 

 

 

 

 

 

 

 

 

 

Technique Resolution Concentration Requirements/Limitations Analysis

X-ray A ~100 um Crystalline samples Complex

diffraction crystal (~1ug)

Electrophor- N/A 2 0.5 mg/ml MW measurement only Simple

esis,HPLC

NMR A 2 1 mg/ml Low MW, soluble proteins Complex

SEM, TEM A-nm 2 0.02 ug/ml 2. 300 kD, dehydrated and Simple
coated

SAXS, SANS A-nm .>_ 1 mg/rnl Synchotron radiation Complex
source

AF M 1 A-20 nm 2 0.02 ug/ml artifacts induced by Simple“
imaging

 

A further limitation of most of the techniques (SEM, TEM and AF M excluded) is

that they require fairly concentrated samples of protein in order to obtain good signal to

noise. The SEM and TEM both require high vacuum and conductive samples, which

means samples must be chemically fixed, dehydrated and coated with graphite or gold.

The only technique that has the potential to observe individual protein molecules at

solid/liquid interfaces, under conditions approaching something close to physiological, is

the AF M.

1.2 Atomic force microscopy (AFM)

Since its recent introduction [11], atomic force microscopy has been used

extensively for imaging biological molecules [12-16]. The AFM is one member of the

5

 

 

 

 

 

 

 

N
4 pm % v X
0.6 um
i 1
180 um
Probe Tip
Figure 2: Scale drawing of Park Scientific Microlever type A cantilever

j

 

 

scanning probe microscope (SPM) family which also includes the scanning tunneling
microscope (STM) and the near-field scanning optical microscope (N SOM). All of these
instruments rely on raster scanning a very small (~10-100 nm) probe tip over the surface
of a sample. In the case of the AFM the probe tip (~10-50 nm radius of curvature) is
mounted on a cantilever arm (see Figure 2). The cantilever bends in response to the
forces which the probe tip senses as it comes close to the sample surface. This interaction
force between the sample and the probe tip may be an attractive or a repulsive force. At
large tip sample separation there is effectively no interaction force. As the probe and
sample are brought closer together (~ 5-15 nm) there is initially an attractive van der
Waals interaction force. As the tip-sample separation becomes smaller the interaction

force becomes repulsive because the electron clouds of the sample and the probe come

6

close enough for electrostatic forces to dominate. The interaction is analogous to the
potential energy surface for a simple diatomic such as H2. The analogy is not completely
sound, however, due to the fact that in the AF M case there is a discontinuity due to the
spring instability of the cantilever (see Figure 3, number 2). In addition the shape of the
force potential for the AFM differs from that of isolated atom case due to the effect of
geometry and the presence of large numbers of atoms. One of the reasons that the AF M
was not developed until the mid 1980’s was that the microfabrication techniques
necessary for the production of small, sharp probe tips mounted on sensitive cantilevers

were still being developed by the electronics industry.

 

 

Cantilever

Sample

Cantilever

Sample

 

 

 

 

 

 

 

 

 

 

 

Approach
1 2 3
v L l I
l 1
Retract
4 5 6
' v 1 fl 1 V

 

 

 

 
 

 

 
 

\5/34

 

-m 0

Tip/Sample Distance

Figure 3: Schematic force vs. distance curve for AF M in air

8

 

 

A schematic diagram of an AF M is given in Figure 4. The key components are
the piezoelectric scanner, the cantilever with integrated probe tip and the diode laser with
position sensitive photodiode. The piezoelectric scanner is typically a lead-ziconate-
titanate (PZT) material. A tube design is often used since it offers x,y and z motion in
one device. In this design an electrode is deposited on the inside of the tube and four
electrodes are deposited on the outer surface of the tube. Lateral motion of the tube is
achieved by control of the voltages applied to opposite outer electrodes, relative to the
inner electrode. The z motion is controlled by application of voltage to all four of the
outer electrodes and the inner electrode. The diode laser and position sensitive
photodiode (PSPD) system allows the measurement of the tip-sample force. The bending
of the cantilever causes a deflection of the laser beam. Initially the laser, cantilever and
PSPD are aligned such that the beam bounces off the end of the cantilever and onto the
center of the two elements of the PSPD. The two elements of the PSPD are referred to as
segments A and B. In order to minimize the effect of noise and fluctuations in the
photodiodes the AF M typically makes use of the so-called error signal for the
measurement of the cantilever deflection or tip-sample force. The error signal is given
by:

A-B
A+B

 

Err =

Where A and B are the voltages measured for each of the photodiode segments.

 

 

Position Sensitive
A Photo Diode

    
 

 

 

 

 

 

 

Tip
Sample Cantilever
z
PZT Piezoelectric
x Tube
y
J
\
PZT Piezoelectric
B Tube
PZT Piezoelectric
material in white
Electrode
material in gray
J

 

Figure 4: Schematic diagram of A) the atomic force microscope (AFM)
and B) PZT scanner tube

10

When one speaks of using an AF M, they might be speaking of any one of a
number of different modes of imaging. The common modes of imaging are: constant and
variable force contact modes, non-contact mode and intermittent contact. Other related

techniques include lateral force and magnetic force microscopy.

The standard mode of imaging is the constant force, contact mode of imaging. In
this method the tip and sample are brought into repulsive force ‘contact’ and a computer
controlled feedback loop adjusts the 2 position of the scanner in order to maintain
constant force. The image formation is based on plotting the 2 position of the scanner at
each x and y point. This mode of imaging is typically used for obtaining topographic
information for a wide variety of samples. Because of the sharp tips and the applied
repulsive forces this mode of imaging is often not suitable for very soft, weakly adsorbed

samples, unless performed under liquid where the adhesion force is negligible.

The variable force contact mode (also known as error mode of imaging) is used
when one wishes to image the topography of lattice scale features such as 0'2 ions in
mica. In this situation one is approaching the noise limit of the instrument. In order to
get information at the lattice or atomic scale the AFM requires an atomically flat
substrate. The tip and sample are first brought into contact. After a successful approach
in a region of the sample which is flat one turns off the force feedback loop. The image
formation in this mode is given by plotting the error signal at each point of the raster scan
in x and y. Figure 5 presents an image of the hexagonal (001) mica lattice taken in the

error mode.
1 1

 

 

 

 

 

Figure 5: Error mode AFM image of the mica lattice. The raw image
is shown in A), while B) is the same image after application
of a Fourier filter.

The non-contact mode of imaging typically utilizes the attractive regime of the
force vs. distance curves. In this mode of imaging the cantilever is mounted on a small
piezoelectric actuator that is used to oscillate the cantilever near its resonant frequency.
The modulation amplitude is typically on the order of tens to hundreds of angstroms. The
force interactions between the tip and sample are monitored either by measuring the
changes in frequency or amplitude of the cantilever oscillation. Because the forces being
monitored are long-range attractive, this method of imaging applies forces on the order of
piconewtons between the tip and sample. While this mode of imaging has somewhat
lower imaging resolution, it often provides a very good way to image soft, elastic or

weakly bound samples without inducing damage.

The intermittent contact mode (also called tapping mode) of imaging is similar to
non-contact in that the cantilever is oscillated near its resonant frequency. In this case,
however, the amplitude is increased and the tip-sample separation is reduced. The
sample is brought close enough to the probe tip so that at the minimum of the cantilever
oscillation, the tip and sample make contact. Because the tip and sample make only
intermittent contact, there is less worry about damage to the sample. Also, in the ideal
case the next point in the raster scan is not reached until the tip has been lified from the
surface. That means that the lateral forces, which arise from the motion of the scanning

in contact mode, are eliminated.

The AFM technique allows the examination of biological specimens without the

need for chemical fixation/dehydration or growth of crystals, which is normally
13

associated with the high-resolution techniques of SEM/TEM or XRD [12,17,18]. In fact,
samples may be imaged with AF M under liquids at physiological concentrations and pH
[19-22]. A number of groups have utilized the high resolution (~1 A) of the AF M for
examination of 2D crystallization of protein molecules [11,18-21]. Others, especially
Hansma, Bustarnante and Henderson, have studied isolated proteins [17,22-28] or DNA
molecules [14,20,23-26] and their complexes [27]. Recently AF M has been used by the
Marchant group to obtain valuable structural information for the multirneric Von

Willebrand factor (MW~260 kD, monomer) [28-30].

In the case of low molecular weight globular proteins and other small three-
dimensional structures, however, AF M has been limited by an inherent loss of resolution
for lateral measurements (10-20 nm, at best). The AF M is routinely used to obtain
angstrom resolution images of crystalline and SAM surfaces [16,31,32]. This angstrom
level resolution, however, is only possible for atomically flat samples with highly ordered
periodic features. When imaging sample features which are not highly ordered and
rough, on the scale of the probe tip, lateral resolution is limited by the finite size of the
probe tip (~10-50 nrn radius of curvature) as well as the sample being examined [30,33-
35] (see Figure 6). Figure 6 shows an ideal probe tip with radius of curvature of 20 nm
and the path it would follow when scanning a small spherical object of radius 6.5 nm.
While the lateral information for the sample object is lost, the height of the object is
measured with a high degree of accuracy. The problem of tip convolution in AF M
imaging manifests itself in a few common ways. Three examples of tip convolution are

illustrated in Figure 7. In Figure 7A an image of tobacco mosaic virus is presented, along
14

 

F 20 nm \

AFM Probe Tip

   
  

Spherical

Probe Trace object

\

 

 

Substrate

 

 

 

 

Figure 6: Lateral tip/sample 'convolution' in AFM

\ J

with a cross sectional line scan taken at the area highlighted by the white line. The

 

tobacco mosaic virus is a cylindrical rod with a radius of 180 A. The line scan shows that
the height is measured to be 180 A, but the lateral dimension is measured to be 350 A.
Another example of a tip convolution manifestation is presented in Figure 7B. In this
case the shape of the probe tip deviates from the ideal spherical shape. This type of tip
produces an image with protrusions in the surface that show a repeating shape (e.g. note
the triangular shape that appears in each of the lightly shaded features). A third
manifestation of the tip convolution effect is presented in Figure 7C. In this image each
raised surface feature seems to be repeated. This is the result of a so-called double tip, in
which the probe tip has been damaged and there are effectively two probe tips that can

interact with the sample surface. While Figures 7B and C highlight artifacts due to bad

probe tips, it is important to remember that this convolution effect leads to limitations for

lateral measurements even for ideal probes (as shown in Figure 7A).

16

 

 

 

 

 

 

-50.'ITTvr-r'
0 2000 4000

Distance (A)

l I

6000 8000

Figure 7A: Image of tobacco mosaic virus on mica illustrating the effect of lateral
convolution of the AF M probe tip.

17

 

Figure 7 B & C: Efi‘ect of tip convolution problem when the probe tip is
B) non-spherical or when there is a C) double tip.

18

This ‘convolution’ effect occurs regardless of imaging conditions (vacuum,
ambient or under liquid) and has limited the applicability of the AF M for the study of
biomolecules at interfaces. Some groups have approached this problem by making
sharper probe tips [35] which has led to commercially available tips with radius of
curvature as low as 10 nm, whereas many of the initially available tips were 40-50 nm in
radius. Still others have examined ways to first characterize the shape of the probe tip
and then mathematically ‘deconvolute’ their AF M data [34-36]. This is still problematic
for our purposes since the probe tips are fragile enough that they may change size or
shape, even over a short period of use. Further, this ‘convolution’ effect is not a true
mathematical convolution of two functions rather it involves a shadowing effect of the tip
as it interacts with the sample. Because of this, the ‘deconvolution’ of data leads to the
determination of an upper limit to the lateral dimensions. The approach which is taken in
the current work is to take advantage of the angstrom level resolution of the AF M for
measurements of height, which is essentially independent of the probe size (please see

Chapter 2.3 for more detail).

1.3 Proposal of SOMS microscopy

This dissertation presents a new approach for making very high resolution (<5 A)
three-dimensional measurements of protein molecules at liquid/solid interfaces, using
atomic force microscopy (AF M). Such measurements are of particular biological
relevance since the preponderance of biological activities of enzymes and their binding

properties occur on membrane or cytoskeletal surfaces. To avoid the loss of resolution
19

that is normally introduced into AF M measurements as a result of probe tip geometry we
have devised a measurement strategy that measures the heights of an ensemble of
individual protein molecules that have adsorbed to the atomically flat surface of mica. A
scale model of an ellipsoid (of dimensions 30 x 45 x 75 A) circumscribed by a
parallelepiped and adsorbed to a mica surface in three different orientations is presented
in Figure 8. Also shown in Figure 8 is a rendering of an AFM probe tip (drawn to scale)
positioned over one of the protein molecules which illustrates that the height of the
molecule is accurately probed by the AFM even though the lateral dimensions will be

overestimated by the convolution effect.

 

 

 

Figure 8: Scale model of AFM probe tip scanning ellipsoidal particles
adsorbed to mica in three different orientations

20

The technique has been termed Surface Oriented Molecular Sizing (SOMS)
because it utilizes the statistical thermodynamic distribution of orientations for adsorbed
molecules to determine the three dimensional structure of proteins. Further, the statistical
thermodynamic nature of the surface adsorption process can be used to estimate the
surface adsorption energy for different protein orientations at the interface. The use of
the projected height of individual adsorbed protein molecules as the defining parameter
for molecular dimensions makes this approach minimally dependent on probe tip
geometry and utilizes the angstrom resolution of the AF M in the z direction. Indeed, it is
advantageous to use a broader tip for making these measurements, thereby minimizing
the applied pressure between the tip and protein and decreasing the chance of
underestimating the height of a molecule due to the digitization inherent in the
microscope (see section 2.2). Furthermore, this technique is generally applicable to
contact and intermittent contact modes of imaging under ambient or liquid conditions.
The relative ease, speed, and resolution with which these molecular dimension
measurements may now be performed make this technique particularly useful for
examining multiple conformational equilibria of adsorbed protein subunits and the
changes in shape and assembly state of multi-subunit proteins resulting from
oligomerization and aggregation at liquid/solid interfaces. Such changes may be
systematically examined as a function of protein concentration and ligand or allosteric
activator binding. Such measurements could help to determine whether single amino acid
substitutions affect protein activity as a consequence of an induced change in the
oligomerization state of the mutant protein or as a result of a chemical change in the

active site. The studies in this dissertation focus primarily on the structure of
21

concanavalin A (Con A) molecules at the water/mica interface, although results for

ferritin are also presented.

22

1.4 References

1. Feng, L.; Andrade, J. D. Structure and Pr0perties of Fibrinogen. In Proteins at
Interfaces 11, Fundamentals and Applications; Horbett, T. A., Brash, J. L., Eds.;
American Chemical Society: Washington, 1995; Vol. ACS Symposium Series 602; pp
66.

2. Horbett, T. A.; Brash, J. L. Proteins at Interfaces: Current Issues and Future
Prospects. In Proteins at Interfaces, Physicochemical and Biochemical Studies; Brash, J.
L., Horbett, T. A., Eds.; American Chemical Society: Washington, 1987; Vol. ACS
Symposium Series 343; pp 1.

3. Norde, W. Adv. Coll. Inter. Sci. 1986, 25, 267.

4. Brash, J. L.; Horbett, T. A. Proteins at Interfaces: An Overview. In Proteins at
Interfaces 11, Fundamentals and Applications; Horbett, T. A., Brash, J. L., Eds.;
American Chemical Society: Washington, 1995; Vol. ACS Symposium Series 602; pp 1.

5. Elwing, H.; Askenda, A.; Ivarsson, B.; Nilsson, U.; Welin, S.; Lundstrom, 1. Protein
Adsorption on Solid Surfaces: Physical Studies and Biological Model Reactions. In
Proteins at Interfaces, Physicochemical and Biochemical Studies; Brash, J. L., Horbett,
T. A., Eds.; American Chemical Society: Washington, 1987; Vol. ACS Symposium
Series 343; pp 468.

6. Nadarajah, A.; Lu, C. F.; Chittur, K. K. Modeling the Dynamics of Protein
Adsorption to Surfaces. In Proteins at Interfaces 11, Fundamentals and Applications;
Horbett, T. A., Brash, J. L., Eds.; American Chemical Society: Washington, 1995; Vol.
ACS Symposium Series 602; pp 181.

7. van Oss, C. J.; Wu, W.; Giese, R. F. Macroscopic and Microscopic Interactions
Between Albumin and Hydrophilic Surfaces. In Proteins at Interfaces 11, Fundamentals
and Applications; Horbett, T. A., Brash, J. L., Eds.; American Chemical Society:
Washington, 1995; Vol. ACS Symposium Series 602; pp 80.

8. Yan, G.; Li, J .-T.; Huang, S.-C.; Caldwell, K. D. Calorimetric Observations of Protein
Conformation at Solid—Liquid Interfaces. In Proteins at Interfaces 11, Fundamentals and
Applications; Horbett, T. A., Brash, J. L., Eds.; American Chemical Society: Washington,
1995; Vol. ACS Symposium Series 602; pp 256.

9. Tilton jr., R. F.; Kuntz, I. D. Biochemistry 1982, 21, 6850.

10. Miller, K. W.; Reo, N. V.; Schoot Uiterkamp, A. J. M.; Stengle, D. P.; Stengle, T.
R.; Williamson, K. L. Proc. Nat. Acad. Sci. USA 1981, 78, 4946.

11. Binnig, G.; Quate, C. F.; Gerber, C. Phys. Rev. Lett. 1986, 56, 930.
23

12. Lal, R.; John, S. A. Am. J. Physiol. 1994, 266, C1.
13. Butt, H.-J.; Downing, K. H.; Hansma, P. K. Biophys. J. 1990, 58, 1473.

14. Bustamante, C.; Vesenka, J.; Tang, C. L.; Rees, W.; Guthold, M.; Keller, R.
Biochemistry 1992, 31, 22.

15. Bustamante, C.; Rivetti, C. Ann. Rev. Biophys. Biomol. Struct. 1996, 25, 395.
16. Louder, D. R.; Parkinson, B. A. Anal. Chem. 1994, 66, 84R.
17. Schnyder, T.; Engel, A.; Lustig, A.; Wallimann, T. J. Biol. Chem. 1988, 263, 16954.

18. Dykstra, M. J. A Manual of Applied Techniques for Biological Electron Microscopy;
Plenum: New York, 1993.

19. Weisenhom, A. L.; Drake, B.; Prater, C. B.; Gould, S. A. C.; Hansma, P. K.;
Ohnesorge, F .; Egger, M.; Heyn, S. P.; Gaub, H. E. Biophys. J. 1990, 58, 1251.

20. Hansma, H. G.; Vesenka, J.; Siegerist, C.; Kelderman, G.; Morret, H.; Sinsheimer,
R. L.; Elings, V.; Bustamante, C.; Hansma, P. K. Science 1992, 256, 1180.

21. Putman, C. A. J.; Van der Werf, K. 0.; De Grooth, B. G.; Van Hulst, N. F.; Greve, J.
Appl. Phys. Lett. 1994, 64, 2454.

22. Weisenhom, A. L.; Maivald, P.; Butt, H.-J.; Hansma, P. K. Phys. Rev. B 1992, 45,
11226.

23. Benzanilla, M.; Manne, S.; Laney, D. E.; Lyubchenko, Y. L.; Hansma, H. G.
Langmuir 1995, 11, 655.

24. Hansma, H. G.; Sinsheimer, R. L.; Li, M. Q.; Hansma, P. K. Nucleic Acids Res.
1992, 20, 3585.

25. Henderson, E. Nucleic Acids Res. 1992, 20, 445.

26. Shaiu, W.-L.; Vesenka, J.; Jondle, D.; Henderson, E.; Larson, D. D. J. Vac. Sci.
Technol. A 1993, 11, 820.

27. Wyman, C.; Rombel, 1.; North, A. K.; Bustamante, C.; Kustu, S. Science 1997, 275,
1658.

28. Marchant, R. E.; Lea, A. S.; Andrade, J. D.; Bockenstedt, P. J. Coll. Inter. Sci. 1992,
148, 261.

24

29. Siedlecki, C. A.; Lestini, B. J.; Kottke-Marchant, K.; Eppel, S. J.; Wilson, D. L.;
Marchant, R. E. Blood 1996, 88, 2939.

30. Eppel, S. J .; Zypman, F. R.; Marchant, R. E. Langmuir 1993, 9, 2281.
31. Alves, C. A.; Smith, E. L.; Porter, M. D. J. Am. Chem. Soc. 1992, 114, 1222.
32. Bottomley, L. A. Anal. Chem. 1998, 70, 425R.

33. Allen, M. J.; Hud, N. V.; Balooch, M.; Tench, R. J.; Seikhaus, W. J.; Balhom, R.
Ultramicroscopy 1992, 42B, 1095.

34. Markiewicz, P.; Goh, M. C. Langmuir 1994, 10, 5.
35. Keller, D. J .; Franke, F. S. Surf. Sci. 1993, 294, 409.

36. Wilson, D. L.; Dalal, P.; Kump, K. S.; Benard, W.; Xue, P.; Marchant, R. E.; Eppell,
S. J. J. Vac. Sci. T echnol. B 1996, 14, 2407.

25

Chapter 2: Conceptual and Theoretical Framework

2.1 Introduction to the Surface Oriented Molecular Sizing technique

The loss of lateral resolution that is normally introduced into AF M measurements
as a result of probe tip geometry has limited the use of the AF M for the measurement of
macromolecular structure. In order to better examine protein structure with the AF M we
have devised a measurement strategy that determines the heights of an ensemble of
individual protein molecules that have adsorbed to the atomically flat surface of mica
(see Figure 8). The technique has been termed Surface Oriented Molecular Sizing
(SOMS) because it utilizes the statistical thermodynamic distribution of orientations for

adsorbed molecules to determine their three dimensional structure.

In order to model the distribution of molecular orientations we needed to develop
a theoretical treatment of the data. The first model used in our understanding [1] was
based on the statistical orientation process that parallelepipeds experienced when being
dropped onto a surface. This model made the simplification that orientation and
adsorption of protein molecules might behave similarly to boxes being dropped on a
surface. Large numbers of parallelepipeds, scaled to the relative dimensions of protein
molecules were dropped onto a surface landing with a statistically determined
distribution. It was found that most of the parallelepipeds landed such that they projected
their shortest dimension from the surface, or in other words they tend to land such that

the side of the box with the largest surface area is in contact with the substrate [1].

26

Although this initial idea was a good place to start, the extrapolation of the ‘box’
model to macromolecular objects, such as proteins, falls short because it fails to account
for the chemical properties of the protein and the surface and the thermodynamic
orientation process occurring at the surface. The model of protein adsorption presented
here takes into account both statistical, as well as thermodynamic properties of the
protein and surface. In this chapter the development of a surface adsorption and
orientation (SAO) model which treats the orientation of protein molecules during

adsorption as a statistical thermodynamic phenomenon is discussed.

2.2 Surface Adsorption and Orientation (SAO) model

In the absence of any localized recognition sites in the protein molecule, which
would tend to lead to a non-statistical orientation of molecules, the ensemble of
molecules is expected to assume a thermodynamically driven distribution of orientations.
A surface adsorption and orientation (SAO) model is proposed, which treats this
interfacial protein orientation as an optimization of free energy. Although in the case of
the current work there is a lack of specific recognition sites on the protein molecules, the
effect of chemical recognition between the surface and the protein molecules will be
discussed. It will be shown that the SAO model can also be used to provide valuable
physical insight for systems that do have preferential orientation due to some recognition

between protein and substrate.

27

The model assumes each protein can be enclosed by a parallelepiped of
dimensions a, b and c, where a _< b _< c (Figure 9 A). The gross structure of many
globular proteins is often approximated by an ellipsoid. A parallelepiped is compatible
with ellipsoidal models in the limit of well-rounded edges. Furthermore, the measured
height is independent of the degree to which the edges are rounded. The most highly
favored interaction between the protein molecules and the hydrophilic mica surface
results in orientations that maximize the surface area of contact. Therefore, a protein
orientation having side be adsorbed on the mica (where b and c are the largest dimensions
by definition) is considered to be the most energetically favorable, yielding a maximum

probability for measured heights corresponding to dimension a (see Figure 9 B).

28

 

 

 

 

of dimensions a, b, c.

 

 

Figure 9 A: Rendering of Con A dimer circumscribed by a parallelepiped
L J

C D

B Probe tip

 

    

c b

Figure 9 B: Schematic of AF M probe tip measuring dimension a as the
L height (side be adsorbed to substrate surface).

 

 

29

The optimization of the orientation free energy, after the initial adsorption step, is
assumed to be dependent on the protein surface in contact with the substrate. The free

energy of orientation for each surface is given by:

AG:;ent E AGite/4M (1)
AGrftZem E Ac;bc‘Aac (2)
Athrllent E AGab Au!) (3)

Where AGbc is the free energy of orientation per unit area of the protein in contact with
the substrate (in this case for side bc), and A is the surface area of the side identified by

the subscript.

The probability of orientation such that height a, b or c are measured depends on
the different free energies of orientation, equations 1-3. These may be estimated by a
Boltzmann distribution using:

bc
2 exp(— AG°"‘%T)

P E 4
(a) Q ()

2 exp[— AG‘:"%T)

Q

ab
2 exp[_ AG°"‘%T)

Q

 

 

P(b) s (5)

P(c) E

 

(6)

30

Where the factor of two arises from the symmetry of the parallelepiped, k is Boltzmann’s
constant, T is the temperature and Q is the canonical partition function. To this point the
model does not make any assumptions about the presence or absence of particular
recognition sites between the protein and the surface, since the free energy of each face of

the protein is independent.

If the system lacks any specific chemical recognition sites then all the free
energies of orientation per unit area are equal (ie. AGO E AGbc = AGac = AGab ). Under
this condition equations 4-6 may be further simplified by taking the ratios of the

probabilities of measuring each orientation.

 

M = exp[ [Aac - Abe lAGO ]

 

 

P(b) kT (7)
£2 = exp( [Aab — Aac IAGO ] (8)
P(c) kT
51a; zeprA... fill/AG.) (9)
P(c) kT

The free energies of orientation have been substituted by their corresponding expression
from equations 1, 2 and 3. These equations make it possible to predict the probability of
each protein orientation on the surface, given the dimensions of the protein molecule.
Conversely, it is also possible to start with the experimentally observed dimensions (a, b

and c) and their probabilities (P(a), P(b) and P(c)) and obtain a value for AG0 using

Equations 7-9. The magnitude of AG0 determines the relative probabilities of measuring

31

each of the three main dimensions of a molecule. Figure 10 illustrates the theoretical
height distribution for a parallelepiped of the same dimensions as Con A dimer for values

of AG() differing by a factor of two.

 

 

 

 

 

 

 

 

 

100 I - I ' I ' ' ' ' ' '
f I
A
80 - G0 2 '
' p K —— ~5 J/mol-A ‘
e 4:
S: 60 . iI -10 J/mol-A2 -
= ' ’
Q)
s. 40 - '
6)
I-
ra. ' '
20 - '
0 ' '
0 20 40 60 80 100
Height (A)

a=30A b=45Ac=59A

Figure 10: Height distribution predicted by SAO model, for given dimensions

The SAO model, therefore, provides a physically based mathematical framework
to analyze the height distribution profile obtained from AF M measurements. This
sensitivity of the observed distribution to the dimensions of the molecule used makes it
possible to also determine the relative amounts of each oligomeric species present (e. g.
monomer, dimer, tetramer). This model is useful for the determination of tertiary and
quaternary structural information for protein molecules at the liquid/solid interface. It is
possible to determine this structural information regardless of other physical data, such as

molecular weight or XRD structure determination, so the technique could be quite useful

32

for the analysis of proteins which have not yielded to structural determination by other

techniques.

Given a large enough sampling, on the order of 103 individual measurements, a
distribution of measured heights is obtained which is characteristic of the molecular
dimensions of the protein. This has been verified using a model height distribution and
different numbers of randomly obtained data points using Mathematica 3.0 (see
Appendix 2). For a system with dimensions a= 30 A, b= 45 A, c= 60 A (similar to
Concanavalin A which is the focus of this work) a theoretical height distribution was
calculated using a sum of three Gaussian functions, of width and relative areas consistent
with the experimental data to be presented in chapter 4. Then the observed height
distributions for different numbers of random measurements of the ideal distribution are
calculated (see Figure 11). It is quite apparent that 100 measurements is insufficient to
elucidate the three-dimensional structure of this system, but by 500-1000 measurements a
reasonable approximation of the actual distribution is obtained. The sum of squares was
calculated for the random sampling of points compared to the ideal height distribution.
The results of this calculation are plotted in Figure 12. It follows the expected

logarithmic dependence on the number of data points sampled.

33

Number of molecules

'Ideal' height distribution

 

 

 

 

 

 

 

 

 

  

 

 

 

g .
g .
a .
0' . F21? '4'0' ' '6'0' ' '8'0' . .160
Height (A)
6 100 points 500 points
5 . 25
4 1 20
3 . 15
2 . 10
1. - l 5
. 11 111111
o 20 4o 60 so 0 20 40 60 80
Height (A) Height (A
1000 points 10000 points
50 400 1
40 300 -j
30 200 j
20
10 100 j
o 20' . '4'0' ' 60' ”so 0 . ' -21)- ' '4'0' ' '6'0' I so
Height (A) Height (A)

Figure 1 1: Experimental data modeled from an ideal height distribution

34

 

r
O

    

- slope = -0.59
-1.5 4

Log(6 2)

 

 

 

I j I ' I ' I ' I

2 3 4 5 6
Log(N)

Figure 12: Log-log plot of variance versus number of measurements.
See Appendix 2 for calculation.

