Citation
Mesoporous silcia nanoparticles as a breast cancer targeting contrast agent for ultrasound imaging

Material Information

Title:
Mesoporous silcia nanoparticles as a breast cancer targeting contrast agent for ultrasound imaging
Creator:
Milgroom, Adnrew Carson
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file. : ;

Subjects

Subjects / Keywords:
Contrast media (Diagnostic imaging) ( lcsh )
Ultrasound contrast media ( lcsh )
Contrast media (Diagnostic imaging) ( fast )
Ultrasound contrast media ( fast )
Genre:
non-fiction ( marcgt )

Notes

Review:
Current clinical use of ultrasound for breast cancer diagnostics is strictly limited to a role as a supplementary detection method to other modalities, such as mammography or MRI. A major reason for ultrasound's role as a secondary method is its inability to discern between cancerous and non-cancerous bodies of similar density, like dense calcifications or benign fibroadenomas. Its detection capabilities are further diminished by the variable density of the surrounding breast tissue with the progression of age. Preliminary studies suggest that mesoporous silica nanoparticles (MSNs) are a good candidate as an in situ contrast agent for ultrasound. By tagging the silica particle surface with the cancer-targeting antibody trastuzumab (Herceptin), suspect regions of interest can be better identified in real time with standard ultrasound equipment. Once the silica-antibody conjugate is injected into the bloodstream and enters the cancerous growth's vasculature, the antibody arm will bind to HER2, a cell surface receptor known to be dysfunctional or overexpressed in certain types of breast cancer. As more particles aggregate at the cell surface, backscatter of the ultrasonic waves increases as a result of the higher porous silica concentration. This translates to an increased contrast around the lesion boundary. Tumor detection through ultrasound contrast enhancement provides a tremendous advantage over current cancer diagnostics because is it significantly cheaper and can be monitored in real time. Characterization of MCM-41 type MSNs suggests that these particles have sufficient stability and particle size distribution to penetrate through fenestrated tumor vasculature and accumulate in HER2+ breast cancer cells through the enhanced permeation and retention (EPR) effect. A study of acoustic properties showed that particle concentration is linearly correlated to image contrast in clinical frequency-range ultrasound, although less pronounced than typical microbubble-type contrast agents. In vitro studies using cells with varied levels of HER2 expression demonstrated the selectivity of the MSN-Herceptin conjugate to cells with HER2 overexpression. Fluorescence imaging suggest these images remain surface-bound and are not incorporated into the cell body. This study demonstrates the potential of MSNs as a stable, safe, and effective imaging contrast agent for ultrasound-based cancer diagnostics. Ultimately this work will contribute towards the improvement of diagnostic alternatives to conventional ionizing radiation-intensive imaging - such as MRI or X-ray - without compromising the specificity of the test.
Thesis:
Thesis (M.S.)--University of Colorado Denver. Bioengineering
Bibliography:
Includes bibliographic references.
General Note:
Department of Bioengineering
Statement of Responsibility:
by Andrew Carson Milgroom.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
862743024 ( OCLC )
ocn862743024

Downloads

This item has the following downloads:


Full Text
MESOPOROUS SILICA NANOPARTICLES AS A BREAST CANCER TARGETING CONTRAST
AGENT FOR ULTRASOUND IMAGING
by
Andrew Carson Milgroom
B.S., Tufts University, 2011
A thesis submitted to the
University of Colorado Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
Bioengineering
2012


This thesis for the Master of Science
Degree by
Andrew Carson Milgroom
has been approved by
Robin Shandas, chair
Daewon Park
Bolin Liu
November 13th, 2012
ii


Milgroom, Andrew Carson (M.S., Bioengineering)
Mesoporous Silica Nanoparticles as a Contrast Agent for Ultrasound Imaging
Thesis directed by Assistant Professor Daewon Park
ABSTRACT
Current clinical use of ultrasound for breast cancer diagnostics is strictly limited
to a role as a supplementary detection method to other modalities, such as
mammography or MRI. A major reason for ultrasound's role as a secondary method is
its inability to discern between cancerous and non-cancerous bodies of similar density,
like dense calcifications or benign fibroadenomas. Its detection capabilities are further
diminished by the variable density of the surrounding breast tissue with the progression
of age.
Preliminary studies suggest that mesoporous silica nanoparticles (MSNs) are a
good candidate as an in situ contrast agent for ultrasound. By tagging the silica particle
surface with the cancer-targeting antibody trastuzumab (Herceptin), suspect regions of
interest can be better identified in real time with standard ultrasound equipment. Once
the silica-antibody conjugate is injected into the bloodstream and enters the cancerous
growth's vasculature, the antibody arm will bind to HER2, a cell surface receptor known
to be dysfunctional or overexpressed in certain types of breast cancer. As more particles
aggregate at the cell surface, backscatter of the ultrasonic waves increases as a result of
the higher porous silica concentration. This translates to an increased contrast around
the lesion boundary. Tumor detection through ultrasound contrast enhancement
provides a tremendous advantage over current cancer diagnostics because is it
significantly cheaper and can be monitored in real time.
Characterization of MCM-41 type MSNs suggests that these particles have
sufficient stability and particle size distribution to penetrate through fenestrated tumor
vasculature and accumulate in HER2+ breast cancer cells through the enhanced
permeation and retention (EPR) effect. A study of acoustic properties showed that
particle concentration is linearly correlated to image contrast in clinical frequency-range
ultrasound, although less pronounced than typical microbubble-type contrast agents. In
vitro studies using cells with varied levels of HER2 expression demonstrated the
selectivity of the MSN-Herceptin conjugate to cells with HER2 overexpression.
Fluorescence imaging suggest these images remain surface-bound and are not
incorporated into the cell body.


This study demonstrates the potential of MSNs as a stable, safe, and effective
imaging contrast agent for ultrasound-based cancer diagnostics. Ultimately this work
will contribute towards the improvement of diagnostic alternatives to conventional
ionizing radiation-intensive imaging such as MRI or X-ray without compromising the
specificity of the test.
The form and content of this abstract are approved. I recommend its publication.
Approved by Robin Shandas


ACKNOWLEDGEMENTS
Foremost, I would like to thank my committee members Dr. Daewon Park, Dr. Bolin
Liu, and Dr. Robin Shandas for taking the time over the past year to provide me with
guidance throughout the project. I would also like to thank Luciano Mazzaro for his
assistance with both ultrasound equipment and contrast agents, and Dr. Qun Li for her
assistance with the cellular assays. Dr. Dunghwa Yun was instrumental in assisting with
chemistry-related advice and solutions. Dr. Lara Hardesty and Dr. Ann Scherzinger (U. of
Colorado Hospital) provided images and knowledge of clinical breast cancer diagnostics.
Figures from Martin E. Anderson (Duke University) were adapted for Faran modeling
and speckle distribution.
v


TABLE OF CONTENTS
CHAPTER
1. Introduction....................................................................1
1.1 Review of breast cancer ultrasound diagnostics..................................1
1.2 Review of Herceptin.............................................................2
1.3 Review of ultrasound contrast...................................................4
1.3.1 Reflection by Rayleigh-like scattering..........................................4
1.3.2 Reflection by variation at media boundaries.....................................5
1.3.3 Reflection by attenuation.......................................................7
1.4 Ultrasound Enhanced Contrast Agent (UECA).......................................8
1.5 Mesoporous silica nanoparticles as an ultrasound contrast agent................10
1.5.1 Composition....................................................................10
1.5.2 Biomedical applications of MSNs................................................11
2. Aims and experimental approach.................................................13
2.1 Specific Aims..................................................................13
2.1.1 Demonstrate the contrast potential of mesoporous silica nanoparticles..........13
2.1.2 Apply MSN as an ultrasound contrast agent through cancer cell targeting........13
2.2 Experimental Obstacles.........................................................14
2.2.1 Particle selection.............................................................14
2.2.2 Sieving........................................................................14
3. Materials and methods..........................................................16
3.1 Materials......................................................................16
3.1.1 Consumables....................................................................16
3.1.2 Equipment......................................................................16
3.2 Methods........................................................................18
3.2.1 Characterization of mesoporous silica nanoparticles............................18
3.2.2 Determination of acoustics properties..........................................18
3.2.3 Herceptin Conjugation to silica nanoparticles..................................19
3.2.5 In vitro studies...............................................................23
4. Results........................................................................25
vi


4.1 Mathematical Justification for MSNs as a contrast agent.........................25
4.1.1 Contrast by backscatter.........................................................25
4.2 Characterization of mesoporous silica nanoparticles.............................27
4.3 Nanoparticle concentration to contrast intensity correlation....................29
4.4 Cellular interaction of nanoparticles...........................................32
4.4.1 Fluorescence microscopy.........................................................32
4.4.2 Flow cytometry..................................................................33
4.4.3 in vitro ultrasound imaging.....................................................35
5. Discussion......................................................................38
5.1 Contrast Efficacy...............................................................38
5.2 Potential Applications..........................................................39
6. Conclusion......................................................................41
6.1. Conclusion......................................................................41
6.2 Future Direction................................................................41
Appendix..............................................................................44
A. MATLAB code: Single-Pulse Processing............................................44
B. MATLAB code: B-mode image processing............................................47
C. Herceptin Protein sequence......................................................49
References............................................................................52
vii


LIST OF FIGURES
FIGURE
1. Ultrasound imaging of cancerous breast tissue.................................2
2. Backscatter intensity as a function of wave frequency and particle diameter...5
3. Incident wave through two media at angle theta................................6
4. Ultrasound image of breast displaying shadowing caused by attenuation........7
5. Diagram of acoustic measurement apparatus....................................17
6. Chemical reaction of MSN-antibody linkage....................................22
7. Comparison of Acoustic Properties............................................26
8. TEM images of MCM-41 mesoporous silica.......................................27
9. SEM imaging of MSNs..........................................................28
10. FTIR analysis................................................................29
11. Comparison of acrylamide plug with and without MCM-41 silica particles.......30
12. Linear interpolation of contrast with respect to MSN concentration...........31
13. Probability distribution function of a sub-resolution scatterer..............32
14. Comparison of cells treated with MSN-Herceptin-FITC..........................33
15. Sample gating of cells for flow cytometry....................................33
16. Dead cells ratio comparison using ethidium homodimer 1.......................34
17. Ultrasound contrast profile of skbr3 cells...................................36
18. Contrast profile cross sections of HER2+ and HER2- cells treated with MSN-Herceptin 37
viii


LIST OF TABLES
TABLE
1. Summary of current contrast agents.............................................9
2. Advantages and disadvantages of silica nanoparticles as a contrast agent in comparison to gas
phase contrast agents.............................................................40
IX


LIST OF ACRONYMS
MSN Mesoporous Silica Nanoparticles
FITC Fluorescein Isothiocyanate
APTES (3-aminopropyl)triethoxysilane
PSD Particle SizeDistribution
MCM Mobil Composition Material
TEOS Tetraethylorthosilicate
CTAB Cetylmethylammonium Bromide
IUPAC International Union of Pure and Applied Chemistry
PBS Phosphate-Buffered Saline
DMEM Dulbecco's Modified Eagle Formula
TEM Tunneling Electron Microscope
LVSEM Low Vacuum Scanning Electron Microscope
NCF Nanomaterials Characterization Facilities
NDT Non-Destructive Testing
PTF Point Target Focus
FBS Fetal Bovine Serum
PS Penecillin-Streptomycin
NHS N-hydroxyl succinimide
EDC l-ethyl-3-(3-dimethylaminopropyl) carbodiimide)
DMSO Dimethylsulfoxide
MWCO Molecular Weight Cutoff
TEMED Tetramethylethylenediamine
AmPs Ammonium Persulfate
BUA Broadband Ultrasound Imaging
UECA Ultrasound Enhanced Contrast Agent
MPI Mean Pixel Intensity
PDF Probability Distribution Function
X


1.
Introduction
1.1 Review of breast cancer ultrasound diagnostics
Ultrasound used for breast cancer detection can accurately determine if a mass is a
benign fluid-filled cyst, or a potentially malignant solid mass. A sonogram of a malignant tumor
is characterized by:
Lack of circumscribed margins
Heterogeneous echo patterns
Increased anteroposterior dimension ('taller than wide')
Unfortunately, it is very difficult to determine if a detected solid mass is truly a malignant
carcinoma, or simply a calcification or fibroadenoma since all of these cases yield similar
densities. Malignant tumors are typically hypoechoic, but may either be hyperechoic or
isoechoic as well (Figure l.A). The detection of the mass diameter may also be difficult to
determine if the surrounding tissue is denser than normal breast tissue, which is common in
women below the age of 50. In the clinical setting, ultrasound is typically administered in
conjunction with another standard method of diagnosis, most commonly mammography. One
study found that ultrasound adjunct to mammography was able to increase the incremental rate
of detection by up to 41%[1]. Remarkably, ultrasound actually has a higher specificity than
mammography, is less affected by surrounding tissue density as a result of age, and can detect
smaller diameter lesions121. However, its inability to stand alone as a primary method of
detection is mostly due to a high false positive rate131. Surgical biopsies of lumps that showed a
positive ultrasound following a negative mammogram were determined to be benign in 3% of
the cases141. Depending on the age group, adding ultrasound to mammography may quadruple
the false positive rate. To further improve detection rates following diagnostic ultrasound,
biopsies are often included. With an average cost around $500, biopsies require time from both
the sonographer and the pathologist. It may also take up to a few days to return the results. Due
to the discomfort of the large bore needles, local anesthetics must also be used. Some patients
have expressed concern with the possibility of contaminating new regions from cancer cells on
the needle as it is pulled out of the tissue.
1


Ultrasound is also commonly used for needle guidance for biopsy as well as localization
for surgical removal. If a lesion has been determined to be cancerous, it is marked for removal
prior to surgery by placing a wire across the diameter of the cancerous region (Figure l.B). The
accuracy of this method of occult lesion localization is dependent upon the sonographer's ability
to distinguish the tumor boundary edges. Again, local anesthetics must be used to
accommodate for the large bore needle. During surgery, the wire is located either visually or
with intraoperative ultrasound. A recent study suggests that around 33% of oncological
surgeons use intraoperative ultrasound regularly151. With or without ultrasound, the extent of
tissue removed around the wire is determined by palpation. Since surgeons must rely on a one-
dimensional marker, excess tissue must be removed as a cautionary measure (Figure l.C).
Figure 1. Ultrasound imaging of cancerous breast tissue. (A) Typical indication of malignant tumor, (B) Needle
inserted through suspected tumor for localization wire injection, (C) MRI of excised tissue with localization
wire and biopsy site clip injected during US imaging
It is evident that ultrasound holds many advantages over current standards of diagnosis.
Once diagnosis specificity is improved upon, ultrasound-based diagnostics may prove to be
adequate as a primary method of detection comparable to mammography. Diagnostic
ultrasound for breast cancer and other HER2 positive cancers can benefit from contrast
enhancement for detection and presurgical occult lesion localization
1.2 Review of Herceptin
Flerceptin is the trade-name given by Genentech for the cancer treatment therapy
trastuzumab. At 155 kDa, trastuzumab is a humanized monoclonal antibody that has been
developed to interfere with the functionality of the FIER2/neu or ErbB2 receptor. When a
secondary receptor such as HER3 binds to an external stimulus, it dimerizes with a FIER2/neu
receptor. The activated FIER2/neu receptor in turn initiates a MAPK and PI3/AKT kinase
2


pathway, subsequently initiating a NF-k8 growth pathway161. By this chain of events, an external
signal can regulate cellular proliferation, migration, differentiation, and adhesion. Certain types
of cancers particularly types of breast cancer have been shown to overexpress the HER2/neu
receptor by 25% up to 100 fold171. Uncontrolled growth as a result of overexpressed HER2/neu
receptors occurs by two mechanisms: (1) A high concentration of HER2/neu receptors on the
cell surface (>2 million/cell) leads to a tendancy for hyperactivity, initiating cell growth pathways
without external mitogen stimulation, or (2) The constitutive activation of the AKT pathway
interrupts the p27Kipl checkpoint181. Since p27Kipl is constantly in a phosphorylated state, it
begins to accumulate in the cytoplasm, unable to reach the nucleus. A loss of inhibition to the
cdk2 pathway by p27Kipl results in an exacerbated effect from the HER2/neu overexpression.
Trastuzumab interferes with HER2/neu hyperactivity by disrupting the dimerization step
prior to pathway initiation. To date, it is FDA approved for treatment of certain breast cancers,
as well as stomach, gastroesphageal junction, and uterine papillary cancers191. Prior to
treatment, the region of interest must be screened to confirm that the specific type of cancer
exhibits overexpression of FIER2/neu receptors on the surface. Immunohistochemistry or
fluorescence in situ hybridization (FISH) of a biopsy section can sufficiently determine the
quantity of FIER2/neu receptors on the cell surface. Despite a high binding affinity of
trastuzumab to the FIER2/neu receptor, a recent study has shown that nearly 70% of patients
undergoing treatment with trastuzumab alone do not respond to the treatment1101. This abysmal
statistic may be explained by patient-to-patient variation in resistance. The resistance may arise
as a de novo primary resistance, or become acquired in a longterm setting. Flerein presents an
opportunity to improve trastuzumab as a cancer treatment option. Currently, trastuzumab is
only applied as an adjuvant treatment, either administered concurrently with
chemotherapeutics or post-surgery. One study found that trastuzumab treatment post-surgery
can reduce relapse in patients by 50%[111.
Since trastuzumab is so highly selective for HER2 receptors, the drug has recently
become a favorable agent for cell targeting. Liu et al. were able to successfully conjugate the
trastuzumab antibody to a liposomal drug delivery vehicle for a camptothecin anticancer drug.
The results showed that immunotargeting with trastuzumab improved cellular uptake from
overexpressing HER2 cancer cells by 50-300 fold1121. Prior studies have also shown that Anti-
HER2 antibodies can maintain their anti-tumor efficacy despite conjugation to another body1131.
3


1.3 Review of ultrasound contrast
Ultrasound waves are longitudinal compression waves caused by pulses of pressure
propagating in a direction away from the source. These pulses create a compression and
expansion (rarefaction) effect that causes ripples to propagate through a medium. Waves are
able to travel through media since most media have some degree of elasticity. These waves are
considered ultrasonic since its frequency is above the audible spectrum of the human ear (~20
kHz).
Depending on the medium, a certain amount of the ultrasound wave returns back in the
direction of the source. The amount of energy returned to the transducer (source) is used to
calculate the elasticity and density of a medium at different depths. By creating an array of
transmitting/receiving transducers, a 2D image can be created. The amount of energy returned
to the transducer is calculated as the pixel intensity at a certain depth. White pixels correspond
to more energy returned to the transducer, whereas black pixels correspond to less energy
returned to the transducer. The extent of returned energy from the initial wave pulse is dictated
by several mechanisms.
1.3.1 Reflection by Rayleigh-like scattering
Most physiologically relevant media exhibit some form of inhomogeneity. On ultrasound
images, the result is a snowy appearance defined a 'speckling'. The speckle pattern of each
tissue is unique. While some groups are currently investigating the possibility of calculating
speckle pattern to determine tissue composition, the observation of speckle pattern in a clinical
setting is still primarily qualitative.
Speckle patterns are caused by inhomogeneities whose diameter is less than the axial
resolution of the imaging system. The interaction of these sub-resolution particles and an
incident sound wave may be defined as Rayleigh-like scatters. First used to describe particle
interactions with light, Rayleigh scattering principles may also be applied to describe the
interaction of sound waves with particles by substituting several parameters. Any
inhomogeneity whose density or compressibility is different from the surrounding medium can
be defined as 'point scatterer'. The interaction of sound waves at an interface of differing
density or compressibility acts as a source for a secondary sound wave. This secondary wave
propagates in a spherical manner concentric to the point scatterer, as described by Huygen's
principal (also originating in the context of light waves)[14]. The magnitude of the secondary
4


wave in the direction of the initial source defines the backscatter. According to the Faran
model1151, the amplitude of the backscatter is primarily dependent on its ka value, or its
wavenumber multiplied by the point scatterer's radius. Under Faran's model of a solid sphere
introduced to planar orthogonal pressure wave, the echo intensity follows a 4th order
dependence when ka is less than 1 similar to Rayleigh scatterers in the context of light waves
(Figure 2, left). As the radius increases, the echo amplitude drastically increases as well.
Assuming that the incident sound wave remains constant (k), it is clear that the echo intensity
will also exponentially decreases as the particle radius decreases (Figure 1, left). Neglecting
other phenomena (such as harmonic oscillation or attenuation), a 1pm diameter particle
observed by a 10 MFIz wave in water (c = 1493m/s) yields an insignificant echo response {ka =
a2n/A = 5e-7*2n/3e-4 = 0.02). As such, contrast agents must take advantage of other
phenomena in order to effectively create backscatter.
Figure 2. Backscatter intensity as a function of wave frequency (k) and particle diameter (a)
The Faran model also reveals a strong frequency dependence for echo intensity (Figure
2, right). When the sub-resolution particle diameter remains constant, echo intensity (dashed
line) displays strong nodes and peaks as a function of frequency (proportional to wavenumber,
k). As frequency increases, resolution improves echo intensity becomes less frequency
dependent.
1.3.2 Reflection by variation at media boundaries
Most relevant ultrasound contrast is introduced at the interface between two media on
a macroscale (>lmm). This is primarily due to the reduction of acoustic mismatch between the
contrast agent and the surround tissue. Image contrast intensity is largely governed by the
5


acoustic mismatch, or difference between acoustic impedance at an interface of two difference
media. The acoustic impedance (Z) can be defined as:
Z = pc;
(1)
The acoustic wave velocity, c, is the quotient of a medium's density and its bulk modulus, B. In
gasses, the bulk modulus is calculated as the product of the gas' inherent adiabatic index (the
coefficient that correlates a change in volume to its responsive change in temperature) and the
ambient pressure. The bulk modulus of a solid is typically determined experimentally.
By knowing the acoustic impedance of two media, the contrast induced at the boundary
between these two media can be theoretically calculated by the intensity reflection coefficient
equation:
(z2-z1y
(z2+Zi)
Where Z2 defines the media through which the incident sound wave initially interacts, and Z2
defines the secondary media of interaction. This form of the equation assumes that the incident
sound wave is perpendicular to the boundary surface. Angle approaches are considered using
Snell's Law:
Ri =
(Z2/cos9t-Z1/cos9i)
(Z2/cos9t+Z1/cos9[)
2
where
sinQi Xx _
sin9t A2 c2
(3)
Figure 3. Incident wave through two media at angle theta
By substituting equation (1) into the Z variables of equation (2), it is apparent that the intensity
reflection coefficient is solely determined by differences in density and stiffness (bulk modulus).
6


