The Characterization of a Mouse Model of
Transient Stroke Using ex vivo MR Microscopy
and in vivo MR Imaging
by
Shuning Huang
Submitted to the Harvard-MIT Division of Health Sciences and
Technology
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Health Sciences and Technology
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2009
@ Massachusetts Institute of Technology 2009. All rights reserved.
. .............
A uthor ........
Harvard-MIT Division of
e
............
h Sciences and Technology
Jan 27, 2009
Certified by....
Bruce R. Rosen, M.D., Ph.D.
Professor of Radiology and Health Sciences & Technology
Harvard-MIT Division of Sciences & Technology
Director, HST Marfrps Center for Biomedical Imaging, MGH
Thesis Supervisor
Certified by...
Young Ro Kim, Ph.D.
Instructor in Radiology at HMS
Apistant in Neuroimaging at MGH/Radiology
Thesis Supervisor
Acrrte-d hx
A
Lee Gehrke, PhD
Hermann von Helmholtz Professor of Health Sciences and Technology
Interim Director
Harvard-MIT Division of Health Sciences and Technology
MASSACHIUSETTS INSTIlTE
OF TECHNOLOGY
APR 1 5 2009
ARCHIVES
The Characterization of a Mouse Model of Transient Stroke
Using ex vivo MR Microscopy and in vivo MR Imaging
by
Shuning Huang
Submitted to the Harvard-MIT Division of Health Sciences and Technology
on Jan 27, 2009, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy in Health Sciences and Technology
Abstract
Disrupted blood-brain barrier after an ischemic attack can cause vasogenic edema
and increase the risk of hemorrhagic transformation. Therefore, early detection and
monitoring of BBB damage is important in the pathological understanding and therapeutic treatment of stroke. Currently, MR contrast agents have been widely used in
clinics for disease diagnosis and treatment evaluation, and in basic research to achieve
better anatomical structure visualization and to understand pathological mechanisms
of various human diseases in animal models. Thus, the central theme of this thesis to
exploit the use of MR contrast agents in the study of ischemic stroke using both in vivo
and ex vivo MR techniques. Specifically, the overall goals of this thesis are twofold: (1)
to exploit the multiple relaxation mechanisms and varying tissue-dependent affinities
of different MR contrast agents for better structure delineation, tissue differentiation,
and image contrast manipulation in magnetic resonance microscopy (MRM) staining,
and (2) to develop an MRI technique that employs intrinsic water as a biomarker for
qualitative and quantitative monitoring of blood-brain barrier (BBB) integrity alteration in a mouse model of stroke using an intravascular long-circulating MRI contrast
agent.
Despite the great success of MRM in anatomical studies, MRM images based on
intrinsic tissue contrast lack the flexibility and target-specificity offered by conventional histological staining. Therefore, the first focus of this thesis was on the development of MRM staining method by utilizing the different tissue relaxation ability
and tissue biophysical/biochemical properties of different MR contrast agents. Two
common MR contrast agents, Gd-DTPA and MnCl 2 were used in this thesis. The
ability of MR contrast agents to increase SNR and enhance image contrast was first
tested in a relatively simple in vitro glioma spheroid (diameter - 400 um) system.
We then fully characterized the relaxation mechanisms and tissue-dependent staining
properties of these contrast agents in the brain tissue, and demonstrated that their
unique relaxation and tissue properties led to differentiated MR staining in the ex
vivo mouse brains, which greatly enhanced the ability of MRM to delineate tissue
structures in addition to providing improved SNR. This MRM staining method was
then applied to the Kif2la knockout mouse model for the anatomical phenotyping of
the new born Kif2la knockout mice.
The BBB damage is usually detected through the spatial leakage profiles of extrinsically administrated markers such as staining dyes, fluoresceins, radiolabeled compounds, or gadolinium based compound, which are only possible when BBB is compromised to the extent that allows extravasation of these markers. It is therefore
desirable to develop a technique that allows the early detection of BBB damage. In
the second part of thesis, we first presented the theoretical background of measuring transvascular water exchange based on a two-compartment water exchange model.
Parameters affecting the quantitative BBB water exchange measurement were initially
characterized using computer simulations. We then performed graded hypercapnia
and Mannitol-induced BBB-opening experiments to test the ability of this novel MRI
technique to detect and monitor the changes of BBB integrity and cerebral blood
volume (CBV). Upon the characterization of this MRI technique, we measured baseline BBB water exchange and other MRI-derived cerebrovascular parameters in the
eNOS knockout mice, and showed that there is basal increase of transvascular water exchange in addition to the morphological changes in the vasculature of eNOS
knockout mice.