Most proteins are asymmetrical or oligomeric, therefore for a given protein,
different dimensions are obtained corresponding to different protein orientations (and/or
oligomers) on the substrate. In order to interpret the height distribution and decompose it
into the dimensions of the protein, the SAO model is used to predict how molecules of
specific dimensions are expected to adsorb on the substrate. With this theoretical
framework it is possible, not only to predict the way that a single oligomer of a protein
will adsorb, but also to model the adsorption of samples containing mixtures of different

oligomeric species.

35

2.3 Other considerations

In order to accurately measure the heights of protein molecules, there are a couple
of other considerations which need to be addressed. The first is that the measurement
may be highly influenced by the nature of the protein and the pressure applied to the
protein by the tip during the imaging process. The second is that there is a possibility of
underestimation of the actual molecular height due to the digitization of the imaging.
Each of these issues will be addressed by experiments in chapter 4, but the theoretical

issues are discussed here.

The AF M image contrast is based on the force interactions between the probe tip,
typically made from silicon or silicon nitride, and the sample. The forces applied during
the AF M imaging process are typically in the range of ~100 pN to hundreds of nN.
While these forces may seem quite small, one must consider that this force is applied
over a very small area at any one time. Therefore, the applied pressure can be quite large.
The applied pressure for a 1 nN force and a 50 nm radius surface area is 1.27 x 105 N/mz,
which is a rather significant pressure. It has been shown that the measurement of
colloidal gold is insensitive to tip/sample applied forces in standard AF M imaging
conditions [2]. The structure of globular proteins, however, is not insensitive to these
forces. In this range of applied pressures one needs to be concerned about the possibility
of compression of the protein structure. Gekko et al. have studied the compressibility of
globular proteins containing a variety of structural types [3]. This work has shown that

compressibility scales with the amount of a-helical structural elements, while proteins

with more B-sheet character are rather incompressible. The Con A protein is primarily B-

36

sheet in nature and has very little (it-helical character (see Figure 21 in Chapter four), and
therefore is expected to be quite incompressible. This will be addressed by experiments
examining the effect of imaging force on the observed height distribution, as well as by
measurements made on succinyl Con A, which has been shown to exist only in the

dimeric form [4].

Because the raster scanning process of the AF M is digitized, one must be
concerned with the possibility that the probe tip is not necessarily making a measurement
at every location of the sample. For example, if a particle of 32 A diameter is on a flat
substrate (see Figure 13) and one acquires a 2 pm x 2 pm image digitized to 256 x 256
pixels it is possible to underestimate the true height of the sample since the probe doesn’t
necessarily interact with the apex of the particle. In this case, each scan line is 7.8 nm
away from the previous line. If a particle happens to lie between two consecutive scan
lines and the radius of curvature of the probe tip is 20 nm, the probe tip will not interact
with the apex of the particle. This problem depends on the digitization, probe tip size and

the size of the sample features of interest.

37

 

 

 

 

 

 

Figure 13: AFM height measurement discrepancy due to digitization

38

In the worst case scenario there is a possibility of completely missing a sample
feature which is actually present. Of more importance for the current work and the
conditions chosen for this work is that one may see the feature, but the height
measurement will be underestimated since the probe tip does not necessarily interact with
the highest feature of the sample particle. The underestimation of the height, or

discrepancy d, is given by:

 

d=(R+r)-\/(R+r)2—X2 (10)
where R is the radius of curvature of the probe tip, r is the radius of a spherical object on

the surface and X is the horizontal distance between the probe and spherical object at the

point of closest contact.

In order to estimate the potential discrepancy in our height measurements, 5 nm
colloidal gold particles were imaged using one probe tip. Because of the tip convolution
effect, the observed lateral dimensions of a spherical object will reflect not only the size
of the object, but also the probe tip. Using the approach of Markiewicz and Goh, the
calculated radius of curvature of the tip was 24 nm (the manufacturer’s nominal value
was given as 20 nm) [5]. Given this radius and a 2 pm, 512 pixel scan range, the
maximum error observed for a 40 A particle would only be 0.7 A, or 1.8 %. The error
expected for a particle of 10 A radius, however, would be 0.8 A or 8 % which is
significant. The error associated with the height measurement of very small particles or
molecules (<20 A) can become significant. It should also be noted that the error expected

for the smaller particles is within the range of the surface roughness of the mica.

39

2.4 References

1. Gilchrist, M. Macromolecular size determination by scanning force microscopy. M.
S., Michigan State University, 1996.

2. Vesenka, J .; Manne, S.; Giberson, R.; Marsh, T.; Henderson, E. Biophys. J. 1993, 65,
992.

3. Gekko, K.; Hasegawa, Y. Biochemistry 1986, 25, 6563.

4. Gunther, G. R.; Wang, J. L.; Yahara, 1.; Cunningham, B. A.; Edelman, G. M. Proc.
Nat. Acad. Sci. USA 1973, 70, 1012.

5. Markiewicz, P.; Goh, M. C. Langmuir 1994, 10, 5.

4O

Chapter 3: Experimental methods

3.1 Materials and instrumentation

Standards. Colloidal gold particles were obtained from Sigma (St. Louis, MO)
and used without further purification. Three sizes of gold particles were used for the
AFM height calibration: 50 d: 8 A (3.8 x 1013 particles/ml), 90 i 12 A (6.0 x 1012
particles/ml), and 180 :1: 8 A (6.7 x 1011 particles/ml). The given dimensions and standard
deviations were determined by the manufacturer using TEM. Tobacco Mosaic Virus
(cylindrical shape; diameter 18 nm) (American Type Culture Collection) was also used to

check the microscope height calibration.

Proteins. Cationized F erritin was obtained from Polysciences (Warrington, PA)
as a 20 mg/ml solution. Concanavalin A (Con A, Canavalia ensiformis) was obtained
from Boehringer-Mannheim (Indianapolis, IN) as lyophilized solid. Succinyl
Concanavalin A was obtained from Sigma (St. Louis, MO) as a 95 % protein lyophilisate.

Proteins used in this work were used as obtained from the commercial source.

SDS-PA GE gel electrophoresis. Gels were prepared using the following: Sodium
dodecyl sulfate (SDS), electrophoretic grade, obtained from Boehringer-Mannheim
(Indianapolis, IN), Bis-Acrylamide (30-40%) purchased from Bio—Rad (Hercules, CA),
TEMED (electrophoresis grade) obtained from Fisher Biotech (Springfield, NJ) and
ammonium persulfate acquired from Gibco BRL (Germany). Tris base (Boehringer-

Mannheim ,Indianapolis, IN) was used as the buffer for gel preparation and sample

41

buffer. A Mini-Protean 11 gel electrophoresis apparatus from Bio-Rad (Hercules, CA)
was used for pouring and running the gels. Kaleidoscope pre-stained molecular weight
standards (7.1-200 kD) for SDS-PAGE gel electrophoresis were obtained from Bio-Rad
(Hercules, CA). Typically a 12% gel was used. The recipe for gel formation and
solutions for electrophoresis was followed, with the following exception. When
preparing the gels, the monomer solution was not degassed prior to adding TEMED and

ammonium persulfate to initiate crosslinking.

Other chemicals. Three different buffer solutions were used for the experiments
found in this work. The three were based on: HEPES obtained from Sigma (St. Louis,
MO), magnesium acetate 99% from Fluka Chemika (Buchs, Switzerland) and Tris from
Boehringer-Mannheim (Indianapolis, IN). The MgClz was ACS grade from Columbus
Chemical Industries (Columbus, OH). The water used for the preparation of solutions
was purified with a MilliQ Type I MilliPore water purification unit (19 MS), resistivity).
Methyl or-D mannopyranoside was obtained as 99% solid from Sigma (St. Louis, MO)
and used without further purification. Muscovite mica used in this work came from two
sources. High quality research grade sheets were a gift of Dr. Richard Schwendeman and
natural samples were obtained from Ward’s Natural Science Establishment. Highly

oriented pyrolytic graphite (HOPG) was a gift of Dr. Thomas J. Pinnavaia.

Instrumentation. The microscope used for this work was an Autoprobe CP

scanning probe microscope (Park Scientific Instruments). Two types of piezoelectric

tube scanners (Park Scientific Instruments) were used for this work. One of these was a

42

five urn high resolution scanner, which had a maximum lateral range of five microns and
2 range of one micron. The second scanner used for some of the studies in chapter 5, was
a 100 micron scanner, which was equipped with a proprietary Scan Master feature which
corrects for non-linearities in the piezoelectric tube. The 100 um scanner had a 2 range

of about 7 pm.

The UV-vis spectrometer used for this work was an ATI Unicam dual beam,
scanning spectrophotometer. It was used both in scanning and fixed wavelength mode,
using air for the background determination, and the given solvent as the reference
sample. The concentrations of Con A and succinyl Con A stock solutions were

determined using A230 with an extinction coefficient of 1.37 ml mg"l cm'1 [1].

3.2 Sample preparation

Measurements of protein height require an atomically flat substrate (less than 5A
corrugations). In most cases we found surface roughness better than 5 A over areas of
tens of microns squared. Mica cleaves very easily along the (001) plane. Along this
direction of the crystal are alternating layers of aluminosilicate and potassium ions. The
(001) plane resulting from the cleavage process consists of a layer of oxygen atoms. This
exposed plane is quite hydrophilic and contains a significant negative surface charge, in
part due to the loss of some endogenous potassium counter ions during the cleavage
process. Treating a freshly cleaved surface of mica with a 5 mM MgC12 solution results

in the replacement of some of the K+ ions with magnesium, thereby reducing the negative

43

surface charge. The magnesium treated surface has been demonstrated to enhance
adsorption of many macromolecules [2]. It is believed that the divalent Mg ions contain
sufficient charge density and size to adequately replace the lost potassium ions [3]. We
have examined the effect of the magnesium treatment on Con A adsorption, and found no

significant changes. These results will be presented in chapter four.

Proteins were dissolved in 10 mM buffer solution and diluted to the working
concentration just prior to application of the sample to the substrate. Twenty five
microliters of protein solution is deposited onto the mica. Twenty minutes later, the
surface is gently rinsed twice with 200 pl portions of MilliQ water and allowed to dry for
an hour and a half to 8 hours at 21°C and 35-50% relative humidity. At sufficiently low
protein concentrations (~10-400 ng/ml for Con A) this preparation yields a dispersed
population of individual protein molecules adsorbed on the magnesium treated mica
surface (Figure 14). The concentration of protein, in each set of experiments has been
chosen to provide a large population of individual protein molecules and to avoid the
seeding of aggregates and monolayers. Overall, the protein coverage for SOMS

measurements is always kept below 1%.

44

 

Figure 14: AFM images of Con A molecules on mica, shown in standard
and three-dimensional renderings.

45

The concentration of stock solutions prepared from the lyophilized Con A
(Boehringer-Mannheim) was found to vary, even when the same mass of lyophilisate and
volume of buffer were used. It was not unusual to observe up to an order of magnitude
difference from preparation to preparation. It is believed that this is due to an
inhomogeneous mixing of the Con A with the residual sodium chloride during the freeze-
drying process. This problem was avoided by careful measurement of concentration by
A230 and by homogenization of the lyophilisate with a spatula prior to weighing out of

sample.

In Chapter 5 data examining the orientation and oligomerization state of Con A
molecules as they aggregate to form monolayers on a surface, will be presented. The

experimental details for these experiments will be discussed in chapter 5.

3.3 AFM imaging

Height measurements may be obtained from AF M studies carried out by contact
mode or intermittent contact mode in ambient or under liquid conditions. For this study
the AF M data were recorded in contact mode under controlled ambient conditions (~35-
50% humidity and 21°C). Under these conditions a monolayer of water exists at the mica
surface [4] and the protein molecules are expected to remain hydrated. This limited
hydration is expected to stabilized the adsorbed protein molecules, and will be discussed
further in chapter four. It was found that the protein molecules adhered to the mica well

enough for measurements only when the humidity was less than ~50%. At higher

46

humidities, multilayers of water at the surface will increase the total attractive capillary
forces between probe and sample [4-7]. Humidities of less than 30% were avoided for
two reasons. First, at much lower humidity the protein may become dehydrated, leading
to possible structural deformations. Also, at humidities below 30%, static repulsive
forces between mica and the probe tip become large enough that the long range forces
dominate the tip-sample interactions. The influence of protein hydration on structure has
been previously studied by scanning tunneling microscopy (STM) [8,9]. The largest
structural changes were found to occur as the humidity dropped below 30%.
Unfortunately those measurements were not able to provide quantitative structural
information because the contrast mechanism (tunneling current) is highly dependent on

the amount of solvent present.

For the measurements presented, sharpened Microlever cantilevers (Park
Scientific Instruments) having a spring constant of approximately 0.05 N/m and an
integrated silicon nitride probe with a radius of curvature of approximately 20 nm,
according to the manufacturer, were used as probes (see Figure 2). The piezoelectric
scanner was calibrated for vertical measurements using Tobacco Mosaic Virus (TMV)
and/or colloidal gold particles. Horizontal measurements were calibrated using a two-

dimensional grating with one um periodicity in x and y.
AFM images were acquired in constant force mode. In this mode of imaging, the

computer control includes a feedback loop that measures the cantilever deflection

(proportional to the A-B/A+B signal from the position sensitive photodiode) and moves

47

the scanner up or down in the z direction, so as to maintain a constant applied tip-sample
force. The minimum applied force required for stable imaging was found to occur within
the electrostatic repulsive regime and was constant for all images. This net force was
~0.l nN (according to the ProScan 1.5 software from Park Scientific Instruments), which
may vary by as much as a factor of two from cantilever to cantilever due variability in the
manufacturing process. The total force applied between the tip and the cantilever under
our imaging conditions, also contains a large attractive or capillary force (~lO'9 - 10'6 N,
depending on the ambient humidity [10]) in addition to the overall repulsive force. The
effect of the tip-sample applied force on the AF M measurements for both ferritin and Con

A will be presented in chapter four.

The scan size for all the data scans presented, was 2.0 x 2.0 pm and 512 x 512
pixels, which was imaged at a linear scan rate of 1.5 Hz (3 urn/s along the fast-scan
direction). This relatively slow scan rate was chosen in order to minimize the lateral
forces which occur between the tip and sample during the raster scanning process. The
lateral forces might be further minimized through the use of intermittent or non-contact

modes of imaging, or by imaging under liquid.

3.4 Data Analysis

The raw images from the microscope contain line to line offset variation in

contrast due to the scaling used by the ProScan data acquisition software (Park Scientific

Instruments) and instrumental drift. Often there is also a slope to the image background.

48

This occurs when a sample is not normal to the x-y plane of the scanner (see Figure 15
A). In addition, there may also be a parabolic curvature to the image data along the scan
direction. This artifact arises, in large part, due to rocking motion of the scanner tube
during its lateral motion (see Figure 15 B). Each of these artifacts can be compensated
for by using a 1St or 2“d order polynomial subtraction made to each scan line of the image.
This background correction is performed using Pro Scan image analysis software from
Park Scientific Instruments. The net effect of this process is that the average background

level of each scan line is brought to the same value.

49

Height (A)

 

First order flattening

 

 

 

 

 

 

 

 

 

 

 

Before
14 r
8
12 '
10 ‘ 6 ‘
8 ., ' :5 I F“ i
6 ‘ .304 1
a l

4 ' ‘ I 2 I I l
2 .
0 I I I 0 r I r I

0 1.0 2.0 0 1.0 2.0

Distance ( p. m) Distance ( u m)

Figure 15 A: Example of image requiring first order flattening

50

Height (A)

Second order flattening

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Before H After

20 4A 14‘

‘5 Ill 1 1::

W1 1 i ii {28.

10 - 1m] l I] ‘ w 3., 6

h 1]“),(1. w. ,

. ‘ ' ‘. ill 1 = 4.

5 WWW“ l“ 2.
0 Distanlc: ( Ll m) 2.0

 

Rocking lateral motion of scanner tube

 

 

 

Figure 15 B: Example of image requiring second order flattening

Each image contained approximately 10-50 molecules per square micron. The
heights of individual protein molecules are measured by taking the maximum height of
the molecule minus height of the local background. A custom data analysis program was
designed in order to automate these height measurements. Two versions of this program
were used for the work of this dissertation. Both of the programs made use of the same
measurement algorithms. The original program was written in Quick Basic 4.0 for DOS
by Dr. Neil Bowlby, with Marcos Dantus and myself. The current version of the
software was written by George Schoendorff and myself in Visual Basic 5.0 for Windows
and made use of a subroutine (supplied by Park Scientific Instruments) which opens the
hierarchical data format (HDF) image files. The Visual Basic source code for the current

version of the program may be found in Appendix 2.

The HDF analysis software first loads the image into memory and prepares a
histogram of the height values for each pixel in the image. If the sample has < 1%
surface coverage (as is the case for the work presented in chapter four), then the most
frequent pixel height value is a good first approximation to the average background level
of the image. This value is used to determine the threshold value to be used for finding
the local maxima contained in the image. The smallest particles that could be accurately
measured were between 6 and 9 angstroms in height, so in general the threshold value
used for analysis was the most frequent pixel height plus 6 to 9 angstroms. Local
maxima may be confined to be within a given area by the specification of a radius value.
For the analysis of the images in this work, I used a radius of 12 pixels. Twelve pixels in

x and y corresponds to a distance of 47 nm, or more than twice the nominal radius of

52

curvature of the probe tip. In order to eliminate large protein aggregates from

consideration, the local maxima search algorithm may be given a cutoff value as well.

The local maxima search finds not only isolated particles, but also anomalous
features of the background and large aggregates. In order to eliminate these from the data
set an interactive peak selection routine has been built in. In this interactive mode, a
magnified gray scale image of the area around each local maximum (:12 pixels in x and
y) is displayed one at a time. Determination of the height of the maximum, relative to the
local background requires a reproducible measurement of the local background. To this
end, a histogram of pixel heights is calculated based on the pixels within 40 pixels along
the y direction and 1 pixel along the x direction. Preference towards the y scan direction
is given since all the images for this work were acquired with the fast scan direction
along the y axis of the scanner. The scanner noise and drift are minimized along the scan
direction since those points were accumulated within a short time of one another. The
most frequent pixel value for the region is taken to be a good estimate of the local
background. Finally, the measured height of the particle of interest is calculated by
taking the maximum height of the particle minus the most frequent local pixel height.
Each maximum may be accepted or rejected on a case-by-case basis, by the program
user. In the work of this dissertation, particles were rejected if they were of irregular
shape (i.e. out-of-round), if the local histogram was anomalous (i.e. non Gaussian,
indicating significant noise) or if a particle is much smaller or larger laterally than other
similar particles in the image. Using these criteria, particles that are composed of large

aggregates or contain anomalously high or low maxima are eliminated.

53

Each data set, generally consisting of more than a thousand individual height
measurements, is tabulated in 3 A bins to produce a height distribution histogram. The
choice of bin size is based primarily on the substrate roughness (which is generally 3-5
A) and is a reasonable, unbiased method of ‘smoothing’ the data. Within the 5 A margin
of error claimed, the results are independent of this bin size choice. A combination of

Gaussian functions is then used to fit the height distribution data.

A x-x 2
y: exp[-2( C] :l (11)
w 7% w

This fitting was done iteratively using constraints on the three parameters: center position

 

 

xc, width w and amplitude A. In all cases, the amplitudes of the functions were allowed
to vary freely. The center positions of the functions were allowed to vary independently.
The widths of the functions were constrained to be between 8 and 13 A. In the case of
the Con A analysis, the functions centered at 59 A (having the lowest amplitude), have
the greatest uncertainty and their positions were fixed while the other parameters were
optimized. The centers of the Gaussian functions are taken as the measured molecular
dimensions, while the area under the curves are proportional to the probability with
which a given molecular orientation is measured (see Chapter 2). Consistency in the
fitting parameters was checked by multiple repetitions of the experiments on different
days. The results presented here conform to this strict criterion, with variations in

measured molecular dimensions of less than 5 A.

54

The final analysis consists of two different procedures: a.) determination of
molecular dimensions, and b.) assignment of quaternary structure (when different
oligomerization states are possible). In order to assign the observed dimensions to the
macromolecule, a model which takes into account the orientations of molecules at
interfaces is required (see Chapter 2). Protein dimensions obtained by this method may
be compared, for consistency, with other biophysical data available for the protein under
investigation, such as molecular weight. For example, the volume of each protein may be
estimated from the dimensions as measured by SOMS. Using the molecular weight of
the protein under study, a volume, based on a density of ~1 g/cm3, may then be
calculated. In the case of the Con A results of this dissertation, the data are shown to be

consistent with the structures determined by x-ray diffraction (XRD).

55

3.5 References

1. Yariv, J .; Kalb, A. J.; Levitzski, A. Biochim. Biophys. Acta 1968, 165, 303.

2. Apell, H. J .; Colchero, J .; Linder, A.; Marti, O. Investigation of the Na, K-ATPase by
SFM. In ST M and AFM in Biology; Marti, O., Amrein, M., Eds.; Academic Press, Inc.:
San Diego, 1993; pp 282.

3. Vesenka, J .; Guthold, M.; Tang, C. L.; Keller, D.; Delaine, E.; Bustamante, C.
Ultramicroscopy 1992, 42-44, 1243.

4. Hu, J .; Xiao, X.-D.; Ogletree, D. F .; Salmeron, M. Science 1995, 268, 267.

5. Bustamante, C.; Vesenka, J.; Tang, C. L.; Rees, W.; Guthold, M.; Keller, R.
Biochemistry 1992, 31, 22.

6. Edstrom, R. D.; Meinke, M. H.; Yang, R.; Elings, V.; Evans, D. F. Biophys. J. 1990,
58, 1437.

7. Thundat, T.; Zheng, X.-Y.; Chen, G. Y.; Warmack, R. J. Surf Sci. Lett. 1993, 294,
L939.

8. Leggett, G. J.; Davies, M. C.; Jackson, D. E.; Roberts, C. J .; Tendler, S. J. B.;
Williams, P. M. J. Phys. Chem. 1993, 97, 8852.

9. Parker, M. C.; Davies, M. C.; Tendler, S. J. B. J. Phys. Chem. 1995, 99, 16155.

10. Bustamante, C.; Rivetti, C. Ann. Rev. Biophys. Biomol. Struct. 1996, 25, 395.

56

Chapter 4: Application of SOMS

This chapter presents measurements of colloidal gold, fenitin and Con A at the
water/mica interface. These samples will be used to show that the SOMS technique is
capable of providing valuable new information about the structure, oligomerization and
orientation of single macromolecules at interfaces. The data obtained from this technique
will also be shown to provide important information regarding the properties of Con A at

the liquid/solid interface

4.1 Analysis of gold and ferritin at the liquid/solid interface

4.1.1 Measurement of colloidal gold at the water/mica interface

One of the concerns with using the AFM to measure the heights of protein
molecules at the liquid/solid interface is that the force applied during imaging might
distort the structure of the protein molecule. While this issue will be examined later in
this chapter, initially, however a proof of principle for the height measurement accuracy
using an incompressible standard will be presented. Colloidal gold has been shown to be
a useful, and incompressible standard for AFM measurements [1]. In addition, since
colloidal gold is spherical one should observe only one orientation for each particle size.
For these reasons, 5, 9 and 18 nm gold particles were measured as a proof of principle for
the SOMS approach. The height distribution for the three sizes of colloidal gold on mica
is plotted in Figure 16. Figure 16 also gives a tabulation of heights and standard

deviations measured by the manufacturer (using TEM) and by the AF M. Note that for

57

the data in this figure the microscope was calibrated based on measurements of the 5 nm
gold colloid and Tobacco Mosaic virus (cylinder with 18 nm diameter). The
manufacturer data is based on the measurement of 100 particles, while the AFM

measurements are based on measuring the number of particles given in Figure 16.

58

 

 

 

 

 

 

 

 

50 - -

m 40 . ~

.2 i 1
.2

t 30 - _
e:
a.
Hut

o 20 _ _
b
E

g 10 - .
Z

0 _ _

0 50 100 150 200 250
Height (A)
TEM AFM Particles
5.0 d: 0.8 5.0 i 0.8 353

9.0 $1.2 10.1 i 1.6 207
18.0 :t 0.8 19.4 i 2.0 87

 

 

 

 

 

measurements are in units of nm

Figure 16: Height distribution measured by AFM for three different sizes of colloidal
gold on mica.

59

It is clear that the agreement between the AF M and TEM measurements is quite
good, but the AF M measurements are overestimating the heights of the larger particles.
The AFM measurements also show a greater standard deviation for the larger particles.
There are a few lines of reasoning that could explain these discrepancies. First, the
calibration of the scanner was probably off by a small amount. Given the results of this
study the microscope was re-calibrated using the data to determine a three-point
calibration. Second, the number of 18 nm particles measured by the AF M was much less
than for the other two sizes. The measurement of fewer particles for the larger colloids
was due to the fact that the adsorption of the particles was found to vary inversely with
the size of the particle. Finally, it has been observed that larger colloidal particles
sometimes vary from the ideal spherical shape. Because the larger particles adhere
weakly to the mica and may posses some ellipticity, it might be expected that the AF M
could have more trouble with their measurement. The agreement of the AF M and TEM
data suggests that the error in the measurement is mostly in the size distribution of the

particles themselves, and not as a result of the measurement technique.

Vesenka et al. have suggested that colloidal gold makes a good incompressible
standard for SPM measurements [1]. This assertion was tested using the AFM and 5 nm
colloidal gold samples. As a means of measuring this compressibility samples were
imaged at different forces and the heights for individual gold particles on the surface
were measured. The data was then plotted as the height measured for a given force,
referenced to the height measured for the minimum applied force. The data for the first

trial of this experiment is presented in Figure 17 A, while a representative image from

60

this data set is given in Figure 17 B. It is readily seen that there is 10% compression
going from a force of 0.05 nN to 0.8 nN, which is quite unexpected. Vesenka et a1.
observed no significant compression of 5-20 nm gold for forces as high as 225 nN [1].
The experiment was repeated the following day with significantly different results. The
results of this second trial are given in Figure 18 A and B. Note that the variation of the
measured height is much less pronounced. The reason for this inconsistency in the data is
found by visual inspection of the images from each trial. The particles observed in
Figure 17 B are broader (laterally) than those of Figure 18 B. The reason for this
difference is a good example of a tip convolution artifact. A new, unused probe tip was
used for the second trial, while a previously used tip was used during trial one. It is likely
that the older tip had some adsorbed material stuck to it, which caused the interaction
region of the tip to be larger. Further, it appears that this adsorbed material is probably
protein material. It is likely that the compressibility behavior observed in the first

experiment (see Figure 17) is characteristic of the adsorbed protein and not the gold.

61

 

l J 1
l

1.04
21.02 . i
1.00 5
0.98 — -
0.96
0.94 — -
0.92
D 0.90 ‘ j

J

0.88 i *

pressibility (H /H

0m

 

 

 

0.0 0.2 0.4 0.6 0.8 1.0
Force (nN)

 

Figure 17: A) Compressibility of 5 nm colloidal gold on mica as a function of force.
B) AFM image from this data set.

62

 

1.04 ‘_
31.02 ‘ '—
5 1.00 —- .. “ -- "
E098 5 ..
£0.96 9 _ n ..
E 0.94 j -—
§0.92 : ..
o 0.90 j --

0.88 -

 

 

 

 

 

 

 

r l ' I ' T ' I ' I

0.0 0.2 0.4 0.6 0.8 1.0
Force (nN)

 

Figure 18: A) Plot of compressibility as a fimction of force for 5 nm colloidal
gold using a new probe tip. B) AFM image from this data set.

63

 

4.1.2 SOMS analysis of ferritin at the water/mica interface

The SOMS approach for the determination of macromolecular dimensions was
first tested using the protein ferritin (cationized, horse spleen). Ferritin is an important
iron storage protein that is ubiquitous in fungi, plants and animals [2]. It has been widely
studied and characterized by a variety of techniques, including electron microscopy and
XRD [2,3]. The basic structure of this protein consists of an apoferritin shell composed
of 24 polypeptide chains organized as a 120 A diameter, hollow sphere containing six 10
A channels that allow the exchange of iron (stored as ferric oxyhydroxide) [2].
Apoferritin has a molecular weight of 450 kD, but when its core is filled with the
maximum number of iron ions (4500) it has a mass of 900 kD. The large size and high
iron content of this protein has allowed the use of electron microscopy for its localization
within cells and for the determination of structural characteristics. The compact structure
and rigidity of the apoferritin shell makes fenitin structure quite resistant to denaturation
[4]. In fact, as part of the processing of commercial ferritin it experiences temperatures
of 80°C [4]. Feder and Giaver have studied the adsorption of ferritin to surfaces using
electron microscopy [4]. The tertiary and quaternary structure of ferritin has been

recently reported by Yau and Zhou using the scanning tunneling microscope (STM) [5].

Since ferritin is large, robust, stable and exhibits multiple oligomeric
organizations, it is an ideal test of the SOMS technique. A 20 rig/ml solution of
cationized horse spleen ferritin was applied to a magnesium treated piece of mica, as

described in Chapter 3. The height distribution obtained from the AFM measurements is

64

given in Figure 19. Three Gaussian populations, centered at 100 A, 116 A and 135 A,
were fit to the data. The dominant population at 116 A (62% of data) is attributed to the
presence of intact ferritin molecules, which have been observed by electron and scanning
tunneling microscopies to be 120 A in diameter [3,5]. The population at 135 A (21 % of
data) is consistent with the presence of ferritin trimer, which was reported by Yau and
Zhou to have a 130 A by 380 A size. The presence of the population centered at 100 A
(17% of data) may be partially denatured or dehydrated ferritin molecules. It should be
noted, however, that there have been reports of ferritin structures, measured by electron
microscopy, between 75 and 125 A. In their studies, Feder and Giaever measured the
diameter of ferritin adsorbed onto a carbon surface, to be 101i8 A [4]. The measurement
of a 100 A population in the current studies is, therefore consistent with previous
observations of ferritin. The study done by F eder and Giaever was done using an
electron microscope, so the ferritin they observed was dehydrated. The population at 100
A in the current data may very well represent a subset of the molecules which may have

dehydrated to a greater extent when adsorbed to the mica surface.