By having a low density and a low acoustic wave velocity (as a result of a small bulk modulus),
gasses provide an excellent acoustic impedance mismatch to soft tissue. Alternatively, by having
a density and acoustic wave velocity higher than soft tissue, solid materials can also yield strong
intensity reflections at a soft tissue boundary.
1.3.3 Reflection by attenuation
The majority of ultrasound imaging is dependent upon backscatter ('echo') caused by
boundary surfaces of different density or compressibility. By this mechanism, the contrast
intensity is determined by how much wave energy is reflected back to the transducer, and how
much energy is either transmitted across the boundary or scattered in a direction away from the
transducer. Ultimately, contrast is determined by the difference between the magnitude of
wave energy received and the original pulse wave energy. Another mechanism of energy loss is
caused by converting acoustic wave energy into another form, such as heat. The result on a
sonogram would be a darker region, defined as 'anechoic'. This loss in energy also affects
regions beneath the anechoic region, resulting in a phenomenon known as shadowing (Figure
4).
Figure 4 Ultrasound image of breast displaying shadowing caused by attenuation. Image obtained from
http://nexradiology.blogspot.com
Attenuation the conversion of acoustic energy to another form can be quantified by the
following equation:
H(f,z) = a0fz
(4)
7


The degree of attenuation is dependent upon the depth of the region of interest, z, as well as
the sound wave frequency,/. As depth and frequency increases, the attenuation of the intitial
pulse also increases. The attenuation coefficient, o0, is a constant value inherent to the medium
commonly expressed in terms of dB/(cm MHz).
Attenuation can be caused by small particles whose diameter is much smaller than the
sound wavelength. In 1972 J.R. Allegra and S.A. Hawley published a landmark paper defining the
parameters that affect attenuation in suspensions and emulsion [16]. Using 0.44-0.653 pm
diameter polystyrene spheres suspended in water or toluene introduced to sound waves across
a 9-165 MHz range, Allegra and Hawley were able to accurately predict the attenuation of the
initial sound wave in each experiment. The proposed model applies six boundary conditions
through six differential equations. The mathematics of these equations is beyond the scope of
this thesis, but several concepts are taken into consideration:
The incident sound wave can be characterized by three wave equations
(compressional, shear, and thermal);
Linearized conservation laws apply to the system. Denser objects with more
inertia will have less spatial displacement from compressional waves. Since
denser solid particles will move less in relation to the surrounding fluid (or soft
tissue), flow is observed across the particle surface. This flow introduces viscous
drag. This viscous flow justifies the addition of a shear wave component to
describe the incident sound wave. As viscosity increases, heat created from
friction increases and acoustic wave energy is lost.
Thermal conduction can be considered, according to thermodynamic laws of
pressure-temperature coupling. Depending of the specific heat of a material,
thermal loss can be related to pressure increases from ultrasonic wave pulses.
1.4 Ultrasound Enhanced Contrast Agent (UECA)
A vast majority of FDA approved contrast agents are gas phase microbubbles that are
encapsulated in either liposomal or protein carriers. Table 1 provides a summary of the three
most widely used contrast agents for ultrasound imaging.
8


Table 1. Summary of current contrast agents
Trade name Sonovue Optison Definity
Gas Sulphur hexafluoride Perfluoropropane Perfluoropropane
Mean bubble size 2-8 pm 3.0-4.5 pm 1.1-2.5 pm
Shell composition Phospholipid Human albumin Phospholipid
Manufacturer Bracco GE healthcare Bristol Myers Squibb
All three contrast agents are applied as an agent to increase the contrast of blood flow
through dynamics systems, particularly through the heart. Figure 5 provides an illustration of
the enhanced contrast of blood when injected with a 1 mL bolus of Sonovue.
Figure 5. Comparison of ventricular filling (A) before and (B) after Sonovue injection
This class of contrast agents is "indicated for use in patients with suboptimal
echocardiograms to opacify the left ventricular chamber and to improve the delineation of the
left ventricular endocardial border"[1?I.
F F F F
Figure 6. Chemical structure of Perfluoropropane
9


All three contrast agents employ some form of heavy gasses (Figure 6). Unlike lighter
gasses (including air), these gasses have a lower solubility which consequentially increases the
half-life in the blood stream. Regardless, the half-life of these bubbles is restricted to several
hours. Due to the relatively large diameter, most bubbles are cleared from the system much
sooner. Since all currently approved microbubbles are larger than 1pm, these contrast agents
are unable to pass through endothelial barriers. 96% of perfluoropropane gas in a 20mL bolus of
Optison is eliminated through the lungs within 10 minutes [1SI. The albumin shell is most likely
eliminated by liver degradation. Heavy gasses such as perfluoropropane provide the contrast
effect typical of gas to solid/liquid phase acoustic mismatch (Section 1.3.2, 'Reflection by
variation at media boundaries'). Another mechanism by which contrast is increased is through
harmonic oscillations created by ultrasound waves close to the bubble resonant frequency [19].
This contrast effect is caused in part by the elasticity of the bubble's geometry. Unfortunately,
this high degree of elasticity also contributes to its susceptibility to rupture. As a result, imaging
time under continues wave pressure is limited to the order of minutes.
Several groups have previously attempted to apply MSNs as a UECA [201. Casciaro et at.
and Liu et at. successfully correlated mean pixel intensity of an image to the concentration of
330-660nm diameter MSNs within a clinically relevant frequency range. In vivo studies
suggested that untargeted MSNs aggregate within Browicz-Kupffer cells as a part of the reticulo-
endothelial system (mononuclear phagocyte system) in the liver. The aggregate concentration
within the liver is significant enough to be detected by radio frequency-range ultrasound. At the
time of this publication, no groups have reported findings on targeted MSN contrast agent
effects. Furthermore, no groups have provided an in-depth computational justification for MSN
usage as a contrast agent.
1.5 Mesoporous silica nanoparticles
1.5.1 Composition
Mesoporous silica is most similar to synthetic zeolites, a class of silicate mineral that is
notable for its porous nature. The porosity of zeolite has made the mineral ideal for commercial
applications such as absorbents and purification filters. The "mesoporous" descriptor of the
silica nanoparticles is defined by the International Union of Pure and Applied Chemistry (IUPAC)
classification of porous materials. Any inorganic structure whose pore diameter is between 20 A
and 500 A is classified as mesoporous. Anything above 500 A would be macroporous, whereas
10


anything below 20 A would be microporous. Although several various microstructures exist,
most follow the general method exemplified from the production of Mobil Composition of
Matter (MCM)-41 (Figure 7). For MCM-41, the surfactant cetylmethylammonium bromide
(CTAB) naturally creates liquid crystalline micelles in water. This will act as the template for the
pore structure. The ceramic sol-gel precursor tetraethylorthosilicate (TEOS) creates a silica
network around the micellular formations through hydrolysis and condensation. Once the CTAB
template is extracted through solvent extraction or calcination (thermal treatment), a
hexagonally ordered silica framework remains.
Figure 7. Production of MCM-41
Compared to other mesoporous silicas, MCM-41 has a large surface area (941 m2/g) and large
pore volume (0.73 mL/g). Up to 80% of its volume can be pore space. The walls of its hexagonal
structure is primarily composed of amorphous silica (Si02), making it less mechanically stable
than other more crystalline zeolites.
1.5.2 Biomedical applications of MSNs
Drug Delivery Vehicle: Since 2001, MSNs have been gaining momentum as a platform for drug
delivery systems. Originally designed as molecular sieves, these particles have several
advantageous characteristics, including:
Large surface area for functionalization and interaction with the environment
Large pore volume for drug loading
Highly ordered pore network for control of loading and release kinetics
11


Silanol-containing surface for easy functionalization
Low immunogenicity, with an ability to naturally be incorporated into the cell by
endocytosis
MSNs as a drug delivery system has been applied to several indications to act as a low release or
site targeting drug reservoir. Recently, MSNs have been applied as an implantable drug delivery
system for bone tissue, thanks to its favorable properties as a bioceramic1211. MSNs have also
been used as a form of biosensors by loading pore space with bioactive dyes, which only
fluoresce under the presence of specific stimuli1221. The functionalization of the pore interiors
allow hydrophobic drugs with low solubility to be delivered to target sites more easily. To avoid
premature release of the loaded drug, methods for capping pores has become a research area
with growing interest. Capping using soft material (antibodies, insulin, pH/temperature sensitive
polymers) or hard materials (gold, cadmium sulfide, iron oxide) provides tight control the drug
delivery package. Cap release can be achieved by both external stimuli (magnetic fields) as well
as internal stimuli (glucose, pH, temperature).
In a previous study, Liong et al. had previously functionalized the surface of MSN
particles by conjugating the surface with folic acid1231. Alpha-folate receptors have been shown
to be upregulated in certain human cancers. Liong was able to show the preferential absorption
of the MSN into the cancer cell lines PANC-1 and BxPC3. Furthermore, the MSNs incorporated
into the cancerous cells were able to release their camptothecin drug payload into the cytosol,
effectively reducing cell survival by 60%. Evidently, MSNs are an efficient platform for cancer
drug delivery systems.
12


2.
Aims and experimental approach
2.1 Specific Aims
2.1.1 Demonstrate the contrast potential of mesoporous silica nanoparticles
Silica as a contrast medium offers several advantages over microbubbles. Above all,
silica offers a multifunctional platform that is easier to modify than its microbubble counterpart.
In order to apply silica nanoparticles as a contrast agent, the particles must demonstrate the
ability to significantly alter the contrast of a region of interest via conventional ultrasound.
Experimental approach: In order to demonstrate the physical effect of ultrasonic waves on
MSNs, the system was initially modeled as a colloidal solid particle distributed throughout a
homogenous phase acoustically similar to soft tissue. The interaction of ultrasonic pressure
waves throughout the system could then be characterized using an Ingaard-Morse model for
acoustic scatterers. By using the Ingaard-Morse model to define thermodynamic and
hydrodynamic interaction between the particle and phase under ultrasonic pressure waves, a
sufficient prediction could be made for the contrast effects during imaging. These predictions
were confirmed by B-mode ultrasound image analysis of agar or acrylamide blocks with varying
concentrations and diameters of silica nanoparticles.
2.1.2 Apply MSN as an ultrasound contrast agent through cancer cell targeting
A major reason ultrasound is not more widely used for breast cancer diagnostics is its
inability to discern between cancerous and non-cancerous masses of the same density. If the
contrast of a cancerous mass can be enhanced in real time, ultrasound will be a far more
effective means of cancer diagnostics. In order to use MSNs as a contrast agent for breast
cancer, the particles must be selective as to where they accumulate. The particle's functionality
is primarily achieved by three mechanisms:
(1) Size: by having a diameter less than 1pm, MSNs are able to pass through the fenestrated
endothelial walls of tumor vasculature.
(2) HER2 receptor targeting: prior research has shown that certain types of breast cancer
overexpress the HER2 surface receptor by 100 fold1241. This overexpression makes the
HER2 surface receptor an ideal targeting antigen for cancerous cells. Herceptin, an anti-
HER2 antibody, was conjugated to the surface of MSNs to ensure accumulation near
tumor regions.
13


(3) Porosity: by maintaining a porous interior structure, variations in density and
compressibility will enhance the reflectivity of incident soundwaves, further increasing
echo contrast during imaging.
By adjusting the size and surface of the MSN, the local volume fraction of MSNs in and around
the tumorous tissue will increase and only the targeted regions of interest will experience an
enhancement in contrast.
Experimental approach: To ensure that both size and surface modification were optimized for
accumulation in vivo, particle size distribution (PSD) and cellular interaction was first
characterized. PSD was analyzed by scanning electron microscope (SEM) and tunneling electron
microscope (TEM) imaging. Surface modification steps were monitored by Fourier transform
infrared spectrometry (FTIR), SEM, and ultraviolet-visible spectroscopy (UV-VIS). To observe
using fluorescence microscopy and fluorescence-activated cell sorting (FACS), the MSN-
Flerceptin conjugates were labeled with fluorescein isothiocyanate (FITC). The FITC-labeled
MSN-Flerceptin was introduced to four different cell lines of increasing expression of the HER2
receptor.
2.2 Experimental Obstacles
2.2.1 Particle selection
In order to be an effective targeting agent for passage through fenestrated tumor
vasculature and target cell attachment, MSNs must fall within a specific diameter range. After
extensive characterization, it is clear that the commercially available MCM-41 yielded a wide
particle size distribution (PSD) with a mean diameter far greater than 8pm. Since MSNs
essentially comprise of stacked hexagonal pore sheets, the mean diameter could be modestly
reduced by sonication, but could not disintegrate fully intact sheets. Thus, the bulk material
must be sieved to collect the particles population with desired diameter.
2.2.2 Sieving
Due to its size, surface charge, and density, MSNs are incapable of dissolving in solution.
Furthermore, these colloidal particles do not remain uniformly suspended in solution longer
than an order of minutes. In general, the issue of insolubility was accommodated by sonication
and use of ethanol as a solvent (more hydrogen bonding with silanol or amine groups).
Flowever, colloid flocculation posed a consistent problem in terms of filtration. Once it was
14


determined that the initially proposed Sephacryl column could not accommodate the larger
silica particles, a simpler paper-based 1pm filter was applied. Unfortunately, less than 1% yield
was obtained after filtration. One suggested explanation for this loss of product was channel
clogging caused by larger diameter particle. Finally, a 1-2 layer woven nylon grid mesh with 1 or
8pm pore diameters was applied. Since pore channels were shorter (75pm) and less circuitous,
the upstream flocculation layer could be more easily disrupted by retrograde flow or sonication.
Regardless, the filtered yield rarely exceeded 40% of the initial mass.
15


3.
Materials and methods
3.1 Materials
3.1.1 Consumables
MCM-41 type (hexagonal) MSNs were purchased through Sigma-Aldrich (Milwuakee,
Wl) and filtered through a 1pm nylon mesh (Elko Filtration lncv Miami, FI). IN hydrochloric acid,
ethanol (200proof), anhydrous toluene, IX Phosphate buffered saline (PBS), (3-
aminopropyl)triethoxysilane (APTES), dimethylsulfoxide (DMSO), and fluorescein isothiocyanate
isomer I (FITC) for MSN functionalization were provided by Sigma-Aldrich (Milwuakee, Wl).
Trastuzumab (Flerceptin) was provided by Dr. Bolin Liu. For ultrasound phantoms, acrylamide,
N,N'-methylene bis(acrylamide), ammonium persulfate (AmPs), and
Tetramethylethylenediamine (TEMED) was used for solid tumor mass mimics.
For in vitro studies, two groups of increasing HER2 expression were used: (1) MDA-MB-
231(TN) has a very low level HER2 expression, and (2) BT474 or SKBR3 yields a moderate level of
HER2 expression. All cell lines were provided courtesy of Dr. Bolin Liu's group (University of
Colorado Department of Pathology. The MDA-MB-231(TN) cell line was cultured in RPMI-1640
with 10% fetal bovine serum (FBS) and 1% Penicillin-Streptomycin (PS). The cancer cell lines
(BT474,SKBR3, and HR20) were cultured in Dulbecco's Modified Eagle Medium (DMEM)/F12
with 10% FBS and 1% PS.
For transmission eletron microscopy (TEM), samples were stained with 2% uranyl
acetate and placed on 400 mesh formvar-coated, carbon evaporated, and glow discharged
copper grids, courtesy of the University of Colorado -Denver Core facilities.
3.1.2 Equipment
To disrupt aggregation prior to use, all MSNs were sonicated in an ultrasonic waterbath
cleaner operating at 35kHz for 10 minutes. All MSN particles were dried in a 100C oven.
For characterization with TEM, samples were imaged on a Tecnai G2 series scope (FEI,
Hillsboro, OR). Low vacuum scanning electron microscopy (SEM) was performed on a JSM-
6480LV W-thermal emission microscope (JEOL Ltd. , Peabody, MA), at the University of
Colorado-Boulder Nanomaterials Characterization Facilities (NCF).
Single-pulse ultrasound measurements were obtained using Non-Destructive Testing
(NDT) transducers with matching frequencies ranging from 0.5-7.5 MHz (GE-Panametric). All
16


transducers were Accuscan type S immersion transducers with point target focus (PTF) and 1"
focus length under water. Two transducers were set facing each other through the round
window of a 12" X 8" X 3" acrylic container (Figure 8). The box was filled with agar gel at the
bottom (or an agar cylinder was placed inside) and filled with deionized water. The water was
retained within the container by covering the window with double layered masking tape.
Curvilinear array transducer probe
Figure 5. Diagram of acoustic measurement apparatus
To ensure clean connectivity, ultrasound gel was applied between the single element
transducers and the masking tape-covered window. The precise distance between transducers
was measured prior to each set of experiments. Each transducer was designated as a pulser or
receiver and attached to a Panametric NDT Model 5800 pulser/receiver. Receiver signal was
acquired on an Infiniium 8000 High Performance Oscilloscope (Agilent Technologies).
Waveform data was sent to MATLAB (The Mathworks, Inc. Natick, MA) for data processing.
B-Mode imaging was acquired on a Sonix SP clinical ultrasound monitor attached to a
L14-5/38 curvilinear probe (Ultrasonix, BC, Canada). All images were taken inside of the acryic
containterr. Images were processed in ImageJ (NIH, Bethesda, MD) and MATLAB (The
Mathworks, Inc, Natick, MA).
Fluorescence-activated cell sorting (FACS) was performed on a FACScan (BD, Franklin
Lakes, NJ) and processed through FlowJO
17


3.2 Methods
3.2.1 Characterization of mesoporous silica nanoparticles
MSN preparation: Prefabricated MCM-41 hexagonal type MSNs were dispersed in 200 proof
ethanol at a concentration of 5 mg mL"1 and bath sonicated for 10 minutes. The dispersion was
then filtered through either a 1,8,10 and 30 pm nylon mesh from a 10 mL syringe.
Retrograde/prograde flow alternation in a 1 sec : 3 sec timing pattern was used to prevent
flocculation on the upstream side of the filter. The filtered particles were dried in a 100C oven
for 48 hours.
TEM imaging: Dried MSN was resuspended in 200 proof ethanol to 500 pg mL"1 and stained in
2% uranyl acetate for 3 minutes. After drying on a 400 mesh formvar coated copper grid, the
samples were imaged at 80kV.
SEM imaging: Dried MSN was resuspended in 200 proof ethanol to 500 pg mL"1. 20pL drops
were placed onto glass slides, allowing for the ethanol to evaporate off. The glass slides were
then sputter coated in a gold/platinum alloy to a 3 nm thickness. Images were acquired at 5.0
kV.
3.2.2 Determination of acoustics properties
Single pulse scan: To quantify the attenuation caused by MSNs, signal loss through an
acrylamideblock was measured. This method was also used to calibrate the acoustic properties
of the acrylamide blocks to better represent soft tissue. Using the apparatus previously
described (Section 3.1.2, "acoustic measurement apparatus"), acrylamide blocks with
incrementally increasing concentrations of 1 pm MSNs (0, 0.1, 0.25, 0.5, 1, 2, 5, and 10 mg mL1)
were placed into the acoustic measurement apparatus. The system apparatus was left
untouched for 5 minutes to allow for all air bubbles and potential contaminants to settle. 25 pJ
pulses were transmitted through the MSN/water samples at 5 kHz. The waveform was averaged
and transferred to a MATLAB program for analysis (Appendix A). To calculate the attenuation, a
fast fourier transform was performed and deconvolved with a reference signal of pure water, as
described by Laugier et at.[25]. The complex spectrum of a signal can be described by the
following equation:
A(f) = A0(ne~2infO
(5)
18


Where A0(f) is the systems response, / is the frequency, L is the distance between the two
transducers, and v0 is the speed of sound through the medium. The frequency-dependent
attenuation of the MSN-water dispersions can be derived from the ratio of its magnitude
spectra to the magnitude spectra from a reference water signal's complex signal, Ar :
a(/)L = Ln
\Mf)\
W)\
(6)
B-mode imaging: To calculate the correlation between image contrast and particle
concentration, 25% polyacrylamide phantoms were created with incrementally increasing
concentrations of 1 pm MSNs at 0, 0.1, 0.25, 0.5, 1, 2, 5, and 10 mg mL"1. Acrylamide was
selected due to its similarity to soft tissue's acoustic properties [26]. The acrylamide phantoms
were created by adding MSN to 25% acrylamide/5% N,N'-methylene bis(acrylamide). Prior to
polymerization, the solution was sonicated for 3 minutes using a probe sonicator set to power
level 3. After adding 0.05% AmPs and 0.025% TEMED, 1 mL was immediately poured into a 5 mL
syringe with the nozzle removed. Complete polymerization occurred between 2-5 minutes. The
polyacrylamide phantoms were submerged under water and placed on a 5% agar platform
approximately 5 cm in height to separate the ultrasound probe's far-field from the highly
reflective bottom of the acoustic measurement apparatus. The phantoms were imaged with a
10 MHz probe at 70dB. Two focal points were set at the top and bottom of the phantom, with
the top of the phantom 2.5 cm from the top of the phantom. 3 images were taken for each
phantom and transferred to a MATLAB program for analysis (Appendix B). The interior of the
phantom was parsed into 50 pixel x 50 pixel squares and measured for mean pixel intensity.
Each square, along with the squares from the other two images of the same concentration, was
then averaged. Each mean gray pixel intensity value was correlated to the MSN concentration.
3.2.3 Herceptin Conjugation to silica nanoparticles
MSN functionalization: Prefabricated MCM-41 hexagonal type MSNs were prepared as
previously described (Section 3.2.1, "MSN preparation"). The 1pm diameter particles were
resuspended in 10% HCI to a concentration of 31.4 mg mL"1, in order to increase the amount of
free silanol groups on the surface (Figure 9, step 1). After stirring for 1 hour at room
temperature, the dispersion was centrifuged at 5,000 rpm for 5 minutes, decanted, and washed
19


in deionized water (d/H20). This centrifugation and wash step was repeated two more times.
After the third decantation, the particles were resuspended in d/H20 and sonicated for 10
minutes to reduce aggregation. The particles were dried in a 100C oven for 48 hours. Silanol
group concentration was predicted by FTIR and pH curve.
For amine functionalization, the hydroxylated MSNs were suspended in anhydrous
toluene to a concentration of 20.7 mg mL"1. APTES was added to bring the final concentration to
6.5% (Figure 9, step 2). The dispersion was placed in a nitrogen environment by injecting a lmL
syringe with an 18G needle through the stopper on a round-bottom flask. The plunger of the
syringe was replaced with a double-layered nitrogen-filled balloon (~10 cm diameter). The lip of
the balloon was secured with parafilm. The dispersion was centrifuged at 5,000 rpm for 5
minutes, decanted, and washed in anhydrous toluene. To purify the particles from unbound
APTES, the dispersion was added dropwise into ether, and again centrifuged 5,000 rpm for 5
minutes. This centrifugation and wash step was repeated two more times. After the third
decantation, the particles were resuspended in 200 proof ethanol and sonicated for 10 minutes
to reduce aggregation. The particles were dried in a 100C oven for 48 hours.
Herceptin/FITC conjugation: To couple the Herceptin antibody with MSNs, amine groups on the
silica surface were coupled to carboxylates from exposed aspartic and glutamic acid residues
and carboxyl termini on the Herceptin antibody (Figure 9, step 3). After preparing a solution of
0.5mg/mL Herceptin in IX PBS, N-hydroxyl succinimide (NHS) and l-ethyl-3-(3-
dimethylaminopropyl) carbodiimide) (EDC) was added as a crosslinking agent (2mM final
concentration for each). Amine-coupled MSNs were resuspended in the solution (4 mg/mL) and
sonicated for 10 minutes to ensure optimal exposure to surface area. The EDC crosslinking
reacted for 4 hours, stirring at room temperature. The product was then centrifuged for 10
minutes at 3,000 rpm and decanted to remove the water-soluble carbodiimide and urea
byproduct. The product was resuspended in IX PBS. The centrifugation, decantation, and
resuspension step was repeated three times.
20