After developing and characterizing these ex vivo and in vivo MR techniques, we
applied the in vivo MRI BBB water exchange detection technique and the ex vivo
MRM staining method to a mouse model of transient stroke. We demonstrated the
importance of CBV restoration in the BBB integrity change at acute stage after reperfusion, and showed that MRM staining may have a great potential in histopathological
studies of ischemic brain injury.
Thesis Supervisor: Bruce R. Rosen, M.D., Ph.D.
Title: Professor of Radiology and Health Sciences & Technology
Harvard-MIT Division of Sciences & Technology
Director, HST Martinos Center for Biomedical Imaging, MGH
Thesis Supervisor: Young Ro Kim, Ph.D.
Title: Instructor in Radiology at HMS
Assistant in Neuroimaging at MGH/Radiology
Acknowledgments
I would like to express my deepest thanks to my advisor and mentor, Professor Bruce
R. Rosen, for his unlimited support, inspiration and encouragement in the past five
years. His great passion and deep insight on scientific research are invaluable to me.
More importantly, his great support as a mentor guided me through the years at MIT
in science and life. Hopefully his ability to identify the essence of scientific problems
will pass on me.
I would like to thank Dr. Young Ro. Kim, who is my thesis co-advisor, for his
endless patience, outstanding instruction, and uncompromisingly high academic standard for which he motivated me to work on this interesting thesis project. Without
the countless discussions with him and the help from him, I could not go this far
successfully.
I have also benefited a lot from the members of my thesis committee, Professor
Alan Jasanoff and Dr. Joseph B. Mandeville. Their invaluable suggestions and deep
insights lead my work to a next higher level after each meeting and discussion.
Many thanks to my collaborators at the Martinos Center and MGH for their help
and support that made my graduate experience a pleasant one. Dr. George Dai,
Dr. Christina Liu, and Philip Liu at the Martinos Center for Biomedical Imaing,
Dr. Dmitriy Atochin, Dr. Hyung-Hwan Kim, Seo Kyoung Hwang, Brian Lee at
the Massachusetts General Hospital provided so much help and insightful discussions
about about my thesis projects.
Last, but by no means least, I would like to express my heartfelt appreciation to
my parents, my husband Xiaofeng, and our lovely daughter Rosanna, for their deepest
and everlasting love, care, and support.
Contents
19
1 Introduction
2
19
.
1.1
MR contrast agents .................................
1.2
Relaxation mechanisms of paramagnetic MR contrast agents in solution 22
22
. . . ... .
1.2.1
Introduction . ....................
1.2.2
Proton paramagnetic relaxation due to dipole-dipole interaction
1.2.3
Proton paramagnetic relaxation due to scalar (contact) coupling 23
1.2.4
Proton paramagnetic relaxation due to Curie relaxation . ...
1.2.5
Comparison of the three relaxation mechanisms
24
25
........
1.3
MR contrast agents in magnetic resonance microscopy .........
1.4
Detection of BBB damage due to stroke using MR contrast agent
28
29
.
31
MR Microscopy of glioma spheroid in collagen I matrix
2.1
2.2
..............
Introduction . ..........
23
. . . . ...
31
. . . ....
33
2.1.1
Methods . ......................
2.1.2
Multicellular tumor spheroid (MTS) and extracellular matrix .
33
2.1.3
Magnetic resonance microscopy . ..........
.
34
2.1.4
Histology
Results.
...........
. ..........
.......
..................
2.2.1
Magnetic resonance microscopy of MTS in collagen I gel
2.2.2
Segmentation and 3D visualization of MTS . ..........
2.2.3
Histopathology
......
.
35
. . . ... .
....
36
. . .
36
37
.
...................
37
2.3
Discussion . .........................
. . . ... .
41
2.4
Conclusions . ........................
. . . ... .
43
3 MRM staining of ex vivo mouse brain at 14T
3.1
The development of MRM staining using Gd-DTPA and MnC12 in the
wild type C57BL/6 mice .........................
3.2
4
3.1.1
Introduction . . . . . . . . . . . . . . ..
3.1.2
.
. . .
.
46
Materials and methods ...........
. . .
.
48
3.1.3
Results .. .. . . .. . . . . . . . . . . . . . .
. . .. .
52
3.1.4
D iscussion . . . . . . . . . . . . . . . . . .
....
MRM of kif2la knockout new born mouse brains
3.2.1
introduction . .
3.2.2
Material and methods
3.2.3
Results and discussion ...........
................
.
..........
.
59
. . . . . . . . .
69
. .
.
69
. . .
.
69
. . .
.
70
MRI detection of BBB water exchange: theoretical background
73
4.1
Introduction . . .
4.2
Two-compartment water exchange model and BBB water exchange
measurement
4.3
4.4
......................
. .
.....................
.
73
. . .
.