65

Number of molecules

 

 

I I r r ' 1 1
50 -
40 '1 ,
i ‘1
-1 . f? 3;
i
30 T if: 1
l 1. °
:3: I
if e
20 j
. i
O! .
10 - .- e e
{f
0 ~ : - ~ -
l I '

 

120 140
Height (A)

 

Figure 19: Height distribution measured for ferritin molecules at the water/
mica interface.

66

The results for ferritin presented to this point are consistent with previous work on
the ferritin system, using STM, TEM and XRD. These data are also consistent with the
proposal of the SOMS technique and the SAC model. Although the intact monomer of
ferritin is spherical, the trimer has been observed to be non-spherical with dimensions of
130 A by 380 A [5]. The data of this section show that the trimer orients such that its
shortest dimension projects from the substrate surface, which is consistent with the SAC

model.

It has been predicted that the compact and rigid structure of the apoferritin core
should make ferritin quite stable at the liquid/solid interface [4]. The above results
indicate that the native structure of ferritin is conserved upon adsorption at the
water/mica interface. The 24 polypeptide chains which make up the apoferritin shell are
composed of 4 or-helical structural units between 35 and 42 A long as well as a fifth short
helix, that lies along a channel of the apoferritin shell [2]. Gekko and Hasegawa have
examined the compressibility of globular proteins as a function of structural
characteristics [6]. They found that proteins containing a significant fraction of or-helical
character were in general more compressible. Although the primary structural
characteristic of the ferritin polypeptides are or-helices, ferritin should be fairly
incompressible due to its compact structure. This hypothesis was tested by measuring the
height distribution of ferritin as a function of imaging force. The results of this study are
presented in Figure 20. As in the gold compressibility study, the results are plotted as the
ratio of the height measured compared to the height measured at the-minimum imaging

force, as a function of imaging force. The ferritin molecules are seen to undergo

67

compression of about 10% in going from an imaging force of 0.05 nN to 1.2 nN. This
leads to an error of about 10-12 A when using a force that’s an order of magnitude
greater than that used for the SOMS experiments of this dissertation. For a structurally
robust protein like ferritin it is clear that contact mode AFM at ambient conditions is
capable of making very precise measurements of structure, as long as the applied force is

kept to a minimum.

 

1.04 J
1.02 . -

1.00 - E] 3
0.98 i i
0.96 i

0.94 4

0.92 3 l "'
0.90 .1
0.88 .. -

I

0.0 0.

4-

1

A l A l I l l

 

 

 

 

 

2 0.4 0.6 0.8 1.0 {.2
Force (nN)

Figure 20: Plot of compressibility as a function of force for ferritin molecules
adsorbed at the water/mica interface.
The SOMS results presented for colloidal gold and ferritin have been shown to be
consistent with previous studies of these systems. These results indicate that the height
measurements of small particles and macromolecules provide a usefirl way to obtain

structural information with the AF M. While the SOMS approach avoids the probe tip

68

convolution problem that reduces the lateral resolution of the AF M for measurements of
these systems, one must also be careful to watch for changes in the average lateral
dimensions of particles that may indicate changes in the probe tip characteristics. The
height measurements made by the AF M appear to be unaffected by the imaging force
near the minimum imaging force required for contact mode imaging of these systems.
The SOMS approach allowed the assignment of different oligomeric species of ferritin
molecules adsorbed at the water/mica interface. While different oligomerization states
were observed for ferritin, the monomeric units of the protein are spherical in shape.
This made ferritin a good first test, but doesn’t provide a very stringent test of the SOMS
technique. The measurement of a protein with an ellipsoidal shape will be investigated in

the next section.

4.2 Analysis of Concanavalin A (Con A) at the liquid/solid interface.

4.2.1 Background information for Con A

Concanavalin A (Con A) is a lectin, obtained from the Jackbean (Canavalia
ensiformis), that binds mannose, glucose, and glycoconjugates containing these
saccharides [7-9]. The strength of the interaction between Con A and saccharides has
been likened to approach that of an antibody interaction, which has led to interest in the
use of Con A adsorbed to surfaces for use in glucose sensor devices [10]. The native
protein exists as a tetramer, composed of four identical subunits with a molecular weight

of 25,500 kDa each [9]. In solutions at pH 27.0, Con A is normally observed as a

69

tetramer [9]. At pH 55.5, however the dimeric form is favored [11]. Gordon and Young
have reported the dissociation of Con A into dimers as a consequence of dilution and
ionic strength at a concentration 55 rig/ml and a pH 27.0 [12]. According to x-ray
diffraction studies, the molecular dimensions of the dimer are approximately 30 A x 45 A
x 75 A, while those of the tetramer are approximately 60 A x 70 A x 70 A (Protein

DataBank, 1CN1, [13]) (see Figure 21).

70

 

 

 

 

 

 

75A

 

Figure 21: Ribbon rendering of the Con A dimer, based on XRD structure of
Shoham et a1. [BNL Protein DataBank, 1CN1]

71

 

4.2.2 SOMS analysis of Con A at the water/mica interface

Preparations of Con A solutions with concentrations in the range of ~10-400
ng/ml yielded dispersed individual protein molecules upon adsorption on magnesium
treated mica surfaces, as described in the experimental section (see Chapter 3, Figure 14,
for example). Analysis of several such images results in the height distribution shown in
Figure 22. This distribution is the result of measuring the heights of 2000 individual
protein molecules. The height distribution is best fit by four Gaussian firnctions centered
at 18, 28, 42 and 59 A (see Table 11). Inspection of the XRD structure of the Con A
dimer (see Figure 21), indicates that the observed height distribution corresponds to
individual Con A dimers. As predicted by the SAO model the most prevalent orientation
is the one in which the largest surface area is in contact with the mica substrate,
projecting the 30 A dimension perpendicular to the surface. The 28 and 42 A dimensions
are in excellent agreement with the x-ray structure (see Figure 21). The 59 A
measurement is significantly shorter than the 75 A expected for Con A dimers projecting
their longest axis normal to the mica surface, but it is quite close to the 60 A dimension
of the tetramer. The assignment of this population of molecules to the dimer will be
verified later in this chapter by measurements made on succinylated Con A. Results to be
presented later in this section will also show that the presence of the 18 A population,

which varies from ~8-3 5% of the total population, can be attributed to denatured Con A.

72

Number of molecules

 

 

 

 

 

 

 

 

400— —
300- .
200- -
100- -
1
0 I l ' l ' d
40 60
Height (A)

Figure 22: Height distribution measured for Con A molecules at the water/
mica interface.

73

80

Table II: SOMS data obtained for Con A at the water/mica interface.

 

Height (A) Width (A) Areaa Probability (%)°

 

* 18 10 489

a 28 10 5079 91.8
b 42 10 409 7.4
c 59 10 45 0.8

AG.) = -9.4 i 2.0 J/mol-A2

8”These measurements correspond to the area under the Gaussian function and are

proportional to the population of molecules with the given height . bThe probability

excludes the particles found to have a dimension of 18-19 A, which are determined to
arise from the presence of denatured protein molecules. The values of AG; given in

Tables 11 and III are the average of three measurements, using equations 7-9 (see Chapter

2).

As in the case of ferritin, there was concern that the imaging forces applied to the
protein molecules might cause distortion of their structure. Experiments were performed
at the minimum imaging force, nominally 0.1 nN, and at forces as high as 1 and 2 nN.
Samples were first imaged at the minimum force for stable contact mode imaging, then
the force was increased and the sample was re-imaged. The results of this study are

presented in Table III.

74

Table III: SOMS data for Con A at the water/mica interface as a function of imaging
force.

0.1 nN 1068 pts.

 

 

 

 

center width Area % of total rel. % a
18 9 932 28.5
28 I2 2094 89.5
45 9 215 9.2
59 12 30 1.3

 

 

 

 

 

 

 

AGo=-8.6 J/mol A2

   

1.5 nN 414 pts. Summary of the three trials
center Area 0 center
18 9 449 :1:
1
12 5 . 59 1
AGo=-10.l J AGo=-9.4 :1: 0.7 J/mol
2.0 nN 432 pts.

center

 

1

AGo=-9.5 J/mol
aRelative percent is the area under the curve for the population of one molecular

orientation relative to the population of all orientations a S b S c

75

The height distributions show no statistically significant differences when data is
acquired at an imaging force between 0.1 nN and 2 nN. This result is consistent with the
work of Gekko and Hasegawa. In their study of the structural factors which influence
compressibility for globular proteins they found a general trend that compressibility
scales with the amount of (rt-helical character of the protein [6]. They observed that
proteins that contain B-sheets tend to be less compressible than those containing primarily
or-helices. The 13 —structure of Con A is perhaps its most apparent feature when
examining the XRD structures. The monomer unit consists primarily of two antiparallel
sheets that together account for over half the amino acids in the sequence [9]. In
addition, Gekko and Hasegawa found a correlation between the presence of certain
residues and an increased or decreased compressibility [6]. A low compressibility was
associated with the Asparagine, Glycine, Serine and Threonine, while a higher
compressibility was associated with the presence of Leucine, Glutamic acid,
Phenylalanine and Histidine [6]. Analysis of the sequence of Con A finds that 33 % of
the amino acids are Asparagine, Glycine, Serine or Threonine, while 18 % are of
Leucine, Glutamic acid, Phenylalanine or Histidine (see Table IV). This result further

indicates that Con A is not expected to exhibit a large compressibility.

76

Table IV: Analysis of the composition of the Con A sequence and the potential influence

on the compressibility of Con A based on Gekko & Hasegawa [6].

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Residue Number % of total

ALA 19 8

ARG 6 2 Decrease Number % of total
Compressibility

ASN 12 5 ASN 12 5

ASP 20 8 GLY 16 7

GLN 5 2 SER 31 13

GLU 7 3 THR 19 8

GLY 16 7 sum 33

HIS 6 2

ILE 15 6

LEU l8 8 Increase Number % of total
Compressibility

LYS 12 5 GLU 7 3

MET 2 1 LEU 18 8

PHE 11 5 HIS 6 2

PRO 11 5 PHE ll 5

SER 31 13 sum 18

THR 19 8

TRP 4 2

TYR 7 3

VAL 16 7

 

 

 

 

77

In order to verify the presence of dimers at the interface, and to rule out the
possibility that Con A was denaturing at the mica surface, two experiments were carried
out. The first involves imaging of succinylated Con A, which is known to exist only as a
dimer, over a wide range of pH [14]. The second experiment was to determine the height

distribution for Con A which has been chemically denatured.

Results from the succinylated Con A samples (1576 individual height
measurements) show a very similar height distribution to that obtained from Con A (see
Figure 23 and Table V). The three main populations in the height distribution are
centered at 29, 45 and 59 A, corresponding closely (within 3 A) to those found for Con
A. These results give further evidence that Con A exists as a dimer and not a tetramer at
the water/mica interface, under these conditions. This result is consistent with the
observations of Gordon and Young, that Con A at a concentration of <5 rig/ml dissociates
to form the dimeric species [12]. This experiment allows an unambiguous assignment of
the oligomeric structure of Con A at the interface. The 59 A population may be
attributed to the dimer oriented such that side ab is in contact with the surface, projecting
c=75 A away from the surface. This dimension is not projected at 0° inclination,
however, which can be reasoned by inspection of the XRD structure of Figure 21. Since
the ends of the molecule are quite rounded, one might expect some angle of inclination
relative to the surface. A simple calculation indicates an inclination of about 38° from

normal.

78

Number of molecules

 

1.60
140-
120?
100;

1

-§ 0\ OO
O O O
1 . 1 1 r

N
O
1

 

‘‘‘‘‘
‘b

 

 

Height (A)

80

Figure 23: Height distribution measured for Succinyl Con A molecules at the

water/mica interface.

79

Table V: SOMS data for succinyl Con A at the water mica interface

 

Height (A) Width (A) Area‘il Probability (%)°

 

* 19 9 1056

a 29 13 1597 87.7
b 45 8 164 9.0
c 59 13 59 3.2

 

AGo = ~6.2 i 0.2 J/mol-Az

a’ bSee notes in Table II

The results of this experiment support the assertion that the SOMS approach has
the capability and sensitivity to measure changes in the oligomerization state of proteins
at the liquid/solid interface. There is a further population of molecules centered at 19 A,
which is similar to that observed at 18 A for Con A, but in this case accounting for 37%
of the total molecular orientations measured. The assignment of this population will be

made in the following experiment.

To investigate the possibility of denaturation of Con A at the water/mica interface
Con A was treated with 2% sodium dodecyl sulfate (SDS) for 2-3 min. at 90-95°C prior
to dilution and preparation of samples. As discussed in Chapter one, the prevailing
model of protein adsorption involves a denaturation step after the initial adsorption [15].
The height distribution from the SDS treated Con A samples, which is the result of 1983
individual height measurements, is shown in Figure 24 (see also Table VI). Note that the
major population for this sample is centered at 18 A, leading to the assignment of the 18-
19 A population to denatured Con A. These results indicate that ~60-70% of the protein

molecules were denatured when treated with SDS at 90-95°C. The rest of the distribution

80

shows populations at 29 A, 46 A and 59 A in excellent agreement with the previous two
sets of data (within 3 A). This indicates that some of the Con A molecules were not
denatured, or possibly underwent refolding upon dilution of the SDS during sample
preparation. This result allows us to assign the peak at 19 A in the succinylated Con A
samples to the presence of ~37% denatured protein in the sample. The degree of
denaturation for the succinylated Con A is slightly higher than the typical ~30%
denaturation observed for Con A on mica. This may be a result of the succinylation

process or improper handling during production.

81

Number of molecules

 

 

—1 i—I N N
O U! O Ur
O O O O
A 1 1 I

 

(II
0
m

L

 

 

O
1

 

 

 

20 40 60 80
Height (A)

Figure 24: Height distribution measured for SDS treated Con A molecules at
the water/mica interface.

82

Table VI: SOMS data for SDS denatured Con A at the water/mica interface

 

Height (A) Witdh (A) Areaa Probability CW)"

 

 

* 18 10 3679

a 29 13 1938 81.2
46 12 350 14.7

c 59 12 100 4.2

8‘ °See notes in Table 11

Further analysis of the results by the SAO model can be used to estimate the free
energy of orientation per unit area, AGO, for Con A and succinylated Con A on Mg”2
treated mica. Using equations 7-9 (see Chapter 2) and values Greight and probability)
from Table II, we calculate average values of -9.4i2.0 J/mol- A2 and -6.2i0.2 J/mol- A2,
for Con A and succinylated Con A, respectively. The quoted uncertainty is the standard
deviation of the three values calculated from equations 7-9. Day to day variability of as
much as 40 % in the calculated value of AGO has been observed. Much of this derives
from the fact that the 42-45 A or 59 A populations are quite small relative to the 28-30 A
population. Because of the exponential dependence in equations 7-9, a small difference
in the 59 A population can make a large difference in the calculated AG. This variability

in the calculated free energy will be discussed again later in this chapter.

The approximate surface area of each face of the protein can be calculated from
the measured dimensions. With these values the free energy of orientation for each of the
three different orientations is calculated (see Table II). The most stable of these being ~ -

20 kJ/mol when the Con A molecules have their largest surface oriented towards the

83

mica. These values are larger than kT (2.5 kJ/mol), indicating this orientation is
relatively stable. Differences in the free energy for each orientation determine the
observed height distribution. The values for Con A and succinylated Con A are similar,
as expected, however, I believe that there may be some small difference related to the
chemical modification of amino acid side chains by the succinylation process.
Succinylation involves the modification of amines in the protein with a concomitant
decrease in positive charge. For Con A, an average of 10 sites per subunit are

succinylated [14], which may cause changes in the interaction with the mica surface.

In conclusion, angstrom resolution measurements from individual protein
molecules at the liquid/solid interface have been observed by a new approach to AF M.
The measurement of populations centered at 28, 42 and 59 A, indicate that Con A
adsorbs onto mica substrates as a dimer. These molecules tend to orient such that the
smallest dimension is projected normal to the surface. This observation is consistent with
the SAO model proposed in Chapter 2. These measurements are in very good agreement
with the x-ray structures determined by Weisgerber et a1. and Shoham et al. [13,16]. The
height distribution observed for Con A at the water/mica interface was shown to be
unaffected by imaging forces between 0.1 and 2.0 nN. Although the tetrarneric form of
Con A is normally observed in solution, the solutions from which samples were prepared
were ~10-400 ng/ml at a pH=7.3, which according to Gordon and Young are expected to
contain primarily the dimeric species [12]. This dimeric state is conserved upon
adsorption of the Con A molecules to the mica surface. The presence of the dimeric

oligomerization state was further confirmed with measurements of succinylated Con A.

84

Application of the SAO model to these measurements gives approximate values for the
free energy of orientation for these molecules. The possibility of protein denaturation
caused by the adsorption process was addressed by measurements of Con A treated by

the detergent SDS. This question will be examined more closely in the following section.

4.2.3 Examination of Con A denaturation at the water/mica interface

It was shown in the previous section that denatured Con A at the water/mica
interface is indicated by the presence of a population of molecules with a height of 18-19
A. The treatment of Con A with SDS and heat was chosen for the previous study
primarily because it is known to effectively denature most protein systems. Additionally
SDS is known to strongly bind to most polypeptides, and the thinking was that upon
dilution of the sample in buffer much of the SDS-protein interactions might remain. In
this way we sought to try to prevent refolding of the Con A before or possibly during the
adsorption process. In order to test this idea a series of experiments were performed to

further examine the denaturation of Con A at the liquid/solid interface.

Comparison of the degree of denaturation typically observed for Con A or
succinyl Con A with that observed for Con A treated with SDS at 90-95°C, shows a
factor of two or more in the population of molecules at 18-19 A. Although significant,
this result was not as dramatic as anticipated. One possibility is that upon dilution of the

denatured stock solution or during the actual adsorption process, some of the Con A

85

molecules may undergo refolding. Con A samples which were denatured by various

methods were prepared and measured in order to test this hypothesis.

The first of these experiments involved treating the Con A with 6 M urea. It was
expected that if refolding were occurring, it might be more likely to happen with urea,
which doesn’t bind to the polypeptide backbone as strongly as SDS. The height
distributions obtained were quite similar to that of the SDS treated Con A (see Figure 25).
The amount of denatured Con A, as determined by taking the area of the population
centered at 17-18 A compared to the total area, was found to be 43-44%. These numbers
are based on 2 sets of data run on two different days, containing 435 and 1220 individual
molecular measurements. Clearly the amount of denaturation observed for the urea
treated Con A is higher than untreated Con A, but much less than SDS treated protein.
This indicates that some refolding of the protein may have occurred during the dilution
process. During the dilution of the sample, the initial concentration of ~6 M urea was

reduced by three orders of magnitude.

86

Number of molecules

 

250 I ‘ l ' I I I

200 ‘

150-

100 -

LII
O
l
,M’f‘.
M

O
1
It
i
‘1

 

 

I ' I ' I

20 40 60
Height (A)

O4

Figure 25: Height distribution measured for Con A molecules treated with 6 M
urea, prior to dilution and adsorption at the water/mica interface.

87

 

80

To examine the effect of heat on the denaturation of Con A, aliquots of Con A in
pH=7.3 HEPES stock solution were treated at 90-95°C for 3-5 minutes. This solution
was then diluted to working concentration and applied to magnesium treated mica. These
samples produced SOMS data that were indistinguishable from the SDS treated samples,
which were also heated to 90-95°C. In this case two sets of samples were examined. The
height distributions showed 65-74% of the molecules to have heights centered between

14 and 18 A.

Upon SDS/heat or temperature denaturation conditions, approximately twice as
many Con A molecules appear to be denatured at the water/mica interface, compared to
untreated molecules. The results from the urea treated samples indicate that Con A
undergoes significant refolding, prior to or during the adsorption process, although it

most likely occurs prior to the adsorption process.

As mentioned in Chapter one, the prevailing model for protein adsorption at the
liquid/solid interface is thought to involve two steps [15]. The first step involves the
collision of a protein molecule with the surface and its layer of adsorbed water, while the
second is thought to involve some conformational change of the protein to maximize its
interaction with the surface. The results to this point have shown that less than one third
of the Con A molecules adsorbed at the water/mica interface appear to be significantly
denatured. In contrast to this case, approximately two thirds of the adsorbed molecules at
the interface are denatured when treated with 2% SDS and/or 90-95°C temperatures.

Because there is $3 3% denaturation in the case of untreated Con A adsorbed to mica,

88

there is reason to believe that this is consistent with the second step of the prevailing
protein adsorption model (see Chapter one, Figure 1). To examine this more closely
samples of Con A on mica were prepared and imaged as before, and then were aged

under ambient conditions (21°C and 35-50% relative hmnidity).

After 24 hours the samples were imaged, producing height distributions that were
not significantly changed from that of freshly prepared samples (1.5 — 8 hours old). It is
a reasonable assumption that after 24 hours, equilibrium of the layer of surface adsorbed
water with the protein and the atmosphere had been reached. Even when imaged after
being adsorbed to the mica surface for 48 hours, there was still not significantly more
denaturation observed. After approximately 72 hours, however, a significant change in
the amount of denatured protein was observed. Prior to aging the samples (prepared on
two different days), they were found to consist of 28-29% denatured molecules. After
three days’ aging on the surface 61 —73% of the molecules measured were found to orient
with heights centered at 18 A. There are two probable explanations for this observation.
First, it is possible that the denaturation of Con A due to its adsorption to the mica surface
is a rather slow process, and this experiment indicates a long time scale for the second
step of the prevailing protein adsorption model for Con A adsorption to mica. This is a
reasonable hypothesis given that after 24-48 hours the samples did not show significantly
more denaturation. The second possibility is that the Con A is denaturing due to its being
kept at 21°C for an extended period. It is also possible that both of these scenarios are
contributing. In either case, however, it is clear that the majority of Con A molecules at

the water/mica interface conserve their native dimeric structure, even when stored for 24

89

hours at ambient conditions. This observation has potential application toward the design

of saccharide sensor devices, or Con A/saccharide based bioassays.

4.2.4 Effect of pH on the structure and adsorption of Con A molecules at the

water/mica interface

The structural characteristics of Con A in solution have been studied under a
variety of conditions. One important factor that influences the structure of Con A in
solution is pH. The effect of various pH conditions on the structure and function of Con
A have been studied by a number of workers [11,17-21]. Some of the key findings of
this work have been that at acid pH, between 2 and 5.5, the dimeric organization is
favored [11], and that between a pH of 8 and 9 there is an irreversible structural transition
[17,18,22]. Using circular dichroism (CD) and optical rotatory dispersion (0RD), Zand

et a1. observed Con A to approach a random coil form at a pH of 9.1 [17].

The structural changes in Con A as a firnction of pH may in part be caused by
changes in the ionic character of the amino acid residues. The isoelectric point of Con A
has been reported to be in the range of 4.5-8 [9,11]. However, a value of 5-5.5 is typical
of the accepted value used in recent literature [23,24] and is consistent with a calculation
of pI=5.00 based on the amino acid sequence [25]. The studies of this dissertation have,
up to this point, looked at Con A deposited on the mica surface from a HEPES buffer at
pH=7.3. At this pH the protein will have a net negative charge, although 11% of the

residues posses some positive character. At a pH of 4.6, however, 89% of the residues

90

have positive charge, leading to a net positive charge for the molecule. A CPK structure
for the Con A dimer, based on XRD data [13] is presented in Figure 26. In this figure the

positively charged amino acids appear shaded.

Observation of Figure 26 indicates that there is a fairly even distribution of the
positively charged amino acids along the surface of the Con A dimer at both pH values.
This observation is consistent with the simplifying assumption made in deriving the SAO
model (see Equations 7-9 in Chapter two). The simplifying assumption was that no face
of the protein contained a unique surface recognition site, which allowed the definition of
AG0 E AG;ya = AGac = 13031,. Since no face of the protein has significantly more charge
than another, the charge would not contribute to making one face of the protein
inherently more favorable for adsorption to the negatively charged mica surface. An
important prediction of the SAO model is that changes in the free energy of the
adsorption process will lead to changes in the observed height distribution. Figure 10
gives theoretically predicted height distributions for a system with the dimensions
observed in our experiments for Con A at two different values of the free energy of
orientation (given in units of energy per unit area). It is apparent that the SAO model
predicts that a more favorable AG0 will produce a higher probability of finding molecules
oriented with their shortest dimension projected normal to the surface. To test this
prediction, a careful study of the height distributions for Con A at the water/mica

interface as a function of pH were carried out.

91

 

     
  
          

  
   

   

  
  
 

. yet»
I” ', 1:304? 934531.35
:ji “1519*? .v , {ks .‘ -4. ~
0 w “ tat-2;» this-West “at:
9" \‘ ‘*r.- r!" 1;.t A,» c :3!” .
a: it" 'Qéd‘fillhfl“ 1.4 ,,
$65 '1’ ‘I .1933“ hie-em
a~ .

pH=73

 

 

 

 

pH=46

 

 

 

Figure 26: Positively charged amino acids in Con A at two pH values. Positive
residues are shaded.

92

 

For this study, solutions of Con A were prepared in 0.01 M HEPES at pH=7.3 and
0.01 M Acetate at pH=4.6. The concentrations were checked, prior to dilution, to verify
that they were equal. To insure good comparison between the measurements, samples at
each pH were run at the same time (i.e. prepared and imaged the same day). The results
presented here are the result of three separate trials. The results from Con A deposited
from a HEPES buffer at pH=7.3 consist of data sets containing 715, 761, 990 and 1047
molecular measurements made on four samples prepared on three different days. The
results for Con A deposited from the acetate buffer at pH=4.6 are taken from data sets
containing 379, 585 and 892 individual measurements made on three samples prepared
on three different days. The measured height distributions, representing the sum of all the
trials, are presented in Figure 27. For purposes of calculating an average and standard
deviation, however, each data set was fit to four Gaussian functions independently. The

results obtained from the Gaussian fits to the data are summarized in Table VII.

93

 

I ' r ' . 1 ' I ' 1 ' r
400 r A \ 1
V1 .‘ l
.2
5 300 . I -
o i
g 1! t:
r .
.5 200 1 E. 1
‘- . . ‘ .1
2 R A I":
E 100 ' 1 ‘
2 ‘ f ’1 -” A ’ Sr
0 q I I “1'1 g .. 3..” k... ‘

 

 

 

0 20 40 60 80 100
THeight(A)'

 

300 I ' I T 1 r
. B 0 .
. . .
2004 _ ~

100 1 1 ' ~

. 1 c
ll . - I *5 4? i- ‘
' ' x
‘ » . Mum-unufig A“
I I ' I I

0 20 40 60 80 1 00
Height (A)

Number of molecules

 

 

 

Figure 27: Height distributions measured for Con A molecules deposited
from solutions at A) pH=7.3 and B) pH=4.6.

94

Table VII: SOMS data for Con A at the water/mica interface deposited from solutions at

two different pH values.

 

 

 

 

 

 

A) pH=7.3
Height (A) Width (A) Probability (%)* % denatured
l9 9 l8 :1: 5
a 29 12 71 i
43 l l 24 i 5
C 59 l l 5 i 4
B) pH=4.6
Height (A) Width (A) Probability (%)* % denatured
l8 9 34 i 5
a 28 12 84 i 14
42 9 13 i" I l
c 59 I l 3 i 3
*See note b in Table 11

Examination of the fitting parameters is in close agreement with the widths and
positions presented in Tables 11, III, V and VI. It is clear that there was more
denaturation of the Con A which had been deposited from the acetate buffer at pH=4.6,
compared to that of Con A in HEPES at pH=7.3. This behavior has not been previously
observed for Con A. The studies of Auer and Schilz have examined Con A in the acidic
pH range [19]. In this work they observed the release of Mn+2 and Ca+2 ions from Con A
beginning at a pH around 6.0 and increasing to a maximum by sz3 [19]. The loss of
the metals from Con A results in a conformational change of the amino acids near the
metal binding sites. While this change is observable in the XRD structures of the native

and apoprotein [13,26], the change in the gross molecular size of the dimer is less than 5

95

A. The transition from native Con A to the demetallized form has been likened to an
“unlocking” of the conformation [13,26]. This is consistent with the observation of more
denatured protein for samples adsorbed to mica at pH=4.6, and may explain why more

denaturation occurs during the adsorption process.

The other striking feature of the results is that there is a tendency for Con A
molecules to more strongly favor the orientation that places side bc on the mica surface,
when deposited from a pH=4.6 solution. This preferential orientation of the Con A dimer
is likely driven by a more favorable interaction of the positively charged Con A dimer
with the net negative charge of the mica surface. This change in the distribution over
orientations is predicted by the SAO model, proposed in chapter 2. From the results of
Table VII, the free energy of orientation is calculated from equations 1-3 and 7 of chapter

2. The results of this calculation are presented in Table VIII.

Table VIII: Free energy of orientation values for Con A at the water/mica interface

deposited from solutions at two different pH values.