MSN modification
Hydrochloric acid
APTF.S
(3-aminopropyl)trie+hoxvsilane
Herceptin modification
Figure 6. Chemical reaction of MSN-antibody linkage
21


Conjugation of modified MSN and Herceptin
Figure 7, continued. Chemical reaction of MSN-antibody linkage
22


After the third resuspension to 5 mg mL"1 in PBS, protein concentration was determined
by Bradford Assay. Once the protein concentration was calculated, 10 mg mL"1 FITC in DMSO
was introduced to a ratio of 80 pg FITC: lmg Herceptin. 1 hour of reaction time stirring at room
temperature in the dark allowed for FITC conjugation to uncoupled amines and sulfhydryl
groups. The solution was dialyzed in 3,500 kDa molecular weight cutoff (MWCO) dialysis tubing
in milliQ water. The water was changed 5 times over a period of 72 hours. Dialysis occurred at
4Cto preserve protein integrity. FITC conjugation was confirmed by fluorescence microscopy
and UV/VIS. FITC conjugation efficiency was determined by the following calculation:
[IgG] =
4(495)
[4(280) 0.31 4(495)]
(7)
Where A(495) corresponds to excitation/emission wavelength of FITC, and A(280) corresponds
to the excitation/emission wavelength of aromatic ring-containing amino acids.
3.2.5 In vitro studies
Fluorescence microscopy: To qualitatively determine binding of the MSN-Herceptin conjugate to
HER2+ cells, three cell lines were incubated and prepared as previously described (Section
3.1.1). The cell lines (MDA-MB-231(TN), BT474, SKBR3, and HR20) were grown to 95%
confluency in 96-well plates. Each cell type was incubated with the MSN-Herceptin-FITC
conjugate in DMEM at varying concentrations (50,100,200,400pg/well) and times (30 minutes, 1
hr, 2hr). At the respective time point, all cells were lifted from the well plate by trypsinization.
The cell suspension was filtered through a 8pm nylon mesh filter. This would allow for the free
MSN to pass, but trap MSN-attached cells on the upstream side of the filter. Similar to MSN
release, accumulated cells were detached from the filter by alternating retrograde/prograde
flow.
Flow cytometry: Preferential cellular uptake of MSN-Herceptin particles were quantified using
flow cytometry. The cell lines were prepared in the same manner used for fluorescence
microscopy. Data were processed by FloJO.
Ultrasonography: To quantitatively show that cellular uptake of MSNs can augment the inherent
contrast in B-Mode ultrasound images, a 2D model was applied. After incubating cells in MSN to
allow for uptake as previously described for fluorescence microscopy, the recovered cell
23


suspension was spread across a 10% agar gel. A 5cm thickness ensured that the focal region of
the ultrasound beam would not detect reverberations from the plate bottom. Images were
acquired as previously described (Section 3.2.2, "B-mode imaging")
24


4.
Results
4.1 Mathematical Justification for MSNs as a contrast agent
4.1.1 Contrast by backscatter
By using the Morse-lngaard model for general ultrasound scatterers, s, through a medium, m,
a= scattering cross section
N= concentration of scatterers per unit volume
k= wave number (equal to 2n/A of the incident soundwave)
r= particle diameter
K adiabatic compressibility (inverse of the bulk modulus, B)
p = density
This model had previously been used to show that gasses exhibit a backscattering cross section
around 1014, whereas solid particles are closer to 2[271. The significant disparity in backscatter
efficiency of the particle is caused by the exponential relationship to scatterer cross section (a)
and its compressibility and density terms (bracketed values). Generally it is assumed that
Km Kg and pm pg. The Morse model also denotes that backscattering cross section is
linearly related to the concentration of scatterers per unit volume. Figure 11 provides an
illustrative comparison of acoustic properties of materials relevant to ultrasound imaging.
Using soft tissue as the exemplary medium, it is clear that the density and acoustic wave
velocity of gasses are much less than soft tissue.
several comparisons between solid and gas particles can be made[RI:
21
25


Figure 8. Comparison of Acoustic Properties
K)
CTn
Intensity reflection coefficient (dimensionless)
o o o o o h-> 1-^
o k> b oo o k>
o o o o o o o
o o o o o o o
soft tissue
muscle
water
fat
skull bone
bulk fused silica
lung
PZT
octofluorpropane
sulfur hexafluoride
air
N*s/mA3
o o o O o o O o o
o o o O o o O o o
m m m m m m m m m
+ + + + + + + + +
o O O O O O O O O
o h-> NJ OJ 1^ on O') 'vj 00
air
sulfur hexafluoride
octofluorpropane
lung
fat
water
soft tissue
muscle
skull bone
bulk fused silica
PZT
3
r+
(D
3
in
<
=
in
O
C
(D
kg/mA3)
h-> NJ OJ 1^ on O') 'vj 00
O O O o O o o o
O O o o O o o o
o O o o O o o o
m/s
onoonoonoonoono
oooooooooo
ooooooooooo
>
o
o
c
(/)
r+
o
3
73
(D
a.
3
n
(D
air
sulfur hexafluoride
octofluorpropane
lung
fat
water
soft tissue
muscle
PZT
skull bone
bulk fused silica
O
(D
3
r+
S'
(/)
Q)
£
o
n
<


If the adiabatic compressibility of a seatterer is not taken into account (such as in equation
2), the disparity in contrast efficiency between gas and solid nanoparticles is now within
one order of magnitude. From the figure above, it is clear that silica is denser than soft tissue
and has a larger acoustic velocity. Since the differences in density and compressibility are
quadratically related to contrast (equation 2), only the absolute difference is critical. For this
reason solid particles create contrast despite an inverse relationship to soft tissue, compared to
gases. To reiterate, the difference in acoustic impedance between medium and seatterer defines
the degree of contrast intensity.
4.2 Characterization of mesoporous silica nanoparticles
TEM imaging (Figure 11) confirmed presence of the hexagonal array typical of MCM-41
type MSNs. Successful TEM imaging proved more difficult than previously anticipated, due to
the flocculating nature of the particles. Figure 11.A reveals the irregular form that has been seen
as typical for commercially available MCM-41. Due to the nature of TEM imaging,
Figure 9. TEM images of MCM-41 mesoporous silica. (A) 100 nm scale of large cluster in aqueous medium.
(B) 20nm scale flake depicting honeycomb-like cluster
particles were placed on the imaging stage while still in solution. This allowed for cluster
formation, as seen in figure ll.A. Prior studies have shown that the degree of aggregation is
directly dependent upon concentration [Z81. For MCM-41 type MSNS with 300 nm diameters, the
critical concentration for an aggregation event to occur was near 500ug/ml.
27


Figure 10. SEM imaging of MSNs. (A) 60,000X zoom, (B) 1,000 zoom
The irregular macrostructure of the MSNs were further illustrated by SEM (Figure 13).
The large macroscale pockets in conjunction with hexagonal nanoscale pores (2.1-2.7nm in
diameter) contributes to a surface area ratio of 1000m2/g.
During surface modification steps prior to Herceptin conjugation, FTIR was used to
monitor the surface groups (Figure 13). Native MSN, hydroxylated MSN (MSN-OFI), and amine-
functionalized MSN (MSN-NFI2) all exhibited the characteristic peaks of aminosilanes, including
H-O-H bending at 1639 cm"1 from residual water. The hydroxylated MSN displayed an increase in
the broad O-FI stretch at 3200-3400 cm"1, as expected. Surprisingly, MSN-OFI displayed more Si-
O-Si bonding than MSN-NFI2. This could be due to shielding effect caused by the additional
ethoxysilane group s seen in MSN-NFI2. As expected, MSN-NFI2 exhibited peaks at 4000 and
1570 cm"1. Any changes in C-H bonding (from APTES) near 3000 cm"1 would be masked by the
considerably larger residual peaks from the instrument cartridge.
28


(acquisition artifact)
ShO-U Stretch
1100 env*
NHjberv)
1S70 cm*
A
4000 3500 3000 2500 2000 1500 1000
Wavenumber (cm-1)
Figure 11. FTIR analysis
4.3 Nanoparticle concentration to contrast intensity correlation
In order to determine the image contrast due to MSN, acrylamide plugs were
polymerized with varying levels of MSN concentration and imaged using conventional clinical
ultrasound (Figure 14, 15). Image contrast never approached a mean pixel intensity (MPI) of 0
since anomalous reflections were detected from imperfections in the acrylamide plug. The
correlation of contrast to MSN concentration is linearly related, as expected from the Morse-
Ingaard model (Equation (8)): The increase of scatterers, N, is directly proportional to the
scattering cross section, o. The Morse-lngaard model also predicts scattering cross section to
have a fourth order dependence on the transducer probe frequency (ie. the wavenumber, k, in
units of Hertz). As frequency increases, the backscatter cross section increases exponentially.
For this set of experiments, imaging was limited to a probe with a central frequency of 7.5 MHz
with a maximum limit of 10MHz.
29



O mg/mL
1 mg/mL
t
0.5 mg/mL 5 mg/mL
Figure 12. Comparison of acrylamide plug with MCM-41 silica particles (left) and without (right). Image was acquired
with a 10 MHz probe at a depth of 2.5 cm.
Exploring higher frequencies was limited by availability of probe types. In a clinical setting,
ultrasound probes may operate at a range between 2-18 MHz.
In the traditional sense, axial resolution (or longitudinal resolution) is defined as the
detectable distance between two objects on a line parallel to the direction of wave propagation.
Axial resolution can be calculated as half of the spatial pulse length, SPL. SPL is defined as the
wavelength, X, multiplied by the number of cycles, 6, emitted per pulse. Wavelength can be
further decomposed as a product of acoustic velocity, c, and the frequency of the incident sound
wave, /. Using the experimental setup for determining the correlation of MSN concentration to
MPI (/= 7.5 MHz, c = 1540 m/s, 6 = 1), the theoretical axial resolution can be calculated:
R
axial
SPL
~2~
X-S
~T~
75pm =
1540ms 1
7.5 lO^-1'1
2
(9)
30


Figure 13. Linear interpolation of contrast with respect to MSN concentration
Clearly the contrast apparent in Figure 14 is significantly smaller than the theoretical axial
resolution, with a diameter closer to l/10th of the projected resolution. Considering the drastic
disparity between these two values, the apparent contrast is not caused by creating an image of
the individual particle boundary, but rather increasing the inhomogeneity of regions in the
microscale. In essence, MSN particles increase the "speckle" noise within a medium. Although
contrast does not reflect the underlying structure of these sub-resolution scatterers, it does
reflect their echogenicity1291. As described previously (Section 4.1.1), the inherent density and
acoustic velocity of silica material in soft tissue yields a substantial degree of reflectivity.
Convention may suggest that increasing speckle noise can decrease the visibility of a
boundary region, as the case with certain hypoechoic lesions. The contrast caused by a sub-
resolution scatterer is calculated by statistical probability. Each scatterer creates a random
phasor along a complex plane (Figure 16). A phasor is a representation of a sinusoidal function.
In this case, the function is the response to an incident sound wave. The contrast of a region can
be defined as the vector sum, r, of each individual scatterer (Figure 16, left). In uniform medium,
when each scatterer population, r, is compiled with scatterer
31


Imaginary Imaginary
Figure 14. Probability distribution function of a sub-resolution scatterer
populations in the same region, the amplitude of contrast within the region equates to a 2D
Gaussian distribution (Figure 16, right). Reverting back to optics theory, this can also be defined
as a Rayleigh probability distribution function (PSD). Since all scatterer population cancel each
other out, a zero-mean phasor results, and any contrast is observed as noise. However, by
targeting the silica particles to specific regions of congregation through flocculation in
phantoms or Herceptin receptor targeting in vitro the periodic arrangement creates a
coherence component that is added to the speckle noise statistics. By adding a coherence
component to a sub-resolution scatterer, a strong phasor is added to the system. This results in
a shift of the PSD either to the right or left along the real axis. Since scatterer arrangement at
the lesion boundary (or flocculation cluster) would not be random, contrast can be detected
despite having a sub-resolution diameter. By having an "ordered arrangement of speckle",
contrast of tumors at the boundary layer can be accentuated.
4.4 Cellular interaction of nanoparticles
4.4.1 Fluorescence microscopy
A preliminary view of HER2-positive and HER2-negative cell lines reveals a modest yet
detectable discrimination between the two cell types (Figure 14). While the MSN-Herceptin
nanoparticles are present in the filtered HER- culture (14.A), most of the particles are either
unbound or attached to the outer wall of the cell membrane. Conversely, most HER2+ cells
(14.B) show attachment of the nanoparticles to the cell. The data does not confirm the precise
location (internal or external) of particle attachment to the cell. Both cell populations were
32


sieved after particle incubation using a 10pm nylon mesh filter, theoretically eliminating all
unbound particles whose diameter is less than 10pm.
A B
100 um
Figure 15 Comparison of cells treated with MSN-Herceptin-FITC (A) HER- cells: 231, (B) HER+ cells: skbr3
4.4.2 Flow cytometry
A FACSCalibur flow cytometer was used to determine the binding of the MSN-Herceptin-
FITC nanoparticle to HER+ cells, as well as determine the cell viability after incubation. Since
compensation was not achievable, the two experiments were not conducted simultaneously.
The binding affinity of the nanoparticles were characterized by the relative absorbance
Observed event population Fluorescence of gated events
Figure 16. Sample gating of cells for flow cytometry. Shown: 231 treated with MSN-Flerceptin
33


near FITC's maximum emission wavelength of 520 nm, using a 530/30 bandpass filter.
The FITC-conjugated molecules were excited using a 488nm laser, and polygon-gated to capture
the event population with a larger forward and side scatter (Figure 15). This gating only captures
approxiately 40% of the event population as a precautionary measure to exclude debris,
unbound residual particles, or dead cells (for the binding assay experiment only).
Following gating, FITC fluorescence was measured for each cell type (Figure 15, right).
The control skbr3 and 231 yielded nearly identical profiles, as expected (231 group not shown).
The MSN-treated 231 cell population (HER2 positive) expressed a detectable degree of
fluorescence from the FITC particles; any fluorescence should indicate any HER2 receptor
expression on the surface. The MDA-MB-231(TN) cell line has been characterized as either a
basal-like and triple negative breast cancer (TNBC) line in literature. Groups such as B. Lehmann
et al. define these TNBC lines as "lacking estrogen receptor (ER) and progesterone receptor (PR)
expression as well as human epidermal growth factor receptor 2 (HER2) amplification". A
normal, non-amplified level of HER2 expression may account for the apparent binding of MSN-
Herceptin-FITC particles to this cell type. Alternatively, indiscriminate binding may be caused by
the silica particle itself, as opposed to the Herceptin antibody. Trewyn et al. have shown than
nude MSNs are naturally incorporated by certain cell types. The accelerated metabolism of
cancer cells increase their cellular uptake, consequentially reducing their partiality for nutrients.
In a qualitative sense, the rate of cell division far exceeds both skbr3 and skbr3pool.
Figure 17. Dead cells ratio comparison using ethidium homodimer 1
34


Skbr3, which indeed overexpresses HER2, shows a strong binding affinity with the nanoparticles.
From an arbitrary cutoff near 102 fluorescence, 96.2% of the gated events from the skbr3 cell
line exceed that cutoff. Alternatively, the 231 cell line exhibits only 15.1% of the gated events
pass the cutoff.
Since silica yields a wide absorption spectra, UV-VIS spectroscopy was skewed by the
concentration of MSNs. For this reason an MTT assay to determine cell proliferation was not
possible (Appendix D). Alternatively, ethidium homodimer-1 (EthD-1) was used to determine the
ratio of dead cells to total cell count. By using a 488nm laser and a 650nm long pass filter, the
tail of the EthD-1 emission profile could be observed without nearly any overlap from FITC's
emission spectra (Appendix E). EthD-1 interference for the binding affinity experiment was not
an issue since those cells were not treated with EthD-1.
A comparison of the fluorescence of skbr3 cells with and without MSN treatment
suggest that the MSN treatment reduces overall cell proliferation. A shift in fluorescence caused
by EthD-1 uptake indicates that the Flerceptin targeting arm also maintains a certain degree of
its therapeutic capability. Flowever, further testing is required to confirm this claim.
4.4.3 in vitro ultrasound imaging
After incubation in MSN-Flerceptin, cells were lifted of the well plate and aliquot onto
agar inside of a 5 ml syringe. After allowing 2 hours for the cells to settle to the agar floor, most
of the media was pipetted off. A second layer of agar was poured over to the remaining cell
suspension (Figure 17). A cross section of the sandwiched cell culture was imaged on a clinical
ultrasound machine at lOMFIz. To quantify the peak pixel intensity and thickness of contrast, the
pixel intensity profile along the y dimension was graphed (yellow line, Figure 17). Five profiles
with variable x coordinated were averaged to produce the mean pixel intensity profile (Figure
17, bottom).
35


(A) untreated
(B)treated
Figure 18. Ultrasound contrast profile of skbr3 cells (A) treated with MSN-Herc and (B) untreated. The
yellow line indicates an example profile cross section.
The presence of MSN-Herceptin results in a spike in contrast by approximately 640%, with a
boundary thickness close to 1mm. The total cell count in the suspension is estimated to be
around 200,000 cells spread across an area of 122mm2.
A comparison of the two cell lines 231 and skbr3 suggests that there is minimal and
insignificant discrepancy in contrast. This observation is partially in conflict with the binding
affinity assay from flow cytometry, which suggests that there is a significant difference between
the binding affinities of MSN-Herceptin to cells that do and do not overexpress the HER2
receptor. This observation is also supported indirectly by UV-VIS spectrometry (Appendix D). The
binding affinity assay does indicate, however, that there is a certain degree of binding to all cell
types. Figure 19 suggests that this indiscriminate binding is enough to create contrast nearly
equivalent to the HER2+ cell line. Possible explanations for this occurrence is provided in the
previous section.
36


250
200
Si 150
a>
x
Q.
ns
ai
skbr3 treated
231 treated
100
50
100
Vertical Length
200
Figure 19. Contrast profile cross sections of HER2+ and HER2- cells treated with MSN-
Herceptin
37


5.
Discussion
5.1 Contrast Efficacy
The potential for silica nanoparticles to act as an effective contrast agent relies on the
delivery and retention method of the particles. It has been well proven that gas phase contrast
agents are ideal for contrast enhancement in dynamic systems (such as cardiovascular flow).
Although silica nanoparticles do provide some degree of contrast in dynamic systems, it is most
effective in static systems. By operating in a static system, the silica nanoparticles can form what
may be termed as a flocculation layer. The formation of this deposition layer is requisite for both
its drug delivery functionality as well as its contrast agent functionality. Due to the relatively
small diameter of the MSN particle with respect to the ultrasonic wavelength, Rayleigh-type
backscatter is diminished. By having substantially high concentrations in at tumor sites,
ultrasound contrast can be augmented.
Point scatterer concentration may fluctuage greatly in dynamic systems, where
concentration is constantly modified by directional flow. Due to a heavy influence from
dispersive flow movement, gasses-filled microbubbles typically only display a half-life in the
region of interest up to a few minutes. This time is further reduced by the rupturing potential of
ultrasound waves. One allure of antibody-conjugated silica contrast agent is that it is extremely
stable in comparison to liposome-enclosed air bubbles, thus eliminating issues of half-life. By
chemically attaching the silica particles to the region of interest the concentration is further
stabilized.
Due to the high porosity of mesoporous silica, air bubbles may be trapped within the
core of the nanoparticle. Part of the contrast effect may be a product of reflectivity
enhancement that occurs by backscatter from these air bubbles trapped in the pores. With a
pore volume ranging from 0.98 cm3/g (MCM-41) up to 2.31 cm3/g (MSU-F), it is possible for air
pockets to comprise more than 33% of the bulk volume (MCM-41). However, the air retention
with the particles was not observed in this study.
Flow cytometry and in vitro ultrasound indicate that the preferential binding of MSN-
Herceptin particles to the different cell types is significant, but not optimal for cell-specific
ultrasound contrast enhancement. One proposed theory is the indiscriminate uptake of silica
particles by numerous cell types. By increasing the count of Herceptin targeting arms on the
surface, more specific binding may be achieved. Since each Herceptin molecule is conjugated to
38


the amine group of the APTES precursor at a 1:1 ratio, the density of Herceptin targeting arms
on the silica surface is limited by the geometric area occupied by the APTES base. Assuming that
the triethoxysilane base is 3.96 x 104 pm2 and the surface area to mass ratio of MCM-41 silica is
1000 m2/g (provided by Sigma), the limit of Herceptin bound to the silica surface is 0.041 moles
per gram of silica.
5.2 Potential Applications
Currently, ultrasound is generally used for two purposes in the context of breast cancer
detection and treatment: secondary diagnosis and presurgical occult lesion localization (Section
1.1, 'Review of breast cancer ultrasound diagnostics'). In both cases, the efficacy of the
procedure is heavily reliant on its ability to discern tumor boundary regions.
In a clinical setting, ultrasound has typically been used as a complementary diagnostic
device with mammography. The greatest disadvantages of mammography is that it is a
radiation-emitting modality and that it is less effective than ultrasound for density
determination. However, even ultrasound is unable to distinguish between benign and
malignant tumors of similar densities. Ultrasound specificity is further diminished if the
surrounding tissue is similar in density, which is common for women below 50 years old. By
injecting MSN+Herceptin particles in tissues regions suspected of cancerous lesions, contrast
enhancement will only occur in cancerous tumor regions. By being able to visualize selective
contrast variations in real time, sonographic images can more effectively distinguish malignant
tumors from benign cysts.
Ultrasound is also commonly used for needle guidance, both for biopsy sampling and
occult lesion localization. If the boundary region of a suspected tumor is difficult to discern, the
result may be poor sampling localization or poor marker placement. By improving visualization
of the boundary regions, biopsy needle placement can be more accurate, or rendered
unnecessary altogether. Since wire length for occult lesion localization is determined by the
length of the lesion, improved visualization of the boundary regions can ensure that the wire
completely passes through the entirety of the lesion.
39


Table 2. Advantages and disadvantages of silica nanoparticles as a contrast agent in comparison to gas phase
contrast agents
Advantages___________________________
Better suited for extravasation
out of leaky tumor capillaries
More stable, allowing longer
half-life in body to take multiple
images
Highly modifiable surface
Potential for reflectivity
enhancement
Large pores allow for delivery
of low solubility drugs
US: portable and cheaper than
Mammography
FDA approved
______Permanent retention within cell
_______possible [depends on cell type]
Disadvantages_______________________
Lower scattering cross section
than microbubbles
Size approaches axial resolution
of conventional ultrasound
transducer frequencies
40