75
. . .
4.2.1
Two-compartment tissue model . . . . . . ..
.
75
4.2.2
Method for water exchange measurement . . . . . . . . . . . .
78
Parameters affecting WEI .
..............
. . .
.
80
. . .
.
82
4.3.1
Effect of imaging related parameters
4.3.2
Effect of contrast agent dose . .........
. . .
.
85
4.3.3
Summary . .
. . .
.
85
. . . . .
...................
Three-compartment model: effect of transcellular water exchange
. .
4.4.1
Three-compartment tissue model .
4.4.2
Effect of transvascular and transcellular water exchange on CBV
and WEI measurements
4.4.3
4.5
. . .
.
. .. .
.....
. . . ...
................
Effect of intravascular volume on WEI measurement . . . . . .
Summary
.................................
86
86
88
90
91
5 The in vivo characterization of the proposed MRI technique and its
application in the eNOS knockout mouse model
93
5.1
In vivo characterization using Mannitol administration and graded hy.................................
percapnia
5.2
5.1.1
Introduction . . . . . . . ..
5.1.2
Materials and method
5.1.3
Results .
5.1.4
Discussion . .........
93
94
. . .
98
...........
103
The change of baseline MRI-derived vascular parameters
knockout mouse model . . . . . . .
Introduction ..
5.2.2
Materials and methods . . .
5.2.3
Results ............
5.2.4
Discussion
OS
. . . . . 106
.
. ......
5.2.1
the
.
.
. . . . . 106
. . . . . 108
. . . . . 110
. . . .
. 115
6 In vivo BBB water exchange measurement and ex vivo MRM stain117
ing of a mouse model of transient ischemic stroke
6.1
The detection of BBB water permeability change in a mouse model of
. ....
. ....
. . ....
. . 1 17
transient ischemic stroke . ...
6.2
6.1.1
Introduction .......
. . . . .. . 117
6.1.2
Material and Method . .
. . . . .
. 118
6.1.3
Results . . . . . . . ...
. . . . .
. 120
6.1.4
Discussion . . . . . . ..
. . . . .
. 123
Ex vivo MRM of ischemic injured mouse brains toward MRM histopathological staining
. . . . . . ...
. . . . .
. 125
6.2.1
Introduction . . . . . . .
. . . . .
. 125
6.2.2
Material and Methods
.
. . . . .
. 126
6.2.3
Results . . . . . . . ...
. . . . .
. 128
6.2.4
Discussion ..
. . . . .
. 138
......
.
10
List of Figures
1-1
Effect of different relaxation mechanisms on proton paramagnetic relaxation (figures are adapted from chapter 3 in [14]).
2-1
........
.
27
The images to the left depict the axial view, whereas those to the
right show an arbitrary angle. These 2D images confirm the surface
heterogeneity that emerges already 12 hrs post placement of the MTS
into the gel, and depict a small group of invasive cells (images on the
right) that can also be seen in Figures 2-2a and 2-3. On the other
hand, images on the left demonstrate that certain parts of the solid
spheroid actually grew into the gel (red circle). . ...........
2-2
.
38
Volumetric rendering and growth dynamics of the MTS in the collagen
I matrix at four consecutive time points. (a) The segmented MTS is
reconstructed in 3D and the color-coding indicates the growth increase
at each time point. The MTS appears to grow anistropic, which may
indicate regional heterogeneities in composition of either MTS or microenvironment, or both, and/or point towards a heterogeneous local
interaction between cells and gel. (b) MTS volume and surface area
are calculated and plotted over time. . ..................
39
2-3
Histopathology findings, comparing 14T MTS with control. (a). H&E
staining of 14T specimen (original magnification x100) showing only
a small number of cells infiltrating the gel. Insert shows infiltrating
cells in higher magnification (original magnification x400). (b). MIB-1
immunostaining of 14T specimen (original magnification x200), highlighting cells that are not in GO as dark brown nuclei.
(c).
H&E
staining of control (original magnification x100) showing a large number of infiltrating cells in adjacent gel. Insert shows the infiltrating
cells in higher magnification (original magnification x400). (e). MIB-1
immunostaining of control (original magnification x200), highlighting
cells that are not in GO as dark brown nuclei. ......
3-1
. ...
.....
40
CNR vs. TE of mouse brains stained with Gd-DTPA or MnC12 at
different concentrations based on 2D MGE sequence. Top panel: brains
in 1, 2.5, 5, or 7.5 mM Gd-DTPA, left: cortex v.s. thalamus, right:
cortex v.s. striatum; bottom panel: brains in 0.12, 0.24, or 0.36 mM
MnC12 showing image contrast between cortex and thalamus on the
left, cortex and striatum on the right. . ..................