 

 

 

 

 

 

 

 

 

 

 

 

A) pH=7.3 B) pH=4.6

A60 -3.4 J/mol A2 AGO -6.5 J/mol A2.
Std. Dev. 0.7 J/mol A2 Std. Dev. 2.8 J/mol A2
AGbc '85 kJ/mol AGi,c -l 6.0 kJ/mol
AGac -5.8 kJ/mol AGac -10.9 kJ/mol
AGab '49- kJ/mol AGab -7.7 kJ/mol

 

 

 

 

 

 

 

 

 

 

96

Note that the standard deviation in the probability of measuring dimension c
(given in Table VII) is on the order of the average, indicating significant uncertainty in
this population. Therefore, the results presented in Table VIII are based solely on the
P(a)/P(b) ratio (see equation 7, chapter 2). The exclusion of the equations which consider
the population of molecules with dimension c (i.e. 59 A) is reasonable, since the
variability of the P(a)/P(c) and P(b)/P(c) is quite large. The data of Table VIII clearly
show an increase in the free energy of orientation (by a factor of nearly two), upon going
from a pH of 7.3 to 4.6. When the data of Table VIII A and Tables 11 and III are
compared one sees a fairly large discrepancy in the calculated free energy of adsorption.
In the earlier data the Con A deposited from a HEPES buffer at pH=7.3 gave AGO ~9
J/mol A2, however the data presented in this section for Con A under the same conditions

gives AGO ~3.5 J/mol A2. This variability will be discussed firrther in section 4.3.

The effect of high pH on Con A structure at the water/mica interface was also
examined. It has been observed that Con A undergoes irreversible loss of its B-structure
when it is exposed to basic media in the pH range 8-9 transition [17,18,22]. A stock
solution of Con A in borate buffer (pH=10) was prepared. Results for the analysis of
three height distributions, containing 709, 1133 and 1962 molecular measurements, are

presented in Table IX.

97

Table IX: A) SOMS data for Con A at the water/mica interface, as deposited from a

pH=10 solution. B) Free energy of orientation values for Con A at the water/mica

interface deposited from a solution at pH=10

 

 

 

 

 

 

 

 

 

 

A)
Height (A) Witdh (A) Probability (%)* % denatured
17 8 51 i 13
a 28 11 71 i- 7
b 42 11 23 i 7
c 59 10 6 i 2
B)
AGO -4.1 J/mol A2
Std. Dev. 0.8 J/mol A2
AG],c -1 0.2 kJ/mol
AGac -6.8 kJ/mol
AGab -4.9 kJ/mol

 

 

 

 

The data for Con A deposited from a pH=10 solution show significantly more

denaturation than Con A deposited from a pH=7.3 solution. The percentage of denatured

molecules lies somewhere between that of Con A treated with 6 M urea (see section

4.2.3) and the SDS and/or heat treated Con A. This observation is consistent with the CD

studies of land et al. that observed Con A to approach a random coil form at a pH of 9.1

[17]. The presence of an irreversible structural transition, occurring somewhere between

a pH of 8 and 9 has also been noted by Pflumm et al. [18,22].

98

As seen from Table IX and Table VII-A, the relative probabilities for the three
molecular orientations of the Con A dimer at the water/mica interface, are not
significantly different for pH=7.3 and pH=10. While the free energy of orientation per
unit area observed for the pH=10 sample is slightly higher than that of the Con A at
pH=7.3 (see Tables VIII and IX), there is only a slight difference in the free energy of
orientation for each protein face. This is in part due to the slight differences in the
average Gaussian peak positions for the data sets. This negligible difference in the free
energy, as compared to the large change in going from pH=7.3 to pH=4.6, is not too
surprising. At a pH of 7.3 Con A has a net negative charge, but at a pH lower than 5 the
charge on the protein becomes positive. In the present case, the protein has become more
negatively charged going from pH=7.3 to pH=10, though the change in charge does not
seem to drastically affect the free energy of the adsorption process onto the net negative

mica surface.

Despite the fact that the surface was treated with Mg+2 prior to the adsorption of
protein, there is reason to believe that the net charge at the surface remains somewhat
negative. This is supported by the data of this section, but this was tested independently
by running a sample of Con A at pH=7 .3 on mica with and without the magnesium
treatment. In this experiment 643 measurements were made for Con A on untreated mica
and 696 measurements of protein on Mg’L2 treated mica. The height distributions
obtained were statistically identical, matching the results presented in Table VII A. The
expectation was that if the negative surface charge is significantly reduced by the

presence of the positive magnesium ions, then the free energy of orientation for

99

negatively charged Con A would favor the 30 A population more strongly. Based on the
results of this section, it is reasonable to assume that the surface charge of the mica

remains a net negative.

This brings up another interesting feature of the data presented in this section. It
was anticipated that the adsorption of Con A from the pH=4.6 solution would progress at
a faster rate than that at pH=7.3-10, where the molecule maintains a net negative charge.
In fact there was no increase in the number of molecules deposited at a pH of 4.6 vs. a pH
of 7.3. Analysis of 7-8 images, acquired on the same day, showed that for Con A
deposited from Acetate buffer (pH=4.6) 76 i 3 molecules/11m2 while 73 j: 6
molecules/11m2 were observed for images of the pH=7.3 deposited samples. The

uncertainties given are :1 o.

4.3 Discussion and conclusions

Height measurements of colloidal gold particles of different sizes (i.e. 5, 9 and 18
nm) made by the AF M in contact mode at ambient conditions showed that these particles
are accurately measured by this technique. The gold particles were observed to be quite
incompressible, showing no significant change in structure within an order of magnitude

of the minimum force required for stable imaging.

The first application of the SOMS technique was the spherical protein ferritin.

The analysis of the height distributions obtained for ferritin showed the presence of three

100

populations of molecules at the water/mica interface. The dominant population, centered
at 116 A is attributed to the native spherical 24-mer, which has been observed by x-ray
and electron microscopy to be 120 A [3,4]. The second most prevalent species, centered
at 135 A is attributed to the presence of a trimer of the native 24-mer, which has been
previously observed by STM [5]. The remaining population at 100 A is consistent with

dehydrated ferritin, as observed in vacuum by electron microscopy [4].

Measurements of individual Con A and succinyl Con A molecules at the
water/mica interface has shown that the Con A dimer is the prevalent structure. The
measurements obtained (29 x 45 x 59 A) for these dimers are in good agreement with the
dimensions determined by x-ray diffraction studies. Further, the results for Con A,
succinyl Con A and SDS treated ConA are self-consistent, and reproducible. In addition
to the structural information obtained, the proposed model also allows us to predict a
value for the free energy of adsorption for Con A at the water/mica interface and the

orientation of the adsorbed proteins.

The denaturation of Con A was examined using the SOMS technique. Typically
about 30% or less of the Con A molecules at the water/mica interface were found to be
denatured. The assignment of molecules with heights of 17-19 A to the presence of
denatured molecules was made based on the observation of Con A treated with SDS and
heated to 90-95°C. This chemical denaturation produced samples with 60—70% of the
observed molecules to have heights centered at 17-19 A. The presence of a population of

denatured molecules for untreated Con A at physiological pH is consistent with the

101

prevailing two step model of protein adsorption (see Figure 1). Measurements of Con A
that was adsorbed to the mica surface and then aged from one to two days showed that
the amount of denatured protein remains nearly constant. After three days, however, we
found that ~60-70% of the Con A had unfolded on the surface. It is interesting to note
that Barnes et al. found that the useful lifetime for their Con A based piezoelectric
glucose sensor was three days [10]. In their work Con A was adsorbed onto siloxane
modified silver electrodes, and the loss of sensor activity was attributed to bacterial attack
[10]. If bacterial attack is the reason for the inactivation of the Con A, we should have
observed the relatively large bacteria with the AFM. In fact, we saw no indication of
bacterial growth with the AFM, which indicates that the structure of Con A at the
water/mica interface is quite stable over the range of 24-48 hours. After that the protein
may denature simply because it cannot maintain its structure without additional hydration
or that the half life of the second step of the protein adsorption model is quite long. Con
A treated with 6 M urea gave a height distribution containing only ~40% denatured
molecules, indicating that some of the molecules may have undergone refolding as the

urea concentration was diluted by three orders of magnitude.

The SOMS model was further tested by comparing Con A deposited onto mica
from buffers of different pH. In this study samples of Con A in 0.01 M HEPES (pH=7.3)
and in 0.01 M acetate (pH=4.6) or 0.01 M borate (pH=10) were prepared in parallel and
imaged within a few hours of one another. At the lower pH, where Con A possesses a net
positive charge, the SOMS analysis showed that the free energy of adsorption was more

favorable than for Con A adsorbed from solutions at pH between 7.3 and 10, where Con

102

A maintains a net negative charge. This observation is consistent with adsorption onto a

negatively charged surface, as predicted by the SAO model presented in Chapter two.

It was expected that the adsorption to the negatively charged mica surface would
be faster for Con A at a pH=4.6than for pH=7.3. This expectation was based on the fact
that 4.6 is below the isoelectric point of Con A (pl=5-5.5). Analysis of the number of
molecules observed at each of the two pH values showed that the number of proteins

adsorbed per square micron lie within the same range (i.e. 73-78 molecules/11m2 and 68-

78 molecules/umz, respectively). The adsorption process appears to be limited by
diffusion of the molecules to the surface, rather than any electrostatic forces present. The
likely explanation for this observation is that in the dilute solutions used (~10-400 ng/ml),
there is good shielding of charge by the solvent. The issue of rate of adsorption as a

function of protein charge will be further addressed in Chapter five.

While these results are shown to be qualitatively consistent with the SAO model,
there are large discrepancies observed in the free energies of orientation predicted. The
results presented in section 4.2.2 predict a value of —9.4 i 2 J/mol A2 while the later
samples of Con A in HEPES (see section 4.2.4) gave a value of —3.4 i 0.7 J/mol A2.
There are a couple of reasons for this discrepancy. First of all the data of section 4.2.2
were for samples prepared using mineral quality mica (Ward’s Natural Science
Establishment, NY), while the later data was taken using high quality mica sheets (gift of
Dr. Richard Schwendeman). A second reason for this discrepancy lies in the way the

SAO model calculates the value of AGO. The ratios of the probabilities of observing each

103

molecular orientation are exponentially related to the free energy. Since the number of
measurements observed at 59 A is much smaller than the number of observations made of
molecules centered at ~30 or ~45 A, there is much more uncertainty in its probability. If
one ignores the AGO calculated from the ratio P(b)/P(c) (as was done in section 4.2.4) the
standard deviation in the calculated value is usually less by a factor of two. If the value
of AGO for Table II is re-calculated one obtains a value of -8.3 i 1 J/mol A2, which is still
more than twice the value obtained for Con A on the optical quality mica. Although it
may seem that sticking with the natural mica would have been better, given the lower
amount of denatured Con A observed, it should be noted that the samples prepared using
this substrate showed a wider range of variability. This variability included measuring
some samples which contained ~10-20% denatured protein, with AGo~-9 J/mol A2, as
well as others with 30% denatured molecules and AGo~-3-4 J/mol A2. The Con A in
HEPES samples deposited onto the high quality mica gave ~30% denatured protein and
AGo~-3-4 J/mol A2. A likely explanation for these observations is that some areas of the
natural mica may have been more positively charged (possibly due to the presence of
excess metal cations) which would explain the increased population of 28-30 A
molecules. This type of shift in the height distribution with increasing positive charge

was shown for Con A adsorbed to mica from the acetate buffer at pH=4.6.

The studies of Con A at the liquid/solid interface presented here not only serve to
validate the SAO model and the SOMS technique, but they also give new insight into the
structure and oligomerization of this lectin. This study marks the first such study of

individual Con A molecules at the liquid/solid interface. The results have important

104

implications to those interested in designing Con A based sensors and assays. The strong
binding of saccharides by Con A has been of interest for use in glucose sensors
[10,27,28]. Further the use of Con A as a model system for antibody-antigen interactions
[27], and the stability of Con A at the liquid/solid interface observed in this work

indicates that Con A may be useful in the design of new bioassays.

We conservatively estimate that the accuracy of this technique is 5 A or less. This
estimate is related to a number of factors. First of all there is 3-5 A roughness in the mica
substrates which we use. Second, although the ultimate resolution of the AF M in the z
dimension is 1 A or less, we are not optimizing our imaging conditions for single point
measurements of height, rather we trading accuracy for acquisition time. For the
purposes of three-dimensional protein oligomerization state and orientation, the accuracy

(< 5 A) is sufficient.

The technique presented here provides a means to measure the molecular
dimensions for the complete size range of monomeric and oligomeric proteins adsorbed
to a surface (natural or synthetic). These measurements may then be utilized to
determine the oligomerization state of the protein. Probe-protein interactions, mainly
lateral and vertical pressure, can cause the molecule to appear shorter [29]. With regard
to protein-substrate interaction, there is an attractive interaction between the polar surface
of the protein molecule and the charged mica surface, possibly deforming the molecule
and causing it to flatten against the mica surface. A study using lysozyme has shown that

it retains its activity when adsorbed onto mica and imaged with AFM, so these

105

deformations may not be extensive enough to alter biological activity [30]. Our
measurements confirm the conservation of native structure for Con A and ferritin, when

compared with available x-ray data.

A thermodynamic model was derived to understand the orientation of proteins at
the liquid/solid interface. Large deviations from this SAO model by adsorbed proteins
would imply a preferred orientation that is presumably related to recognition between a
protein domain and the surface. The extent of deviation from the orientation model may
therefore be useful in defining preferred molecular orientation for oligomers or multiple
equilibria for dissociated subunits of oligomers. In these cases one is able to extend the
simple model presented here, by allowing each face of the protein to have a different
value for AGo (Equations 1-3). In this way a quantitative value may be assigned to the

energy difference associated with each orientation.

An examination of the preferred orientation of the molecules also provides the
opportunity to predict the organization of individual subunits within the protein oligomer
and potentially evaluate the conformational flexibility of dissociated individual subunits.
Measurements with angstrom resolution may be performed on functional complexes that
are associated with a surface, a more common biological interaction within and between
cells. These results present an approach which extends the analytical power and the
resolution of AF M to provide direct physical characterization (i.e. dimensions,
orientation and free energy of adsorption) of biomolecules at interfaces, and is minimally

dependent on the geometry of the probe tip. This approach provides the capability to

106

examine properties of protein structure and oligomerization at liquid/solid interfaces
using picomole amounts of protein. Further, it offers the ability to distinguish between
the possible quaternary forms and their interconversions resulting from the binding of
ligands, substrates, or allosteric effectors. Such measurements can be most useful in
studies of enzyme regulation, DNA/RNA organization, gene regulation, protein
biosynthesis and chaperonin function, bacterial attachment and adhesion, and the effects

of post-translational modification on protein organization and activity.

107

4.4 References

1. Vesenka, J.; Manne, S.; Giberson, R.; Marsh, T.; Henderson, E. Biophys. J. 1993, 65,
992.

2. Munro, H. N.; Linder, M. C. 581978, 2.

3 . Massover, W. H. Micron 1993, 24, 389.

4. Feder, J .; Giaever, I. J. Coll. Inter. Sci. 1980, 78, 144.
5. Yau, S.-T.; Zhou, Y. Mod. Phys. Lett. B 1995, 9, 187.
6. Gekko, K.; Hasegawa, Y. Biochemistry 1986, 25, 6563.

7. Concanavalin A as a Tool; Bittiger, H.; Schnebli, H. P., Eds.; John Wiley & Sons:,
1 976.

8. Sharon, N.; Lis, H. Science 1989, 246, 227.

9. Reeke, G. N., Jr.; Becker, J. W.; Cunningham, B. A.; Wang, J. L.; Yahara, 1.;
Edelman, G. M. Structure and Function of Concanvalin A. In Concanavalin A;
Chowdhury, T. K., Weiss, A. K., Eds.; Plenum Publishing Corp.: New York, 1975.

10. Barnes, C.; D'Silva, C.; Jones, J. R; Lewis, T. J. Sens. Act. B 1991, 3, 295.

11. Liener, I. E. Isolation and Properties of Concanavalin A. In Concanavalin A as a
T 001; Bittiger, H., Schnebli, H. P., Eds.; John Wiley & Sons:, 1976; pp 2.

12. Gordon, J. A.; Young, R. K. J. Biol. Chem. 1979, 254, 1932.

13. Shoham, M.; Yonath, A.; Sussman, J. L.; Moult, J.; Traub, W.; Kalb, A. J. J. Mol.
Biol. 1979, 131, 137.

14. Gunther, G. R.; Wang, J. L.; Yahara, L; Cunningham, B. A.; Edelman, G. M. Proc.
Nat. Acad. Sci. USA 1973, 70, 1012.

15. Brash, J. L.; Horbett, T. A. Proteins at Interfaces: An Overview. In Proteins at
Interfaces 11, Fundamentals and Applications; Horbett, T. A., Brash, J. L., Eds.;
American Chemical Society: Washington, 1995; Vol. ACS Symposium Series 602; pp 1.
16. Weisgerber, S.; Helliwell, J. R. J.Chem. Soc., Faraday Trans. 1993, 89, 2667.

17. land, R.; Agrawal, B. B. L.; Goldstein, I. J. Proc. Nat. Acad. Sci. USA 1971, 68,
2173.

108

l8.

19.

20.

21.

22.

23.

24.

25.

Pflumm, M. N.; Beychok, S. Biochemistry 1974, 13, 4982.

Auer, H. E.; Schilz, T. Int. J. Peptide Protein Res. 1984, 24, 462.

Lis, H.; Sharon, N. Ann. Rev. Biochem. 1986, 55, 35.

Sharon, N.; Lis, H. FASEB J. 1990, 4, 3198.

Pflumm, M. N.; Wang, J. L.; Edelman, G. M. J. Biol. Chem. 1971, 246, 4369.
Lvov, Y.; Ariga, K.; Ichinose, 1.; Kunitake, T. Thin Solid Films 1996, 284-285, 797.
Gallinet, J .-P.; Gauthier-Manuel, B. Eur. Biophys. J. 1993, 22, 195.

Swiss Institute of Bioinforrnatics. Compute pI/MW Tool. Available

http://expasy.hcuge.ch/ch2d/pi_tool.htrnl, June 1998.

26.

Reeke, G. N. j.; Becker, J. W.; Edelman, G. M. Proc. Nat. Acad Sci. USA 1978, 75,

2286.

27. Janata, J. J. Amer. Chem. Soc. 1975, 97, 2914.

28.

Kremer, F. J. B.; Engbersen, J. F. J.; Kruise, J.; Bergveld, P.; Starmans, D. A. J.;

Feijen, J .; Reinhoudt, D. N. Sensors and Actuators B 1993, 13-14, 176.

29. Radmacher, M.; Fritz, M.; Cleveland, J. P.; Walters, D. A.; Hansma, P. K. Langmuir
1994, 10, 3809.

30. Radmacher, M.; Fritz, M.; Hansma, H. G.; Hansma, P. K. Science 1994, 265, 1577.

109

Chapter 5: Study of Con A Orientation at the Liquid/Solid Interface: From

Isolated Molecules to Thin Films

5.1 Introduction to Con A film formation

While the majority of papers dealing with protein structure at liquid/solid
interfaces deal with measurements of monolayer or multilayer films, and 2D crystals,
very little is known about the mechanisms of adsorption or the microstructure of thin
protein films. The data presented thus far in this dissertation have shown that the dimeric
oligomerization state of Con A is favored at the water/mica interface, and that Con A
molecules tend to orient such that the surface area in contact with the mica is maximized.
One of the issues that hasn’t been addressed is whether the Con A simply dissociated to
the dimeric form in solution as a result of concentration, pH or ionic strength [1], or
whether the interfacial chemistry favors the dimeric form. Would tetrarners of Con A be
observed as the molecules adsorb to the interface from a concentrated solution? If so,
how would they orient? In order to investigate these issues a series of experiments was
designed to examine Con A molecules at the water/mica interface as deposited from more

concentrated protein solutions.

The interest in thin films of Con A dates back to the mid-seventies. Janata
published a study of an ‘immunoelectrode’, composed of a PVC coated Pt wire that was
subsequently layered with Con A [2]. This device was shown to be sensitive to the

binding of yeast mannan. The description of this device as an ‘immunoelectrode’ was

110

justified in that Con A-saccharide interactions, while not a true antibody-antigen
interaction, are often used as a model for ligand binding and antibody-antigen
interactions [2]. From the mid-eighties to the present there have been a number of studies

of Con A associated in thin films, studied by a variety of techniques [3-10].

The findings of these studies are indicative of one of the problems in the proteins
at interfaces field, namely that there is significant variability in conditions and results,
and many unanswered questions. Afshar-Rad et a1. deposited films of Con A from a
pH=5-5.5 buffer onto mica [3]. Using the surface force apparatus, they observed an
apparent layer thickness of 38 A, consistent with the presence of dimers [3]. Gallinet and
Gauthier-Manuel also used a surface force apparatus and found 12.5 A, 23 A, 39 A and
46 A layers within their multilayer films on mica [6]. The thickness of expelled layers in
their study was dependent on the pH of the deposition solution used. The same year
Safar et al. prepared Con A films on glass and mica [7]. Using TEM they investigated
the morphology, finding rough ($400 A), thick films (455 i 72 A), but did not determine
the number of layers present [7]. Using CD and FTIR spectroscopies Safar et al. found
that Con A retained its B-structure in the films they deposited [7]. Lvov et a1. deposited
Con A onto polyethyleneimine cation layers. Using a quartz crystal microbalance
(QCM) they determined the mean layer thickness to be 57 A i10% when depositing from
pH=7 solution and 17 A i10% depositing from a pH=5.6 solution [8]. Lvov et a1. viewed
this result as consistent with studies that have found preference for the dimer at pH~5.6,
but tetramer formation at pH~7[11,12]. Fitzpatrick et a1. deposited Con A films onto

mica at pH=7 and examined them using x-ray photoelectron spectroscopy (XPS) [10].

111

The film thickness was determined by monitoring the N Is signal as a function of angle
and x-ray energy to determine a depth profile [10]. The measured thickness for films
varied between 14 and 41 A, but the average thickness quoted was 21 i 4 A (based on the
angle resolved depth profile) [10]. Although they did not make measurements of film
thickness or molecular orientations, both Barnes et al. and Haas and Mohwald, note that
they had difficulty in controlling the state, stability and evenness of Con A coatings on
quartz and phospholipid layers, respectively [3,4]. The data and preparation conditions
for each of these studies is summarized in Table X. It is quite apparent from the

variability in the measurements of these studies that there is not a clear picture of the

structure and growth of Con A films.

Table X: Summary of literature results for Con A thin films deposited on surfaces.

 

 

 

 

 

 

 

 

 

Ref. Method Concentration Deposition pH of substrate Thickness
time solution (A)

[6] SFA 40 rig/ml 60 min. 3.9 Mica 12.5

[3] SFA Not specified 2 hr. 5.5 Mica 38

[8] QCM 1 mg/ml 20-30 min. 5.6 “PEIT on Ag 57 (i10%)

[6] SFA 40 ug/ml 60 min. 6 Mica 39

[8] QCM 1 mg/ml 20-30 min. 7 alPEI+ on Ag 17 (i10%)

[10] XPS 1-10 rig/ml 17-19 hr. 7 Mica 21 :t 4

[6] SFA 40 ug/ml 60 min. 7.4 Mica 23, 46

 

 

 

 

 

 

 

a poly (ethyleneimine)+ modified Ag electrode on QCM

112

 

A better understanding of the structure and molecular orientation of Con A thin
films is paramount to the fundamental understanding of protein-protein and protein-
surface interactions. The results of Chapter four present a well-characterized picture of
isolated Con A molecules at the liquid/solid interface. A logical next step would be to
ask questions about how this early stage of adsorption might progress as the adsorption
continues toward formation of monolayer and thin films. Additionally the potential
commercial application of Con A-saccharide based sensors and bioassays, makes Con A

a good candidate for this study.

Results presented in Chapter four show that Con A maintains much of its native
dimeric structure at the water/mica interface and appears to be little affected by the
imaging forces applied by contact mode AFM. As the Con A molecules begin to
aggregate at the interface the problem of tip convolution will preclude the determination
of individual molecular positions and orientation, but the height of local domains will be
accessible. While techniques such as neutron reflection and ellipsometry are capable of
determining overall film thickness, the AF M allows one to make local morphology and

topography with angstrom height resolution and nanometer lateral resolution.

The approach to the problem of making high-resolution measurements of protein
films is essentially the same as in the isolated molecule case (i.e. SOMS). Though it is
hindered by reduced lateral resolution, the AF M can measure the height of regions of the

film to 1 A accuracy, limited by the substrate flatness and roughness. The height

113

distribution obtained for images of sub-monolayer protein films will reflect populations
of heights corresponding to both the mica substrate and the protein molecules that make
up the film. The heights of various regions of the film may then be measured relative to
the background. This approach may be extended to complete monolayer coverage as
well. As will be shown, the height of the mica substrate may be determined either by
producing sub-monolayer films or by physically cutting away a small section of the film

from the surface with the AFM.

This chapter will present studies of Con A at the water/mica interface as
monolayer films are formed. Data for the early stages of growth, as well as morphology
of complete monolayers will be presented. These results will be compared to the
observed orientation and structure of individual Con A molecules presented in Chapter
four, and to the results of previous studies for Con A films at liquid/solid interfaces.
Further, the implications to understanding the film growth mechanism and the design of

sensor/assay devices will be discussed as well.

5.2 Experimental and theoretical details

The materials and instrumentation are the same as given in Chapter three,

however, the preparation and imaging of samples, as well as the experimental analysis

and theory are somewhat different from that presented in Chapters two and three. A brief

description of the additional experimental details will be given in this section.

114

Preparation of Con A Films. 25 111 of solution containing ~120 rig/ml Con A in
0.01 M buffer (HEPES at pH=7.3 or Acetate at pH=4.6), was applied directly to mica,
which was freshly cleaved and treated with magnesium ions (as described in Chapter
three). Samples were incubated at 21°C for the duration of the deposition time and then
rinsed with two 200 111 portions of milli Q water. The deposition time was controlled
using a wristwatch and was varied between 1 min. and 60 min. The samples were ‘dried’
(see Chapter three) at 21°C and 35-50% RH. Films were imaged beginning 1 hour after

preparation and were discarded after 12 hours.

AFM imaging of Con A at the liquid/solid interface. Two different types of
Microlever cantilevers (Park Scientific Instruments) were used. Measurements were
made using either sharpened Microlevers, having a spring constant of approximately 0.05
N/m with an integrated silicon nitride probe (20 nm radius of curvature), or Microlevers
with a spring constant of 0.5 N/m containing 50 nm radius of curvature tips. The higher
spring constant was desirable for this application since high forces ~10 nN were useful
when scraping away portions of the film for the measurement of the background mica
height. It was also found that somewhat lower vertical forces, ~1.5 nN and higher lateral
forces were capable of cutting away sections of Con A film. The 5 11m and 100 um
scanners were calibrated for vertical measurements with colloidal gold particles (5 and 9
nm) (Sigma, St. Louis). Horizontal measurements are calibrated using a two-dimensional

grating with 1 pm periodic spacing in both the x and y directions.

115

AF M images were acquired in the constant force mode, as discussed in Chapter
three. As the films grow to the point of nearly complete coverage, a measurement of film
thickness is not made directly, since the mica background is not known. In cases of
monolayer or near monolayer coverage a small area was scanned at high force and at a
high scan rate. The high scan rate was used in order to cause increased lateral forces
between the probe tip and the protein film during the raster scanning process. After
repeated scanning, the forces were sufficient to scrape away the Con A molecules from
the mica surface (see Figures 33,36,37). The molecules that are scraped away deposit at
the edges of the scan area, or in some cases lead to fouling of the probe tip with adsorbed

proteins.

Analysis of Con A thin films: Analysis of the Con A thin films was similar to that
of the isolated molecules in that we focused on the height distribution. In the case of the
films, however, we do not have the lateral resolution necessary to determine the
orientation of each molecule in the film. As discussed in Chapter one, the lateral
resolution is limited by the size and shape of the probe tip, as well as the sample
morphology. In order to get an idea of the best lateral resolution to be expected a scale
model was drawn (see Figure 28). As can be seen from Figure 28, the lateral resolution
for the protein films will depend on both the film and the probe. Given ideal probe tip
size and geometry (i.e. Microlever spherical probe tip with radius of curvature of 20 nm),
one could expect lateral resolution of 11 nm for a 30 A thick discontinuous film. A 60 A
film would have lateral resolution limited to 14 nm, while an 80 A thick film would limit

it to 16 nm. This means that for the Con A films one might expect to be able to

116

determine morphology for local areas of 11 x 11 nm to 16 x 16 nm. The broader,
unsharpened Microlever probe tips, having a radius of curvature of 50 nm, would have

resolution limitations of 32 x 32 nm to 56 x 56 nm.

117

20 nm

llnm

20nm

  
     

.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A. .A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.A.

 
 

14nm

Figure 28: Model illustrating the AF M lateral resolution expected for measurement of
thin films on mica.

118

The same considerations used for the SOMS analysis, namely that the height
resolution is on the order of a few angstroms or better and that this height information
conveys the molecular orientation information will be used for analysis of the AF M
images. In the case of isolated molecules we construct a height distribution from
measurements of individual molecular heights. In the case of thin films, where the
heights for individual molecules cannot be determined, a slightly different approach
needs to be taken to examine the distribution of molecular orientations. Therefore, the
height of each pixel in the image (262144 pixels total) was placed into 2 A bins. For
images in which a large area of the protein film was scraped away, or one of a
discontinuous film (see Figures 29-37), a distribution of heights containing information
about the mica surface and the adsorbed protein molecules is obtained. The population of
pixels at lower height values corresponds to the mica substrate, while the higher
distributions correspond to the different orientations of molecules in the film. The data
were fit with Gaussian functions, in the same way as the height distributions of Chapter
four. The orientations of the protein molecules were deduced by taking the heights

relative to the mica substrate.