6.
Conclusion
6.1. Conclusion
Observation of the acoustic properties of 1pm unfunctionalized mesoporous silica nanoparticles
(MSNs) revealed that the backscatter coefficient of these particles are sufficient to be
considered as an ultrasound enhanced contrast agent (UECA). By conjugating MSN to the HER2-
targetting antibody Herceptin (trastuzumab), local concentrations near tumor sites
overexpressing HER2 receptors increased and further improved the backscatter cross-section of
the MSNs. In vitro studies suggest the binding affinity of the MSN-Herceptin particles is
preferential to cancer cells that overexpress the HER2 surface receptor. Due to the large size of
the particle conjugate, the particles are not incorporated into the cell by endocytosis. Targeted
MSN are ideal for ultrasound contrast enhancement, ultimately improving the reliability of
cheaper, more user friendly, and non-radiating imaging modality.
6.2 Future Direction
6.2.1 In vivo studies
The next objective of the project is to prove contrast enhancement in vivo. Nude athymic
BALB/c female mice will be used for a tumor xenograft model. The breast cancer cell lines used
for in vitro studies will be cultured and subcutaneously injected into the right flank of 4-6 week
old nude mice. After tumor formation up to 65mm3, the mice will be given intraperitoneal
injections of MSN+Herceptin+FITC. Following injection, B-mode images will be taken at the
tumor site using a 40MFIz transducer probe. Prior to study, LD50 of MSN+Flerceptin and the
final endpoint of the study must be determined. Injection frequency and ultrasound imaging is
dependent upon cellular uptake and filtration time.
6.2.2 MSN as a theranostic agent
A great deal of research has been dedicated towards the application of MSN as a drug delivery
vehicle [301. Flowever, the innate properties of MSN also allow them to be applied as an agent
recently termed as a theranostic the combination of therapeutics and diagnostics into a single
agent. Already MSNs have shown their potential as a therapeutic agent (Section 1.4.2,
'Biomedical Applications of MSNs'). As evidence for MSNs as a UECA builds, these particles will
increasingly become more appealing as a theranostic agent as well.
41


The current conjugation motif of MSN+Herceptin is already in essence a theranostic
agent: the original purpose of Herceptin is not to act as a targeting agent, but rather to act as a
treatment for a HER2+ breast cancer types. Further study of cell proliferation under
MSN+Herceptin inoculation must be taken to determine how MSN conjugation affects the
therapeutic efficacy of Herceptin. It is worth noting that the conjugation motif of this study did
not include any form of drug loading into the cells; this may provide a second 'double punch' for
its therapeutic characteristics.
6.2.3 Incorporating perfluoropentane into the MSN pores
A high degree of porosity allows the MSN particle cavity to be loaded with material and used as
a carrying agent. Similar to drug loading, MSNs may also be loaded with a gas. The introduction
of a gas phase would theoretically enhance the backscatter coefficient in a way similar to
microbubbles (solid-to-gas or liquid-to-gas interfaces yield large backscatter). The rigidity of the
MSN pores may restrict the oscillatory motion typically seen in microbubble contrast agents.
Alternatively, this restriction of motion will also provide for greater stability and half-life. As
mentioned previously, extended half-life is beneficial for long term tumor tracking or extended
imaging time.
Like drug loading, gas retention within the particles may be solved by capping. Several
groups have applied hard caps (iron oxide Fe304, cadmium sulfide CdS, gold Au) as well as soft
caps (G-lnsulin)[31]. Most of these capping agents can be removed by environmental stimuli.
Fortunately cap release may not be necessary, thus sidestepping a major issue with this field of
research. However, cap release under cancer cell stimuli may have an advantage as well. Heavy
gasses such as perfluoropentane are used in most clinically approved contrast agents due to
their low partition coefficient in blood. By incorporating a gas phase into MSNs, backscatter may
be drastically increased.
6.2.4 Dye loading for presurgical occult lesion localization
Currently, wire placement is the standard for tumor marking prior to surgery. This method is
inefficient, since only one dimension of the tumor is marked for surgical removal.
Consequentially, an excessive amount of tissue around the lesion must be removed as a
precautionary measure. By functionalizing the surface or loading the MSN pores with a slow-
release dye, targeted cells can be easily stained. Upon scission of the breast during surgery,
42


stained cancer cells can easily be visualized and removed. Dyes such as cyanine can be easily
excited under a blacklight lamp (low energy UVA).
43


APPENDIX
A. MATLAB code: Single-Pulse Processing
% ID Signal Processing Protocol
% By: Andrew Milgroom
% Translational Biomaterials Research Labs
% University of Colorado- Anschutz Medical Campus
% Date: 09/17/2012
% This protol describes the attenuation of an ultrasound signal by
% deconvoluting the image with a signal through pure water through the same
% distance.
% Set variables, figure dimensions, and get screen size
signal = [];
signalAUC = 0;
signalDC = [];
water = [];
fx =1000;
fy = 600;
Size = get(0,'screensize');
% Create the figure and center it on the screen
handles.hfig2 = figure('NumberTitle','off', ...
'Position',[floor((Size(3)-fx)/2) ...
floor((Size(4)-fy)/2) ...
fx fy] . .
'NameSingle Pulse Processing GUI',...
'Color',[1 1 1],...
PaperPositionMode,auto,...
1InvertHardcopy,'off',...
'DoubleBuffer','On',...
'MenuBar','none');
% Create file menu, graph window, data box and buttons
menu = uimenu(handles.hfig2,'Label','File');
uimenu(menu,'Label','Load image...','Callback','[signal water] =
loadsignal(signal,water) ');
uimenu(menu,'Label','Close','Callback','close');
aucbox = uicontrol(handles.hfig2,'Style','edit','Units','Normalized','Position',..
[.05 .35 .2 .07],'Tag','aucbox');
auctitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position',
[.05 0.43 .2 .03],'String','Cumulative Voltage Difference');
grptitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position',
[.55 0.53 .15 .03],'String','Signals');
sigtitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position',
[.1 0.96 .15 .03],'String','Signal');
watertitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position
[.4 0.96 .15 .03],'String','Water');
dctitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position',.
[.7 0.96 .15 .03],'String','Deconvoluted signal');
denoisebtn =
uicontrol(handles.hfig2,'Style','PushButton','Units','Normalized' ,'Position' ,...
[.07 .26 .15 .05],'String','Denoise and Smooth','Callback','[signal water] =
denoise(signal water);');
analysisbtn =
uicontrol (handles. hfig2, Style ' PushButton ', 'Units ', 'Normalized', Position
[.07 .19 .15 .05],'String','Run Analysis','Callback','signalAUC signalDC =
analyze(signal water signalAUC signalDC);');
44


savebtn =
uicontrol(handles.hfig2,'Style','PushButton','Units','Normalized','Position.
[.07 .12 .15 .05],'String','Save to file','Callback','Data = save(signal signalAUC
signalDC); ') ;
handles.mainaxes = axes('Parenthandles.hfig2);
axes(handles.mainaxes);
set(gca,'Title',text ('String
set(handles.mainaxes,'YTickLabel',[]);
set(handles.mainaxes,'FontSize',8,...
'Color',[1 1 1],...
'Box','On',...
'YColor',[0 0 0],...
'XColor',[0 0 0],...
'XTichLabel
'Units','Normalized', . .
'Position',[.3 .12 .6 .4]);
handles.signalaxes = axes(1 Parent1,handles.hfig2);
axes(handles.signalaxes);
set(gca,Title 1,text('String
set(handles.signalaxes,'YTickLabel',[]);
set(handles.signalaxes,'FontSize',8,...
'Color',[1 1 1],...
'Box','On',...
'YColor',[0 0 0],...
'XColor',[0 0 0],...
'XTickLabel',[],...
'Units','Normalized',...
'Position',[.05 .5 .25 .35]);
handles.wateraxes = axes('Parent',handles.hfig2);
axes(handles.wateraxes);
set(gca,'Title',text ('String;
set(handles.wateraxes,'YTickLabel',[]);
set(handles.wateraxes,'FontSize',8,...
'Color',[1 1 1],...
'Box','On',.. .
'YColor', [0 0 0], . .
'XColor',[0 0 0],...
' XTickLabel',[],...
'Units','Normalized',...
'Position',[.35 .6 .25 .35]);
handles.deaxes = axes('Parent',handles.hfig2);
axes(handles.deaxes);
set(gca,'Title',text ('String',''));
set(handles.deaxes,'YTickLabel',[]);
set(handles.deaxes,'FontSize',8,...
'Color',[1 1 1],...
'Box','On',...
'YColor',[0 0 0],...
'XColor',[0 0 0],...
'XTickLabel',[],...
'Units','Normalized',...
'Position',[.65 .6 .25 .35]);
function [signal water] = loadsignal(signal,water)
% This function loads the signal from the user, and grabs the water file
% from a previously described path
function [signal water] = denoise(signal water)
% This function denoises AND smooths the signals
function signalAUC signalDC = analyze(signal water signalAUC signalDC
% This function analyzes the smoothed and denoised signal by deconvoluting
% It and looking at the area difference between water and the signal. The
% the display for 'DC' will appear
45


function Data = save(signal signalAUC signalDC)
% This function saves the original signal, the cumulative difference (AUG)
% and the deconvoluted signal. The data is added to a previously described
% workspace
46


B. MATLAB code: B-mode image processing
% B-Mode Image Processing Protocol
% By: Andrew Milgroom
% Translational Biomaterials Research Labs
% University of Colorado- Anschutz Medical Campus
% Date: 08/01/2012
% This protocol imports a grayscale image of a B-Mode ultrasound scan from
% selecte filepath. The image automatically detects the boundary of the
% region of interest (ROI). The region of interest is parsed into 50 pixel
% x 50 pixel squares. For each square, mean pixel intensity distribution
% (w/st. dev.) and 2D Fourier transform is acquired. All squares are then
% averaged and saved to a predefined workspace, where it can be compared
% with other data
% Set variables, figure dimensions, and get screen size
Bscan = [];
tempdata = [];
fx =800;
fy = 600;
Size = get(0,'screensize');
% Create the figure and center it on the screen
handles.hfig2 = figure('NumberTitleoff', ...
'Position',[floor((Size(3)-fx)/2) ...
floor((Size(4)-fy)/2) ...
fx fy] . .
'Name','B-Mode Image Processing GUI',...
'Color',[1 1 1],...
'PaperPositionMode','auto',...
'InvertHardcopy','off',...
'DoubleBuffer','On',...
'MenuBar','none');
% Create file menu, graph window, data box and buttons
menu = uimenu(handles.hfig2,'Label','File');
uirnenu(menu,'Label','Load Image...','Callback','Bscan = loadimage(Bscan)');
uimenu(menu,'Label','Close','Callback','close');
mgsbox = uicontrol(handles.hfig2,'Style', 'edit', 'Units' ,'Normalized', 'Position'
[.05 .85 .2 .07] ,'Tag', 'mgsbox');
stdbox = uicontrol(handles.hfig2,'Style', 'edit', 'Units' ,'Normalized', 'Position'
[.05 .70 .2 .07] ,'Tag', 'stdbox');
mgstitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position' .
[.08 0.93 .15 .03],'String','Mean Gray Scale');
stdtitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position',.
[.08 0.78 .15 .03],'String','Standard Deviation');
bscantitle = uicontrol(handles.hfig2,'Style','text','Units','Normalized','Position'
[.55 0.93 .15 .03],'String','B-Mode Image');
analysisbtn = uicontrol(handles.hfig2,'Style','PushButton','Units','Normalized',...
'Position',[.07 .19 .15 .05],'String','Run Analysis
'Callback','tempdata =
analyze(Mtrace,Atrace,Bscan,tempdata);');
savebtn =
uicontrol(handles.hfig2,'Style','PushButton','Units','Normalized','Position',...
[.07 .12 .15 .05],'String','Save to file','Callback','Data =
save(Data,tempdata);');
handles.maimaxes = axes('Parent',handles.hfig2);
axes(handles.mainaxes);
set(gca,'Title',text ('String',''));
47


set(handles.mainaxes,'YTickLabel',[]);
set(handles.mainaxes,'FontSize',8.
'Color',[1 1 1],...
'Box','On',.. .
'YColor', [0 0 0],. .
'XColor [0 0 0], . .
'XTickLabel
'Units','Normalized', . .
'Position',[.3 .12 .6 .8]);
function Bscan = loadimage(Bscan)
% Loads image from JPG or TIFF file, saves image as matrix, 'image', and
% displays it on the screen
Bscan = uigetfile
imshow(Bscan)
function tempdata = analyze(Mtrace,Atrace,Bscan,tempdata)
% Finds mean grayscale pixel intensity, its distribution, as well as the
% fourier transform of the image (currently unused). Avoids bias by
% sampling the ROI in 50 pixel x 50 pixel boxes then averages the values.
% Total box count varies.
function Data = save(Data, tempdata)
% Adds new image data to the previous workspace
48


C. Herceptin Protein sequence
Human trastuzumab peptide sequence
>Anti-HER2 Light chain 1
DIQMTQSPSSLSASVGDRVTITCRASQDVNTAVAWYQQKPGKAPKLLIYSASFLYSGVPS
RFSGSRSGTDFTLTISSLQPEDFATYYCQQHYTTPPTFGQGTKVEIKRTVAAPSVFIFPP
SDEQLKSGTASWCLLNNFYPREAKVQWKVDNALQSGNSQESVTEQDSKDSTYSLSSTLT
LSKADYEKHKVYACEVTHQGLSSPVTKSFNRGEC
>Anti-HER2 Heavy chain 1
EVQLVESGGGLVQPGGSLRLSCAASGFNIKDTYIHWVRQAPGKGLEWVARIYPTNGYTRY
ADSVKGRFTISADT S KNTAYLQMNS LRAEDTAVYYCS RWGGDGFYAMDYWGQGTLVTVS S
ASTKGPSVFPLAPSSKSTSGGTAALGCLVKDYFPEPVTVSWNSGALTSGVHTFPAVLQSS
GLYSLSSWTVPSSSLGTQTYICNWHKPSNTKVDKKVEPPKSCDKTHTCPPCPAPELLG
GPSVFLFPPKPKDTLMISRTPEVTCVWDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQY
NSTYRWSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLPPSRD
ELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSR
WQQGNVFSCSVMHEALHNHYTQKSLSLSPGK
>Anti-HER2 Light chain 2
DIQMTQSPSSLSASVGDRVTITCRASQDWTAVAWYQQKPGKAPKLLIYSASFLYSGVPS
RFSGSRSGTDFTLTISSLQPEDFATYYCQQHYTTPPTFGQGTKVEIKRTVAAPSVFIFPP
SDEQLKSGTASWCLLNNFYPREAKVQWKVDNALQSGNSQESVTEQDSKDSTYSLSSTLT
LSKADYEKHKVYACEVTHQGLSSPVTKSFNRGEC
>Anti-HER2 Heavy chain 2
EVQLVESGGGLVQPGGSLRLSCAASGFNIKDTYIHWVRQAPGKGLEWVARIYPTNGYTRY
ADSVKGRFTISADT S KNTAYLQMNS LRAEDTAVYYCS RWGGDGFYAMDYWGQGTLVTVS S
ASTKGPSVFPLAPSSKSTSGGTAALGCLVKDYFPEPVTVSWNSGALTSGVHTFPAVLQSS
GLYSLSSWTVPSSSLGTQTYICNVNHKPSNTKVDKKVEPPKSCDKTHTCPPCPAPELLG
GPSVFLFPPKPKDTLMISRTPEVTCVWDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQY
NSTYRWSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLPPSRD
ELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSR
WQQGNVFSCSVMHEALHNHYTQKSLSLSPGK
Total AA count: 1,330
Molecular Weight 145531.5 g/mol
Lysine: 90/1,330 (6.77%)
Aspartic Acid: 58/1,330 (4.36%)
Glumatic Acid: 60/1,330 (4.51%)
Amines (plus termini): 7.07%, 10,285.7 g/mol
Carboxylates (plus termini): 9.17% 13,349.5 g/mol
49


D. UV-VIS spectroscopy
Cells were incubated overnight at 37C in 0.125 mg/ml MSN-Herceptin. Afterwards the
cells were washed, trypsinized, and sieved with a 10pm nylon mesh. The samples were treated
with lOpI of 12mM MTT solution for 30 minutes at room temperature. Absorbance at 540nm
was measured (top figure). The graph below suggests that not only does the MSN-Herceptin
induce cell proliferation with respect to HER2 expression, but may increase proliferation beyond
the untreated samples. A more plausible explanation would be that the MSN has a stronger
binding affinity to HER2+ cells. As MSN concentration increases, more light is absorbed in the
well. Theoretically, retention of 0.025 mg in a lOOpI volume could increase absorbance by ~0.07
(bottom figure).
50


Absorption
E. FITC and EthD-1 excitation and emission spectra
Wavelength (nm)
FITC excitation
emission
EthD-1 exciation
emission
Excitation and emission spectra of 7-aminoactinomycin-D
546 647
Wavelength (nm)
c
o
m
£
4)
4)
O
C
4)
O
to
4)
t_
O
D
Li.
51


REFERENCES
1. Al-Douri, Y., Ahmed, N.M., Bouarissa, N. & Bouhemadou, a. Investigated optical
and elastic properties of Porous silicon: Theoretical study. Materials & Design 32,
4088-4093 (2011).
2. Alunni, J. Imaging inflammatory breast cancer. Diagnostic and Interventional
Imaging 93, 96-104 (2012).
3. Alunni, J.-P. Imaging inflammatory breast cancer. Diagnostic and interventional
imaging 93, 95-103 (2012).
4. Anderson, C.R. et al. Ultrasound molecular imaging of tumor angiogenesis with an
integrin targeted microbubble contrast agent. Investigative radiology 46, 215-24
(2011).
5. Casciaro, S. etal. Optimal enhancement configuration of silica nanoparticles for
ultrasound imaging and automatic detection at conventional diagnostic
lfrequencies. Investigative radiology 45, 715-24 (2010).
6. Chung, T.-H. et al. The effect of surface charge on the uptake and biological
function of mesoporous silica nanoparticles in 3T3-L1 cells and human
mesenchymal stem cells. Biomaterials 28, 2959-66 (2007).
7. Corsetti, V. et al. Breast screening with ultrasound in women with mammography-
negative dense breasts: evidence on incremental cancer detection and false
positives, and associated cost. European journal of cancer (Oxford, England:
1990) 44, 539-44 (2008).
8. Deshpande, N., Needles, a & Willmann, J.K. Molecular ultrasound imaging: current
status and future directions. Clinical radiology 65, 567-81 (2010).
9. Fan, H. et al. Modulus-density scaling behaviour and framework architecture of
nanoporous self-assembled silicas. Nature materials 6, 418-23 (2007).
10. Friedrich, J., Seidel, C., Ebner, R. & Kunz-Schughart, L. a Spheroid-based drug
screen: considerations and practical approach. Nature protocols 4, 309-24 (2009).
11. Gao, Z.-G., Fain, H.D. & Rapoport, N. Controlled and targeted tumor
chemotherapy by micellar-encapsulated drug and ultrasound. Journal of controlled
release : official journal of the Controlled Release Society 102, 203-22 (2005).
12. Gary-Bobo, M. et al. Cancer therapy improvement with mesoporous silica
nanoparticles combining targeting, drug delivery and PDT. International journal of
pharmaceutics 423, 509-15 (2012).
13. Goldberg, B.B., Liu, J.B. & Forsberg, F. Ultrasound contrast agents: a review.
Ultrasound in medicine & biology 20, 319-33 (1994).
52


14. Haw, A D A M S., Ay, N.A.M.P., Reston, R.O.Y.C.P. & Ond, A.N.D.B. Original
Contribution PROPOSED STANDARD THERMAL TEST OBJECT FOR
MEDICAL. 25, 121-132 (1999).
15. Hocine, O. et al. Silicalites and Mesoporous Silica Nanoparticles for photodynamic
therapy. International journal of pharmaceutics 402, 221-30 (2010).
16. Hocine, O. et al. Silicalites and Mesoporous Silica Nanoparticles for photodynamic
therapy. International journal of pharmaceutics 402, 221-30 (2010).
17. Hristopher, D.O.A.C. & Umbull, D.A.H.T. Review ADVANCES IN
ULTRASOUND BIOMICROSCOPY. 26, 1-27 (2000).
18. Jeong, J.S., Cannata, J.M. & Shung, K.K. Adaptive HIFU noise cancellation for
simultaneous therapy and imaging using an integrated HIFU/imaging transducer.
Physics in medicine and biology 55, 1889-902 (2010).
19. Kim, J. Investigation of the Formation and Structure of APTES Films on Silicon
Substrates. 14222, 1-2
20. Koole, R. et al. Paramagnetic lipid-coated silica nanoparticles with a fluorescent
quantum dot core: a new contrast agent platform for multimodality imaging.
Bioconjugate chemistry 19, 2471-9 (2008).
21. Lavisse, S., Adotevi, C., Opolon, P., Bourget, P. & Perricaudet, M. In Vitro
Echogenicity Characterization of Poly.lactide- Agent Application. 40, 536-544
(2005).
22. Lin, Y.-S. et al. Well-Ordered Mesoporous Silica Nanoparticles as Cell Markers.
Chemistry of Materials 17, 4570-4573 (2005).
23. Liong, M. et al. Multifunctional inorganic nanoparticles for imaging, targeting, and
drug delivery. ACS nano 2, 889-96 (2008).
24. Liu, J. et al. Nanoparticles as image enhancing agents for ultrasonography. Physics
in medicine and biology 51, 2179-89 (2006).
25. Liu, Y., Mi, Y., Zhao, J. & Feng, S.-S. Multifunctional silica nanoparticles for
targeted delivery of hydrophobic imaging and therapeutic agents. International
journal of pharmaceutics 421, 370-8 (2011).
26. Neill, T.P.O. & Winkler, A.J. OOriginal Contribution. 20, 579-588 (1994).
27. Poincloux, R. et al. Contractility of the cell rear drives invasion of breast tumor
cells in 3D Matrigel. Proceedings of the National Academy of Sciences of the
United States of America 108, 1943-8 (2011).
53


28. Saunders, a E. Heating of bone and soft tissue by ultrasound. Imaging Science
Journal, The 55, 52-56 (2007).
29. Search, H., Journals, C., Contact, A., Iopscience, M. & Address, I.P. Equivalence
between three scattering formulations for ultrasonic wave propagation in particulate
mixtures. 3481, (1998).
30. Speed, C. a Therapeutic ultrasound in soft tissue lesions. Rheumatology (Oxford,
England) 40, 1331-6 (2001).
31. Tang, H. et al. Facile synthesis of pH sensitive polymer-coated mesoporous silica
nanoparticles and their application in drug delivery. International Journal of
Pharmaceutics 421, 388-396 (2011).
32. Tang, H. et al. Facile synthesis of pH sensitive polymer-coated mesoporous silica
nanoparticles and their application in drug delivery. International journal of
pharmaceutics 421, 388-96 (2011).
33. Umchid, S. FREQUENCY DEPENDENT ULTRASONIC ATTENUATION
COEFFICIENT. 234-238 (2008).
34. Zou, M. & Yang, D. Nanoindentation of silica nanoparticles attached to a silicon
substrate. Tribology Letters 22, 189-196 (2006).
35. ULTRASONIC MONITORING OF PARTICULATE SUSPENSIONS IN-PROCESS:
PHYSICS, TECHNOLOGY AND APPLICATIONS R. E. Challis, A. K. Holmes and A. N.
Kalashnikov University of Nottingham, United Kingdom.
36. Mohsine, a. & El Hami, a. A robust study of reliability-based optimization methods
under eigen-frequency. Computer Methods in Applied Mechanics and Engineering
199, 1006-1018 (2010).
37. Trewyn, B.G., Nieweg, J. a., Zhao, Y. & Lin, V.S.-Y. Biocompatible mesoporous
silica nanoparticles with different morphologies for animal cell membrane
penetration. Chemical Engineering Journal 137, 23-29 (2008).
38. Han, A., Abuhabsah, R., Blue, J.P., Sarwate, S. & OBrien, W.D. Ultrasonic
backscatter coefficient quantitative estimates from high-concentration Chinese
Hamster Ovary cell pellet biophantoms. The Journal of the Acoustical Society of
America 130, 4139-47 (2011).
39. Ophir, J. & Parker, K.J. OOriginal Contribution. 15, 319-333 (1989).
40. Laugier, P., Droin, P., Laval-Jeantet, a M. & Berger, G. In vitro assessment of the
relationship between acoustic properties and bone mass density of the calcaneus by
comparison of ultrasound parametric imaging and quantitative computed
tomography. Bone 20, 157-65 (1997).
54