5.3 Results

5.3.1 Growth and orientation of Con A films from pH=7.3 solutions

The first set of experiments involved deposition of Con A from a HEPES buffer

(pH=7.3) onto magnesium treated mica. Deposition time was varied between 1 minute

and 1 hour. Figures 29-33 give the AFM images and pixel height histograms for these

119

films. After 1 minute, the Con A molecules have begun to form islands on the mica, but
the surface coverage is quite low (see Figure 29). The height histogram given in figure
29 B shows that there are at least three populations of heights for this image. The
population centered ~150 A, which accounts for 53 i 10 % of the image, corresponds to
the height of the mica substrate or background level. One should note that the height
values obtained by the constant force mode of imaging are relative heights, due to
changes made in the height of the PZT scanner in response to changes in force. If the
height of the mica substrate is normalized to 0 A (which will be done with all of the
results given in this chapter), we see that the other two populations are centered at 12 i 3
A and 32 i 2 A (see Table XI). Surprisingly these values are quite close to the 18-19 A
and 28-30 A populations observed for ensembles of single Con A molecules on mica (see
Chapter four). These data are consistent with the presence of Con A dimers and unfolded
protein. The samples of Chapter four were prepared from solutions that were ~10-400
ng/ml Con A. It is known for Con A that the dimeric form is favored when the
concentration is below 5 ug/ml [1]. The solutions used to prepare the samples of this
section were twenty times more concentrated than this, and were at a pH of 7.3, which is
also favorable for the formation of Con A tetrarners [11]. This observation of heights
corresponding to Con A dimers and denatured protein suggests that the interface, and/or
intermolecular forces between Con A molecules are favoring one orientation on the mica
surface. The orientation of Con A dimers, which projects the 30 A dimension normal to
the surface is consistent with the observations of Chapter four, however in the case of
aggregates of Con A at the interface we observed no 42-45 A or 59 A heights

corresponding to the other dimensions observed for Con A dimers at the interface. The

120

other striking feature of this data is that the population of heights centered at 12 i 3 A
occurs with the same frequency as the population at 32 i 2 A. In Chapter four we saw
that it was common to observe about 30% of all Con A molecules, deposited from a
pH=7.3 solution, as denatured (population centered at 18-19 A). In the present data there
is nearly a fifty-fifty mix of 12 :1: 3 A and 32 i 2 A. Much of this 12 A population may
be an artifact of the imaging process (i.e. tip ‘convolution’). When the image is scaled
such that the population lying 12 A above the background is highlighted one sees that
much of these areas are accounted for by the edges of the islands, where tip convolution
is expected. There are, however, some areas which are large enough not to be a tip
convolution artifact, but rather to the presence of denatured protein. Figure 30 renders a
1 x 1 11m region of the image from Figure 29 as a contour plot. In Figure 30 the range of
the contours is set to be equivalent to the three populations observed in the height
distribution of Figure 29. Also in this figure a 15 nm square is given to illustrate the
approximate lateral resolution expected for this film. It is clear that many of the 12 A
regions are halos around 30 A islands, but there are other 12 A tall regions that are
significant. This indicates that either the interface, or the intermolecular forces between

Con A molecules is somehow favoring a change in the native dimer conformation.

121

 

I 5001un I

 

 

 

 

 

35000...,...,.,.,
30000]
25000'
20000:
15000‘

Number of pixels
5
8
9’

5000-

 

 

 

 

100 120 140 160 180 200 220
Height (A)

Figure 29: 1 min. deposition of Con A in HEPES on mica. A) AFM image
B) Height histogram.

122

 

[:3 178.3 -- 200.0A
156.7 -- 178.3A
- 135.0 -- 156.7A

 

 

 

 

 

15 nm

 

 

 

Figure 30: Portion of image from Figure 29, rendered as a contour plot.
Approximate lateral resolution is represented by a square drawn
to scale of image.

123

Figure 31 gives a representative AF M image and height histogram for a 5 minute
deposition of Con A (HEPES buffer, pH=7.3) onto mica. Again there are three
populations observed in the height distribution, corresponding to the mica substrate, and
different populations of Con A heights. One should note that there are a few places on
the image where there are fairly high heights (i.e. pixels show up as white). The number
of pixels associated with these areas are much less than 1% of the total number of pixels
and are not apparent in the histogram. It appears that these areas are anomalous, and may
be areas where there has been deposition of aggregates, which formed in solution and not
on the surface, or possibly due to local differences in the mica surface. The data obtained
from this figure are summarized in Table XI. The numbers are quite similar to those
observed for the 1 minute deposition, however, there is more surface coverage in this
case. Note again that there is a fifty-fifty distribution of 13 i 1 A and 30 A heights,

consistent with denatured and dimeric Con A, respectively.

124

 

 

 

 

 

rxels

Number of p

 

 

 

 

0 20 40 60 80 100 120
Height (A)
Figure 31: 5 min. deposition of Con A in HEPES on mica. A) AFM image
B) Height histogram.

125

Figure 32 presents representative data for a 10 minute deposition of Con A. The
major observation here is that only two distinct populations of heights are observed. The
film is still submonolayer and growing by aggregation of the islands seen after 1 minute
and 5 minutes. Again, while there are a few anomalous aggregates, the height
distribution contains only heights corresponding to the mica substrate and a population of
Con A at 18 i 3 A. These data were consistent between samples prepared on the same
day, as well as ones prepared 1 month apart. This seems to indicate some kind of
collapse of the Con A dimer on the surface, or perhaps an imaging artifact. In either case,
it is likely that the cause is a cooperative intermolecular force between Con A molecules
as it forms a layer at the interface. Because the constant force mode of imaging was
used, the height can be anomalously high or low due to large differences in the
interaction potential of the probe tip and the sample. In the present case, it is possible
(though improbable, as will be discussed later) that as the surface coverage of Con A
increases that the probe sample interaction changes leading to a lower measured height.
For this to be the case, the interaction potential energy would have to decrease (become
less repulsive), which would cause the feedback mechanism to underestimate the true
height. This is likely not the case, however, and will be addressed in the following set of

experiments.

126

 

 

 

 

 

Number of pixels

 

 

 

 

0 20 40 60 80 l 00 120
Height (A)
Figure 32: 10 min. deposition of Con A in HEPES on mica. A) AFM image
B) Height histogram.

127

Figure 33 illustrates representative data obtained for a Con A film which was
deposited over a 1 hour period of time. In this case it is apparent that there has been
complete coverage of the surface with the Con A film. In this case we needed to
physically scrape away a small section of the film with the AF M probe, in order to
determine the height of the mica background (as described in the experimental section of
this Chapter). The height distribution plotted in Figure 33 B shows the presence of at
least three different populations of heights. The relative heights obtained, and their
relative probabilities are given in Table XI. The two height populations, corresponding
to the Con A molecules, are observed at 47 i 3 A (87 i 3%) and 67 :t 6 A (8 i 3%). It is
quite interesting that after the formation of a continuous film, that the unfolded
conformation is not observed, nor is the 30-32 A high dimeric conformation. There are a
couple of possibilities that could explain these observations. One of the explanations is
that under these conditions the tetrarneric oligomerization state has formed at the
interface. The dimensions of the Con A tetramer are 60 A x 70 A x 70 A (Protein
DataBank, 1CN1, [13]), which could explain the 67 i 6 A population. The 67 A
population is relatively uncommon, only ~8% of the film. It is possible that the large
population at 47 i 3 A is tetramer which has denatured. Another explanation of the data,
which is consistent with the earlier observations, is that a complete monolayer of
denatured Con A dimers 18 A high was produced (consistent with the observation of the
18 A population after 10 minute deposition). Once this layer formed, a second layer of
Con A dimers may have seeded on it. These dimers adsorbing onto the already deposited
layer of Con A would likely favor the orientation that maximized interactions with the

film (i.e. 30 A height). This would account for the 47 i 3 A population.

128

 

|<,—>|

50nm

 

 

 

 

Number of pixels

 

 

 

0 20 40 60 80 100 i 0
Height (A)

Figure 33: 60 nrin. deposition of Con A in HEPES on mica. A) AFM image
B) Height histogram.

129

Table XI: Data for Con A films deposited in HEPES buffer (pH=7.3).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Deposition time Thickness Relative
# of samples (A) probability (%)
1 min. 0 53 i 10
5 samples 12 i 3 28 i 8
32 i 2 20 i 3
5 min. 0 41 i 5
2 samples 13 i 1 30 i1
30 i 0 29 i 4
10 min. 0 37 i 11
5samples 18:3 63i11
60 min. 0 5*
3 samples 47 i 3 87 :1: 3
67 i 6 8 i 3

 

 

 

* Background in this data is due to an area of protein that was physically scraped away,

as discussed in the experimental section

5.3.2 Growth and orientation of Con A films at acidic pH

This section presents data obtained for the deposition of Con A from a 0.01 M

acetate buffer (pH=4.6) onto magnesium treated mica.

130

Figures 34-37 present

representative AF M images and the corresponding pixel height distributions for

deposition times ranging from 1 minute to 1 hour.

Figure 34 presents data obtained for a deposition time of 1 minute. It is quite
apparent that after only one minute there is significant grth of the Con A film. It has
already progressed to the stage of island coalescence. The height distribution of Figure
34 B clearly shows the presence of at least three distinct populations of heights in the
image. The data, which include other trials of the experiment, are summarized in Table
XII. The two populations observed are centered at 15 i: 2 A and 36 :I: 2 A, which are
within 5 A of those observed during the growth of films from the pH=7.3 HEPES buffer.
It is interesting that the population centered at 36 i 2 A is observed 3-4 times more often
than that at 15 i 2 A. It appears that the Con A is again present as the dimeric and
denatured state, however there is a much higher preference for the native dimer in the
energetically most favorable orientation. This is consistent with the observations of
Chapter four, in which the orientation projecting the 30 A height was highly favored at

acidic pH.

131

 

 

 

 

 

 

 

 

 

17500 ‘ ' ' ' ' ' ' ' ' ' '
- '
.2 15000 ' . -
Q
.5 12500 ‘ -
‘5' 10000“- , -
B 7500 - l -
E i .
E 5000 . A! -
2500 - ft A _
. ,2; ‘3“ ,, ,1 \
0 1 , ......._.... . , - , a
0 20 40 60 80 100 120
Height (A)
Figure 34: 1 min. deposition of Con A in Acetate on mica. A) AFM image
B) Height histogram.

132

Figure 35 presents representative data of Con A deposited for 5 minutes on mica.
It is clear that the film growth has nearly completed the first monolayer. The height
distribution (see Figure 35 B) again shows three populations, the relative heights of
which are presented in Table XII. Similarly to the data of the previous section (see
Figure 32 and Table XI), the height distribution shows a predominance of the 18 A, or
denatured, population. In this case, however, there is a small (~6% of the pixels in the
image) population centered at 34 A, which correlates well with the presence of the Con A

dimer.

133

 

 

I 500 nm I

 

 

 

 

 
  
    

 

 

 

30000 - u - r - I r I .
' O. 1
25000 ‘ 4 ‘
.9. i
Q
.5 20000 ~ —
D- .
c...
c 15000 - .
i-
2 ' l
E 10000 '
= .
Z 5000 -
0 . . . I
0 20 40 60 80 100 120
Height (A)
Figure 35: 5 min. deposition of Con A in Acetate on mica. A) AFM image

B) Height histogram.

134

Figure 36 presents data obtained for a 10 minute deposition on mica. In this case
the layer was nearly continuous and required a physical scraping away of a small area, for
the determination of the mica background. The data for these samples is summarized in
Table XII. Besides the population due to the mica background, there are populations
centered at 41 and 55 A. The predominant population is that centered at 41 A. This
population is similar in height to that observed for the one hour deposition of Con A from

the HEPES buffer (pH=7.3), and might be explained in the same way.

135

 

 

 

 

 

 

40000 t - . . w . - - - - .
35000 - -
30000 -

25000‘

 
  
  

20000"

Number of pixels
{2
O
O
C

10000-
5

O

O

O
l

 

 

O

0 20 40 6O 80 100 120
Height (A)

Figure 36: 10 min. deposition of Con A in Acetate on mica. A) AFM image
B) Height histogram.

136

The data obtained for a 1 hour deposition of Con A from the acetate buffer
(pH=4.6) is given in Figure 37. Figure 36 A is an AFM image of a 1.5 x 1.5 pm area
showing a small area of the film scraped away. There are four populations necessary to
obtain a reasonable fit to the data. The three populations corresponding to the film are
centered at 47 i 8 A, 68 i 5 A and 92 i 1 A. The first two populations correspond quite
well with those observed for the Con A in HEPES film (1 hour deposition) of the

previous section. The 92 A population is likely due to the presence of a multilayer film.

137

 

 

 

 

 

 

Number of pixels
00
O
O
O

 

 

 

 

 

0 20 40 60 80 100 120 140 160
Height (A)
Figure 37: 60 min. deposition of Con A in Acetate on mica. A) AFM image
B) Height histogram.

138

Table XII: Data for Con A films deposited in Acetate buffer (pH=4.6).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Deposition time Thickness Relative
# of samples (A) probability (%)
l min. 0 26 i 4
4samples 15:2 16i3
36 i 2 58 i 6
5 min. 0 20
1 sample 18 74
34 6
10 min. 0 3 i 1*
2 samples 41 86 i 2
55 j: 1 11 i 1
60 min 0 4 i 2*
2 samples 47 i 8 16 i 4
68 + 5 62 i 9
92 i 1 18 :t 3

 

 

 

 

 

* Background in this data is due to an area of protein that was physically scraped away,

as discussed in the experimental section.

139

5.3.3 Kinetics of the deposition process at physiological and acidic pH

Examination of Figures 29-37, and Tables XI and X11 clearly shows that the rate
of adsorption of the Con A to the mica surface is faster at the lower pH, where the protein
molecules are positively charged. In Chapter four it was noted that the rate of adsorption
did not appear to change with pH, for deposition of isolated molecules from very dilute
solutions, however, under the present conditions the pH has a significant effect on the
rate. For very dilute solutions, it is hypothesized that the electrostatic interactions
between the protein molecules and the mica surface are shielded by the solvent. In the
case of more concentrated protein solutions, the electrostatic interactions between the
protein and the substrate appear to become important. This may be simply a fact of the
larger number of molecules which would be in close proximity to the surface at any given
instant, as compared to the case of the solutions of Chapter four. In the case of the dilute
solutions the rate of adsorption is likely limited by the Brownian motion, or diffusion of
the protein molecules toward the surface. In the present case, however, one of the

dominant rate determining factors appears to be electrostatic attraction.

The kinetics of the adsorption process were investigated by plotting the amount of
amount of substrate surface (i.e. 100% - % surface coverage) present as a function of
deposition time. It was found that a plot of the natural logarithm of the % uncoated
surface observed versus time gave a straight line (see Figure 38). This indicates that this

adsorption process is first order or pseudo-first order. The negative slope of the plot

140

gives the first order rate constant in units of min". The data were fit by linear regression,

and only the submonolayer films were included in the plot.

141

 

 

5 i O pH=7.3

 

 

00-h

ln [l-coverage]
- [\J

 

 

 

 

Time (min.)

Figure 38: First order kinetics plot of Con A data at two different pH values.

142

The rate constant for deposition from the pH=7.3 HEPES solution was found to
be 0.08 min", while the rate constant for the pH=4.6 acetate buffer was 0.30 min]. This

is a difference by nearly a factor of four in the effective rate of film growth.

5.3.4 Morphology of Con A thin films at the water/mica interface

It has been previously reported that thin films of Con A are quite rough [4,5,7].
The current study has found them to be significantly rougher than what would be
expected if the molecules were forming ordered arrays on the mica substrate. The
Proscan (Park Scientific Instruments) image analysis program was used to calculate RMS
and average roughness for the continuous thin films deposited on mica. The results of
this analysis are given in Table XIII. Typically values of 3-5 A of average roughness can
be expected for mica, over the typical 2 x 2 pm scans acquired with the AFM. The
roughness data presented in Table XIII was calculated for 025-] pm areas, where the
film was seen to be continuous and not containing any large aggregates on the film
surface. It is clear that the films are typically rougher than the mica substrate, and that
the films deposited from the acetate buffer are significantly rougher than those deposited

from HEPES.

143

Table XIII: RMS and Average roughness for Con A thin films.

 

 

 

 

Solvent: HEPES pH=7.3 Acetate pH=4.6
Roughness: RMS Average RMS Average
(A) 7 5 12 9
0' 2 1 2 1

 

 

 

 

 

 

 

In order to get a better idea of the film morphology, cross-sectional slices of the
film image were plotted. Representative line profiles, for fihns deposited from the
different buffers, are plotted in Figure 39. The data presented in these figures has been
normalized, such that the minimum height value of the line is set to be zero. The films
deposited from the HEPES buffer tend to have heights that vary between 0 and 30 A,

while those deposited from acetate typically vary from 0 to 60 or 70 A.

144

 

A 80, ~ . - . 4
70-

”‘60-:
V50:

 

i
'53 30-
m 20{

 

 

 

0 0.25 0.50 0.75 1.00
Distance (u m)

 

 

 

 

 

 

 

 

 

 

 

0 ' 0.25 I 0.50 f 0.'75 ' 1.00
Distance (14 m)

Figure 39: Line profiles taken from AF M images of Con A thin films, deposited at
A) pH=7.3 and B) pH=4.6.

145

The origin of the differences in the morphology of these films can be understood
in part by consideration of the difference in the deposition kinetics between the two
buffer solutions used. As discussed in section 5.3.4 Con A deposits onto mica at a
significantly faster rate from a buffer at pH=4.6 than it does from a solution at pH=7.3.
Because the molecules are depositing at a much faster rate from the pH=4.6 solution,
there is less time to maximize interactions with the surface and adjacent Con A
molecules. This may partly explain why these films appear rougher than those deposited

at a slower rate.

Another possible explanation for the morphological differences may in part be
due to the formation of multi-layer films. The data presented in Tables XI and X11 show
that films formed from one hour depositions show features at approximately 47 A and 67
A heights in both buffer systems used. The film deposited from the acetate buffer,
however, has another feature in its height distribution, appearing at 92 A. Additionally
there is a difference in the relative proportions of each orientation. In the case of the
HEPES system the 47 A heights account for ~87 % of the film, while in the acetate
system only ~16 % of the film has this orientation. This inverse proportionality also
appears in the relative amount of the 67 A populations. This data is consistent with the
faster kinetics observed, and the formation of multi-layers of Con A on the mica surface,

which appears to account for the differences in the observed film morphology.

146

5.4 Discussion

The data presented in section 5.3.1 and 5.3.2 examined the distribution of
orientations for Con A molecules at the water/mica interface as the growth of thin films
progressed. Multiple heights were measured, even at very early times in the film
formation, where islands of the protein molecules had not yet grown together to form a
continuous layer. The heights which are observed at the various deposition times are
consistent with previous studies of Con A film thicknesses [3,6,8,10]. In the previous
studies of Con A films adsorbed to solid supports, the techniques used (i.e. SFA, QCM,
XPS) provided only average thickness of layers. The current study has used the AF M to
examine layer thickness during film formation, providing not just average thickness, but
also showing the presence of multiple conformations or orientations of Con A at the

interface.

The growth of the Con A films appeared to go through stages. The three stages
observed in this study are schematically illustrated in Figure 40. In this figure Con A
dimers are represented by ellipsoids, with dimensions corresponding to the dimensions of

the native and denatured form of the Con A dimer.

147

 

 

A
A @3311V

 

 

%@ 0]) 1%

 

 

Beeeiiiéég

 

 

 

 

 

 

 

 

 

15x30A

 

15x45A 30x45A 30x70A

 

Figure 40: Possible orientations of Con A molecules at the water/mica interface, as a
function of time: A) early growth, B) intermediate growth, C) continuous
film (as discussed in the text).

148

 

In the first stage (see Figure 40A) there was formation of two predominant
orientations or conformations observed at 12—15 A and 30-36 A. Although the relative
amount of each of these orientations was nearly 50%, much of the 12-15 A population
can be attributed to the tip convolution artifact. This is consistent with the presence of
Con A dimers, as was observed in Chapter four for Con A deposited from very dilute
solutions. The observation of dimer species was expected under acidic conditions, but
the presence of dimers deposited from a HEPES buffer at pH=7.3 was surprising. It is
well known that in solution Con A favors the tetrarneric organization when the pH is 7 or
above, while the dimeric species is favored below pH=6 [11,12]. Con A is also known to
form dimers when it is diluted below 5 ug/ml [1], as was the case in the solutions used in
Chapter four, but observation of dimer species under the present conditions is not

generally predicted.

The second stage (see Figure 40B) of film growth involves a change in the
orientation of the Con A molecules, where the predominant species have a height of
about 18 A. This is consistent with the presence of unfolded or denatured Con A (see

Chapter four).

The third stage of growth (see Figure 40C) occurs when the protein molecules
have formed a continuous monolayer, and/or they have begun to seed multi-layers. The
predominance of areas with heights centered around 47 A and 67 A could be due to a few
possible orientations of Con A dimers on the surface, as illustrated in Figure 39C. The 67

A measurement, however may also be indicative of the formation of Con A tetramer at

149

the interface. The current results cannot distinguish between the different possible modes
of Con A surface orientation for these films. This could be examined in the future by
careful studies of saccharide binding affinity as a function of film growth. Since the
location of the metal centers (a.k.a. sugar binding sites) are well known, it may be
possible to determine the orientation of the Con A molecules, and possibly whether there
are tetrarneric Con A molecules present. These results indicate that Con A tends to exist
at the water/mica interface as a dimer. It appears that this variation from the tetrarneric
form of Con A, which is typically observed for Con A solutions at neutral pH, is a result

of the chemistry of the interface affecting the stability of the tetramer.

The kinetics of the deposition process is dramatically affected by the pH of the
deposition solution. The rate of deposition appears to be almost four times faster when
deposition takes place from an acetate buffer at pH=4.6, as opposed to the HEPES buffer
at pH=7.3. This is in contrast to the results presented in Chapter four, where the rate of
deposition from dilute solutions ~102 ng/ml was not significantly different for the pH=7.3
and pH=4.6 solutions. The likely explanation for this is that in dilute solution, the
electrostatic interactions between the protein and surface are shielded sufficiently that the
deposition process is diffusion limited. In the case of more concentrated solutions (i.e.
~102 ug/ml) the deposition process is not diffusion limited, and the electrostatic attraction
between the positively charged Con A (at pH=4.6) and the negative surface charge of the

mica substrate increases the rate of deposition.

150

This difference in the rate of adsorption with pH can also begin to explain the
difference in the surface morphology of the films. The films produced by deposition
from the acetate buffer system were significantly rougher than those deposited from
HEPES buffer. The faster deposition may not allow sufficient time for the Con A
molecules to maximize their orientation on the surface before more molecules can come

in and deposit.

The study of Con A thin films by AF M has not been previously published,
although this system has been interrogated and characterized by a number of other
techniques. This study links together what was learned about the orientation and
structure of individual Con A molecules at the water/mica interface (see Chapter four and
Appendix 1), and what has been determined by previous studies of the bulk properties of
Con A thin films. It is hoped that this information can be used to better understand the
function of and to design better saccharide sensing devices, and possibly develop new
Con A-saccharide based bioassays. Furthermore, the approach used in this study
demonstrates that valuable information can be obtained from AF M data, despite the

lateral tip convolution problem.

151

5.5 References
1. Gordon, J. A.; Young, R. K. J. Biol. Chem. 1979, 254, 1932.
2. Janata, J. J. Amer. Chem. Soc. 1975, 97, 2914.

3. Afshar-Rad, T.; Bailey, A. 1.; Luckham, P. F.; MacNaughtan, W.; Chapman, D.
Biochim. Biophys. Acta 1987, 915, 101.

4. Barnes, C.; D'Silva, C.; Jones, J. P.; Lewis, T. J. Sens. Act. B 1991, 3, 295.
5. Haas, H.; Mohwald, H. Thin Solid Films 1989, 180, 101.
6. Gallinet, J.-P.; Gauthier-Manuel, B. Eur. Biophys. J. 1993, 22, 195.

7. Safar, J.; Roller, P. P.; Ruben, G. C.; Gajdusek, D. C.; Gibbs, C. J. j. Biopolymers
1993, 33, 1461.

8. Lvov, Y.; Ariga, K.; Ichinose, I.; Kunitake, T. Thin Solid Films 1996, 284-285, 797.

9. Kremer, F. J. B.; Engbersen, J. F. J.; Kruise, J.; Bergveld, P.; Starmans, D. A. J.;
F eijen, J .; Reinhoudt, D. N. Sensors and Actuators B 1993, 13-14, 176.

10. Fitzpatrick, H.; Luckham, P. F.; Eriksen, S.; Hammond, K. J. Coll. Inter. Sci. 1992,
149, 1.

11. Reeke, G. N., Jr.; Becker, J. W.; Cunningham, B. A.; Wang, J. L.; Yahara, 1.;
Edelman, G. M. Structure and Function of Concanvalin A. In Concanavalin A;

Chowdhury, T. K., Weiss, A. K., Eds.; Plenum Publishing Corp.: New York, 1975.

12. Liener, I. E. Isolation and Properties of Concanavalin A. In Concanavalin A as a
Tool; Bittiger, H., Schnebli, H. P., Eds.; John Wiley & Sons:, 1976; pp 2.

13. Shoham, M.; Yonath, A.; Sussman, J. L.; Moult, J .; Traub, W.; Kalb, A. J. J. Mol.
Biol. 1979, 131, 137.

152

APPENDIX 1

Re-print of publication

153

J. Phys. Chem. B 1998, 102, 1649—1657 1649

Imaging the Molecular Dimensions and Oligomerization of Proteins at Liquid/Solid

Individual Concanavalin A (ConA) molecules have been imaged at the liquid/solid interface with an atomic
force microscope (AFM). Three- dimensional sizing with very high resolution (<5 A) has been obtained by
a novel approach based on height distributions. which avoids the tip convolution effects which normally
affect scanning probe microscopy techniques. Each height measurement correlates to a particular molecular
orientation on the surface. A large number of such measurements provide a statistical ensemble of orientations.
The complete height distribution reflects the three-dimensional size of the protein sample and hence its ternary
and quaternary structure. A surface adsorption and orientation model. based on a minimization of surface
adsorption energy, is proposed. This model is in good agreement with the observed height distribution of
Con A molecules at the liquid/solid interface. Analysis of Con A and succinylated Con A molecules on
mica demonstrates that Con A dimers are the prevalent species at the liquid/solid interface. This is in contrast
to the tetrameric organization of Con A normally observed in solution. The new possibilities opened by
height distribution analysis on the physical characterization of biomolecules at interfaces are also discussed.

Interfaces
Mark J. Waner,+ Martha Gilchrist,’ Melvin Schindler} and Marcus Dantus“
Deparonenrs of Chemistry and Biochemistry; Michigan State University. East taming. Michigan 48824
Received: October 2, 1997; In Final Fonn: December )0. 1997

Introduction

The behaviorofproteinsatthe solid/liquid interfacehasbeen
ofgreatinterestbeeauseoftheftmdarnemalrolethatmernbranes.
cytoskeletons. and other interfaces play in cells. as well as for
the more practical aspects related to the development of
biological assays. biocompatibility of materials. and protein
processing."3 The number of techniques available to quanti-
tativelymeastnestructtneandstrucmralchangesofindividual
proteinsatinterfaees. however. isverylirnited. lnthispaper
we present very high- resolution (<5 A) three-dimensional
measurements of Concanavalin A (Con A) molecules at the
water/mica interface. The measurement of the molecular size
allowsustodeterminetheoligomerizationstateandanestimate
ofthestnfaceuismptimener'gyfordiffereruproteinoriemations
at the interim.