41. Liu, J., Li, J., Rosol, T.J., Pan, X. & Voorhees, J.L. Biodegradable nanoparticles for
targeted ultrasound imaging of breast cancer cells in vitro. Physics in medicine and
biology 52, 4739-47 (2007).
42. Zhao, Y., Vivero-Escoto, J.L., Slowing, I.I., Trewyn, B.G. & Lin, V.S.-Y. Capped
mesoporous silica nanoparticles as stimuli-responsive controlled release systems for
intracellular drug/gene delivery. Expert opinion on drug delivery 7, 1013-29 (2010).
43. Liong, M. et al. Multifunctional inorganic nanoparticles for imaging, targeting, and
drug delivery. ACS nano 2, 889-96 (2008).
44. Nielsen, U.B. et al. Therapeutic efficacy of anti-ErbB2 immunoliposomes targeted
by a phage antibody selected for cellular endocytosis. Biochimica et biophysica
acta 1591, 109-118 (2002).
45. Liberman, A. et al. Hollow silica and silica-boron nano/microparticles for contrast-
enhanced ultrasound to detect small tumors. Biomaterials 33, 5124-9 (2012).
46. Arvazyan, A.R.P.S., Udenko, O.L.E.G.V.R., Wanson, S.C.D.S. & Owlkes, J.B.R.F.
Original Contribution SHEAR WAVE ELASTICITY IMAGING : A NEW
ULTRASONIC TECHNOLOGY OF MEDICAL DIAGNOSTICS. 24, 1419-1435
(1998).
47. Republic, C. Ultrasound attenuation imaging. 55, 180-187 (2004).
55


Full Text

PAGE 1

MESOPOROUS SILICA NANOPARTICLES AS A BREAST CANCER TARGETING CONTRAST AGENT FOR ULTRASOUND IMAGING by Andrew Carson Milgroom B.S., Tufts University, 2011 A thesis submitted to the University of Colorado Denver in partial fulfillment of the require ments for the degree of Master of Science Bioengineering 2012

PAGE 2

ii This thesis for the Master of Science Degree by Andrew Carson Milgroom has been approved by Robin Shandas chair Daewon Park Bolin Liu November 13 th 2012

PAGE 3

iii Milgroom, Andrew Carson (M .S., Bioengineering) Mesoporous S ilica N anoparticles a s a C ontrast A gent for U ltrasound I maging Thesis directed by Assistant Professor Daewon Park ABSTRACT Current clinical use of ultrasound for breast cancer diagnostics is strictly limited to a rol e as a supplementary detection method to other modalities such as mammography its inability to discern between cancerous and non cancerous bodies of similar density, like dense calcific ation s or benign fibroadenomas. Its detection capabilities are further diminished by the variable density o f the surrounding breast tissue with the progression of age. Preliminary studies suggest that mesoporous silica nanoparticles (MSNs) are a good cand idate as an in situ contrast agent for ultrasound. By tagging the silica particle surface with the cancer targeting antibody trastuzumab (Herceptin) suspect regions of interest can be better identified in real time with standard ultrasound equipment. Once the silica antibody conjugate is injected into the bloodstream and enters the cancerous to be dysfunctional or overexpressed in certain types of breast cancer. As more particles aggregate at the cell surface, backscatter of the ultrasonic waves increases as a result of the higher porous silica concentration This translates to an increased contrast around the lesion boundary. Tumor detection through ultrasound contrast enhancement provides a tremendous advantage over current cancer diagnostics because is it significantly cheaper and can be monitored in real time. Characterization of MCM 41 type MSNs suggests that these particles have sufficient stability and particle si ze distribution to penetrate through fenestrated tumor vasculature and accumulate in HER2+ breast cancer cells through the enhanced permeation and retention (EPR) effect. A st udy of acoustic properties show ed that particle concentration is linearly correla ted to image contrast in clinical frequency range ultrasound although less pronounced than typical microbubble type contrast agents In vitro studies using cells with varied levels of HER2 expression demonstrated the selectivity of the MSN Herceptin conju gate to cells with HER2 overexpression. Fluorescence imaging suggest these images remain surface bound and are not incorporated into the cell body.

PAGE 4

iv This study demonstrates the potential of MSNs as a stable, safe, and effective imaging contrast agent for u ltrasound based cancer diagnostics. Ultimately this work will contribute towards the improvement of diagnostic alternatives to conventional ionizing radiation intensive imaging such as MRI or X ray without com promising the specificity of the test T he form and content of this abstract are approved. I recommend its publication. Approved by Robin Shandas

PAGE 5

v ACKNOWLEDGEMENTS Foremost, I would li ke to thank my committee members Dr. Daewon Park, Dr. Bolin Liu, and Dr. Robin Shandas f or taking the ti me over the past year to provide me with guidance throughout the project. I would also like to thank Luciano Mazzaro for his assistance with both ultrasound equipment and contrast agents, and Dr. Qun Li for her assistance with the cellular assays. Dr. Dun ghwa Yun was instrumental in assisting with chemistry related advice and solutions. Dr. Lara Hardesty and Dr. Ann Scherzinger (U. of Colorado Hospital) provided images and knowledge of clinical breast cancer diagnostics. Figures from Martin E. Anderson (Du ke University) were adapted for Faran modeling and speckle distribution.

PAGE 6

vi TABLE OF CONTENTS C HAPTER 1. Introduction ................................ ................................ ................................ ................... 1 1.1 Review of breast cancer ultrasound diagnostics ................................ ........................... 1 1.2 Review of Herceptin ................................ ................................ ................................ ...... 2 1.3 Review of ultrasound contrast ................................ ................................ ...................... 4 1.3.1 Reflection by Rayleigh like scattering ................................ ................................ ........... 4 1.3.2 Reflection by variation at media boundaries ................................ ................................ 5 1.3.3 Refl ection by attenuation ................................ ................................ .............................. 7 1.4 Ultrasound Enhanced Contrast Agent (UECA) ................................ ............................... 8 1.5 Mesoporous silica nanoparticles as an ultrasound cont rast agent ............................. 10 1.5.1 Composition ................................ ................................ ................................ ................ 10 1.5.2 Biomedical applications of MSNs ................................ ................................ ................ 11 2. Aims and experimental approach ................................ ................................ ............... 13 2.1 Specific Aims ................................ ................................ ................................ ................ 13 2.1.1 Demonstrate the contrast potential of mesoporous silica nanoparticles .................. 13 2.1.2 Apply MSN as an ultrasound contrast agent through cancer cell targeting ............... 13 2.2 Experimental Obstacles ................................ ................................ ............................... 14 2.2.1 Particle selection ................................ ................................ ................................ ......... 14 2.2.2 Sieving ................................ ................................ ................................ .......................... 14 3. Materials and me thods ................................ ................................ ............................... 16 3.1 Materials ................................ ................................ ................................ ...................... 16 3.1.1 Consumables ................................ ................................ ................................ ............... 16 3.1.2 Equipment ................................ ................................ ................................ ................... 16 3.2 Methods ................................ ................................ ................................ ...................... 18 3.2.1 Characterization of mesoporous silica nanoparticles ................................ ................. 18 3.2.2 Determination of acoustics properties ................................ ................................ ........ 18 3.2.3 Herceptin Conjugation to silica nanoparticles ................................ ............................ 19 3.2.5 In vitro stud ies ................................ ................................ ................................ ............. 23 4. Results ................................ ................................ ................................ ......................... 25

PAGE 7

vii 4.1 Mathematical Justification for MSNs as a contrast agent ................................ ........... 25 4.1.1 Contrast by backscatter ................................ ................................ ............................... 25 4.2 Characterization of mesoporous silica nanoparticles ................................ ................. 27 4.3 Nanoparticle co ncentration to contrast intensity correlation ................................ .... 29 4.4 Cellular interaction of nanoparticles ................................ ................................ ........... 32 4.4.1 Fluorescence microscopy ................................ ................................ ............................ 32 4.4.2 Flow cytometry ................................ ................................ ................................ ............ 33 4.4.3 in vitro ultrasound imaging ................................ ................................ ......................... 35 5. Discu ssion ................................ ................................ ................................ .................... 38 5.1 Contrast Efficacy ................................ ................................ ................................ .......... 38 5.2 Potential Applications ................................ ................................ ................................ 39 6. Conclusion ................................ ................................ ................................ ................... 41 6.1. Conclusion ................................ ................................ ................................ ................... 41 6.2 Future Direction ................................ ................................ ................................ .......... 41 Appendix ................................ ................................ ................................ ............................ 44 A. MATLAB code: Single Pulse Processing ................................ ................................ ....... 44 B. MATLAB code: B mode image processing ................................ ................................ ... 47 C. Herceptin Protein sequence ................................ ................................ ........................ 49 R eferences .. ................................ ................................ ................................ ............................ 52

PAGE 8

viii LIST OF FIGURES FIGURE 1. Ultrasound imag ing of cancerous breast tissue. ................................ ................................ ... 2 2. Backscatter intensity as a function of wave frequency an d particle di ameter .................... 5 3. Incident wave through two media at angle theta ................................ ................................ 6 4 Ultrasound image of breast displaying shadowing cause d by attenuation .......................... 7 5. Diagram of acoustic measurement apparatus ................................ ................................ .... 17 6. Chemical reaction of MSN antibody linkage ................................ ................................ ...... 22 7. Comparison of Acoustic Properties ................................ ................................ .................... 26 8. TEM imag es of MCM 41 mesoporous silica ................................ ................................ ........ 27 9. SEM imaging of MSNs ................................ ................................ ................................ ......... 28 10 FTIR analysis ................................ ................................ ................................ ...................... 29 11. Comparison of acrylamide pl ug with and without MCM 41 silica particles ..................... 30 12. Linear interpolation of contrast with respect to MSN concentration .............................. 31 13. Probability distribution function of a sub resolution scatterer ................................ ........ 32 14 Comparison of cells trea ted with MSN Herceptin FITC ................................ .................... 33 15. Sample gating of cells for flow cytometry. ................................ ................................ ....... 33 16. Dead cells ratio comparison using ethidium homodimer 1 ................................ .............. 34 17. Ultrasound contrast profile of skbr3 cells. ................................ ................................ ........ 36 18. Contrast profile cross sections of HER2+ and HER2 cells treated with MSN Herceptin 37

PAGE 9

ix LIST OF TABLES TABLE 1. Summary of current contrast agents ................................ ................................ .................... 9 2. Advantages and disadvantages of silica nanoparticles as a contrast agent in comparison to gas phase contrast agents ................................ ................................ ................................ ............. 40

PAGE 10

x LIST OF ACRONYMS MSN Mesoporous Si lica Nanoparticles FITC Fluorescein Isothiocyanate APTES (3 aminopropyl)triethoxysilane PSD Particle SizeDistribution MCM Mobil Composition Material TEOS Tetraethylorthosilicate CTAB Cetylmethylammonium Bromide IUPAC International Union of Pure and Applied Chemistry PBS Phosphate Buffered Saline DMEM TEM Tunneling Electron Microscope LV SEM Low Vacuum Scanning Electron Microscope NCF Nanomaterials Characterization Facilities NDT Non Destructive Testing PTF Poin t Target Focus FBS Fetal Bovine Serum PS Penecillin Streptomycin NHS N hydroxyl succinimide EDC 1 ethyl 3 (3 dimethylaminopropyl) carbodiimide) DMSO Dimethylsulfoxide MWCO Molecular Weight Cutoff TEMED Tetramethylethylenediamine AmPs Ammonium Per sulfate BUA Broadband Ultrasound Imaging UECA Ultrasound Enhanced Contrast Agent MPI Mean Pixel Intensity PDF Probability Distribution Function

PAGE 11

1 1. Introduction 1.1 Review of breast cancer ultrasound diagnostics Ultrasound used for breast cancer dete ction can accurately determine if a mass is a benign fluid filled cyst, or a potentially malignant solid mass. A sonogram of a malignant tumor is characterized by: Lack of circumscribed margins Heterogeneous echo patterns Increased anteroposterior dimensio Unfortunately, it is ve ry difficult to determine if a detected solid mass is truly a malignant carcinoma, or simply a calcification or fibroadenoma since all of these cases yield similar densities. Malignant tumors are typically hypo echoic, but may either be hyperechoic or isoechoic as well (Figure 1.A) The detection of the mass diameter may also be difficult to determine if the surround ing tissue is denser than normal breast tissue which is common in women below the age of 50. In t he clinical setting, ultrasound is typically administered in conjunction with another standard method of diagnosis, most commonly mammography. One study found that ultrasound adjunct to mammography was able to increase the incremental rate of detection by up to 41 % [1 ] Remarkably, ultrasound actually has a higher specificity than mammography, is less affected by surrounding tissue density as a result of age, and can detect smaller diameter lesions [ 2 ] However, its inability to stand alone as a primary meth od of detection is mostly due to a high false positive rate [ 3 ] Surgical biopsies of lumps that showed a positive ultrasound follow ing a negative mammogram were determined to be benign in 3% of the cases [ 4 ] Depending on the age group, adding ultrasound to mammography may quadruple the false positive rate. To further improve detection rates following diagnostic ultrasound, biopsies are often included. With a n average cost around $500, biopsies require time from both the sonographer and the pathologist. It m ay also take up to a few days to return the results. Due to the discomfort of the large bore needles, local anesthetics must also be used. Some patients have express ed concern with the possibility o f contaminating new regions from cancer cells on the needl e as it is pulled out of the tissue.

PAGE 12

2 Ultrasound is also commonly used for needle g uidance for biopsy as well as localization for surgical removal If a lesion has been determined t o be cancerous, it is marked for removal prior to surgery by placing a wire across the di ameter of the cancerous region (Figure 1.B) The accuracy of this method of occult lesion localization to distinguish the tumor boundary edges. Again, local anesthetics must be used to accommoda te for the large bore needle. During surgery, the wire is located either visually or with intraoperative ultrasound A recent study suggests that around 33% of oncological surgeons use intraoperative ultrasound regularly [5] With or without ultrasound, t he extent of tissue removed around the wire is determined by palpation. Since surgeons must rely on a one dimensional marker, excess tissue must be removed as a cautionary measure (Figure 1.C) It is evident that ultrasound holds many advantages over curre nt standards of diagnosis. Once diagnosis specificity is improved upon, ultrasound based diagnostics may prove to be adequate as a pr imary method of detection comparable to mammography. Diagnostic ultrasound for breast cancer and other HER2 positive cancer s can benefit from contrast enhancement for detection and presurgical occult lesion localization 1. 2 Review of Herceptin Herceptin is the trade trastuzumab. At 155 kDa, t rastuzumab is a humanized mo noclonal antibody that has been developed to interfere with the functionality of the HER2/neu or ErbB2 receptor. When a secondary receptor such as HER 3 binds to an external stimulus it dimerizes with a HER2/neu receptor. The activated HER2/neu receptor in turn initiates a MAPK and PI3 /AKT kinase Figure 1 Ultrasound imaging of cancerous breast tissue. (A) Typical indication of malignant tumor, (B) Needle inserted through suspected tumor for localization wire injection, (C) MRI of excised tissue with localization wire and biopsy site cli p injected during US imaging A B C

PAGE 13

3 pathway, subsequently initiating a NF k8 growth pathway [6 ] By this chain of events, an external signal can regulate cellular proliferation, migration, differentiation, and adhesion. Certain typ es of cancers part icularly types of breast cancer have been shown to overexpress the HER2/neu receptor by 25% up to 100 fold [7 ] Uncontrolled growth as a result of overexpressed HER2/neu receptors occurs by two mechanisms: (1) A high concentration of HER2/neu receptors on the cell surface (>2 million/cell) leads to a tendancy for hyperactivity, initiating cell growth pathways without external mitogen stimulation, or (2) The constitutive activation of the AKT pathway interrupts the p27Kip1 checkpoint [ 8 ] Since p27Kip1 is co nstantly in a phosphorylated state, it begins to accumulate in the cytoplasm, unable to reach the nucleus. A loss of inhibition to the cdk2 pathway by p27Kip1 results in an exacerbated effect from the HER2/neu overexpression. Trastuzumab interferes with HE R2/neu hyperactivity by disrupting the dimerization step prior to pathway initiation. To date, it is FDA approved for treatment of certain breast cancers, as well as stomach, gastroesphageal junction, and uterine papillary cancers [ 9 ] Prior to treatment, t he region of interest must be screened to confirm that the specific type of cancer exhibits overexpression of HER2/neu receptors on the surface. I mmunohistochemistry or fluorescence in situ hybridization (FISH) of a biopsy section can sufficiently determin e the quantity of HER2/neu receptors on the cell surface. Despite a high binding affinity of trastuzumab to the HER2/neu receptor, a recent study has shown that nearly 70% of patients undergoing treatment with trastuzumab alone do not respond to the treatm ent [ 10 ] This abysmal statistic may be explained by patient to patient variation in resistance The resistance may arise as a de novo primary resistance, or become acquired in a longterm setting. Herein presents an opportunity to improve trastuzumab as a c ancer treatment option Currently, trastuzumab is only applied as an adjuvant treatment, either administered concurrently with chemotherapeutics or post surgery. One study found that trastuzumab treatment post surgery can reduce relapse in patients by 50% [ 11 ] Since trastuzumab is so highly selective for HER2 receptors, the drug has recently become a favorable agent for cell targeting. Liu et al were able to successfully conjugate the trastuzumab antibody to a liposomal drug delivery vehicle for a camptot hecin anticancer drug. The results showed that immunotargeting with trastuzumab improved cellular uptake from overexpressing HER2 cancer cells by 50 300 fold [12] Prior studies have also shown that Anti HER2 antibodies can maintain their anti tumor efficac y despite conjugation to another body [13]

PAGE 14

4 1.3 Review of ultrasound contrast Ultrasound waves are longitudinal compression waves caused by pulses of pressure propagating in a direction away from the source. These pulses create a compression and expansion (rarefaction) effect that causes ripples to propagate through a medium. Waves are able to travel through media since most media have some degree of elasticity. These waves are considered ultrasonic since its frequency is above the audible spectrum of the h uman ear (~20 kHz). Depending on the medium, a certain amount of the ultrasound wave returns back in the direction of the source. The amount of energy returned to the transduce r (source) is used to calculate the elasticity and density of a medium at diffe rent depths. By creating an array of transmitting/receiving transducers, a 2D image can be created. The amount of energy returned to the transducer is calculated as the pix el intensity at a certain depth. White pixels correspond to more energy returned to the transducer, whereas black pixels correspond to less energy returned to the transducer. The extent of returned energy from the initial wave pulse is dictated by several mechanisms. 1.3.1 Reflection by Rayleigh like scattering Most physiological ly relev ant media exhibit some form of inhomogeneity. On ultrasound images, the result is a snowy tissue is unique. While some groups are currently investigating the possibility of calculating speckle p attern to determine tissue composition, the observation of speckle pattern in a clinical setting is still primarily qualitative Speckle patterns are caused by inhomogeneities whose diameter is less than the axial resolution of the imaging system. The inte raction of these sub resolution particles and an incident sound wave may be defined as Rayleigh like scatters. First used to describe particle interactions with light, Rayleigh scattering principles may also be applied to describe the interaction of sound waves with particles by substituting several parameters Any inhomogeneity whose density or compressibility is different from the surrounding medium can density or compressibility acts as a source for a secondary sound wave. This secondary wave principal (also originating in the context of li ght waves) [14] The magnitude o f the secondary

PAGE 15

5 wave in the direction of the initial source defines the backscatter. According to the Faran model [15] the amplitude of the backscatter is primarily dependent on its ka value or its Un introduced to planar orthogonal pressure wave, the echo intensity follows a 4 th order dependence when ka is less than 1 s imilar to Rayleigh scatterers in the context of light waves (Figure 2, left ). As the radius incre ases, the echo amplitude drastically increases as well. Assuming that the incident sound wave remains constant (k), it is clear that the echo intensity will also exponentially decreases as the particle radius decreases (Figure 1, left). Neglecting other p henomena (such as harmonic oscillation or attenuation), a 1 m diameter particle observed by a 10 MHz wave in water (c = 1493m/s) yields an insignificant echo response ( ka = a2 / 4 = 0.02). As such, contrast agents must take advantage of other phenomena in order to effectively create backscatter. Figure 2 Backscatter intensity as a function of wave frequency (k) and particle diamet er (a) The Faran model also reveals a strong frequency dependence for echo intensity (Figure 2, right). When the sub resolution particle diameter remains constant, echo intensity (dashed line) displays strong nodes and peaks as a function of frequency (pr oportional to wavenumber, k ). As frequency increases, resolution improves echo intensity becomes less frequency dependent. 1.3. 2 Reflection by variation at media boundaries Most relevant ultrasound contrast is introduced at the interface between two media on a macroscale (>1mm) This is primarily due to the reduction of acoustic mismatch between the contrast agent and the surround tissue. Image contrast intensity is largely governed by the

PAGE 16

6 acoustic mismatch, or difference between acoustic imped a nce at an interface of two difference media. The acoustic impedance ( Z ) can be defined as: The acoustic wave velocity, c s density and its bulk modulus, B In gasses rent adiabatic index (the coefficient that correlates a change in volume to its responsive change in temperature) and the ambient pressure. The bulk modulus of a solid is typi cally determined experimentally. By knowing the acoustic impedance of two media, the contrast induced at the b oundary between these two media can be theoretically calculated by the intensity reflection coefficient equation: Where Z 1 defines the media through which the incident sound wave initially interacts, and Z 2 defines the secondary media of interaction. This fo rm of the equation assumes that the incident sound wave is perpendicular to the boundary surface. Angle approaches are considered using where Figure 3 Incident wave through two media at angle theta By substituting equation (1) into the Z variables of equation (2) it is apparent that the intensity reflection coefficient is solely determined by differences in density and stiffness (bulk modulus).