'l'headsorptionofaproteinonasolidsurfaceisaeomplex
process.‘-s Ctnrentmodelsidentifytwostepsintheadsorption
process. The first step involves the eollision of the protein.
which is translating by Brownian motion. with the surface. If
the energetic benefit of the removal of surface adsorbed water
isgreaterthanthethennalenergyoftheproteinmoleculethen
itrernainsonthe strrfaeemtherwiseitrenn'nstosolution’ Once
theproteinisadsorbedtheseeondstepisthwghttoinvolvea
conformational change that maximizes electrostatic interactions.
local interactions between electron donor moieties in the potein
withelectron meeptorsonthe Stu-face. andmaximizationofvan
der Waals attractions." It is often assumed that this second step
involves changes in the protein structure. akin to denaturation.
‘lhere is now some experimental evidence that “compact"
globular proteins maintain their native structure when adsorbed
and that only some “soft/flexible" proteins suffer some degree
of denattu'ation upon adsorption. especially on hydrophobic
surfaces." The wood step. therefore. involves primarily an

 

' Corresponding anther. e- -mail: danmseeernmsuedu.
'Deparmof

ancestry
*Deparuutt of Biochemistry.

orientation process that is thermodynamically driven and yields
proteins in the most energetically favorable orientation. Inde-
pendent evidence of protein orientation at interfaces has been
obtained by Lee and Saavedra'

Since its recent introduction.9 atomic force microscopy (AFM)
has been extensively used for imaging biological mlearles.'°‘"
The AFM technique allows the examination of biological
specimens without the need for chemical fixation/dehydration
or growth of crystals, which IS normally associated with other
high-resolution techniques. "“5" In fact. samples may be
imaged with AFM tmder liquids at physiological concentmions
and pH. ‘7 A ntunber ofgroups have utilized the high resolution
(~1 A) of the AFM for examination of 20 crystallization of
protein molecules.”""’“ Others. especially Hansma. Busta-
mante, and Henderson. have studied isolated proteins'uz'” or
DNA molecules‘z-29‘33 and their complexes.“ Recently. AFM
hasbeenusedbytheMarchantgrouptoobtainvaluable
structural information for the multimeric Von Willelnmi factor
(MW ~ 260 kD. monomer). ”-3537

lnthecaseofsmallerglobularproteinsandothersrnalltrnee-
dimensionalstructures. however. AFMhasbeenlimitedbyits
inherent low resolution (IO-20 run) in lateral measurements.
The difference between height and lateral resolution (~l vs
~100A.respectively)isduetofinitesizeoftheprobetip
(~10-50mradiusofcurvanne)andthesamplebeing
examined"‘” (see Figtn'e 1A). This “convolution” efl'ect
occurs regardless of imaging conditions (vacuum. ambient. or
under liquid). Some groups have proposed making sharper
probetipstoreducethisproblem."whileothershaveexamined
waystofirstcharacterizetheshapeoftheprobetipandthen
mathematically “deconvolute” the AFM data” Because the
heightmeasmementsobtainedbyAFMaremaffectedbytbe
sizeoftheprobc.resolutionoflerlesscanberealizedin
this dimension.

lnthissmdy.weextmdtbemilityofAPMtothequantimive
measurement of molecular dimensions with angstrom resohrtion

81089-5647(97)03221-5 CCC: 515.“) O 1998 American Chemical Society
Published on Web 07/04/1998

154

1650 J. Phys. Chem. 8, VOL 102, No. 9. I998

 

 

 

i1gue1.(A)SchmaticrqnesentationoftheAl-‘M
dratheightmeastuementsarenotafi’ectedbytip
convohrnom'l'haecan nunorindiscrwancy themsuredheight
(awhmflnprobeofndi‘uHRHsdisplacedbyadimceuHrom
dreapexoftbeprotdn.ofradius (r). lnourmeasurementsthrs
amountsloaoj‘irmflce text). (B)ltepresentative
drawingofConcanavalinAdimasar-nngedin urthree
micastn‘facewithanAmpobeposirionedontopofoneofthc

 

molearles.The_uien_tar:ionmodel proteinsnnybe
crudely ‘-~‘rxedicrstheir
orientationonmestnfaeemordingtotherelaiivedimensionsofthe
rectan parallelqriped.The sinofthemolecules.tlreprobe.and

themicalanicearedrawntoseale.

atliquid/solid interfaces. Such measurementsareof particular
biological relevance since the preponderance of biological
activitiesofenzymesandtlxirbindingpropertiesoccnron
membrane or cytoskeletal surfaces. To avoid the loss of
resohrtionthatisnormallyintroducedintoAl-‘Mmeastnements
uaresultofprobetipgeomeny.wehavedeviwdameasrne-
mentstrategythatmeasurestheherghtsofanemembleof

flat surface of mica (Figure 13.11:: techniquelras beentermed
surface oriented molecular sizing (SOMS) because it utilizes
the statistical c distribution of orientations for

’ l ‘ ‘ ‘ermmm-
ofproteins. 'l‘heuseoftheprojectedheightofindividual
adsorbed protein molecules I“as the defining parameter for

i'qacué'nn
onrnobetipgeometryandutilizestheangstromresolutionof
theAFMinthezdirection. Mitisadvantageoustouse
abrudernpfmmakingthesemeasnemanstlxrebyminimiz-
ingtheappliedpresstnebetweenthetipandtaoteinand
cleansing the chance of underestimating the height of a
urolecule duetothedigitintioninherentinthemiaoseope.
Furthermore. this technique is generally applicabletoeontact
andinterminentcontactmodesofimagingunderambientor
liquid «auditions. The relativeease.speed.andresolution with
whichthesemessurementsofmoleculardinrensionmaynow
be performed make this technique particularly useful for
examining multiple conformational equilibria of adsorbed

155

Waner et al.

subunits and the changes in shape and assembly state
of multi-subturit proteins resulting from oligomerization and
aggregation at liquid/solid interfaces. Such changes may be
systematically examined as a function of protein concentration
and ligand or allosteric activator binding. Such measurements
could help to determine whether single amino acid substitutions
affect protein activity as a consequence of an induced change
in the oligomerization state of the mutant protein or as a result
of a chemical change in the active site. We will present
evidence of this phenomenon in a future publication.

We present in this article a study of Con A at the liquid!
solid interface. Our findings demonstrate that. at the water/
mica interface. Con A exists primarily as a dimer. which is in
contrast to the tetramaic form which is typically observed in
solution. The dimeric structure is confirmed by measurements
on succinylated Con A. We also explored the possibility of
denattrration at the liquid/solid interface by compmison to
measurements on sodium domeyl sulfate (SDS) neared Con A
samples. Confirmation is also obtained horn measurements on
ferritin protein molecules. which are found to agree well with
molecular dimensions derived from images obtained with X-ray
difl'nction and electron microscopy. The application of a
sin-face adsorption and orientation (SAO) model in conjunction
withAFMtothesutdyofproteinsu'ucnueandoligomerization
at the liquid/solid interface in the presence of activators.
inhibitors. and denaturing agents will be dismissed

Experimental Section
Materials. Colloidal gold particles were obtained from
Sigma (St Louis MO) and used without further ptnification
Three sins of gold particles were used for the calibration: 50
a: 8 A (3.8 x 10” par'ticles/mL). 90 :t 12 A (6.0 x to‘2
particles/mm. and 180 :k s A (6.7 x10“particlesImL). 'nie
cbtermined

Polysciences (Warrington, PA) as a 20 mglml. solution.
ConcanavalinA(ConA)wasobtainedfromBoehringer-
Mannheim (Indianapolis. IN) as lyophilized solid. Succinyl
Concanavalin A was obtained from Sigma (St. Louis. MO) as
a95%proteinlyophilisate. Proteinptnity waschecltedbyS DS
polyacrylamidege le esislecnophore .l-IEES wasobtainedfrom
Sigma (St Louis. MO). and the MgClz was ACS grade from
Columbus Chemical Industries (Columbus. OH). The water-
used tn preparing solutiors was purified with a MilliQ Type 1
MilliPore water purification unit (19 MW resistivity). The
eoncennxtionsofConAandsuccinleonAstocksolutions
were ' using Am with an extimtion coefficient of
1.37 ml. mg" cm". ‘°

Preparation of Samples for Height Measuremems with

usedmuscovite micaobtained fromWard's Nannal Science
Establishment. Inc. (Rochester. NY.)asthesubstrate. Inmcst
caseswefotmdaflamessbetterthanZleverareasoftensof
nricrmsquared. Whifreshlycleavedmricahasanetnegative
surfaeecharge. Treatingafreshlyclcavedsurfaceofmicawith
aSmMMof‘L L

K’iorrswithMg“nraldngthesurfacemorepodtivelycharged
The been

solvedin lOli-IEPESsolurionatpH73. (Them
used in each case is specified for cub protein.) Twenty-five
microliters of protein solution is depodted onto the mica
Twerrtyminmeslater.tlresmfaceisgentlyrinsedtwicewith

Oligomerization of Proteins at Liquid/Solid Interfaces

 

x50) cureapond Indifferent orientations ofCon A dimers (see text).

200 ml. portions of MilliQ water and allowed to dry for several
hours (at room temperauire and 35—50% relative humidity).
This preparation yields a dispersed population of individual
protein molecules when imaged by AFM (Figure 2). The
concentrationofproteinbasbeenchosentoprovidealarge
population of individual protein molecttles and to avoid the
mding of aggregates and monolayers. Overall. the protein
coverageisalwayskeptbelow 1%. Intbepresertceofarelative
humidityof~30%. ithasbeenshownthatthereisamonolayer
ofwatermaintainedontlnmicasurfacewhichanacttohythte

forthisworkwasanAutoprobeCP scanningprobemicroscope
(Park Scientific Insuumertts). with a 5 pm high- -resolution
scanner. Height measurements may be obtainedfrom AFM
studiescarriedoutbycontactmodeoriraerrrtiuentcontactmode
in ambient or under liquid conditions. For this smdy the AFM
dataweremordedincontactmodeunderconnolledambiettt
conditiats (~40% humidity and 21° C). Underthese conditions
amonolayerofwaterexistsatthemicasurface.“andtleprotein
moleculesarefullyhydrated Itwasfoundthattheprotein
molecules adhered to the mica well enough for measurements
only when the humidity was less than 50%. probably because
high humidity increases the attractive capillary forces between
probe and sample.”-""7-‘3 Humidities of less than 30% were
avoiMd to prevent protein dehydration and possible structural
deformatiorts. The effects of controlled hydration on proteins
have been studied by sunning nrnneling microscopy (STM).““5
Thelargest ctnngeswereformdtooccurasthehumidity
dropped below 30%. Unfortunately those measurements did
not provide quantitative structural information because the
contrast mechanim (tunneling current) ts highly dependent on
the amount of solvent.

Sbarpemd Microlever cantilevers (Park Scientific Instru—
ments) having a spring constant of approximame 0.05 Nlrn and
an integrated silicon nitride probe with a radius of curvature of
approxirnatelyZOnm.accordingtothetmnufacnner.wereused
astaobes. Thescannerwascalibratedforvertical measurements
using Tobacco Mosaic Virus (cylindrical shape: diameter 18
am) (Amaican Type Culnne Collection) and colloidal gold
particles (5 nm) (Sigma. St. Louis). Horizontal measurements
are calibrated using a twodimensional grating with 1000
feantres per millimeter.

AFM tmageswere acquired in constartt force trtode. in which
thePZ'I'scannermovesinthezdirectionsoastomaintaina
oonstantforcebetweenthesampleandtheprobetip. To

156

J. Phys. Chem. B, VoL 102. No. 9. I998 165]

determine the nrinimtmt force to achieve stable imaging with
negligible or no measurement distortion. we imaged ferritin
protein molecules with a range of applied forces. At low
imaging forces we accurately the well- characterized
dimensions (a sphere with 120 A diameter) of this protein ‘7
At higher imaging forces a deformation of ~40% was observed
We compared the force meastiremerrts on ferritin with similar
ones on gold particles. As expected. the gold particles were
more resistant to deformation due to applied pressure. The
minimum force for stable imaging was found to octntr witltin
the electrostatic repulsive regime and was constant for all
images. The total force applied between the tip and the
cantilever under otir imaging conditions contains a large
attractive or capillary force (~lO"-lO'° N. depending on the
ambient humidity") in addition to the overall repulsive force.
The capillary forces between tip and sample may be minimized
by imaging under liquid. These typeso of measurements in

ththe proposed SOMS technique may be ideally
suitedforthemsnrdyofproteinsatinterfacesandareofgreat
interest in our group.

Giventhattheoverallacctnacyofourmethodislimitedby
the roughness ofthe substrate to ~3—5 A. force distortions can
be neglected except for-very soft and large (>100 A) feanires
ofproteins(notthecase intheresultspresentedinthisarticle).
lnthosecaseswecanuseintermitientcontactmodescanning
techniques unda liquid. where the tip-protein irtteraction forces
canbeftntherminimized. ”ScansizeforinrageswasZOx
2.0mmd512 x 512pixelsatalinearscanrateofljliz.
which is equivalent to 3 urn/s along the fast scan The fairly
slowlinearscanratewaschosentohelpminimizethelateral
foraesthatoccurbetwcenthetipandsample duringtheraster
scanning process. The lateral forces can also be fttrther
minimiud by using intermittent or nonconiact modes of
imaging. which are currently being investigated.

Each image of individual protein molecules corttained ap-
proximnely IO-SO molecules/um! Individual protein mol-
eculesaremeasuredbytaltingthemaximurnheightofthe
molecule minustheheightofthelocalbackground Fachdata
set. consisting of more than 1000 individual height measure-
ments. is tabulated in 3 A birts resulting in a height distribution
histogram Thechoiceofbinsizeisbasedprimarilyonthe
‘independent of
this choice. A combination of Gaussian functions is then used
tofitthedata. 'l'hisfittingwasdoneiterativelyusingcotntraints
onthethreeparametersofatypicalGaussianfunction(i..e.
center position. width. and amplitude).
amplitudes of the ftmctiorts were allowed to vary freely. The
centerpositionsofthefunctionswereallowedtovaryinde-
pendently. 'I‘hewidthsofthefunctionswereconstrainedtobe
between8and 13A. 'I'hefttnctr'onscenteredat59khaving
the lowest amplitude. have the greatest uncertainty. and their
positionwasfixedwhiletheotherparameterswereoptimiud
ThecentersoftheGaussianftmctionsaretakenasthemeanned
molecular dimensions. while the area tracks the curves is
proportional to the probability with which a givert moleailar
orientatiort is measured Consistency in the fitting paratrteters
ischeckedbymultiplerepetitionsoftheexperimunsondifferurt
days. 'l'herestiltspresentedhereconformtothisstrictcriteria
to less than 5

The final analysis consists of two different procedures: (a)
determinatiort of molecular dimensions and (b) assignment of
quaternary strucnne (when different oligomers are observed).
Toassigntheobser'veddimensionstothemacrornoleuilea
modellhattakesintcaccounttheorientationsofmolecubsat

non-It:

1652 J. Phys. Chem B. Vol. 102. No. 9, I998

 

 

 

A 80 L B — .5 Janet-A1 ‘
°\° — -l0JI-ol-K1
"’ t
>. 60
o
5 .
= 40
5'
E .
0 L . l J I A

 

 

 

0 20 40 60 80 100

Height (A)

parallel of
45AM:=75A.(§)Calntheight

asACon aspedicted
bytheSAOmdelforvahtesol‘AGoof-Sand -lOJ/(molA’) Notice
thatfmhighapoteinmbsnteaffinityproteinstendtoorieotsothat
thenlngestnnfaceuuhesonmemicathaebyprojecdngthcir
sbmest dimension perpendicular to the surface
interfaces is required (vide infra). Protein dimensions obtained
by this method are evaluated for consistency with other
biophysical data available for the protein under investigation.
such as molecular weight For example. the volume of each
oligomer is calculated from the assumed dimensions. Using
thehtownniolecuhrweightoftheprmdnbeingsutdicda
volume is then calculated (assuming a density of ~1 glcrn’)
andcornparedtothevolumepredictedbytheassumeddimen-
stons.

Intheabsenceofchemicallyspecificinteractionsofthe
proteinmolecules withthesufamtheensemble of molecules
isexpectedtoasstnneathermodynamieally drivendistribution
of orientations A smface adsorption and orientation (SAO)
ntodelisproposedwhichtreatsthis interfacial protein orienta-
tionasanoptimizationoffreeenergy.

'l'hernodelasstnneseachproteincanbeenclosedbya

pedofdirnensionsmb.andc.whereasb_<_c
(Figure 3 A). This geometrical shape is compatible with
ellipsoidal models in the limit of well- rounded edgesand
corners. F
thedegreetowhichtheedgesareroundecLA favorable
interactionbetweentheproteinmoleculesandthehydrophilic
micasurfacereaultsinorientationsthatmaximizethesurface
areaofcontact. Thereforeaproteinorientationbavingthebc
madmbedonthemicaWherebandcarethelargest
dimsionsbydefinition)isconsideredtobethemostenerged-
cally favorable.yieldingamaximum probability formeasured
heights correspondingtodimensionauee Figure 3B).

'l'heotientationfreeenergyaftertheirtitialadsorptionstep
isasrumedtobedependentontheproteinsurfaceincontact

157

themeasured height is independento of

Waner et al.

withthesubsu-ate. Intheabserweofparticularsurface recogni-
tion motifs. the free energy of orientation for each surface is
given by

15ka a 13ng (I)
ACKM a AGmA“ (2)
A013,... = ACM» (3)

where AG,t is the free energy of orientation per urtit area of
the protein in contact with the substrate (in this case bc). andA
is the surface area of the side identified by the subscript.

The probability of orientation such that height a. b. or c is
measured depends on the different free energies of orientation.
eqs 1-3. These may be estimated by a Boltzmann distribution
“$138

2 —AG"‘. k
P(a)=—————-°xp( Q “‘1 n (4)
—A ”IN/k
P(b)=2exp( QC“ 7) (5)
—A k
P(c) =———2 a“ 96:“! 7) (6)

wherethefactoronarisesfromthesymmetryofthe
parallelepipedkisBoltznurm'sconstanLTisthe

and Q is the canonicalpartition function. 'l'hesecxpressions
aresimpliftedbytakingtheratio betweentheseprobabilities
andhyasstnningthatallorientationenergiesperunitareaare
equal(asexpectedinthcabsenceofspecificrecognitionsites
inthesurfaceortheprotein)sothatAGosAGk=AG.=
AGJWELIIIISM

A —A AC
P(c)/P(b) = exp([“—k;i]-——°) (7)

[A — A,c AG
P(b)/P(c) = cxp(—$TT—]——°) (8)

[A —A ]AG
P(c)/P(c) = cxp("’—kT"‘—°) (9)

wherethefreeenergiesoforientationhavebeensubstiurtedby

‘ccneeporaiingexpreasion(eqs12.and3). Giventhese
equations it is possible to predictthe probabilities of each
protein orientation on the surface given its dimensions. It is
also possible to start with the experimentally observed dimen-
sions (a. b and c) and their probabilities (P. P... and P,) and
obtainavalueforAGousingeqs 7—9. Themagnitude ofAGo

’ the relative probabilities of measuring each of the
threentaindimensionsofaprotein figure3illtmratesthe
theoretical height distribution for a parallelepiped. of the same
dintensiortsasConAdimer.forvaluesofAGodifferirtgby
only a factor of 2.

The SAO model. therefore. provides a mathematical frame-
worktoanalyzetheheightdistributionprofllesandobtainthe
relativeamotuusofeadtoligomaicstate. c.g..monomer dimer.
tetramer (see the data analysis section) In addition to the
obvious height differences for the oligomers. we observe that
the probability of measuring those values varies extensively
depending on the configuration. Therefore. for proteins of

Oligomerization of Proteins at Liquid/Solid Interfaces

 

Y I I I T

20-

Number of molecules
1 A l 44 A l A_

10-

 

 

C
a

 

 

 

100 150 200 250
Eda-HA)

Figure 4. AFM height distribution data for colloidal gold particles
adsmbedonminReaultsoftheleast-squareaGaussianfitstothedata
maybefoundin‘l'ablel.

TABLEI: AFMBeightMeasurementsofColltidalGold
ParticlesCernparedtotheValueaObtainedUsingTEM

V

 

sine determined sine determined particles

by TEM (nm)' by AFM (am) measured
5.0 :h 0.8 5.0 :L- 0.8 353
9.0 i 1.2 10.1 i 1.6 207
18.0 :t 0.8 19.4 d: 2.0 87

‘TEManalysissuppliedbythemanufacnuer.

unknownsuttcnnethismodelisusefulforthedetermination
of tertiary and quaternary structure.

Most proteins are asymmetrical or oligomeric; therefore. for
agivatptmcimdifferentdimensionsareobtainedconeaponding
to different protein orientations (and/or oligomers) on the
substrate. Given alarge enough sarnpling.on theorderof 103
individual measurements. a distribution of measured heights is
obtained that is characteristic of the molecular dimensions of
theprotein Tointerprettheheightdisnibutionanddecompose
itintotbedimensionsoftheproteinweusetheSAOmodelto
predicthowmoleculesofspecificdimensionsadsorbtothe
substrate.

Results

Height Measurements of Collddfl Gold with AFM.
Monodispersed spherical colloidal gold particles were initially
employed to demonstrate the subnanometer height measure-
rnentsofAFMthatareminimallyinfluerudbyprobegeom
etry.“ The particles were chosen because their sizes were
cornparabletothatofproteinmolearles. Twentyfivemiaoliters
of sample (10‘1- 10” pmicles/ml.) was deposited onto a freshly
cleaved.nngnesiumion-treated~lun3pieoeofmicaandrinsed
twicewichOOmLofMilliQwater.aftera20minincubation
time. Thispreparationresultedinauniformdistributionofgold
particles on the mica surface. The results of height measure-
memsontheseparticlesarepresentedinFigme4andTable 1.
'l'hethreedominantpeaksinthehistogramareingood
agreement with the TEM analysis performed by the manufac-
turer. Forthel8nmgoldwefoundagreaterdeviation
presumably because some particles had slightly different shapes
(nonspherical). These types of variations are more common
for larger particles. In general. the similarity between the
standarddeviationscalculated forbothTEMandSOMS
measurementsCl'able l)ntggeststhatthedominantsomeeof
errorinthesemeasurementsmayresideinthesizedisuibution
of the particles themselves rather than the measurement
techniqueorAFMprobegeometryasdiscussedinMaterials
andMethods.

ArtalyaisofFer-r'itinattheuqmdsolidlaterface. Totest
thisapproachforthedeterminationofthemacromolecular

158

J. Phys. Chem. 8. VOL 102. No. 9. 1998 1653

 

l A A g; a I

A

s-a
C
L

4

Number of molecules

 

 

 

 

 

I) ran no m 150 ill
Height (A)
Figure 5. AFM height distribution data for ferritin. The data are best
fit by three Gaussian functions centered at 97 A. 115 A (the intact
protein). 133 A (ferritin mm). and 144 A. The two assigned peaks
account for 85% of the data. .

dimensions ofproteins. we performed measurementson ferritin
for which high-resolution structures had been previously
obtained by both electron microscopy" and X-ray crystal-
lography.“ Ferritin (horse spleen) has been demonstrated to
be a spherical molecule comprised of 24 polypeptides (ratios
of homologous H and L chains) and bound iron (Fe’+)."’
FerritinhasamolecularweightofMSkDaandithasbeen
shown to have a diameter of 120 A by electron microscopy"
and X-ray crystallography.‘8 AFM measurements of samples
prepared from a 20 ug/mL solution of ferritin yielded a
distribution of measured heights with a dominant peak at 115
AMgmeStmexcelleruagreetnentwiththepreviatslyreponed
results for ferritin. The additional peaks in the height distribu-
tionindicatethepresenceoffenitin trimers (133 A)andptoteins
of smaller oligomeric composition. (Such ferritin strucntres
have been previously observed with adsorbed ferritin samples
onpolishedTiandMousingSTM.”) Otndatashowthe
presence of three main molecular species: intact ferritin (74%).
trimer: (11%). and some smaller species at 97 A (9%). All
otherfenitinspeciesaccountfor~6%oftheobservedvahres.

AnalysistoncamvalinA(ConA)atthequnld/Sdid
Interface. ConcanavalinA(ConA)isalectin.obtainedfrom
the jackbean (Cmamlia ensifonnis). that binds to mannose.
glucose. and glycoconjugates containing these saccharides-”’53
The native protein is composed of four identical subunits. each
with molecular weight 25 500 kDrL’2 Con A in solutiat is
normally observedasatetrameratpli 7.00rabove.’2 However.
itwasshownbyGordonandYoung”thatConAdissociates
intodimersasaconsequenceofdilutionandionicstrengthat
a concentration SSpg/mLandapl-l 27.0. According to X-ray
diffraction sntdies.themoleculardimensions ofthedimerare
approxirnately3OA at 45A x 75A.whiletboseofthetetramer
areapproximatelyGOA x 70A x 70A(ProteinDataBank.
lCNl“) (Figure 7).

Preparations ofConA solutions (~400nglmL).asdeacribed
inthe Experimental Section. yieldedmonodispenedindivichtal
proteinmoleculesadsorbedonthemicasurface(seeFigtne2.
for example). Analysis of several such images results in the
height distribution shown in Figure 6. This distribution is the
result of 2000 individual protein molecule measurements. The
height distribution is best fit by four Gaussian functions cemered
a 18.28.42. and 59A(seeTable2A). lnspectionofthe
x-ray structure ofthe Con A dimer (see Figure 7) indicates
that the observed height distribution corresponds to individtal
ConAdimersorientedsuchthattheirlargestsurfaceareaisin
contact with the mica substrate. projecting the 30 A dimension
perpendiculartothesurface. ‘l'hisorientationisexpectedand
canbeunderstoodonthebasisofSAOmodel. ‘l'heZBand42

1654 J. Phys. Chem. B. Vol. 102. N0. 9. I998

 

I ' I Y ' U ' I

V
L

A

Number of molecules
H
8 §

0
v .

 

 

 

o 20 40 60 ll 100
Heist-NA)

l-‘lgnre‘. AFMheightdisuibutionthtaQNOheightmeasmements)

forConA’l‘hedataarebestfnbyfourGutssianfunctiooseenteredat

18.28.42.and59A.‘l‘hereaultsobtainedfromthisfitmaybefound

in‘l'ableZA.

 

 

 

 

 

 

 

 

 

Pure“). RibbonrenderingoftheCooAd'nnerMoteinDataBmk.
1CN1")-(A)Sideviewahowingtheshmaxis.30A(v/hichdoesnot
inchtdethehairpinntrns).andthelongestaxis.7SAforendtoend.
(B)Sidefiw¢ofingtheintermediateaxismeasmingamoximately
45A(negleetingthetwohairpinturnsattheextremes).
AdinmsionsareinexcelbntagreementwiththeX-raystrucnne
(seeFigtn'e7). 'l'he59Ameasurementissha1erthanthe75
AexpectedasstmtingtlmindividualConAdimerstaojecttlnir
longestaxisawayfromthemica(0°fromnormalinclination).
’1‘his0‘orientationisnotfavorableonthebasisofprotein
strucnrreshowninl-“tgure7;tbaef0te.onewouldexpectsome
angleofinclination. Ourmeamuementsindicateaninclination
ofabout38°fromnormal. Wewillshowthatthepresenceof
the18Apopulation(~8%ofthetotalpopulation)canbe
attributed to denantred Con A in the following experiments.
TontleoutthepossibilitythatConAwasdenatutingatthe
micasurface.twoexperimentswerecarriedout. Theftratone
irtvolvesimgingofsuccinylatedConAwhichislmowntoexist
onlyasadimer,overawiderangeofpli.” Thesectmd
experimentwastoptn'poselydenanuetheConAandthtermine
the height distribution with our technique.
ResultsfromtbesuccinylatedConA(1576individualheigbt
measurementshhowaverysimilarheightdistributiontotlnt
obtainedfromConA(seeFigtne8andTableZB). Thethree
mainpeaksfmtndareat29.45.and59A.carespondingcloaely
(within3A)tothoaefoundforConA. Thesereaultsindicate

159

Waneretal.

TABLEZ: Ieaultsfrornfla‘ghtnistrihutionAnalysisand
Least-Squareslr‘lttoGaun’anl-‘unctious'

 

 

height (A) fwhm (Ar m probability (%Y
(A) Concanavalin A

at: 18 10 489

a 28 10 5079 91.8

b 42 10 409 7.4

c 59 10 45 0.8
(B) Succinylated Con A

at: 19 9 1056

a 29 13 1597 87.7

b 45 8 164 9.0

c 59 13 $9 3.2
(C) SDS-Treated Con A

at 18 10 3679

a 29 13 1938 81.2

b 46 12 350 14.7

c 59 12 100 4.2

"lbevalueaintbesetablesarethereaultoftheheightdisnibution
analysisofrneasurements madeon 200a 1576. mid 1983 indvidual
proteinsforA.B.andC.respectively. "l'heaevaluearepreaentlhefull
widthathalf-maximum(fwhm)fm’eachofthe0aussianfunctionsin
thefit‘ThesemeasurementscureapoodtotheareaundertheGauasian
ftmctionandareproportionaltothetotalmunberofmoleculeswithin
arangeofheightsinA.‘Theprobabilitycxcludesthcpafiiclesfctmd
tohaveathmeasionofls-19Alabeledbymwhicharedetamined
toarisefrornthcrleaenceofdenaturedproteinmolecules. Thevalues
ofAGuuetheavaageofmmemeaanemenuusingeqs7-9uee
teat). ForConA(A)andsuecinylatedConA(B)weobtain-9.4:t
2.0 ll(mol A!) m -6.2 :h 02 tumor A2). respectively.

 

T W a r v *1

.n.
O

88 88 8g 8
AJA‘A‘ALA“.

 

Number of molecules

1
1
1
1
I

 

 

 

o T i. 4'. Jo ' {after
WM)

Fined. AFMheightdilributiondata(1576heightmea-trements)

famechlemATbedmuebestfitbyfotuGauaaianfunctioas

canedatl9.29.45.and$9k1‘hereaultsobtainedfromtheleast-

squaresfitmaybeformdinTableZB.

thatConAexistsasadimerandnotatetrameratthewaterl
observtimsofGordonandYormg‘Hndprofidesauppmtfor
thecapabilityandsensitivity ofthisapproachformeasming
changesintheoligomerintionstateofproteins Theonly
difl'erenceisanadditionalpopulationofmoleculeawithafight
of19A. Anexplanationofthispopulationwithasmaller
dimensionisgivenbelow.
ToinvestigatetbeprilityoftbnattnationofConAat
thewaterlnricainterfacewetreatedtheproteinwich'lisoditun
dodecylsulfate(SDS)for2-3minat90-95°Cpiortodilution
and peparation of samples. ‘lhe height distribution from the
SDS-treated Con A samples. from 1983 individual height
masurements.isshowninFigtne9(seealsoTable2C). Note
MthemajorpopulatimiscenteredatlSA. Thereaults
indicatetlnt~61$oftheptoteinmoleculesweredenatttred
Therestofthedistributionsbowspeaksat29.46.and59A.in
excellentagreement(<5A)withtheptevimtstwosetsofdata
ThisindicatesthatsomeoftheConAmoleailesweremt

Oligomerization of Proteins at Liquid/Solid Interfaces

 

' fi' 7 V Y

I
1-1;-.1

Number of molecules
335.-.???5 ‘5

AJA‘A‘

 

 

 

o to 40 6'on loo
fleithA)

Figure 9. AFM height distribution data (1983 height measurements)

forSDS-ueatedConAThedataarebestfitbyforn'Gaussianfunctions

centctedat18 29 4o m59A. 'I'heresultsobtainedfromthisleaat
squaresfit maybefoundin'fable 2C.

denannedortmderwentrefoldingupondilutionoftheSDS
during 1e preparation. This result allowsustoassigntbe
peakat19 inthesuccinylatedConAsamplestothepresence
of ~37%denannedproteininthesample.

FtntheranalysisoftheresultsbytheSAOmodelcanbeuaed
todeterminethefreeenergyoforiemationperunithGo.
for Con A and succinylated Con A on Mg2‘*-treated mica.
Usingeqs7-9andvalues (heightandpr'obability)from‘l‘able
Zwecalculateaveragevalmof-9.4:h2.0and-6.2:l:0.2
ltJ/(nnl A2). respectively. The uncertainty is the standard
deviationofthethreevaluescalculatedfromeqs7-9. Day-
to-dayvariabilityofasmuchasmmthecalatlatedvahteof
AGohasbeenobserved.“ 'lheapproximatesurfaceareaofeach
faceoftheproteincanbecalculatedfromthemeasmed
dimensions. Withthesevalueswecalculatethefreeenergyof
orientation for each of thethreedifi‘erent orientations (see Table
2).themost stable of these being ~-20 kJ/mol whentheCon
Amolcculeshavetheirlargestan‘faceorientedtowardthemica.
These values are larger than kT (2.5 kJ/mol). indicating this
orientationisrelativelystable. Thefreeenergyissmallenough
thatdistortionsinthenativestructurearenotexpected
Diflerencesinthefreeenergyforeachorientationdetermine
theobservedheightdistribution. ThevaluesforConAand
suecinylatedConAaresimilar.asexpecled;however.we
believethattheremaybesomesmalldifl'erencerelatedto
modificationofaminoacidsidechainsofConAbytbe
succinylation process. Succinylation involvesthemodification
ofaminesinthemoteinwithaconcomitantdecreaseinpositive
charge. ForConA.anaverageof10sitespersubunitare
succinylated.”causingchangesinitsinteractionwiththemica
surface.

In conclusion. we have obtained angstrom resolution mea-
surements directly from individual proteins at the liquid/solid
interface. Ourmeasurements. 28, 42. and 59 A. indicatethat
ConAadsorbsontomicasubstratesasadirner.primar-ilyhaving
itssmallestdimensionprojectednormaltothestn'face. Our
measurements are in very good agreement with the X-ray
strucnnedeterminedbySholmnetal.“ Althoughtbetetranuic
formofConAisnormallyobservedinsolutionthesolution
fromwhichwepreparedoursarnpleswas~400nglmLatapH
=7.3. which accordingtoGordonandYormg is expectedto
corttainprimar'ilythedirrterspecles.53 Thisdimericstateis
conserveduponadsorptionoftheConAmobculestothemica
surface. Thedimericstrucnrrewasconfirmedwithmeasme-
mentsonsuccinylatedCorlA. Evidencetotheoriemationof
theConAdimersonthemicacornesfromtheprevalenceof
28Ameasurementsinthedataandmalysisofthedifferent

160

J. Phys. Chem. B. VoL 102. No. 9, 1998 1655

heights (42 and 59 A) using the $110 model. The possibility
ofproteindenantrationcausedbytheadsorptionprocesswas
addressedby measurementsonConAtreatedbythedenanrring
agent SDS. which indicated that the degree of denaturation upon
adsorption is minimal (<5 A changes). Applying the SAC
model to our measurements results in approximate values for
the free energy of orientation for these molecules. We are
crmentlyinvestigatingtheeffectsoflargechangesintheprotein/
surface interaction on these values.

Disendon

Measurements of individual Concanavalin A and succinyl
ConAmoleculesatthewaterfmica interface havedeterrnined
thattheConAdimeristheprevalentstrucmre. Themeasure-
memsobtained(29x45x59A)forthesedimersareingood
agreement with the dimensions determinedby X-ray diffraction
studies. Further. the results for Con A. succinyl Con A. and
SDS-treated ConA are self—consistent and reproducible. In
addition to the structural information obtained. the proposed
modelalsoallowsustopredictavalueforthefreeenergyof
adsorptionforConAatthcwatedmicainterfaceandthe
orientation oftheadsorbedproteins.

Weconaervativelyestimatethattheaccuracyofthistechnique
isSAorleas. ‘I‘hisestimateisrelatedtoanumberoffactors.
FustofalLthereisS-SAroughnessinthemicasubstrates
whichweuse. Secondalthou htheultimateresolutionofthc
AFMinthezdimensionisl orless.wearenotoptimizing
ornimaging conditiom for single-pointmeasuremenu of height:
rather. wearetradingaccuracyforacquisitiontime. Forthe
purposesoftlnee-dimensionalproteinoligomerizationstateand
orientatiorhtheaectn'acy(<5 A)is sufficient.

'l'hetechniqucpresentedhereprovidesameanstomeasure
themoleculardimensionsforthecompletesizerangeof
monomericandoligomericproteinsadsorbedtoastu'face
(unnalorsynthetic). 'l'heaemeasurementsnnythenbeutiliced
todeterrninetheoligomerizationoraggregationstateofthe
protein. This approach extends the nanometer resolution of
Ammthecomactorimermittentcomactmndesrmderambient
orhqtndcondiuonstotlntofangstromsbyavoidingexceasive
imagebroadeningofbiologicalsamplesthatisassociatedwith
thelater'aldistortionsintroducedbytheprobetipgeornetry.""39
Probe-pmteininteracdonanninlylateralandverticalpreestne.
cancausethemoleculetoappearshorter."5 Withregardto
protein-substrate interactiorhthereisanamactive interaction
betweenthepolarsurfaceoftheproteinmoleatleandthe
charged mica surface. possibly deforming the molecule and
causingittoflattenagainstthemicasmface. Astudyusing
lysozyme hasshown that itretainsits activity when adstlbed
ornomicaandimagedwithAf-M.sothesedeformationsmay
not be extensive enough to alter biological activity.” Our
Wounfirmtheconservationofmtivestrucnnewhen
compared with available X-ray data

Athermodynamicrnodelwasthrivedtounderatandthe
orientationofproteinsattheliquidlsolidiruerface. Large
deviationsfromthisSAOmodelbyadsorbedproteinswould
imply a preferred orientation that is westnnably related to
recognitionbetweenaproteindomainandthesrn'face. The
extent of deviation from tlte orientation model may therefae
be useful in defining preferred molecular orientation for
oligomers or multiple equilibria for dissociated submits of
oligomers. lnthesecasesoneisabletoextendthesimplemodel
presentedhere. byallowingeachfaceofthcproteintohavea
difi’erentvalueforAGueqs 1-3). lnthiswayaquantitative
valuemaybeassignedtotheenergydifferenceassociatedwitb

1656 J. Phys. Chem. B, VoL 102, N0. 9, I998

eachoricnution. Indeedbinding studies with streptavidinand
free biotin provide strong evidence for a ligand-mediated
preferential orientation for the adsorption of streptavidin dimers
and tetramers to the mica surface (work in progress).
An examination of the preferred orientation of the molecules
also provides the opportunity to predict the organization of
individual subunits within the protein oligomer and porentially
evaluate the conformational flexibility of dissociated individual
subunits. Measurements with angstrom resolution may be
performed on functional complexes that are associated with a
surface. a more common biological interaction within and
between cells. This paper presents an approach that extends
themlyticalpowerandtheresolution ofAFM toprovide direct
physical characterization (both dimensions and orientation) of
biomolecules at interfaces and is ntinimally dependent on the
geometry of the probe tip. This approach provides (1) the
capability to examine the concentration- and ligand-dependent
properties of protein oligomerization using picomole amounts
of protein. (2) offers the capability to distinguish between the
possible quaternary forms of oligomers and their interconver-
sions resulting from the binding of ligands. substrates. or
allosteric effectors. and (3) determines whether adsorption of
oligomeric proteins to surfaces (natural or synthetic) can modify
protein assembly and aggregation state. Such meastnemcnts
can be mosr useful in studies of enzyme regulation. DNA/RNA
organization. gene regulation. protein biosynthesis and chap-
aoninfiimnomhacterialmachmentandadhesionandthe
effects of postnamlational modification on protein organization
and activity.

Acknowledgment. 'I‘hisresearchwaspaniallyftmdedby
the REF-Protein Structure. Function and Design Award (MSU).
the Center for Fundamental Materials Research (MSU). and a
Camille and Henry Dreyfus New Faculty Award (MD). We
are very grateful to Dr. Neil Bowlby for programming the data
analysis software and Drs. Leslie Kuhn. Michael Garavito. John
Wang. and Jack Preiss of the Department of Biochemistry for
their incisive critiques. MD. is a Becltman Young Investigator

and a Packard Science and Engineering Fellow.

“mend Notes

(I) Hm‘l‘. A.; Brash. 1.1. InProreinsarlnlerfaccs.Pkys-icochan-
real and Biochemical Shelia. Brash. J. L. Horbett. T. A.. Eta; ACS
Symposium Series 343; American Chenn‘eal Society: Washington. DC.
1987; p 1.

(2) Norde, W. Adv. Colloid Interface Sci 1’6. 25. 267.

(3) Brush. J. I..; Herbert. T. A. In Proteins at Interfaces ll. Finde-
ml: and Applications; Horbett. T. A.. Brash. J. I... Eds.; ACS

SymosinmSeriesGOZ;AmerieanOemiealSociety: WashingtonDC.
1995; p l.

(4) Elwing, H.; Askenda, A.; Ivarsson, B.; Nilsson. U.; Welin. 8.;

I. In Proteins or Interfaces. Physicoclianical ardBioclremical
Studies; Brush. 1. I... Hubert. T. A.. Eds.; ACS Sympoaitmr Series 343;
American Orenheal Society: Washington. DC. 1987; p 468.
(5) NachnjahA.;Lu.CF..Chimu.lLlL1nProteuuarlnmfoccs
ll. meApplicaians; Horbetr.T. A.. BrashJ. I... Eds.; ACS
Symposium Scream; AmericanChemiealSociety: thingtmDC.
1995; p 181.
(6) van Oss.C. J.; Wu.W.; 0iese.R. F. Inhorciruorlnmfocesll.

1995:1380.

(‘7) YamGn Li.J.-.;T Httang.S..-C; CaldwelLK.D.lnProrcinsat
Mecca”. FWMWWMWTAMMJL"
Eds.; ACS SymposiumSeries602.Ameriean0remicalSociety: Wash-
' DC. 1995 p256

(8)1.ec.1.E;Saavedr-LSHSInProrcimarlm¢rfnc¢sIl.F-td¢namb

W;WT.A.;MJ.L.E&.; ACSSynmmiumSer'ns

American Washagton.DC.1995 p269

(9) Ema; QmeCHF; GerheruCPlrystchculflx

161

Waner et al.

(10) hi. R.; John. S. A Am J. PhysroL 1994. 266. CI.

(11) Bun. H.-J.; Downing. IL H.: Hausm P. K. Biophys. J. 1990. 58.
1473.

(12) Bustamante. C.; Vesenka. J.; Tang. C. I..; Rees. W.; Guthold. M.;
Keller. R. Biocharu’srry I992. 31. 22.

(13) Bustanmte. C.; Rivetti. C. AMI. Rcu. Biophys. Blow-cl Since
1996. 25. 395.

(14) Louder. D. R.; Parlcimon. B. A. Anal Chm. I994. 66. MR.

(15) Schnyder. T.; Engcl. A.; Lustig. A.; Wallimann. T. J. Biol. Chem
1”.263. 16954

(16) Dykstra. M J. A Marisol of Applied Technique: for Biological
Electron Microscopes; Plenum: New York. 1993.

(l7)Weisenhorn.A..;I. Drake..;B Prater...:CB Gould.SA..C:
Hansma. P. R.; Ohnesorge, F.; Eager. M.; Heyn. S. P.; Gaub. H. E. Biophys
J. I990. 58. 1251.

(18) Schahert. F.; Hefti. A.; Goldie. R.; Stemnzr. A.; Engel. A.; Meyer
E; Overney. R.; Gilntherodt. H.-J. Ultramicroscopy 1992. 428. 1118.

(19) SchaberL F. A.; Bean. 0: Engel. A. Science I995. 268. 92.

(20) Ohnishi. 8.; Han. M.; Furuno. T.; Saahc. H. Biophys. J. I992.
63. 1425.

(21) Mallrin. A. J.; Kumetsov. Y. 0.; McPlerson. A. J. Struct Biol.
1996. H7. 124.

(22) Roberts. C. J.; Williams. P. M.; Davies. J.; Dawkes. A. C.: Sefton.
J.; Edward. J. C.: Haymes. A. 0.; Beatka C.; Davies. M. C.; Tendler. S.
J. B. ngnusir 1995. u. 182.

(23) Ill. C. R.; Keivens. V. M.; Hale. J. E; Nakamura. K. It: Joe. R.
A.; Chang. S.; Melcher. E. D.; Drake. B. Biophys. J. 1993. 64. 919.

(24) Edstrom. R. D.; Meinke. M. H.; Yang. X.; Yang. R.; Elings. V.;
Evans. D. F. Biophys. J. 1990. 58. 1437.

(25) Marchant. R. E; Lea. A. S.. Andrade. J. D.; Bockerrstedt. P. J.
Colloid Interface Sci. 1992. I48. 261.

(26) Radrmcher. M.; Fritz. M.; Clevehnd. J. P.; Walus. D. A.; Hanson.
P. K largmm'r I994. 10. 3809.

(27) Radmacher.M.;Fritz.M;Hanstna.H.0.;Hansnn.P.KScia-cc
1994. 265. 1577.

(28) Henderson. R. M.; Schneider. 5.: Li. 0.. L: Hornsby. D.; White.
S. J.; Oherleithner. H. Proc. Natl Acad. Sci USA 1996. 93. 8756.

(29) Bennnilla. M.; Mature. 5.; Laney. D. E; Lyubchenlto. Y. L;
Hansma. H. G. Immir 1995. II. 655.

(30) Hansma H. 0.: Sinsheim. R. L.; Li. M. 0.; Hanson. P. K. Nucl
Acids Res. 1992. 20. 3585.

(31) Hanson. H. 0.; Vesenka. J.; Siegerist. C.; Kelderm. 0.; Monet.
H.;Sinsheimer.R.L.;Elings.V.;Bustamte.C;Hansrm.P.KSciac¢
I992. 256. 1180.

(32) Henderson. E. Nucl. Acids Res. 1992. 20. 445.

(33) Shaiu. W.-l..: Vesenka. J.; Jondle. D.: Henderson. 5.; Laram. D.
D. J. Vac. Sci. Tech-0L A 1993. ll (4). 820.

(34) Wyman. 0.; Rental. 1.; North. A. R.; Bums. C.; Kustu. S.
Sciatcc I997. 275. 1658.

(35) Siedlecki. C. A.; Iestini. B. J.; KottkeoMarcllnt. K.; Ewell. S. J.;
Wilson. D. I..; Marci-rt. R. E Blood I996. 88 (8). 2939.

(%)KellerhaDJ FrankeF...SSurgSci....l993294409

(37) EppclLS.J.;Zypnnn.F.R.;Marchant.R.Eng-arir1993.9

228(38)Allen.M.J.; HudN..;V Balooch.M.. Tench.R..;J Seikhaus.W.
J.; Balhoru.R Ulmcroscopylmafl. 1095.

(39) Martiewicz.P.;00h.M. C. largo-rut I994. IO. 5.
£40) Yariv. J.; Kalb.A.J.; Levrtnkr A. 8m Bioplm.AcraI98.
l .303.

(41) ApeILH.J.:Colchero.J.;Linder.A;Marti.O.In51Nr-rdAl-’M
inggBiologyanLO..Amrein.M.Eds.;Acaderchreasz SanDiego.CA.
l 3;p282.

(42) Hu.J.;Xiao.X.-D.;Oglenec.D.F.;Salnlron.M.Scicnccl995.
268.267.

(43)1hundat.T.;Diarg.X.-Y.;Oien.0. Y.;WuuchRJJmISc-i.
mimzwusw.

(44)m0.l.;Davies.MC.;Jacbon.D.E.;Roherts.C.J.;Tardh.
S. J. 8.; Williams. P. M. J. Phys. Chem 1993. 97. 8852.

(45) Parker. M. 0.; Davies. M. C.; Tendler, S. J. B. J. Phys. Client.
1995.99.16155.

(46)HeightnusmernentsofsmallspheriealparticleswithAm-e
slightlyafl’eacd.~1%.beeauseofthegeometryoftheprobeandbecatae
individualmlinesmaybeseparatedbydistanceseompuablemthesire
ofparuclesbeingmeastned.Tlnserroroectnawhentheendoftheprohe.
whicheanhernodelednahemisphere. tracteaactlyattheapexofthe

proternForcxmnplema2x2umaquleimagedigitiaedt0512x512
pinktheendoftlnprobeeouldbeufaruZOAfromtheqrexoftl:
moleculeifitisloeatedexaalybetweentwoscanlines. InFigmelAJhe
idealizedshapeaoftheendoftheprobeandaspheriealproteinmolecule
areiflusnatedBeeauaeofurefinitenumherofpointsthatarerecorded.th
apeaofdreproteinwinusuallynotbeeaactlyimderthemofurepobe;
thereforethemeasmedheightoftlnpanidewillbesliglulymllerJhe

Oligomerintion of Proteins at liquid/Solid Interfaces
discrepancy. d. can be calculated 8 follows:

d=(R+r)-\/(R+r)2-X2 (3)

whereRistheradiusofctnvanueoftheprohe.ristheradiusofaspherieal
Mmdxuthehmiaontaldisuncebetweeotheproheandmeprotem
whentheprobeisclosesttotheprotein(seeFigure lA).Probestnsfor
our measurements had approximately a 24 um radius of curvantre. as
calculated from the width and height measurements of colloidal gold
particles ofknown dimensions.” For particles withadiameter of4OA.the
mimum height deviation can be as much as 0.7 A. We have modeled
this effect by calculating the average error expected given 5(1) randomly
pheed4OAdiameterspheres.(Spheriealgeonetrymxrmizesthis resolution
effect.) The average error obtamod from our simulation was 02 A. which
amounts to 0.5% error in the height measurement of the particles. This
mismallertlmnflnnannalmugbneasofatonfieaflyflatnnathcrefoc.
nonemptwasmadetomininuneitorcorrectforit.

(47) Mmsover. W. H. Micron I993. 24. 389.

(48) Harrison. P. M. J. Mol. Biol 1963. 6. 402.

162

J. Phys. Chem. B. Vol. 102. No. 9. I998 1657

(49) Yau. S.-T.; Zion. Y. Mod Phys. Lett. B 1995. 9. I87.

(50) Concanavalin A as a Tool; Bittiger. H.. Schnebli. H. P.. Eds: John
Wiley & Sons: New York. 1976.

(51) Sharon. R.; Lis, H. Science 1989. 246. 227. .

(52) Reeke. 0. R.. Jr.; Becker. J. W.; Cunningham. B. A.; Wang. J. 1...:
Yam 1.: Edelman. 0. M. In Concanvalin A; Chowdhury. T. K.. Weiss.
A. K.. Eds.; Plenum Publishing Corp.: New York. 1975.

(53) Gordon. J. A.; Young. R. K. J. Biol Client 1979. 254. 1932.

(54) Shoham M.; Yonath. A.; Stmnran. J. L; Moult. J.; Traub. W.; Kalb.
A. J. J. Mol Biol 1979. I31. I37.

(55) Gtmther. 0. R.; Wang. J. L; Yahara. 1.: Cunningham. B. A.;
Edelman. 0. M. Proc. Natl. Acad Sci. USA I973. 70 (4). 1012.

(56) The origin ofthis variability is currently under investigation. Some
ofthe variation may be attributed to the amount of hackgrormd noise.
perhapsduetodifferencesinthemicasubsuatcorchangesininsuument
performance. Capillary form may also play a role in this variability. Further
experiments. carried out under liquid conditions. will help us make this
assessment

APPENDIX 2

Computer program source code

163

APPENDIX 2.1

Source code for Mathematica calculation of Chapter 2

164

(a Parameters *)

Nbins = 100;

8izeRange= 100; (* in Angstroms *)
Convert = Nbins / SizeRange;

(* Set positions of the Gaussians *)
nOne = 30 Convert;

n'l‘wo = 45 Convert;

n'l‘hree = 60 Convert;

(a The width and amplitude of the three Gaussians *)
width = 10 Convert;

popOne = '70;

pop'l‘wo = 20;

pop'l‘hree = 10;

(a The data simulated is the sum of three Gaussians *)

Table [H [n] = IntegerPart[popOne *Exp[- ( (n - nOne) / width) A2] ] +
IntegerPart[pop'1‘wo * Exp[- ( (n - n'l‘wo) / width) “2] ] +
IntegerPart[popThreea Exp[- ( (n - n'l'hree) / width) " 2] ] , {n, 1 , Nbins}] ;

 

 

ListPlot[

Table[l-I[n] , {n, 1, Nbins}] , PlotRange -> All, PlotJoined -> True]

70-

60[

50»

40-

30*

20-

10’

20 40 60 8b 160
-Graphics-
imnu

Area = Z H[n]

nsl

1698

Tab1e[U[i] =0. {1. 1, Areal]:

For[test=0;n=1;i=1, n<=Nbins,

n++, Tab1e[U[i] =n, (i, test, test+H[n]}]; test +=H[n]]
test

1698

165

(* This is the distribution of all the values *)
Distrib=Table[U[i], {i, 1, Area”;
ListPlot[Distrib]
100i '
80»
60~

40~

20~

 

 

250 500 750 1000 1250 1500

-Graphics-

166

Nsamples = 100;

Tab1e[Discrete[n] = 0, {n, 1, Nbins}];
temp := Random[Integer, Area - 1] + 1

measurement a Table[Distrib[[temp]] , {i, 1, Nsamplesfl;

For[suma = 0; j = 1, j <= Nsamples,
j++, Discrete[measurement[[j]]] += 1; some 4»: l]
suma
For[am =0; j: 1, j <= Nbins, j++, If[H[j] >Hm, Hm=H[j], Hm=Hm]]
Hm
For[Dm = 0; j = 1, j <3 Nbins,
j++, If[Discrete[j] >Dm, Dm=Discrete[j], Dm=Dm]]
Dm
Por[j = 1, j <= Nbins, j++,
Dnorm[j] = (Discrete[j] IDm); Hnorm[j] = (H[j] lI-Im)]
For[sumsq= 0; j = l, j <= Nbins, j++, sumsq += (Dnorm[j] -Hnorm[j]) A2]
N[sumsq]

ListPlot[
Table[Diacrete[n] , {n, 1, Nbins}] , PlotRange -> All, PlotJoined -> True]

100

72

3.2102

 

 

 

 

 

 

 

 

 

 

 

 

 

20 4O 60 80 100

- Graphics -

167

Nsamples = 500;

Table[Discrete[n] = 0, {n, 1, Nbinsn;
temp := Random[Integer, Area - 1] + 1

measurement = Tab1e[Distrib[[temp]] , {i, 1, Nsamplesfl;

For[suma = 0; j = 1, j <= Nsamples,
j++, Discrete[measurement[[j]]] += 1; some += 1]
sum
For[Hm =0; j =1, j<= Nbins, j++, If[H[j] >Hm, Hm=B[j], Hm=Hm]]
Hm
For[Dm = 0; j = 1, j <= Nbins,
j++, I£[Discrete[j] >Dm, Dm=Disorete[j], Dm=Dm]]
Dm
For[j = 1, j <= Nbins, j++,
Dnorm[j] = (Disorete[j] IDm); Hnormlj] a (H[j] /Hm)]
For[sumsq= 0; j = 1, j <= Nbina, j++, sumsq +2 (Dnorm[j] -Hnorm[j]) A2]
lemq]

ListPlot[
Table[Discrete[n] , {n, 1, Nbins}] , PlotRange -> All, PlotJoined -> True]

500
72
26

1.2382

25)

20~

 

ISL

 

10»

 

 

20 4O 6O 80 100

- Graphics -

168

Nsamples = 1000;

Table[Discrete[n] = 0, {n, 1, Nbins}];
temp := Random[Integer, Area - 1] + 1

measurement = Tab1e[Distrib[[temp]], {i, 1, Nsamples}];

For[suma = 0; j = 1, j <= Nsamples,
j++, Discrete[measurement[[j]]] +=1; suma += 1]
suma
For[Hm =0; j = 1, j <= Nbins, j++, If[I-l[j] >Hm, Hm=H[j], Hm=Hm]]
Hm
For[Dm = 0; j = 1, j <= Nbins,
j++, If[Discrete[j] >Dm, Dm=Discrete[j], Dm=Dm]]
Dm
For[j = 1, j <= Nbins, j++,
Dnorm[j] = (Discrete[j] IDm); Hnorm[j] = (H[j] /Hm)]
For[sumsq= 0; j = 1, j <= Nbins, j++, sumsq += (Dnorm[j] -Hnorm[j]) A2]
leumsq}

ListPlot[
Table [Discrete[n] , {n, 1, Nbins}] , PlotRange -> All, PlotJoined -> True]

1000
72
41

0.4693

40»

30>

20~

10»

 

20 4O 6O 80 100

- Graphics -

169

APPENDIX 2.2

Source code for Image Analysis software in Visual Basic 5.0

170

HDF analysis program source code for Visual Basic 5.0
Main.frm Code

'******** HDF Analysis Visual Basic 5.0 program ********
'**************** VCI'SIOIIZ.I 6/12/98 ****************

'*** Written by George Schoendorff and Mark J. Waner ***
'Analyzes Park Scientific Instruments AFM images to find heights of particles.

'*** Uses ScanData (supplied by Park Scientific Instruments)to open and display image.
'*** Based on hdfanal5i QuickBasic 4.0 program written by Dr. Neil Bowlby.
'*** Original Program and algorithms were formulated by:

'*** Dr. Marcos Dantus, Mark J. Waner and Dr. Neil Bowlby.