PAGE 17

7 By having a low density and a low acoustic wave velocity (as a result of a small bulk modulus), gasses provi de an excellent acoustic impedance mismatch to soft tissue. Alternatively, by having a density and acoustic wave velocity higher than soft tissue, solid materials can also yield strong intensity reflections at a soft tissue boundary. 1.3. 3 Reflection by at tenuation boundary surfaces of different density or compressibility. By this mechanism, the contrast intensity is determined by how much wave energy is reflected back to th e transducer, and how much energy is either transmit ted across the boundary or scattered in a direction away from the transducer. Ultimately, contrast is determined by the difference between the magnitude of wave energy received and the original p ulse wave energy. Another mechanism of energy loss is caused by converting acoustic wave energy into another form, such as heat The result on a This loss in energy also affects regions beneath the anechoic region, resulting in a phenome non known as shadowing (Figure 4 ). Figure 4 Ultrasound image of breast displaying shadowing caused by attenuation. I mage obtained from http://nexradiology.blogspot.com Attenuation the conversion of acoustic energy to another form can be quantified by the following equation:

PAGE 18

8 The degree of attenuation is dependent upon the depth of the region of interest, z a s well as the sound wave frequency, f As depth and frequency increases, the attenuation of the intitial pulse also increases. The attenuation coefficient, a o is a constant value inherent to the medium commonly expressed in terms of dB/(cm MHz). Attenu ation can be caused by small particles whose diameter is much smaller than the sound wavelength. In 1972 J.R. Allegra and S.A. Hawley published a landmark paper defining the parameters that affect attenuation in suspensions and emulsion [16] Using 0.44 0. 653 m diameter polystyrene spheres suspended in water or toluene introduced to sound waves across a 9 165 MHz range, Allegra and Hawley were able to accurately predict the attenuation of the initial sound wave in each experiment. The proposed model applie s six boundary conditions through six differential equations. The mathematics of these equations is beyond the scope of this thesis, but several concepts are taken into consideration: The incident sound wave can be characterized by three wave equations (c ompressional, shear, and thermal); L inearized conservation laws apply to the system. Denser objects with more inertia will have less spatial displacement from compressional waves. Since denser solid p articles will move less in relation to the surrounding fluid (or soft tissue), flow is observed across the particle surface. This flow introduces viscous drag. This viscous flow justifies the addition of a shear wave component to describe the incident sound wave. As viscosity increases, heat created from frict ion increases and acoustic wave energy is lost. Thermal conduction can be considered, according to thermodynamic laws of pressure temperature coupling. Depending of the specific heat of a material, thermal loss can be related to pressure increases from ult rasonic wave pulses. 1.4 Ultrasound Enhanced Contrast Agent (UECA) A vast majority of FDA approved contrast agents are gas phase microbubbles that are encapsulated in either liposomal or protein carriers. Table 1 provides a summary of the three most wide ly used contrast agents for ultrasound imaging.

PAGE 19

9 Table 1 Summary of current contrast agents Trade name Sonovue Optison Definity Gas Sulphur hexafluoride Perfluoropropane Perfluoropropane Mean bubble size 2 8 m 3.0 4.5 m 1.1 2. 5 m Shell composition Phospholipid Human albumin Phospholipid Manufacturer Bracco GE healthcare Bristol Myers Squibb All three contrast agents are applied as an agent to increase the contrast of blood flow through dynamics systems, particularly throug h the heart. Figure 5 provides an illustration of the enhanced contrast of blood when injected with a 1 mL bolus of Sonovue. Figure 5. Comparison of ventricular filling (A) before and (B) after Sonovue injection cated for use in patients with suboptimal echocardiograms to opacify the left ventricular chamber and to improve the delineation of the [17] Figure 6. Chemical structure of Perfluoropropane

PAGE 20

10 All three contrast agents employ some form of heavy gasses (Figure 6 ). Unlike lighter gasses (including air), these gasses have a lower solubility which consequentially increases the half life in the blood stream. Regardless, the half life of these bubbles is restricted to several hours. Due to the relatively large diamet er, most bubbles are cleared from the system much sooner. Since all currently approved microbubbles are larger than 1 m, these contrast agents are unable to pass through endothelial barriers. 96% of perfluoropropane gas in a 20mL bolus of nated through the lungs within 10 minutes [18] The albumin shell is most likely eliminated by liver degradation. Heavy gasses such as perfluoropropane provide the contrast ection by harmonic oscillations created by ultrasound waves close to the bubble resonant frequency [19] This contrast effect is caused in part by the elasticity o f the bubble Unfortunately, thi s high degree of elasticity also contributes to its susceptibility to rupture. As a result, imaging time under con tinues wave pressure is limited to the order of minutes. Several groups have previously attempted to apply MSNs as a UECA [20] Casciaro et al and Liu et al. successfully correlated mean pixel intensity of an image to the concentration of 330 660nm diameter MSNs within a clinically relevant frequency range. In vivo studies suggested that untargeted M SNs aggregate within Browicz Kupffer cells as a part of the reticulo endothelial system (mononuclear phagocyte system) in the liver. The aggregate concentration within the liver is significant enough to be detected by radio frequency range ultrasound. At t he time of this publication, no groups have reported findings on targeted MSN contrast agent effects. Furthermore, no groups have provided an in depth computational justification for MSN usage as a contrast agent. 1. 5 Mesoporous silica nanoparticles 1. 5 .1 Composition Mesoporous silica is most similar to synthetic zeolites, a class of silicate mineral that i s notable for its porous nature. The porosity of zeolite has made the mineral ideal for commercial applications such as absorbents and purification filt of the silica nanoparticles is defined by the International Union of Pure and Applied Chemistry (IUPAC) classification of porous materials. Any inorganic structure whose pore diameter is between 20 and 500 is classified as mesoporous. Anything above 500 would be macroporous, whereas

PAGE 21

11 anything below 20 would be microporous. Although several various microstructures exist, most follow the general method exemplified from the production of Mobil Composition of Matter ( MCM ) 41 (Figure 7 ). For MCM 41, the surfactant cetylmethylammonium bromi de (CTAB) naturally create s liquid crystalline micelles in water. This will act as the template for the pore structure. The ceramic sol gel precursor tetraethylorthosilicate (TEOS) creates a silica network around the micellular formations through hydrolysis and condensation. Once the CTAB template is extracted through solvent extraction or calcination (thermal treatment), a hexagonally ordered silica framework remains. TEOS CTAB Compared to other mesoporous silicas, MCM 41 has a large surface area (941 m 2 /g) and large pore volume (0.73 mL/g). Up to 80% of its v olume can be pore space. The walls of its hexagonal structure is primarily composed of amorphous silica (SiO 2 ), making it less mechanically stable than other more crystalline zeolites. 1. 5 .2 Biomedical applications of MSNs Drug Delivery Vehicle : Since 2001 MSNs have been gaining momentum as a platform for drug delivery systems. Originally designed as molecular sieves, these particles have several advantageous characteristics, including: Large surface area for functionalization and interaction with the envi ronment Large pore volume for drug loading Highly ordered pore network for control of loading and release kinetics Figure 7. Production of MCM 41

PAGE 22

12 Silanol containing surface for easy functionalization Low immunogenicity, with an ability to naturally be incorporated into the cell by endoc ytosis MSNs as a drug delivery system has been applied to several indications to act as a low release or site targeting drug reservoir. Recently, MSNs have been applied as an implantable drug delivery system for bone tissue, thanks to its favorable propert ies as a bioceramic [21] MSNs have also been used as a form of biosensors by loading pore space with bioactive dyes, which only fluoresce under the presence of specific stimuli [22] The functionalization of the pore interiors allow hydrophobic drugs with low solubility to be delivered to target sites more easily. To avoid premature release of the loaded drug, methods for capping pores has become a research area with growing interest. Capping using soft material (antibodies, insulin, pH/temperature sensitiv e polymers) or hard materials (gold, cadmium sulfide, iron oxide) provides tight control the drug delivery package. Cap release can be achieved by both external stimuli (magnetic fields) as well as internal stimuli (glucose, pH, temperature). In a previo us study, Liong et al had previously functionalized the surface of MSN particles by conjugati ng the surface with folic acid [23] A lpha folate receptors have been shown to be upreg ulated in certain human cancers Liong was able to show the preferential abs orption of the MSN into the cancer cell lines PANC 1 and BxPC3. Furthermore, the MSNs incorporated into the cancerous cells were able to release their camptothecin drug payload into the cytosol, effectively reducing cell survival by 60%. Evidently, MSNs ar e an eff icient platform for cancer drug delivery systems.

PAGE 23

13 2. Aims and experimental approach 2.1 Specific Aims 2.1. 1 Demonstrate the contrast potential of mesoporous silica nanoparticles Silica as a contrast medium offers several advantages over microbubb les. Above all silica offers a multifunctional platform that is easier to modify than its microbubble counterpart. In order to apply silica nanoparticles as a contrast agent, the particles must demonstrate the ability to significantly alter the contrast o f a region of interest via co nventional ultrasound Experimental approach : In order to demonstrate the physical effect of ultrasonic waves on MSNs, the system was initially modeled as a colloidal solid particle distributed through out a homogenous phase ac oustically similar to soft tissue The interaction of ultrasonic pressure waves throughout the system could then be characterized using an Ingaard Morse model for acoustic scatterers. By using the Ingaard Morse model to define thermodynamic and hydrodynami c interaction between t he particle and phase under ultrasonic pressure waves, a sufficient prediction could be made for the contrast effects during imaging. These prediction s were confirmed by B mode ultrasound image analysis of agar or acrylamide blocks w ith varying concentrations and diameters of silica nanoparticles. 2.1.2 Apply MSN as an ultrasou nd contrast agent through cancer cell targeting A major reason ultrasound is not more widely used for breast cancer diagnostics is its inability to discern betw een cancerous and non cancerous masses of the same density. If the contrast of a cancerous mass can be enhanced in real time, ultrasound will be a far more effective means of cancer diagnostics. In order to use MSNs as a contrast agent for breast cancer, t he particles must be selective as to where they accumulate. Th is primarily achieved by t hree mechanisms: (1) Size: by having a diameter less than 1 m MSNs are able to pass through the fenestrated endothelial walls of tumor vasculat ure. (2) HER2 receptor targeting: prior research has shown that certain types of breast cancer overexpress the HER2 surface receptor by 100 fold [24] This overexpression makes the HER2 surface receptor an ideal targeting antigen for cancerous cells. Herceptin an anti HER2 antibody, was conjugated to the surface of MSNs to ensure accumulation near tumor regions.

PAGE 24

14 (3) Porosity: by maintaining a porous interior structure, variations in density and compressibility will enhance the reflectivity of incident soundwaves, further increasing echo contrast during imaging. By adjusting the size and surface of the MSN, the local volume fraction of MSNs in and around the tumorous tissue will increase and only the targeted regions of interest will experience an enhancement in con trast. Experimental approach : To ensure that both size and surface modification were optimized for accumulation in vivo, p article size distribution (PSD) and cellular interaction was first characterized. PSD was analyzed by scanning electron microscope (SE M) and tunneling electron microscope (TEM) imaging. Surface modification steps were monitored by Fourier transform infrared spectrometry (FTIR), SEM, and ultraviolet visible spectroscopy (UV VIS). To observe using fluorescence microscopy and fluorescence a ctivated cell sorting (FACS) the MSN Herceptin conjugate s were labeled with fluorescein isothiocyanate (FITC) The FITC labeled MSN Herceptin was introduced to four different cell lines of increasing expression of the HER2 receptor. 2.2 Experimental Obst acles 2.2. 1 Particle selection In order to be an effective targeting agent for passage through fenestrated tumor vasculature and target cell attachment MSNs must fall within a specific diameter range. After extensive characterization, it is clear that th e commercially available MCM 41 yield ed a wide particle size distribution (PSD) with a mean diameter far greater than 8 m. Since MSNs essentially comprise of stacked hexagonal pore sheets, the mean diameter could be modestly reduced by sonication, but coul d not disintegrate fully intact sheets. Thus, the bulk material must be sieved to collect the particles population with desired diameter. 2.2. 2 Sieving Due to its size, surface charge, and density, MSNs are incapable of dissolving in solution. Furthermore these colloidal particles do not remain uniformly suspended in solution longer than an order of minutes. In general, the issue of insolubility was accommodated by sonication and use of ethanol as a solvent (more hydrogen bonding with silanol or amine gro ups). However, colloid flocculation posed a consistent problem in terms of filtration Once it was

PAGE 25

15 determined that the initially proposed Sephacryl column could not accommodate the larger silica particles, a simpler paper based 1 m filter was applied. Unf ortunately, less than 1% yield was obtained after filtration. One suggested explanation for this loss of product was channel clogging caused by larger diameter particle. Finally, a 1 2 layer woven nylon grid mesh with 1 or 8 m pore diameters was applied. Since pore channels were shorter (75 m) and less circuitous, the upstream flocculation layer could be more easily disrupted by retrograde flow or sonication. Regardless, the filtered yield rarely exceeded 40% of the initial mass.

PAGE 26

16 3. Materials and methods 3.1 Materials 3.1.1 Consumables MCM 41 type (h exagonal) MSNs were purchased through Sigma Aldrich (Milwuakee, WI) and filtered through a 1 m nylon mesh (Elko Filtration Inc., Miami, Fl). 1N hydrochloric acid, ethanol (200proof), anhydrous toluene, 1X Phosp hate buffered saline (PBS), (3 aminopropyl)triethoxysilane (APTES), dimethylsulfoxide (DMSO), and fluorescein isothiocyanate isomer I (FITC) for MSN functionalization were provided by Sigma Aldrich (Milwuakee, WI). Trastuzumab (Herceptin ) was provided by Dr. Bolin Liu. For ultrasound phantoms, acrylamide, methylene bis(acrylamide), ammonium persulfate (AmPs), and Tetramethylethylenediamine (TEMED) was used for solid tumor mass mimics. For in vitro studies, t wo groups of increasing HER2 expression were used: (1) MDA MB 231(TN) has a ver y low level HER2 expression, and (2) BT474 or SKBR3 yields a moderate level of HER2 expression Colorado Department of Pathology. The MDA MB 231(TN) cell line was cultured in RPMI 1640 with 10% fetal bovine serum (FBS) and 1% Penicillin Streptomycin (PS). The cancer cell lines with 10% FBS and 1% PS. For trans mission eletron microscopy (TEM), samples were stained with 2 % uranyl acetate and placed on 400 mesh formvar coated, carbon evaporated, and glow discharged copper grids, courtesy of the University of Colorado Denver Core facilities. 3.1.2 Equipment To dis rupt aggregation prior to use, all MSNs were sonicated in a n ultrasonic waterbath cleaner operating at 35kHz for 10 minutes. All MSN particles were dried in a 100 0 C oven. For characterization with TEM samples were imaged on a Tecnai G2 series scope (FE I Hillsboro, OR) Low vacuum scanning electron micr oscopy (SEM) was performed on a JSM 6480LV W thermal emission microscope (JEOL Ltd. Peabody, MA), at the University of Colorado Boulder Nanomaterials Characterization Facilities (NCF). Single pulse u ltrasound measurements were obtained using Non Destructive Testing (NDT) transducers with matching frequencies ranging from 0.5 7.5 MHz ( GE Panametric ). All

PAGE 27

17 Figure 5 Diagram of acoustic measurement apparatus focus length under water. Two transducers were set facing each other through the round (Figure 8 ). The box was filled with agar gel at the bottom (or an agar cylinder was placed inside) and filled with deionized water The water was retained within the container by covering the window with double layered masking tape. To ensure clean connectivity, ultrasound gel was applied between the single element transducers and the ma s king tape covered window. The precise distance between transducers was measured prior to each set of experiments. Each transducer was designated as a pulser or receiver and attached to a Panametric NDT Model 5800 pulser/receiver. Receiver signal was acquired on an Infiniium 8000 High Performan ce Oscilloscope (Agilent Technologies ). Waveform data was sent to MATLAB (The Mathworks, Inc Natick, MA) for data processing. B Mode imaging was acquired on a Sonix SP clinical ultrasound monitor attached to a L14 5/38 curvilinear probe (Ultrasonix, B C, Canada). All images were taken inside of the acryic containterr. Images were processed in ImageJ (NIH, Bethesda, MD) and MATLAB (The Mathworks, Inc, Natick, MA). Fluorescence activated cell sorting (FACS) was performed on a FACScan (BD, Franklin Lakes NJ) and processed through FlowJO

PAGE 28

18 3.2 Methods 3.2.1 Characterization of m esoporous s ilica n anoparticles MSN preparation: Prefabricated MCM 41 hexagonal type MSNs were dispersed in 200 proof ethanol at a concentration of 5 mg mL 1 and bath sonicated for 10 minutes. The dispersion was then filtered through either a 1 ,8,10 and 30 m nylon mesh from a 10 mL syringe. R etrograde/prograde flow alternation in a 1 sec : 3 sec timing pattern was used t o prevent flocculation on the upstream sid e of the filter. T he filtered particles were dried in a 100 0 C oven for 48 hours. TEM i maging : Dried MSN was resuspended in 200 proof ethanol to 500 g mL 1 and stained in 2% uranyl acetate for 3 minutes. After drying on a 400 mesh formvar coated copper grid, the samples we re imaged at 80kV. SEM i maging : Dried MSN was resuspended in 200 proof ethanol to 500 g mL 1 20 L drops were placed onto glass slides, a llowing for the ethanol to evaporate off. The glass slides were then sputter coated in a gold/platinum alloy to a 3 n m thickness. Images were acquired at 5.0 kV. 3.2.2 Determination of acoustics properties Single pulse scan: To quantify the attenuation caused by MSNs signal loss through a n acrylamideblock was measured. This method was also used to calibrate the acous tic properties of the acrylamide blocks to better represent soft tissue. Using the apparatus previously acrylamide blocks with incrementally increasing concentrations of 1 m MSNs (0, 0.1, 0.25, 0.5, 1, 2, 5, and 10 mg mL 1 ) were placed into the acoustic measurement apparatus. The system apparatus was left untouched for 5 minutes to allow for all air bubbles and potential contaminants to settle. 25 J pulses were transmitted through the MSN/water samples at 5 kHz. The waveform was averaged and transferred to a MATLAB program for analysis (Appendix A) To calculate the attenuation, a fast fourier transform was performed and deconvolved with a reference signal of pure water, as described by Laugier et al [25] The complex spectrum of a signal can be described by the following equation:

PAGE 29

19 Wher e is the systems response, is the frequency, is the distance between the two transducers, and is the speed of sound through the medium. The frequency dependent attenuation of the MSN water dispersions can be derived from the ratio of its magnitude : B mode imaging: To calculate the correlation between image contrast and particle concentration, 25% polyacrylamide phantoms were created with incrementally increasing concentrations of 1 m MSNs at 0, 0.1, 0.25, 0.5, 1, 2, 5, and 10 mg mL 1 Acrylamide was [26] The acrylamide phantoms methylene bis(acrylamide). Prior to polymerization, the solution wa s sonicated for 3 minutes using a probe sonicator set to power level 3. After adding 0.05% AmPs and 0.025% TEMED, 1 mL was immediately poured into a 5 mL syringe with the nozzle removed. Complete polymerization occurred between 2 5 minutes. The polyacryla mide phantoms were submerged under wate r and placed on a 5% agar platform approximately field from the highly reflective bottom of the acoustic measurement apparatus The phantoms were imaged with a 10 MHz probe at 70dB. Two focal points were set at the top and bottom of the phantom, with the top of the phantom 2.5 cm from the top of the phantom. 3 images were taken for each phantom and transferred to a MATLAB program for analysis (Appendix B ). The inter ior of the phantom was parsed into 50 pixel x 50 pixel squares and measured for mean pixel intensity. Each square, along with the squares from the other two images of the same concentration, was then averaged. Each mean gray pixel intensity value was corre lated to the MSN concentration. 3.2. 3 Herceptin Co njugation to silica nanoparticles MSN functionalization : Prefabricated MCM 41 hexagonal type MSNs were prepared as ). The 1 m diameter particles were re suspended in 10% HCl to a concentration of 31.4 mg mL 1 in order to increase the amount of free silanol groups on the surface ( Figure 9 step 1). After stirring for 1 hour at room temperature, the dispersion was centrifuged at 5,000 rpm for 5 minutes, dec anted, and washed

PAGE 30

20 in deionized water ( di H 2 O ). This centrifugation and wash step was repeated two more times. After the third decantation, the particles were resuspended in di H 2 O and sonicated for 10 minutes to reduce aggregation. The particles were dried i n a 100 0 C oven for 48 hours. Silanol group concentration was predicted by FTIR and pH curve For amine functionalization, the h ydroxylated MSNs were suspended in anhydrous toluene to a concentration of 20.7 mg mL 1 APTES was added to bring the final conce ntration to 6.5% (Figure 9 step 2). The dispersion was placed in a nitrogen environment by injecting a 1mL syringe with an 18G needle through the stopper on a round bottom flask. The plunger of the syringe was replaced with a double layered nitrogen fille d balloon (~10 cm diameter) The lip of the balloon was secured with parafilm. T he dispersion was centrifuged at 5,000 rpm for 5 minutes, decanted, and washed in anhydrous toluene. To purify the particles from unbound APTES the dispersion was added dropw ise into ether, and again centrifuged 5,000 rpm for 5 minutes. This centrifugation and wash step was repeated two more times. After the third decantation, the particles were resuspended in 200 proof ethanol and sonicated for 10 minutes to reduce aggregatio n. The particles were dried in a 1 00 o C oven for 48 hours. Herceptin/FITC conjugation : To couple the Herceptin antibody with MSNs, amine groups on the silica surface w ere coupled to carboxylates from exposed aspartic and glutamic acid residues and carboxyl termini on the Herceptin antibody (Figure 9 step 3). A fte r preparing a solution of 0.5mg/mL Herceptin in 1X PBS N hydroxyl succinimide (NHS) and 1 ethyl 3 (3 dimethylaminopropyl) carbodiimide) (EDC) was added as a crosslinking agent (2mM final concentra tion for each). Amine coupled MSNs were resuspended in the solution (4 mg/mL ) and sonicated for 10 minutes to ensure optimal exposure to surface area. The EDC crosslinking reacted for 4 hours, stirring at room temperature. The product was then centri fu ged for 10 minutes at 3,000 rpm and decanted to remove the water soluble carbodiimide and urea byproduct. The product was resuspended in 1X PBS. The centrifugation, decantation, and resuspension step was repeated three times.