'Synopsis of program:

' 1) File is read and conversion of raw data (Integers) are converted into anstroms.
' 2) The height data is then put into a 2 angstrom bin histogram.
' 3) The Analysis routine (click button) will then find all local maxima (within
' given radius) that are above the threshold value, but below the cutoff.
' 4) The x,y position and the height of each particle can then be saved to a text
' file or you can go into an interactive peak selection/rejection mode (analysisfrm)
' 5) The height of a particle is determined by making a histogram of all pixel heights
' within 40 pixels (in Y) and 1 pixel (in X) and using the most frequent height
' as the local background
Dim Zgain, MaxPixel, MinPixel As Single

Dim DataArray(l To 512, 1 To 512) As Single

Dim xpeaks(1 To 5000), ypeaks(1 To 5000), zpeaks(1 To 5000) As Integer

Dim peakno, radius, threshold, cutoff, MostFreq, p, pl, acceptedpeaks, size As Integer
Dim Zspan As Long

Dim fileName, defaultfilename As String 'Strings to store open and save file names
Dim xpeaksl(1 To 5000), ypeaksl(1 To 5000), zpeaksl(1 To 5000) As Integer

Public Sub About_Click()
frmAbout.Show
End Sub

Public Sub buttonGrayScale_Click()
ScanData.GrayScaleSetting
End Sub

Public Sub textMax_ChangeO

textCutoff.Text = textMax.Text
End Sub

171

Public Sub buttonMinAna1_Click()

'Do analysis to look for minima
textError.Text = ""

Dim MinExclude As Integer
MinExclude = textMinExclude.Text

'Clear the arrays to prepare for new analysis
Erase xpeaks

Erase ypeaks

Erase zpeaks

peakno = 0

'change cursor to hourglass
Forml.MousePointer = 11

'Locate the local minima
radius = textRadius.Text
For x = radius + 1 To size - radius
For y = radius + 1 To size - radius

For rx = -radius To radius
For ry = -radius To radius
Ifrx=OThenIfry=OThenGoTo 16
If DataArray(x + rx, y + ry) < DataArray(x, y) Then GoTo 20
'j ump out if there's a point (MinExclude) Angstroms higher within the radius
If DataArray(x + rx, y + ry) > (DataArray(x, y) + MinExclude) Then GoTo 20
If DataArray(x + rx, y + ry) = DataArray(x, y) Then If (rx <= 0) Then If (ry <=
0) Then GoTo 20
16 Next ry
Next rx
peakno = peakno + 1
xpeaks(peakno) = x: ypeaks(peakno) = y: zpeaks(peakno) = DataArray(x, y)
If peakno > 4998 Then textError.Text = "Too many valleys. Increase radius or Min.
Exclude": peakno = 0: textNumPeaks.Text = O: GoTo 30

20 Next y
Next x

If peakno = 0 Then textError.Text = "No valleys detected. Decrease radius or Min.
Exclude": peakno = 0: textNumPeaks.Text = O: GoTo 3O

3O textPeakValley.Text = "# Valleys"
textNumPeaks.Text = peakno

172

 

buttonPeakAnalysis.Enabled = True
buttonAnalysis.Default = False
buttonPeakAnalysisDefault = True

Form1.MousePointer = 0 'Default mouse pointer
End Sub

Public Sub buttonPeakAnalysis_Click()
Forml .MousePointer = 11 'or VB hourglass

'Write all data to file in order to pass info to analysis form
Open "variable" For Output As #2
Write #2, peakno, MaxPixel, MostFreq, cutoff, threshold, radius, size
Write #2, fileName, textPeakValley.Text, textMinExclude.Text, defaultfilename
For x = 1 To size
For y =1 To size
Write #2, DataArray(x, y)
Next y
Next x
For peaks = 1 To peakno
Write #2, xpeaks(peaks), ypeaks(peaks), zpeaks(peaks)
Next peaks
Close #2

buttonPeakAnalysis.Defau1t = False
buttonSave.Default = True

analysis.Show
Forml .MousePointer = 0 'Default mouse pointer
End Sub

Public Sub buttonSave_Click()
Form1.MousePointer = 11 'or VB hourglass
Dim SaveFileName
'Write peak data to file
SaveFileName = textSaveF i1eName.Text
Open SaveFileName For Output As #1
Write #1, fileName
Write #1, textPeakValley.Text; acceptedpeaks
Write #1, "maximum="; MaxPixel, "most frequent="; MostFreq, "Min. exclude=";
textMinExclude.Text
Write #1, "cutoff="; cutoff, "threshold="; threshold, "radius="; radius
For i = 1 To acceptedpeaks
Write #1, xpeaksl(i), ypeaksl(i), zpeaksl(i)
Next i

173

Close #1
F orml .MousePointer = 0 'Default mouse pointer
End Sub

Public Sub buttonScales_Click()

'Turn on and off display scales

If ScanData.DisplayScales Then
buttonScales.Caption = "Scales On"
ScanData.DisplayScales = False
Else

buttonScales.Caption = "Scales Off"
ScanData.DisplayScales = True
EndIf

End Sub

Public Sub Exit_Click()
Kill "data"

End 'Exit application
End Sub

Public Sub buttonAnalysis_Click()

textError.Text =

'Clear the arrays to prepare for new analysis
Erase xpeaks

Erase ypeaks

Erase zpeaks

peakno = 0

'Get peak-finding parameters
radius = textRadius.Text
cutoff = textCutoff.Text
threshold = textThreshold.Text

'Change cursor to hourglass
Forml .MousePointer = 11

'Locate the local maxima
For x = radius + 1 To size - radius
For y = radius + 1 To size - radius
If DataArray(x, y) < threshold Then GoTo 20
If DataArray(x, y) > cutoff Then GoTo 20

174

For rx = -radius To radius

For ry = -radius To radius

Ifrx = 0 Then Ifry = 0 Then GOT016

'jump out of the loop if there's anything taller within the radius

If DataArray(x + rx, y + ry) > DataArray(x, y) Then GoTo 20

If DataArray(x + rx, y + ry) = DataArray(x, y) Then If (rx <= 0) Then If (ry <= 0)
Then GoTo 20
16 Next ry

Next rx
peakno = peakno + 1
xpeaks(peakno) = x: ypeaks(peakno) = y: zpeaks(peakno) = DataArray(x, y)
'Indicate overflow error
If peakno > 4998 Then textError.Text = "Too many peaks. Increase threshold": peakno
= O: textNumPeaks.Text = O: GoTo 30
20 Next y
Next x
'Indicate lack of particles found
If peakno = 0 Then textError.Text = "No peaks detected. Decrease threshold": peakno =
O: textNumPeaks.Text = O: GoTo 3O
3O textPeakValley.Text = "# peaks"
textNumPeaks.Text = peakno

'Calculate histogram for local background around each peak found
Dim lowx, highx, lowy, highy, maxpeakheight, maxpeakhistbar, maxbin, maxzcoord,
peakheight, bin, HistBar(l To 600) As Integer
For p = 1 To peakno
lowx = xpeaks(p) - 1: If lowx < 1 Then lowx = 1
highx = xpeaks(p) + 1: If highx > size Then highx = size
lowy = ypeaks(p) - 40: If lowy < 1 Then lowy = 1
highy = ypeaks(p) + 40: If highy > size Then highy = size
maxpeakheight = O
For x = lowx To highx
maxpeakhistbar = O: maxbin = O: maxzcoord = O: peakheight = 0
For bin = 1 To zpeaks(p): HistBar(bin) = 0: Next bin
For y = lowy To highy
If DataArray(x, y) > maxzcoord Then maxzcoord = DataArray(x, y)
bin = DataArray(x, y)
If bin > zpeaks(p) Then GoTo 4020
If bin < 1 Then GoTo 4020
HistBar(bin) = HistBar(bin) + 1
4020 Next y
For bin = 1 To zpeaks(p)
If HistBar(bin) > maxpeakhistbar Then maxpeakhistbar = HistBar(bin): maxbin = bin
Next bin
'calculate height of particle relative to local background
peakheight = maxzcoord - maxbin

175

If peakheight > maxpeakheight Then maxpeakheight = peakheight
Next x

xpeaksl(p) = xpeaks(p)

ypeaksl (p) = ypeaksm)
zpeaksl(p) = zpeaks(p) - maxbin
Next p

acceptedpeaks = peakno

buttonPeakAnalysis.Enabled = True
buttonAnalysisDefault = False
buttonPeakAnalysis.Default = True

Forml .MousePointer = 0 'Default mouse pointer
End Sub

Public Sub Open_C1ick()

'CommonDialog is used to show Open and Save Dialog Boxes

'Microsoft Common Dialog Control 5.0 has to be checked in Project/Components...
'and then installed on Form

Form1.MousePointer = 11 'or VB hourglass

textPeakValley.Text = "": textNumPeaks.Text = "": textComments.Text = ""
textCutofT.Text = "": textError.Text = "": textFilename.Text = ""
textMaxPixel.Text = "": textMinPixel = "": textMostFreq.Text = ""
textSaveFileName.Text = "": textScanSize.Text = "": textThreshold.Text = ""
peakno = 0: threshold = O: cutoff = 0: MostFreq = O: MaxPixel = 0: MinPixel = O
fileName = ""

Erase xpeaks: Erase ypeaks: Erase zpeaks: Erase xpeaksl: Erase ypeaks]: Erase
zpeaks]

Erase DataArray

buttonPeakAnalysis.Enabled = False

buttonAnalysis.Enabled = False

buttonMinAna1.Enabled = False

buttonSave.Enabled = False

buttonAnalysis.Default = False

CommonDialog.ShowOpen

fileName = CommonDialog.fileName

ScanData.DisplayScales = False

ScanData.LoadHDF (fileName) 'Load HDF file with name from Open File Dialog box
textFilename.Text = fileName

textScanSize.Text = Str(ScanData.YScanSize) + "x" + Str(ScanData.XScanSize)
textComments.Text = ScanData.Comments

size = ScanData.GetNumCols

'Determine default filename for peak height data file
Dim tempstr(1 To 500) As String

176

Dim store, store2 As Integer

For i = 1 To Len(fileName)
tempstr(i) = Mid(fileName, i, 1)
If tempstr(i) = "\" Then store = i
If tempstr(i) = "." Then store2 = i
Next i
textSaveFileName.Text = Mid(fileName, store + 1, store2 - store) + "txt"
defaultfilename = textSaveFileName.Text

'Here we will take image data which is loaded into Scan Data Control and put it
'into an array

Dim DataGain As Single

Dim TempArray(1 To 512, 1 To 512) As Single

Dim MinDataAt As Single

DataGain = ScanData.DataGain
MinDataAt = 32767

MaxPixel = -32767

MinPixel = 32767

Form1.MousePointer = 11 'or VB hourglass

'Put data from column counterl -1 and row counter2-1 in HDF file
'into array (DataAt array is from O to 511, not 1 to 512)
'Rows & columns are transposed in DataAt l!!!!!!
If optionY.Va1ue = True Then
For counterl = 1 To size
For counter2 = 1 To size
TempArray(counterZ, counterl) = (CSng(ScanData.DataAt(counter1 - l, counter2 - 1))
* DataGain)
If TempArray(counterZ, counterl) < MinDataAt Then MinDataAt =
TempArray(counterZ, counterl)
Next counter2
Next counterl
Else
For counterl = 1 To 512
For counter2 = 1 To 512
'Rows & Columns may be transposed l!!!!!!!!!!!!!!
TempArray(counterl, counter2) = (ScanData.DataAt(counter1 - 1, counter2 - 1) *
DataGain)
If TempArray(counterl , counter2) < MinDataAt Then MinDataAt =
TempArray(counterl , counter2)
Next counter2
Next counterl
End If

177

'calibrate pixel values in angstroms and put into array
For counterl = 1 To size

For counter2 = 1 To size

DataArray(counterl , counter2) =(TempArray(counter1, counter2) — MinDataAt) *
10000

If DataArray(counterl , counter2) > MaxPixel Then MaxPixel = DataArray(counterl ,
counter2)

If DataArray(counterl , counter2) < MinPixel Then MinPixel = DataArray(counterl ,
counter2)

Next counter2
Next counterl

'Calculate histogram for entire image

Dim Sizecategory, MaxSizeCategory, TotalHistBars As Integer
Dim binsize, BarLength As Integer

Dim MaxHistBar As Long

Dim HistBar(l To 300) As Long

MaxHistBar = 0

'define the histogram

Zspan = MaxPixel - MinPixel
TotalHistBars = 100

binsize = Zspan / TotalHistBars

For x = 1 To size
For y = 1 To size
Sizecategory = (1 + ((DataArray(x, y) - MinPixel) / Zspan) * (TotalHistBars - 1))
HistBar(Sizecategory) = HistBar(Sizecategory) + 1
If HistBar(Sizecategory) > MaxHistBar Then MaxHistBar = HistBar(Sizecategory):
MaxSizeCategory = Sizecategory
Next y
Next x
MostFreq = Int(MinPixel + (MaxSizeCategory * Zspan / TotalHistBars))

'Find histogram range

Dim LastHistBar, check As Integer
check = 0

LastHistBar = 100

Fori=100 To 1 Step -1
BarLength = Int((HistBar(i) / MaxHistBar) * 90)
If BarLength = 0 And check = 0 Then GoTo 50
If BarLength > 1 Then GoTo 6O

40 Next i
GoTo 70

178

50 LastHistBar =i
GOTO 40

60 check = 1
GoTo 40

70 'Draw the histogram for the entire image

Dim Width As Integer

'Width is total "# pixels" / "# relevant histogram bars" - 1 pixel spacing between bars
Width = Int((280 / LastHistBar) - 1)

pictureImageHist.Cls
pictureImageHist.DrawWidth = Width
pictureImageHistForeColor = 255
pictureImageHist.DrawMode = 13
For i = 1 To LastHistBar

If HistBar(i) = 0 Then GoTo 8O

BarLength = Int((HistBar(i) / MaxHistBar) * 90)

x =10 + ((Width+ 1) * (i - 1))

y = 100 - BarLength

pictureImageHist.Line (x, y)-(x, 100)

80 Next i

'Assign values to text boxes
textMaxPixel.Text = Int(MaxPixel)
textMinPixe1.Text = Int(MinPixel)
textCutoff.Text = Int(MaxPixel)
textMostFreq.Text = MostFreq
textThreshold.Text = MostFreq + 9

buttonAnalysis.Enabled = True
buttonMinAnal.Enabled = True
buttonSave.Enabled = True
buttonAnalysis.Default = True

Form1.MousePointer = 0 'Default mouse pointer
End Sub

Public Sub Save_C1ick()

CommonDialog.ShowSave

fileName = CommonDialog.fileName

ScanData.SaveHDF (fileName) 'Saving HDF file with name from Save File Dialog box
End Sub

179

Analysis.frm Code

Dim DataArray(l To 512, 1 To 512), zpeaks(l To 5000), zpeaksl(l To 5000) As Single
Dim MaxPixel As Single

Dim xpeaks(l To 5000), ypeaks(l To 5000) As Integer

Dim peakno, p, p1, acceptedpeaks, size As Integer

Dim xpeaksl(l To 5000), ypeaksl(l To 5000) As Integer

Dim xzoomcenter, yzoomcenter, x1, y1 As Integer

Dim HistBar(l To 300) As Integer

Dim lowx, highx, lowy, highy, maxbin, maxpeakheight As Integer

Dim peakheight, maxzcoord, phb, bin, maxpeakhistbar As Integer

Dim bkc As Long

Dim Last As Boolean

Dim MostFreq, cutoff, threshold, radius As Integer

Dim textPeakValley, textMinExclude, fileName, defaultfilename As String

Public Sub buttonAccept_;Click()

'indicate that last peak was accepted and enable back button
Last = True
buttonBack.Enabled = True

tethHeight.ForeColor = O

xpeaksl(pl) = xpeaks(p)
ypeaksl(pl) = ypeaks(p)
zpeaksl(p1)= zpeaks(p) - maxbin
p=p+1

p1 = p1 + 1

acceptedpeaks = acceptedpeaks + 1

If p > peakno Then
analysis.Cls
pictureHistogram.Cls
Label4.Caption = "Save points or Quit?"
LabelS.Caption = "Enter Filename."
Exit Sub

Else

xzoomcenter = xpeaks(p): yzoomcenter = ypeaks(p)

'Clear form
analysis.Cls

180

'Zoom in on point

Dim heightmax, heightmin, step, xcount, ycount As Integer
Dim ptcol(l To 25, 1 To 25) As Long

heightmax = DataArray(xpeaks(p), ypeaks(p))

heightmin = heightmax

For x = xzoomcenter - 12 To xzoomcenter + 12 'Find minimum height
x1 = x

Ifx<1Thenx1=1

If x > size Then x1 = size

For y = yzoomcenter - 12 To yzoomcenter + 12

yl = y

Ify<1Theny1=1

If y > size Then yl = size

If DataArray(xl, yl) < heightmin Then heightmin = DataArray(x], yl)
Next y

Next x

step = Int(255 / (heightmax - heightmin»

xcount = 1: ycount = 1

For x = xzoomcenter - 12 To xzoomcenter + 12

x1 = x

Ifx<1Thenx1=1

Ifx > size Then X] = size

For y = yzoomcenter - 12 To yzoomcenter + 12

yl = y

Ify< 1 Thenyl =1

If y > size Then y1 = size

Color = Int((DataArray(x1, yl) - heightmin) * step)

If Color > 255 Then Color = 255

ptcol(xcount, ycount) = RGB(Color, Color, Color)

'Set all points >= the peak to green

If DataArray(x] , y1)>= DataArray(xpeaks(p), ypeaks(p)) Then ptcol(xcount, ycount) =
RGB(O, 255, O)

ycount = ycount + 1

Next y

ycount = 1

xcount = xcount + 1
Next x

'Set the peak to red

ptcol(13, 13) = 255

Dim xconv, yconv As Single

xconv = 3#
For px = 1 To 25
yconv = 3#

For py =1 To 25

181

'analysis.PSet (xconv, yconv), ptcol(px, py)
analysis.PSet (xconv, yconv), ptcol(px, py)
yconv = CSng(yconv + 5)

Next py

xconv = CSng(xconv + 5)

Next px

'Calculate histogram for local background around each peak found
lowx = xpeaks(p) - 1: If lowx < 1 Then lowx = 1

highx = xpeaks(p) + 1: If highx > size Then highx = size

lowy = ypeaks(p) - 40: If lowy < 1 Then lowy = 1

highy = ypeaks(p) + 40: If highy > size Then highy = size
maxpeakheight = 0

pictureHistogram.ForeColor = 255

pictureHistogramDrawMode = 13

pictureHistogramDrawWidth = 1

pictureHistogram.Cls

For x = lowx To highx
maxpeakhistbar = 0: maxbin = O: maxzcoord = O: peakheight = O
For bin = 1 To zpeaks(p): HistBar(bin) = 0: Next bin
For y = lowy To highy
If DataArray(x, y) > maxzcoord Then maxzcoord = DataArray(x, y)
bin = DataArray(x, y)
If bin > zpeaks(p) Then GoTo 4020
If bin < 1 Then GoTo 4020
HistBar(bin) = HistBar(bin) + 1
phb = HistBar(bin)
prhb >15 Then phb =15
pictureHistogram.Line (bin * 5, 141)-((bin + 1) * 5, 141 - phb * 9), , BF
4020 Next y
For bin = 1 To zpeaks(p)
If HistBar(bin) > maxpeakhistbar Then maxpeakhistbar = HistBar(bin): maxbin = bin
Next bin
peakheight = maxzcoord - maxbin
If peakheight > maxpeakheight Then maxpeakheight = peakheight
Next x
pictureHistogram.ForeColor = RGB(O, O, 255)
pictureHistogram.Line (maxbin * 5, 141)-((maxbin + 1) * 5, 0), , BF

textCurrentPeak.Text = p

textAcceptedPeaks.Text = acceptedpeaks

If DataArray(xpeaks(p), ypeaks(p)) - heightmin <= 5 Then tethHeight.ForeColor =
RGB(255, O, O)

tethI-Ieight.Text = Int(DataArray(xpeaks(p), ypeaks(p)) - maxbin)

182

Form1.MousePointer = 0 'Default mouse pointer
End If
End Sub

Private Sub buttonBack_Click()
tethHeight.ForeColor = 0

If Last = False Then GoTo 100

p1 = pl - 1

acceptedpeaks = acceptedpeaks - 1
xpeaksl(pl) = O

ypeaksl(pl) = 0

zpeakleJI) = 0

100 p = p - 1
buttonBack.Enabled = False

xzoomcenter = xpeaks(p): yzoomcenter = ypeaks(p)

'clear form
analysis.Cls

'Zoom in on point

Dim heightmax, heightmin, step, xcount, ycount As Integer
Dim ptcol(l To 25, 1 To 25) As Long

heightmax = DataArray(xpeaks(p), ypeaks(p))

heightmin = heightmax

For x = xzoomcenter - 12 To xzoomcenter + 12 'Find minimum height
x1 = x

Ifx<1Thenx1=1

If x > size Then x1 = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1 = y

Ify< l Thenyl =1

If y > size Then y1 = size

If DataArray(x] , yl) < heightmin Then heightmin = DataArray(x] , yl)
Next y

Next x

step = Int(255 / (heightmax - heightmin»

xcount = 1: ycount = 1

For x = xzoomcenter - 12 To xzoomcenter + 12

x1 = x

Ifx< 1 Thenxl =1

Ifx > size Then X] = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1=y

183

Ify<1Theny1 =1

If y > size Then yl = size

Color = Int((DataArray(x1, yl) - heightmin) * step)
If Color > 255 Then Color = 255

ptcol(xcount, ycount) = RGB(Color, Color, Color)
'Set all points >= the peak to green

If DataArray(x] , y1)>= DataArray(xpeastfi, ypeaks(p)) Then ptcol(xcount, ycount) =
RGB(O, 255, O)

ycount = ycount + 1

Next y

ycount = 1

xcount = xcount + 1

Next x

'Set the peak to red
ptcol(l3, 13) = 255

Dim xconv, yconv As Single

xconv = 3#

For px = 1 To 25

yconv = 3#

For py = 1 To 25

analysis.PSet (xconv, yconv), ptcol(px, py)
yconv = CSng(yconv + 5)

Next py

xconv = CSng(xconv + 5)

Next px

'Calculate histogram for local background around each peak found
lowx = xpeaks(p) - 1: If lowx < 1 Then lowx = l

highx = xpeaks(p) + 1: If highx > size Then highx = size

lowy = ypeaks(p) - 40: If lowy < 1 Then lowy = 1

highy = ypeaks(p) + 40: If highy > size Then highy = size
maxpeakheight = O

pictureHistogram.ForeColor = 25 5

pictureHistogramDrawMode = 13

pictureHistogramDrawWidth = 1

pictureHistogram.Cls

For x = lowx To highx
maxpeakhistbar = 0: maxbin = O: maxzcoord = O: peakheight = O
For bin = 1 To zpeaks(p): HistBar(bin) = 0: Next bin
For y = lowy To highy
If DataArray(x, y) > maxzcoord Then maxzcoord = DataArray(x, y)
bin = DataArray(x, y)
If bin > zpeaks(p) Then GoTo 4020

184

If bin < 1 Then GoTo 4020
HistBar(bin) = HistBar(bin) + 1
phb = HistBar(bin)
prhb >15 Then phb =15
pictureHistogram.Line (bin * 5, 141)-((bin + 1) * 5, 141 - phb * 9), , BF
4020 Next y
For bin = 1 To zpeaks(p)
If HistBar(bin) > maxpeakhistbar Then maxpeakhistbar = HistBar(bin): maxbin = bin
Next bin
peakheight = maxzcoord - maxbin
If peakheight > maxpeakheight Then maxpeakheight = peakheight
Next x
pictureHistogram.ForeColor = RGB(O, O, 255)
pictureHistogram.Line (maxbin * 5, 14l)—((maxbin + 1) * 5, O), , BF

textCurrentPeak.Text = p

textAcceptedPeaks.Text = acceptedpeaks

If DataArray(xpeaks(p), ypeaks(p)) - heightmin <= 5 Then tethHeight.ForeColor =
RGB(255, O, 0)

tethHeight.Text = Int(DataArray(xpeaks(p), ypeaks(p)) - maxbin)

Form1.MousePointer = 0 'Default mouse pointer
End Sub

Public Sub buttonQuit_Click()
Unload Me
End Sub

Public Sub buttonReject_Click()

'indicate Last peak was reject for back routine and enable back button
Last = False

buttonBack.Enabled = True

tethHeight.ForeColor = O

p=p+1

If p > peakno Then
analysis.Cls
pictureHistogram.Cls
Label4.Caption = "Save points or Quit?"
LabelS.Caption = "Enter F ilename."
Exit Sub

Else

xzoomcenter = xpeaks(p): yzoomcenter = ypeaks(p)

185

'Clear form
analysis.Cls

'Zoom in on point

Dim heightmax, heightmin, step, xcount, ycount As Integer
Dim ptcol(l To 25, 1 To 25) As Long

heightrnax = DataArray(xpeaks(p), ypeaks(p))

heightmin = heightmax

For x = xzoomcenter - 12 To xzoomcenter + 12 'Find minimum height
x1 = x

Ifx<1Thenx1=1

If x > size Then X] = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1 = y

Ify<1Theny1=1

If y > size Then yl = size

If DataArray(x] , yl) < heightmin Then heightmin = DataArray(xl , yl)
Next y

Next x

step = Int(255 / (heightmax - heightmin»

xcount = 1: ycount = 1

For x = xzoomcenter - 12 To xzoomcenter + 12

x1 = x

Ifx<1Thenx1=1

If x > size Then x1 = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1 = y

Ify<1Theny1=1

If y > size Then yl = size

Color = Int((DataArray(x1, yl) - heightmin) * step)

If Color > 255 Then Color = 255

ptcol(xcount, ycount) = RGB(Color, Color, Color)

'Set all points >= the peak to green

If DataArray(xl, y1) >= DataArray(xpeaks(p), ypeaks(p)) Then ptcol(xcount, ycount) =
RGB(O, 255, 0)

ycount = ycount + 1

Next y

ycount = 1

xcount = xcount + 1
Next x

’Set the peak to red

ptcol(l3, 13) = 255

Dim xconv, yconv As Single

186

xconv = 3#

For px = 1 To 25

yconv = 3#

For py = 1 To 25

analysis.PSet (xconv, yconv), ptcol(px, py)
yconv = CSng(yconv + 5)

Next py

xconv = CSng(xconv + 5)

Next px

'Calculate histogram for local background around each peak found
lowx = xpeaks(p) - 1: If lowx < 1 Then lowx = 1

highx = xpeaks(p) + 1: If highx > size Then highx = size

lowy = ypeaks(p) — 40: If lowy < 1 Then lowy = 1

highy = ypeaks(p) + 40: If highy > size Then highy = size
maxpeakheight = O

pictureHistogram.ForeColor = 255

pictureHistogram.DrawMode = 13

pictureHistogram.DrawWidth = 1

pictureHistogram.Cls

For x = lowx To highx
maxpeakhistbar = 0: maxbin = O: maxzcoord = O: peakheight = O
For bin = 1 To zpeaks(p): HistBar(bin) = 0: Next bin
For y = lowy To highy
If DataArray(x, y) > maxzcoord Then maxzcoord = DataArray(x, y)
bin = DataArray(x, y)
If bin > zpeaks(p) Then GoTo 4020
If bin < 1 Then GoTo 4020
HistBar(bin) = HistBar(bin) + 1
phb = HistBar(bin)
prhb >15 Then phb =15
pictureHistogram.Line (bin * 5, 141)-((bin + 1) * 5, 141 - phb * 9), , BF
4020 Next y
For bin = 1 To zpeaks(p)
If HistBar(bin) > maxpeakhistbar Then maxpeakhistbar = HistBar(bin): maxbin = bin
Next bin
peakheight = maxzcoord - maxbin
If peakheight > maxpeakheight Then maxpeakheight = peakheight
Next x
pictureHistogram.ForeColor = RGB(O, O, 255)
pictureHistogram.Line (maxbin * 5, 141)-((maxbin + 1) * 5, 0), , BF

textCurrentPeak.Text = p
textAcceptedPeaks.Text = acceptedpeaks

187

If DataArray(xpeaks(p), ypeaks(p)) - heightmin <= 5 Then tethHeight.ForeColor =
RGB(255, O, O)
tethHeight.Text = Int(DataArray(xpeaks(p), ypeaks(p)) - maxbin)

Forml .MousePointer = 0 'Default mouse pointer
End If
End Sub

Public Sub buttonSave_Click()
Form1.MousePointer = 11 'or VB hourglass
Dim SaveFileName
Dim peakcount As Integer
'Write peak data to file
SaveFileName = textSaveFileName.Text
Open SaveFileName For Output As #15
Write #15, fileName
Write #15, textPeakValley; acceptedpeaks
Write #15, "maximum="; MaxPixel, "most frequent="; MostFreq, "Min. exclude=";
textMinExclude
Write #15, "cutoff="; cutoff, "threshold="; threshold, "radius="; radius
For peakcount = 1 To acceptedpeaks
Write #15, xpeaksl(peakcount), ypeaksl(peakcount), zpeaksl(peakcount)
Next peakcount
Close #15
Forml .MousePointer = 0 'Default mouse pointer
Unload Me
End Sub

Private Sub Form_Activate()
buttonBack.Enabled = False
tethHeight.ForeColor = O

acceptedpeaks = O

p = 1

p1 = 1

textSaveFileName.Text = defaultfilename
xzoomcenter = xpeaks(p): yzoomcenter = ypeaks(p)

'clear form
analysis.Cls

'Zoom in on point

Dim heightmax, heightmin, step, xcount, ycount As Integer
Dim ptcol(l To 25, 1 To 25) As Long

188

heightmax = DataArray(xpeaks(p), ypeaks(p))

heightmin = heightmax

For x = xzoomcenter - 12 To xzoomcenter + 12 'Find minimum height
x1 = x

Ifx<1Thenx1 =1

If x > size Then x1 = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1 = y

Ify< l Thenyl =1

If y > size Then yl = size

If DataArray(xl, yl) < heightmin Then heightmin = DataArray(x] , yl)
Next y

Next x

step = Int(255 / (heightmax - heightmin))

xcount = 1: ycount = 1

For x = xzoomcenter - 12 To xzoomcenter + 12

x1 = x

Ifx<1Thenx1=1

If x > size Then X] = size

For y = yzoomcenter - 12 To yzoomcenter + 12

y1 = y

Ify<1Theny1 =1

If y > size Then yl = size

Color = Int((DataArray(x1, yl) - heightmin) * step)

If Color > 255 Then Color = 255

ptcol(xcount, ycount) = RGB(Color, Color, Color)

'Set all points >= the peak to green

If DataArray(x] , y1)>= DataArray(xpeaks(p), ypeaks(p)) Then ptcol(xcount, ycount) =
RGB(O, 255, O)

ycount = ycount + 1

Next y

ycount = 1

xcount = xcount + 1
Next x

'Set the peak to red

ptcol(l3, 13) = 255

Dim xconv, yconv As Single

xconv = 3#
For px = 1 To 25
yconv = 3#

For py = 1 To 25

analysis.PSet (xconv, yconv), ptcol(px, py)
yconv = CSng(yconv + 5)

Next py

189

xconv = CSng(xconv + 5)
Next px

'Calculate histogram for local background around each peak found
lowx = xpeaks(p) - 1: If lowx < 1 Then lowx = 1

highx = xpeaksw) + 1: If highx > size Then highx = size

lowy = ypeaksfli) - 40: If lowy < 1 Then lowy = 1

highy = ypeaks(p) + 40: If highy > size Then highy = size
maxpeakheight = O

pictureHistogram.ForeColor = 255

pictureHistogram.DrawMode = 13

pictureHistogram.DrawWidth = l

pictureHistogram.Cls

For x = lowx To highx
maxpeakhistbar = 0: maxbin = O: maxzcoord = O: peakheight = 0
For bin = 1 To zpeaks(p): HistBar(bin) = 0: Next bin
For y = lowy To highy
If DataArray(x, y) > maxzcoord Then maxzcoord = DataArray(x, y)
bin = DataArray(x, y)
If bin > zpeaks(p) Then GoTo 4020
If bin < 1 Then GoTo 4020
HistBar(bin) = HistBar(bin) + 1
phb = HistBar(bin)
prhb >15 Then phb =15
pictureHistogram.Line (bin * 5, 141)-((bin + 1) * 5, 141 - phb * 9), , BF
4020 Next y
For bin = 1 To zpeaks(p)
If HistBar(bin) > maxpeakhistbar Then maxpeakhistbar = HistBar(bin): maxbin = bin
Next bin
peakheight = maxzcoord - maxbin
If peakheight > maxpeakheight Then maxpeakheight = peakheight
Next x
pictureHistogram.ForeColor = RGB(O, 0, 255)
pictureHistogram.Line (maxbin * 5, 141)-((maxbin + 1) * 5, 0), , BF

textCurrentPeak.Text = p

textAcceptedPeaks.Text = acceptedpeaks

If DataArray(xpeaks(p), ypeaks(p)) - heightmin <= 5 Then tethHeight.ForeColor =
RGB(255, 0, 0)

tethHeight.Text = Int(DataArray(xpeaks(p), ypeaks(p)) - maxbin)

Form1.MousePointer = 0 'Default mouse pointer
End Sub

Public Sub Form_LoadO

190

Open "variable" For Input As #3
Input #3, peakno, MaxPixel, MostFreq, cutoff, threshold, radius, size
Input #3, fileName, textPeakValley, textMinExclude, defaultfilename
For x =1 To size
For y = 1 To size
Input #3, DataArray(x, y)
Next y
Next x
For peaks = 1 To peakno
Input #3, xpeaks(peaks), ypeaks(peaks), zpeaks(peaks)
Next peaks
Close #3
Kill "variable"

Forml .MousePointer = 0 'Default mouse pointer
End Sub

191

"‘hilttutu