PAGE 31

21 F igure 6 Chemical reaction of MSN antibody linkage Herceptin modification MSN modification

PAGE 32

22 Figure 7 continued Chemical reaction of MSN antibody linkage Conjugation of modified MSN and Herceptin

PAGE 33

23 After the third resuspension to 5 mg mL 1 in PBS, protein concentration was determined by Bradford Assay Once the protein concentration was calculated, 10 mg mL 1 FITC in DMSO was introduced to a ratio of 80 g FITC: 1mg Herceptin. 1 hour of reaction time stirring at room temperature in the dark allowed for FITC conjugation to uncoupled amin es and sulfhydryl groups. The solution was dialyzed in 3,500 kDa molecular weight cutoff (MWCO) dialysis tubing in milliQ water. The water was changed 5 times over a period of 72 hours. Dialysis occurred at 4 0 C to preserve protein integrity. FITC conjugati on was confirmed by fluorescence microscopy and UV/VIS. FITC conjugation efficiency was determined by the following calculation: Where A(495) corresponds to excitati on/emission wavelength of FITC, and A(280) corresponds to the excitation/emission wavelength of aromatic ring containing amino acids. 3.2.5 In vi tr o studies Fluorescence microscopy: To qualitatively determine binding of the MSN Herceptin conjugate to HER2 + cells, three cell lines were incubated and prepared as previously described (Section 3.1.1). The cell lines (MDA MB 231(TN), BT474, SKBR3, and HR20) were grown to 95% confluency in 96 well plates. Each cell type was incubated with the MSN Herceptin FITC conjugate in DMEM at varying concentrations (50,100,200,400 g/well ) and times (30 minutes, 1 hr, 2hr). At the respective time point, all cells were lifted from the well plate by trypsinization. The cell suspension was filtered through a 8 m nylon mesh filt er. This would allow for the free MSN to pass, but trap MSN attached cells on the upstream side of the filter. Similar to MSN release, accumulated cells were detached from the filter by alternating retrograde/prograde flow. Flow cytometry: Preferential cel lular uptake of MSN Herceptin particles were quantified using flow cytometry. The cell lines were prepared in the same manner used for fluorescence microscopy. Data were processed by FloJO. Ultrasonography: To quantitatively show that cellular uptake of M SNs can augment the inherent contrast in B Mode ultrasound images, a 2D model was applied. After incubating cells in MSN to allow for uptake as previously described for fluorescence microscopy, t he recovered cell

PAGE 34

24 suspension was sp read across a 10% agar gel A 5 cm thickness ensured that the focal region of the ultrasound beam would not detect reverberations from the plate bottom. Images were

PAGE 35

25 4. Results 4.1 Mathematical Justification for MSNs as a contrast agent 4.1.1 Contrast by backscatter By using the Morse Ingaard model for general ultrasound scatterers, s, through a medium, m several comparisons between solid and gas particles can be made [R] : = scattering cross section N = concentration of scatterers per unit volume k = wave number (equal to of the incident soundwave) r = particle diameter = adiabatic compressibility (inverse of the bulk modulus, B ) = density This model had previously been used to show that gasses exhibit a backscattering cross section around 10 14 whereas solid particles are closer to 2 [27] Th e significant disparity in backscatter efficiency of the particle is caused by the exponential relationship to scatterer cross section ( ) and its compressibility and density term s (bracket ed values). Generally it is assumed that and T he Morse model also denotes that backscattering cross section is linearly related to the concentration of scatterers per unit volume Figure 11 provides an illustrative comparison of acoustic properties of materials r elevant to ultrasound imaging. Using soft tissue as the exemplary medium, it is clear that the density and acoustic wave velocity of gasses are much less than soft tissue.

PAGE 36

26 0 1000 2000 3000 4000 5000 6000 7000 8000 air sulfur hexafluoride octofluorpropane lung fat water soft tissue muscle skull bone bulk fused silica PZT kg/m^3) Densities 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 air sulfur hexafluoride octofluorpropane lung fat water soft tissue muscle PZT skull bone bulk fused silica m/s Wave Velocity 0.000 0.200 0.400 0.600 0.800 1.000 1.200 soft tissue muscle water fat skull bone bulk fused silica lung PZT octofluorpropane sulfur hexafluoride air Intensity reflection coefficient (dimensionless) Intensity v. soft tissue 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 1.00E+05 1.00E+06 1.00E+07 1.00E+08 air sulfur hexafluoride octofluorpropane lung fat water soft tissue muscle skull bone bulk fused silica PZT N*s/m^3 Acoustic Impedance Figure 8 Comparison of Acoustic Properties

PAGE 37

27 If the adiabatic compressibility of a scatterer is not taken into accou nt (such as in equation 2), the disparity in contrast efficiency between gas and solid nanoparticles is now within one order of magnitude. From the figure above, it is clear that silica is denser than soft tissue and has a larger acoustic velocity. Since the differences in density and compressibility are quadratically related to contrast (equation 2), only the absolute difference is critical For this reason solid particles create contrast despite an inverse relationship to soft tissue, compared to gases. To reiterate, the difference in acoustic impedance between medium and scatterer defines the degree of contrast intensity. 4. 2 Characterization of mesoporous silica nanoparticles TEM imaging (Figure 1 1 ) confirmed presence of the hexagonal array typical of MCM 41 type MSNs. Successful TEM imaging proved more difficult than previously anticipated, due to the flocculating nature of the particles. Figure 1 1 .A reveals the irregular form that has been seen as typical for commercially available MCM 41. Due to the nature of TEM imaging, particles were placed on the imaging stage while still in solution. This allowed for cluster formation, as seen in figure 1 1 .A. Prior studies have shown that the degree of aggregation is directly dependent upon concentration [28] For MCM 41 type MSNS with 300 nm diameters, the critical concentration for an aggregation event to occur was near 500ug/ml. Figure 9 TEM images of MCM 41 mesoporous silica. (A) 100 nm scale of large cluster in aqueous medium. (B) 20nm scale flake depicting honeycomb like cluster A B

PAGE 38

28 The irregular macrostructure of the MSNs were further illustrated by SEM (Figure 13). The large macroscale pockets i n conjunction with hexagonal nanoscale pores (2.1 2.7nm in diameter) contributes to a surface area ratio of 1000m 2 /g. During surface modification steps prior to Herceptin conjugation, FTIR was used to monitor the surface groups (Figure 1 3 ). Native MSN, h ydroxylated MSN (MSN OH), and amine functionalized MSN (MSN NH2) all exhibited the characteristic peaks of aminosilanes, including H O H bending at 1639 cm 1 from residual water. The hydroxylated MSN displayed an increase in the broad O H stretch at 3200 3 400 cm 1 as expected. Surprisingly, MSN OH displayed more Si O Si bonding than MSN NH2. This could be due to shielding effect caused by the additional ethoxysilane group s seen in MSN NH2. As expected, MSN NH2 exhibited peaks at 4000 and 1570 cm 1 Any c hanges in C H bonding (from APTES) near 3000 cm 1 would be masked by the considerably larger residual peaks from the instrument cartridge. A B Figure 10 SEM imaging of MSNs (A) 60,000X zoom, (B) 1,000 zoom

PAGE 39

29 Figure 11 FTIR analysis 4 3 Nanoparticle concentration to contrast intensity correla tion In order to determine the image contrast due to MSN, acrylamide plugs were polymerized with v ar ying levels of MSN concentration and imaged using conventional clinical ultrasound ( Figure 1 4, 15 ). Image contrast never approached a mean pixel intensity ( MPI) of 0 since anomalous reflections were detected from imperfections in the acrylamide plug. The correlation of contrast to MSN concentration is linear ly related, as expected from the Morse Ingaard model (Equation (8)): The increase of scatterers, N is directly proportional t o the scattering cross section, The Morse Ingaard model also predicts scattering cross section to have a fourth order dependence on the transducer probe frequency (ie. the wavenumber, k, in units of Hertz). As frequency increases, the backscatter cross section increases exponentially. For this set of experiments, imaging was limited to a probe with a central frequency of 7.5 MHz with a maximum limit of 10MHz.

PAGE 40

30 Exploring higher frequencies was limited by availability of probe types. In a clinical setting, ultrasound probes may operat e at a range between 2 18 MHz. In the traditional sense, axial resolution (or longitudinal resoluti o n) is defined as the detectable distance between two objects on a line parallel to the direction of wave propagation. Axial resolution can be calculated a s half of the spatial pulse length, SPL. SPL is defined as the wavelength, multiplied by the number of cycles , emitted per pulse. Wavelength can be further decomposed as a product of acoustic velocity, c and the frequency of the incident sound wave, f Using the experimental setup for determining the correlation of MSN concentration to MPI ( f = 7.5 MHz, c = 1540 m/s, = 1 ), the theoretical axial resolution can be calculated: Figure 12 Comparison of acrylamide plug wit h MCM 41 silica particles (left) and without (right). Image was acquired with a 10 MHz probe at a depth of 2.5 cm.

PAGE 41

31 Clearly the contrast apparent in Figure 1 4 is significantly smaller than the theoretical axial resolution with a diameter closer to 1/10 th of the proje cted resolution. Considering the drastic disparity between these two values, the apparent contrast is not caused by creating an image of the individual particle boundary, but rather increasing the inhomogeneity of regions in the microscale. In essence, MSN Although contrast does not reflect the underlying structure of these sub resolution scatterers, it does reflect their echogenicity [29] As described previously (Section 4.1.1), the inherent density and acoustic velocity of silica material in soft tissue yields a substantial degree of reflectivity. Convention may suggest that increasing speckle noise can decrease the visibilit y of a boundary region, as the case with certain hypoechoic lesion s The co ntrast caused by a sub resolution scatterer is calculated by statistical probability. Each scatterer creates a random phasor along a complex plane (Figure 16). A phasor is a representation of a sinusoidal function. In this case, the function is the respons e to an incident sound wave. The contrast of a region can be defined as the vector sum r of each individual scatterer ( Figure 16, left). In uniform medium, when each scatterer population, r is compiled with scatterer Figure 13 Linear interpolation of contrast with respect to MSN concentration

PAGE 42

32 populations in the sa me region, the amplitude of contrast within the region equates to a 2D Gaussian distribution (Figure 16, right). Reverting back to optics theory, this can also be defined as a Rayleigh probability distribution function (PSD) Since all scatterer population cancel each other out, a zero mean phasor results, and any contrast is observed as noise. However, by targeting the silica particles to specific regions of congregation through flocculation in phantoms or Herceptin receptor targeting in vitro the peri odic arrangement creates a coherence component that is added to the speckle noise statistics. By adding a coherence component to a sub resolution scatterer, a strong phasor is added to the system. This results in a shift of the PSD either to the right or l eft along the real axis Since scatterer arrangement at the lesion boundary (or flocculation cluster) would not be random, contrast can be detected despite having a sub resolution diameter. contrast of tumors at the boundary layer can be accentuated. 4. 4 Cellular interaction of nanoparticles 4.4.1 Fluorescence microscopy A preliminary view of HER2 positive and HER2 negative cell lines reveals a modest yet detectable discrimination between the two cell types (F igure 14). While the MSN Herceptin nanoparticles are present in the filtered H ER culture (14.A), most of the particles are either unbound or attached to the outer wall of the cell membrane. Conversely, most HER2+ cells (14.B) show at tachment of the nanopa rticles to the cell. The data does not confirm the precise location (internal or external) of particle attachment to the cell. Both cell populations were Figure 14 Probability distribution function of a sub resolution scatterer

PAGE 43

33 sieved after particle incubation using a 10 m nylon mesh filter, theoretically eliminating all unbound particles whose diameter is less than 10 m. 4.4.2 Flow cytometry A FACS Calibur flow cytometer was used to determine the binding of the MSN Herceptin FITC nanoparticle to HER+ cells, as well as determine the cell viability after incubation Since compensation was not achievable, the two experiments were not conducted simultaneously. The binding affinity of the nanoparticles were characterized by the relative absorbance Figure 15 Comparison of cells treated with MSN Herceptin FITC (A) HER cells: 231, (B) HER+ cells: skbr3 A B Figure 16 Sample gating of cells for flow cytometry. Shown: 231 treated with MSN Herceptin 100 um

PAGE 44

34 /30 ban dpass filter. The FITC conjugated molecules were excited usi ng a 488nm laser, and polygon gated to capture the event population with a larger forward and side scatter (Figure 15). This gating only captures approxiately 40% of the event population as a prec autionary measure to exclude debris, unbound residual particles, or dead cells (for the binding assay experiment only ). Following gating, FITC fluorescence was measured for each cell type (Figure 1 5, right ). The control skbr3 and 231 yielded nearly identi cal profiles, as expected (231 group not shown) The MSN treated 231 cell population (HER 2 positive ) expressed a detectable degree of fluorescence from the FITC particles ; any fluorescence should indicate any HER2 receptor expression on the surface. The MD A MB 231(TN) cell line has been characterized as either a basal like and triple negative breast cancer (TNBC) line in literature. Groups such as B. Lehmann et al. exp normal, non amplified level of HER2 expression may account for the apparent binding of MSN Herceptin FITC particles to this cell type. Alternatively, indiscriminate bindin g may be caused by the silica particle itself, as opposed to the Herceptin antibody. Trewyn et al have shown than nude MSNs are naturally incorporated by certain cell types. The accelerated metabolism of cancer cells increase their cellular uptake, conseq uentially reducing their partiality for nutrients. In a qualitative sense, the rate of cell division far exceeds both skbr3 and skbr3pool. Figure 17 Dead cells ratio comparison using ethidium homodimer 1

PAGE 45

35 Skbr3 which indeed overexpresses HER2, shows a strong binding affinity with the nanoparticles. From an arbitra ry cutoff near 10 2 fluorescence, 96.2% of the gated events from the skbr3 cell line exceed that cutoff. Alternatively, the 231 cell line exhibits only 15.1% of the gated events pass the cutoff. Since silica yields a wide absorption spectra, UV VIS spectr oscopy was skewed by the concentration of MSNs. For this reason an MTT assay to determine cell proliferation was not possible (Appendix D). Alternatively, ethidium homodimer 1 (EthD 1) was used to determine the ratio of dead cells to total cell count. By u sing a 488nm laser and a 650nm long pass filter, the tail of the EthD emission spectra (Appendix E). EthD 1 interference for the binding affinity experiment was not an issue since those cells were not treated with EthD 1. A comparison of the fluorescence of skbr3 cells with and without MSN treatment suggest that the MSN treatment reduces overall cell proliferation. A shift in fluorescence caused by EthD 1 uptake indicates that the Herceptin targeting arm also maintains a certain degree of its therapeutic capability. However, further testing is required to confirm this claim. 4.4.3 in vitro ultrasound imaging After incubation in MSN Herceptin, cells were lifted of the well plate and aliquot onto agar inside of a 5 ml syringe. After allowing 2 hours for the cells to settle to the agar floor, most of the media was pipetted off. A second layer of agar was poured over to the remaining cell suspension (Figure 1 7 ) A cross section of the s andwiched cell culture was imaged on a clinical ultrasound machine at 10MHz. To quantify the peak pixel intensity and thickness of contrast the pixel intensity profile along the y dimension was graphed (yellow line, Figure 1 7 ). Five profiles with variable x coordinated were averaged to produce the mean pixel intensity profile (Figure 1 7 bottom).

PAGE 46

36 0.0 50.0 100.0 150.0 200.0 250.0 0 50 100 150 200 Mean Pixel Intensity Vertical length MSN-Herc treated untreated The presence of MSN Herceptin results in a spike in contrast by approximately 640%, with a boundary thickness close to 1mm. The tot al cell count in the suspension is estimated to be around 200,000 cells spread across an area of 122mm 2 A comparison of the two cell lines 231 and skbr3 s uggests that there is minimal and insignificant discrepancy in contrast This obse rvation is par tially i n conflict with the binding affinity assay from flow cytometry which suggests that there is a significant difference between the binding affinit ies of MSN Herceptin to cells that do and do not overexpress the HER2 receptor. This observation is als o supported indirectly by UV VIS spectrometry (Appendix D). The binding affinity assay does indicate, however, that there is a certain degree of binding to all cell types. Figure 1 9 suggests that this indiscriminate binding is enough to create contrast nea rly equivalent to the HER2+ cell line. Possible explanations for this occurrence is provided in the previous section. Figure 18 Ultrasound contrast profile of skbr3 cells (A) treated with MSN Herc and (B) untreated. The yellow line indicates an example profile cross section. (A ) untreated (B) treated

PAGE 47

37 0 50 100 150 200 250 0 50 100 150 200 Mean Pixel Intensity Vertical Length skbr3 treated 231 treated Figure 19 Contrast profile cross sections o f HER2+ and HER2 cells treated with MSN Herceptin

PAGE 48

38 5. Discussion 5.1 Contrast Efficacy The potential for silica nanoparticles to act as an effective contrast agent relies on the del ivery and retention method of the particles. It has been well proven that gas phase contrast agents are ideal for contrast enhancement in dynamic systems (such as cardiovascular flow) Although silica nanoparticles do provide some degree of contrast in dyn amic systems, it is most effective in static systems. By operating in a static system, the silica nanoparticles can form what may be termed as a flocculation layer. The formation of this deposition layer is requisite for both its drug delivery functionalit y as well as its contrast agent functionality. Due to the relatively small diameter of the MSN particle with respect to the ultrasonic wavelength, Rayleigh type backscatter is diminished By having substantially h igh concentrations in at tumor sites ultr asound contrast can be augmented Point scatterer concentration may fluctuage greatly in dynam ic systems, where concentration is constantly modified by directional flow Due to a heavy influence from dispersive flow movement gasses filled microbubbles ty pically only display a half li fe i n the region of interest up to a few minutes. This time is further reduced by the rupturing potential of ultrasound waves. One allure of antibody conjugated silica contrast agent is that it is extremely stable in compariso n to liposome enclosed air bubbles, thus eliminating issues of half life. By chemically attaching the silica particles to the region of interest the concentration is further stabilized. Due to the high porosity of mesoporous silica, air bubbles may be tra pped within the core of the nanoparticle. Part of the contrast effect may be a product of reflectivity enhancement that occurs by backscatter from these air bubbles trapped in the pores. With a pore volume ranging from 0.98 cm 3 /g (MCM 41) up to 2.31 cm 3 /g (MSU F), it is possible for air pockets to comprise more than 33% of the bulk volume (MCM 41). However, the air retention with the particles was not observed in this study. Flow cytometry and in vitro ultrasound indicate that the preferential binding of MS N Herceptin particles to the different cell types is significant, but not optimal for cell specific ultrasound contrast enhancement. One proposed theory is the indiscriminate uptake of silica particles by numerous cell types. By increasing the count of Her ceptin targeting arms on the surface, more specific binding may be achieved. Since each Herceptin molecule is conjugated to

PAGE 49

39 the amine group of the APTES precursor at a 1:1 ratio, the density of Herceptin targeting arms on the silica surface is limited by t he geometric area occupied by the APTES base. Assuming that the triethoxysilane base is 3.96 x 10 4 pm 2 and the surface area to mass ratio of MCM 41 silica is 1000 m 2 /g (provided by Sigma), the limit of Herceptin bound to the silica surface is 0.041 moles p er gram of silica. 5.2 Potential Applications Currently, ultrasound is generally used for two purposes in the context of breast cancer detection and treatment: secondary diagnosis and presurgical occult lesion localization (Section proc edure is heavily reliant on its ability to discern tumor boundary regions. In a clinical setting, ultrasound has typically been used as a c omplementary diagnostic device with mammogra phy. The greatest disadvantages of mammography is that it is a radiation emitting modality and that it is less effective than ultrasound for density determination. However, even ultrasound is unable to distinguish between benign and malignant tumors of sim ilar densities. Ultrasound specificity is further diminished if the surrounding tissue is similar in density, which is common for women below 50 years old. By injecting MSN+Herceptin particles in tissues regions suspected of cancerous lesions, contrast enh ancement will only occur in cancerous tumor regions. By being able to visualize selective contrast variations in real time, sonographic images can more effectively distinguish malignant tumors from benign cysts. Ultrasound is also commonly used for needle guidance, both for biopsy sampling and occult lesion localization. If the boundary region of a suspected tumor is difficult to discern, the result may be poor sampling localization or poor marker placement. By improving visualization of the boundary regio ns, biopsy needle placement can be more accurate, or rendered unnecessary altogether. Since wire length for occult lesion localization is determined by the length of the lesion, improved visualization of the boundary regions can ensure that the wire comple tely passes through the entirety of the lesion.

PAGE 50

40 Table 2 Advantages and disadvantages of silica nanoparticles as a contrast agent in comparison to gas phase contrast agents Advantages Disadvantages Better suited for extravasation out of leaky tumor capillaries Lower scattering cross section than microbubbles M ore stable, allowing longer half life in body to take multiple images Size approaches axial resolution of conventional ultrasound transducer frequencies Highly modifiable su rface Potential for reflectivity enhancement Large pores allow for delivery of low solubility drugs US: portable and cheaper than Mammography FDA approved Permanent retention within cell possible (depends on cell type)

PAGE 51

41 6. Conclusion 6.1. Con c l usion Observation of the acoustic properties of 1 m unfunctionalized mesoporous silica nanoparticles (MSNs) revealed that the backscatter coefficient of these particles are sufficient to be considered as an ultrasound enhanced contrast agent (UECA). By c onjugating MSN to the HER2 targetting antibody Herceptin (trastuzumab), local concentrations near tumor sites overexpressing HER2 receptors increased and further improved the backscatter cross section of the MSNs. In vitro studies suggest the binding affin ity of the MSN Herceptin particles is preferential to cancer cells that overexpress the HER2 surface receptor. Due to the large size of the particle conjugate, the particles are not incorporated into the cell by endocytosis. Targeted MSN are ideal for ultr asound contrast enhancement, ultimately improving the reliability of cheaper, more user friendly, and non radiating imaging modality. 6.2 Future Direction 6.2.1 In vivo studies The next objective of the project is to prove contrast enhancement in vivo. Nu de athymic BALB/c female mice will be used for a tumor xenograft model. The breast cancer cell lines used for in vitro studies will be cultured and subcutaneously injected into the right flank of 4 6 wee k old nude mice. After tumor formation up to 65mm 3 t he mice will be given i ntraperitoneal injection s of MSN+Herceptin+FITC. Following injection, B mode images will be taken at the tumor site using a 40MHz transducer probe. Prior to study, LD50 of MSN+Herceptin and the final endpoint of the study must be det ermined. Injection frequency and ultrasound imaging is dependent upon cellular uptake and filtration time. 6.2.2 MSN as a theranostic agent A great deal of research has been dedicated towards the application of MSN as a drug delivery vehicle [30] Howeve r, the innate properties of MSN also allow them to be applied as an agent recently termed as a the ranostic the combination of therapeutics and diagnostics into a single agent. Already MSNs have shown their potential as a therapeutic agent (Section 1.4.2, increasingly become more appealing as a theranostic agent as well.

PAGE 52

42 The current conjugation motif of MSN+Herceptin is already in essence a theranostic agent: the original purpose of Herceptin is not to act as a targeting agent, but rather to act as a treatment for a HER2+ breast cancer types. Further study of cell proliferation under MSN+Herceptin inoculation must be taken to determine how MSN conjugation affec ts the therapeutic efficacy of Herceptin. It is worth noting that the conjugation motif of this study did its therapeutic characteristics. 6.2. 3 Incorporati ng perfluoropentane into the MSN pores A high degree of porosity allows the MSN particle cavity to be loaded with material and used as a carrying agent. Similar to drug loading, MSNs may also be loaded with a gas. The introduction of a gas phase would theo retically enhance the backscatter coefficient in a way similar to microbubbles (solid to gas or liquid to gas interfaces yield large backscatter). The rigidity of the MSN pores may restrict the oscillatory motion typically seen in microbubble contrast agen ts. Alternatively, this restriction of motion will also provide for greater stability and half life. As mentioned previously, extended half life is beneficial for long term tumor tracking or extended imaging time. Like drug loading, gas retention within the particles may be solved by capping. Several groups have applied hard caps (iron oxide Fe 3 O 4 cadmium sulfide CdS, gold Au) as well as soft caps (G Insulin ) [31] Most of these capping agents can be removed by environmental stimuli. Fortunately cap r elease may not be necessary, thus sidestepping a major issue with this field of research. However, cap release under cancer cell stimuli may have an advantage as well. Heavy gasses such as perfluoropentane are used in most clinically approved contrast agen ts due to their low partition coefficient in blood. By incorporating a gas phase into MSNs, backscatter may be drastically increased. 6.2.4 Dye loading for presurgical occult lesion localization Currently, wire placement is the standard for tumor marking prior to surgery. This method is inefficient, since only one dimension of the tumor is marked for surgical removal. Consequentially, an excessive amount of tissue around the lesion must be removed as a precautionary measure By functionalizing the surface or loading the MSN pores with a slow release dye, targeted cells can be easily stained. Upon scission of the breast during surgery,

PAGE 53

43 stained cancer cells can easily be visualized and removed. Dyes such as cyanine can be easily excited under a blacklight lam p (low energy UVA).

PAGE 54

44 APPENDIX A. MATLAB code: Single Pulse Processing % 1D Signal Processing Protocol % By: Andrew Milgroom % Translational Biomaterials Research Labs % University of Colorado Anschutz Medical Campus % Date: 09/17/2012 % This protol des cribes the attenuation of an ultrasound signal by % deconvoluting the image with a signal through pure water through the same % distance. % Set variables, figure dimensions, and get screen size signal = []; signalAUC = 0; signalDC = []; water = []; fx = 1000; fy = 600; Size = get(0, 'screensize' ); % Create the figure and center it on the screen handles.hfig2 = figure( 'NumberTitle' 'off' ... 'Position' ,[floor((Size(3) fx)/2) ... floor((Size(4) fy)/2) ... fx fy], ... 'Name' 'Single Pulse Processing GUI' ... 'Color' ,[1 1 1], ... 'PaperPositionMode' 'auto' ... 'InvertHardcopy' 'off' ... 'DoubleBuffer' 'On' ... 'MenuBar' 'none' ); % Create file menu, graph window, data box and buttons menu = uimenu(handles.hfig2, 'Label' 'File' ); uimenu(menu, 'Label' 'Load Image...' 'Callback' '[signal water] = loadsignal(signal,water)' ); uimenu(menu, 'Lab el' 'Close' 'Callback' 'close' ); aucbox = uicontrol(handles.hfig2, 'Style' 'edit' 'Units' 'Normalized' 'Position' ... [.05 .35 .2 .07], 'Tag' 'aucbox' ); auctitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [ .05 0.43 .2 .03], 'String' 'Cumulative Voltage Difference' ); grptitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [.55 0.53 .15 .03], 'String' 'Signals' ); sigtitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Nor malized' 'Position' ... [.1 0.96 .15 .03], 'String' 'Signal' ); watertitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [.4 0.96 .15 .03], 'String' 'Water' ); dctitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [.7 0.96 .15 .03], 'String' 'Deconvoluted signal' ); denoisebtn = uicontrol(handles.hfig2, 'Style' 'PushButton' 'Units' 'Normalized' 'Position' ... [.07 .26 .15 .05], 'String' 'Denoise and Smooth' 'C allback' '[signal water] = denoise(signal water);' ); analysisbtn = uicontrol(handles.hfig2, 'Style' 'PushButton' 'Units' 'Normalized' 'Position' ... [.07 .19 .15 .05], 'String' 'Run Analysis' 'Callback' 'signalAUC signal DC = analyze(signal water signalAUC signalDC);' );

PAGE 55

45 savebtn = uicontrol(handles.hfig2, 'Style' 'PushButton' 'Units' 'Normalized' 'Position' ... [.07 .12 .15 .05], 'String' 'Save to file' 'Callback' 'Data = save(signal signalAUC signalDC);' ); handles.mainaxes = axes( 'Parent' ,handles.hfig2); axes(handles.mainaxes); set(gca, 'Title' ,text( 'String' '' )); set(handles.mainaxes, 'YTickLabel' ,[]); set(handles.mainaxes, 'FontSize' ,8, ... 'Color' ,[1 1 1], ... 'Box' 'On' ... 'YColor' ,[0 0 0], ... 'XColor' ,[0 0 0], ... 'XTickLabel' ,[], ... 'Units' 'Normalized' ... 'Position' ,[.3 .12 .6 .4]); handles.signalaxes = axes( 'Parent' ,handles.hfig2); axes(handles.signalaxes); set(gca, 'Title' ,text( 'String' '' )); set(handles.signalaxes, 'YTickLabel' ,[]); set(handles.signalaxes, 'FontSize' ,8, ... 'Color' ,[1 1 1], ... 'Box' 'On' ... 'YCol or' ,[0 0 0], ... 'XColor' ,[0 0 0], ... 'XTickLabel' ,[], ... 'Units' 'Normalized' ... 'Position' ,[.05 .6 .25 .35]); handles.wateraxes = axes( 'Parent' ,handles.hfig2); axes(handles.wate raxes); set(gca, 'Title' ,text( 'String' '' )); set(handles.wateraxes, 'YTickLabel' ,[]); set(handles.wateraxes, 'FontSize' ,8, ... 'Color' ,[1 1 1], ... 'Box' 'On' ... 'YColor' ,[0 0 0], ... 'XColor' ,[0 0 0], ... 'XTickLabel' ,[], ... 'Units' 'Normalized' ... 'Position' ,[.35 .6 .25 .35]); handles.dcaxes = axes( 'Parent' ,handles.hfig2); axes(handles.dcaxes); set(gca, 'Title' ,text( 'String' '' )); set(han dles.dcaxes, 'YTickLabel' ,[]); set(handles.dcaxes, 'FontSize' ,8, ... 'Color' ,[1 1 1], ... 'Box' 'On' ... 'YColor' ,[0 0 0], ... 'XColor' ,[0 0 0], ... 'XTickLabel' ,[], ... 'Units' 'Normalized' ... 'Position' ,[.65 .6 .25 .35]); function [signal water] = loadsignal(signal,water) % This function loads the signal from the user, and grabs the water file % from a previously described path function [signal wa ter] = denoise(signal water) % This function denoises AND smooths the signals function signalAUC signalDC = analyze(signal water signalAUC signalDC % This function analyzes the smoothed and denoised signal by deconvoluting % it and looking at the area di fference between water and the signal. The % the display for 'DC' will appear

PAGE 56

46 function Data = save(signal signalAUC signalDC) % This function saves the original signal, the cumulative difference (AUC) % and the deconvoluted signal. The data is added to a previously described % workspace

PAGE 57

47 B. MATLAB code: B mode image processing % B Mode Image Processing Protocol % By: Andrew Milgroom % Translational Biomaterials Research Labs % University of Colorado Anschutz Medical Campus % Date: 08/01/2012 % This p rotocol imports a grayscale image of a B Mode ultrasound scan from % selecte filepath. The image automatically detects the boundary of the % region of interest (ROI). The region of interest is parsed into 50 pixel % x 50 pixel squares. For each square, m ean pixel intensity distribution % (w/st. dev.) and 2D Fourier transform is acquired. All squares are then % averaged and saved to a predefined workspace, where it can be compared % with other data % Set variables, figure dimensions, and get screen siz e Bscan = []; tempdata = []; fx = 800; fy = 600; Size = get(0, 'screensize' ); % Create the figure and center it on the screen handles.hfig2 = figure( 'NumberTitle' 'off' ... 'Position' ,[floor((Size(3) fx)/2) ... floor((Size(4) fy)/2) ... fx fy], ... 'Name' 'B Mode Image Processing GUI' ... 'Color' ,[1 1 1], ... 'PaperPositionMode' 'auto' ... 'InvertHardcopy' 'off' ... 'Do ubleBuffer' 'On' ... 'MenuBar' 'none' ); % Create file menu, graph window, data box and buttons menu = uimenu(handles.hfig2, 'Label' 'File' ); uimenu(menu, 'Label' 'Load Image...' 'Callback' 'Bscan = loadimage(Bscan)' ); u imenu(menu, 'Label' 'Close' 'Callback' 'close' ); mgsbox = uicontrol(handles.hfig2, 'Style' 'edit' 'Units' 'Normalized' 'Position' ... [.05 .85 .2 .07], 'Tag' 'mgsbox' ); stdbox = uicontrol(handles.hfig2, 'Style' 'edit' 'Units' 'Normalized' 'Position' ... [.05 .70 .2 .07], 'Tag' 'stdbox' ); mgstitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [.08 0.93 .15 .03], 'String' 'Mean Gray Scale' ); stdtitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normalized' 'Position' ... [.08 0.78 .15 .03], 'String' 'Standard Deviation' ); bscantitle = uicontrol(handles.hfig2, 'Style' 'text' 'Units' 'Normaliz ed' 'Position' ... [.55 0.93 .15 .03], 'String' 'B Mode Image' ); analysisbtn = uicontrol(handles.hfig2, 'Style' 'PushButton' 'Units' 'Normalized' ... 'Position' ,[.07 .19 .1 5 .05], 'String' 'Run Analysis' ... 'Callback' 'tempdata = analyze(Mtrace,Atrace,Bscan,tempdata);' ); savebtn = uicontrol(handles.hfig2, 'Style' 'PushButton' 'Units' 'Normalized' 'Position' ... [.07 .12 .15 .05], 'String' 'Save to file' 'Callback' 'Data = save(Data,tempdata);' ); handles.mainaxes = axes( 'Parent' ,handles.hfig2); axes(handles.mainaxes); set(gca, 'Title' ,text( 'String' '' ));

PAGE 58

48 set(handles.mainaxes, 'YTickLabel' []); set(handles.mainaxes, 'FontSize' ,8, ... 'Color' ,[1 1 1], ... 'Box' 'On' ... 'YColor' ,[0 0 0], ... 'XColor' ,[0 0 0], ... 'XTickLabel' ,[], ... 'Units' 'Normalized' ... 'Position' ,[.3 .12 .6 .8]); function Bscan = loadimage(Bscan) % Loads image from JPG or TIFF file, saves image as matrix, 'image', and % displays it on the screen Bscan = uigetfile imshow(Bscan) function tempdata = analyze(Mtrace ,Atrace,Bscan,tempdata) % Finds mean grayscale pixel intensity, its distribution, as well as the % fourier transform of the image (currently unused). Avoids bias by % sampling the ROI in 50 pixel x 50 pixel boxes then averages the values. % Total box count varies. function Data = save(Data, tempdata) % Adds new image data to the previous workspace

PAGE 59

49 C. Herceptin Protein sequence Human trastuzumab peptide sequence >Anti HER2 Light chain 1 DIQMTQSPSSLSASVGDRVTITCRASQDVNTAVAWYQQKPGKAPKLLIYSASFLYSGVPS RF SGSRSGTDFTLTISSLQPEDFATYYCQQHYTTPPTFGQGTKVEIKRTVAAPSVFIFPP SDEQLKSGTASVVCLLNNFYPREAKVQWKVDNALQSGNSQESVTEQDSKDSTYSLSSTLT LSKADYEKHKVYACEVTHQGLSSPVTKSFNRGEC >Anti HER2 Heavy chain 1 EVQLVESGGGLVQPGGSLRLSCAASGFNIKDTYIHWVRQAPGKGLEWVARIYPTNGYTRY ADSVKGRFTISADT SKNTAYLQMNSLRAEDTAVYYCSRWGGDGFYAMDYWGQGTLVTVSS ASTKGPSVFPLAPSSKSTSGGTAALGCLVKDYFPEPVTVSWNSGALTSGVHTFPAVLQSS GLYSLSSVVTVPSSSLGTQTYICNVNHKPSNTKVDKKVEPPKSCDKTHTCPPCPAPELLG GPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQY NSTYRVVSVLTVLHQDWLNGKEYKCK VSNKALPAPIEKTISKAKGQPREPQVYTLPPSRD ELTKNQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSR WQQGNVFSCSVMHEALHNHYTQKSLSLSPGK >Anti HER2 Light chain 2 DIQMTQSPSSLSASVGDRVTITCRASQDVNTAVAWYQQKPGKAPKLLIYSASFLYSGVPS RFSGSRSGTDFTLTISSLQPEDFATYYCQQHYTTPPTFGQG TKVEIKRTVAAPSVFIFPP SDEQLKSGTASVVCLLNNFYPREAKVQWKVDNALQSGNSQESVTEQDSKDSTYSLSSTLT LSKADYEKHKVYACEVTHQGLSSPVTKSFNRGEC >Anti HER2 Heavy chain 2 EVQLVESGGGLVQPGGSLRLSCAASGFNIKDTYIHWVRQAPGKGLEWVARIYPTNGYTRY ADSVKGRFTISADTSKNTAYLQMNSLRAEDTAVYYCSRWGGDGFYAMDYWGQG TLVTVSS ASTKGPSVFPLAPSSKSTSGGTAALGCLVKDYFPEPVTVSWNSGALTSGVHTFPAVLQSS GLYSLSSVVTVPSSSLGTQTYICNVNHKPSNTKVDKKVEPPKSCDKTHTCPPCPAPELLG GPSVFLFPPKPKDTLMISRTPEVTCVVVDVSHEDPEVKFNWYVDGVEVHNAKTKPREEQY NSTYRVVSVLTVLHQDWLNGKEYKCKVSNKALPAPIEKTISKAKGQPREPQVYTLPPSRD ELTK NQVSLTCLVKGFYPSDIAVEWESNGQPENNYKTTPPVLDSDGSFFLYSKLTVDKSR WQQGNVFSCSVMHEALHNHYTQKSLSLSPGK Total AA count: 1,330 Molecular Weight 145531.5 g/mo l Lysine: 90/1,330 (6.77%) Aspartic Acid: 58/1,330 (4.36%) Glumatic Acid: 60/1,330 (4.51%) Amines (plus termini): 7.07%, 10,285.7 g/mol Carboxylates (plus termini): 9.17% 13,349.5 g/mol

PAGE 60

50 D. UV VIS spectroscopy Cells were incubated overnight at 37 o C in 0.125 mg/ml MSN Herceptin. Afterwards the cells were washed, trypsinized, and sieved with a 10 m nylon mesh. The samp les were treated with 10 l of 12mM MTT solution for 30 minutes at room temperature. Absorbance at 540nm was measured (top figure). The graph below suggests that not only does the MSN Herceptin induce cell proliferation with respect to HER2 expression, but may increase proliferation beyond the untreated samples. A more plausible explanation would be that the MSN has a stronger binding affinity to HER2+ cells. As MSN concentration increases, more light is absorbed in the well. Theoretically, retention of 0.0 25 mg in a 100 l volume could increase absorbance by ~0.07 (bottom figure). 0.000 0.050 0.100 0.150 0.200 0.250 0.300 0.350 231 skbr3 skbr3 pool A(540) MSN-HERC treatment untreated 0 0.1 0.2 0.3 0.4 0.5 0.6 2 1 0.5 0.25 0.1 0 A(540) Concentration (mg/mL) MSN Herceptin MSN-Herc

PAGE 61

51 E. FITC and EthD 1 excitation and emission spectra Excitation and emission spectra of 7 aminoactinomycin D 546 647

PAGE 62

52 REFERENCES 1. Al Douri, Y., Ahmed, N.M., Bouarissa, N. & Bouh emadou, a. Investigated optical and elastic properties of Porous silicon: Theoretical study. Materials & Design 32 4088 4093 (2011). 2. Alunni, J. Imaging inflammatory breast cancer. Diagnostic and Interventional Imaging 93 96 104 (2012). 3. Alunni, J. P. Imag ing inflammatory breast cancer. Diagnostic and interventional imaging 93 95 103 (2012). 4. Anderson, C.R. et al. Ultrasound molecular imaging of tumor angiogenesis with an integrin targeted microbubble contrast agent. Investigative radiology 46 215 24 (2011 ). 5. Casciaro, S. et al. Optimal enhancement configuration of silica nanoparticles for ultrasound imaging and automatic detection at conventional diagnostic 1 frequencies. Investigative radiology 45 715 24 (2010). 6. Chung, T. H. et al. The effect of surface ch arge on the uptake and biological function of mesoporous silica nanoparticles in 3T3 L1 cells and human mesenchymal stem cells. Biomaterials 28 2959 66 (2007). 7. Corsetti, V. et al. Breast screening with ultrasound in women with mammography negative dense b reasts: evidence on incremental cancer detection and false positives, and associated cost. 1990) 44 539 44 (2008). 8. Deshpande, N., Needles, a & Willmann, J.K. Molecular ultrasound imaging: current status and fu ture directions. Clinical radiology 65 567 81 (2010). 9. Fan, H. et al. Modulus density scaling behaviour and framework architecture of nanoporous self assembled silicas. Nature materials 6 418 23 (2007). 10. Friedrich, J., Seidel, C., Ebner, R. & Kunz Schughar t, L. a Spheroid based drug screen: considerations and practical approach. Nature protocols 4 309 24 (2009). 11. Gao, Z. G., Fain, H.D. & Rapoport, N. Controlled and targeted tumor chemotherapy by micellar encapsulated drug and ultrasound. Journal of controll ed 102 203 22 (2005). 12. Gary Bobo, M. et al. Cancer therapy improvement with mesoporous silica nanoparticles combining targeting, drug delivery and PDT. International journal of pharmaceutics 423 509 15 (2012). 13. Goldberg, B.B., Liu, J.B. & Forsberg, F. Ultrasound contrast agents: a review. Ultrasound in medicine & biology 20 319 33 (1994).

PAGE 63

53 14. Contribution PROPOSED STANDARD T HERMAL TEST OBJECT FOR MEDICAL. 25 121 132 (1999). 15. Hocine, O. et al. Silicalites and Mesoporous Silica Nanoparticles for photodynamic therapy. International journal of pharmaceutics 402 221 30 (2010). 16. Hocine, O. et al. Silicalites and Mesoporous Silica N anoparticles for photodynamic therapy. International journal of pharmaceutics 402 221 30 (2010). 17. ULTRASOUND BIOMICROSCOPY. 26 1 27 (2000). 18. Jeong, J.S., Cannata, J.M. & Shung, K.K. Adaptive HIFU noise cancellation for simultaneous therapy and imaging using an integrated HIFU/imaging transdu cer. Physics in medicine and biology 55 1889 902 (2010). 19. Kim, J. Investigation of the Formation and Structure of APTES Films on Silicon Substrates. 14222 1 2 20. Koole, R. et al. Paramagnetic lipid coated silica nanoparticles with a fluorescent quantum dot c ore: a new contrast agent platform for multimodality imaging. Bioconjugate chemistry 19 2471 9 (2008). 21. Lavisse, S., Adotevi, C., Opolon, P Agent Application. 40 536 544 (2005). 22. Lin, Y. S. et al. Well Ordered Mesoporous Silica Nanoparticles as Cell Markers. Chemistry of Materials 17 4570 4573 (2005). 23. L iong, M. et al. Multifunctional inorganic nanoparticles for imaging, targeting, and drug delivery. ACS nano 2 889 96 (2008). 24. Liu, J. et al. Nanoparticles as image enhancing agents for ultrasonography. Physics in medicine and biology 51 2179 89 (2006). 25. Li u, Y., Mi, Y., Zhao, J. & Feng, S. S. Multifunctional silica nanoparticles for targeted delivery of hydrophobic imaging and therapeutic agents. International journal of pharmaceutics 421 370 8 (2011). 26. Neill, T.P.O. & Winkler, A.J. OOriginal Contribution. 20 579 588 (1994). 27. Poincloux, R. et al. Contractility of the cell rear drives invasion of breast tumor cells in 3D Matrigel. Proceedings of the National Academy of Sciences of the United States of America 108 1943 8 (2011).

PAGE 64

54 28. Saunders, a E. Heating of bone and soft tissue by ultrasound. Imaging Science Journal, The 55 52 56 (2007). 29. Search, H., Journals, C., Contact, A., Iopscience, M. & Address, I.P. Equivalence between three scattering formulations for ultrasonic wave propagation in particulate mixtures. 3481 (1998). 30. Speed, C. a Therapeutic ultrasound in soft tissue lesions. Rheumatology (Oxford, England) 40 1331 6 (2001). 31. Tang, H. et al. Facile synthesis of pH sensitive polymer coated mesoporous silica nanoparticles and their application in drug deliver y. International Journal of Pharmaceutics 421 388 396 (2011). 32. Tang, H. et al. Facile synthesis of pH sensitive polymer coated mesoporous silica nanoparticles and their application in drug delivery. International journal of pharmaceutics 421 388 96 (2011) 33. Umchid, S. FREQUENCY DEPENDENT ULTRASONIC ATTENUATION COEFFICIENT. 234 238 (2008). 34. Zou, M. & Yang, D. Nanoindentation of silica nanoparticles attached to a silicon substrate. Tribology Letters 22 189 196 (2006). 35. ULTRASONIC MONITORING OF PARTICULATE SUSP ENSIONS IN PROCESS: PHYSICS, TECHNOLOGY AND APPLICATIONS R. E. Challis, A. K. Holmes and A. N. Kalashnikov University of Nottingham, United Kingdom. 36. Mohsine, a. & El Hami, a. A robust study of reliability based optimization methods under eigen frequency. Computer Methods in Applied Mechanics and Engineering 199 1006 1018 (2010). 37. Trewyn, B.G., Nieweg, J. a., Zhao, Y. & Lin, V.S. Y. Biocompatible mesoporous silica nanoparticles with different morphologies for animal cell membrane penetration. Chemical Engi neering Journal 137 23 29 (2008). 38. backscatter coefficient quantitative estimates from high concentration Chinese Hamster Ovary cell pellet biophantoms. The Journal of the Acoustic al Society of America 130 4139 47 (2011). 39. Ophir, J. & Parker, K.J. OOriginal Contribution. 15 319 333 (1989). 40. Laugier, P., Droin, P., Laval Jeantet, a M. & Berger, G. In vitro assessment of the relationship between acoustic properties and bone mass den sity of the calcaneus by comparison of ultrasound parametric imaging and quantitative computed tomography. Bone 20 157 65 (1997).

PAGE 65

55 41. Liu, J., Li, J., Rosol, T.J., Pan, X. & Voorhees, J.L. Biodegradable nanoparticles for targeted ultrasound imaging of breast cancer cells in vitro. Physics in medicine and biology 52 4739 47 (2007). 42. Zhao, Y., Vivero Escoto, J.L., Slowing, I.I., Trewyn, B.G. & Lin, V.S. Y. Capped mesoporous silica nanoparticles as stimuli responsive controlled release systems for intracellular drug/gene delivery. Expert opinion on drug delivery 7 1013 29 (2010). 43. Liong, M. et al. Multifunctional inorganic nanoparticles for imaging, targeting, and drug delivery. ACS nano 2 889 96 (2008). 44. Nielsen, U.B. et al. Therapeutic efficacy of anti ErbB2 immunoliposomes targeted by a phage antibody selected for cellular endocytosis. Biochimica et biophysica acta 1591 109 118 (2002). 45. Liberman, A. et al. Hollow silica and silica boron nano/microparticles for contrast enhanced ultrasound to detect small tum ors. Biomaterials 33 5124 9 (2012). 46. Arvazyan, A.R.P.S., Udenko, O.L.E.G.V.R., Wanson, S.C.D.S. & Owlkes, J.B.R.F. ULTRASONIC TECHNOLOGY OF MEDICAL DIAGNOSTICS. 24 1419 1435 (1998). 47. Republic, C. Ultrasound attenuation imaging. 55 180 187 (2004).