3-2
56
Coronal sections of mouse brains stained with 5 mM Gd-DTPA (a) or
0.24 mM MnC12 (b). Images were acquired using 3D FLASH sequence
with different flip angles (from left to right, 100, 250, 550, 700, and 90')
and TEs (top: TE = 4 ms, bottom: TE = 8 ms).
3-3
. ...
. ...
.....
57
Changes in tissue T 1 relaxation time over time for mouse brains stained
with 5 mM Gd-DTPA and 0.24 mM MnCl 2 . These data demonstrate
that at least three days are needed for the contrast agent to fully penetrate brain tissue. To ensure good tissue staining, approximately five
days are needed before imaging.
. ...
.........
..
. . .
58
3-4 high-resolution MRM of ex vivo mouse brain stained with 5 mM GdDTPA2 (a) or 0.36 mM MnCi (b) reveals detailed anatomical structure
with different image contrast in certain regions brain, e.g., cerebellum,
cortex, and hippocampus.
3-5
...................
.....
60
high-resolution MRM of ex vivo mouse brain stained with 0.24 mM
MnC12 (a) or 5 mM Gd- DTPA (b) reveals detailed anatomical structure with different image contrast in certain regions brain, e.g., cerebellum, cortex, and hippocampus. .................
3-6
. . .
61
Changes in R 1, and R2 * values in cerebellar molecular and granular
layers of mouse brain after stained with 0.36 mM MnCl 2 . The higher
ARI and AR 2 * in granular layer indicate that Mn 2+ is possibly compartmentalized into neuronal cells in this layer.
3-7
. ............
65
Images of ex vivo mouse brains fixed without (a) and with (b) Mn 2+
in 4% PFA during transcardial perfusion. Image contrast is not significantly affected by the different staining protocols. . ........
3-8
. .
66
Regions in the cerebellum of brains stained with MnC12 or Gd-DTPA
showed correlation with gene expression patterns of Ca 2+ binding protein calretinin (calb2) in cerebellum and lectithin cholesterol acyltransferase (Lcat). Calb2 and Lcat expression images were obtained from
online Allen Brain Atlas (http://www.brain-map.org) .........
3-9
.
68
High resolution MR Images of wild type (top panel) and kif2la KO
(bottom panel) mice revealed anatomical differences in multiple brain
regions.
...................
..............
..
71
4-1
Diagram of a cerebral capillary enclosed in astrocyte end-feet. Characteristics of the blood-brain barrier are indicated: (1) tight junctions
that seal the pathway between the capillary (endothelial) cells; (2) the
lipid nature of the cell membranes of the capillary wall which makes it
a barrier to water-soluble molecules; (3), (4), and (5) represent some
of the carriers and ion channels; (6) the "enzymatic barrier" that removes molecules from the blood; (7) the effilux pumps which extrude
fat-soluble molecules that have crossed into the cells.
4-2
75
. .........
76
Time course of intravascular MR signal after intravenous administration of Gd-PGC ......
4-4
...
Two-compartment tissue model: assuming immediate and complete
mixing of protons within each of the compartments
4-3
. .....
......................
80
Dependency of Vapp on flip angle, TR, and water exchange rate. Measurement at larger flip angle reflects true cerebral blood volume; while
those at smaller flip angles are sensitive to transvascular water exchange rate. ............
4-5
......
........
.....
81
Dependency of calculated WEI on contrast agent dose (i.e. TR (a) and
intravascular Rpost) (b) at different WER. Increasing contrast agent
dose and TR can increase the sensitivity of WEI. However, the linearity
relationship between WEI and WEI is reduced as TR increases.
4-6
84
...............
86
Effect of water exchange between interstitial and extracelllular spaces
on CBV ...........
4-9
....
Three-compartment tissue model assuming no water exchange between
intravascular and intracellular spaces ...
4-8
83
Error in CBV and WEI quantification due to the inaccuracy in flip
angles at different transvascular water exchange rates.. .. ....
4-7
. . .
...........
....
......
89
Effect of transcellular water exchange on WEI quantification.. ....
90
4-10 Effect of intravascular blood volume on WEI under different BBB water
permeabilities. ..................
............
..
92
5-1
Experiment Paradigm: intravenous Mannitol administration (a) and
5-2
97
..........
CO 2 challenge (b) .........
Representative images of mouse brain acquired before (top panel) and
after (bottom panel) contrast agent administration (a), and temporal
evolution of vascular MRI signal showing the long half life of this intravascular MRI contrast agent (b), 3D angiography revealing vascular
structure (c).
5-3
.....
...........
..........
...............
100
Changes of CBV and WEI after intravenous Mannitol administration:
CBV (a) and WEI (b) .............
5-5
.....
.......
101
Plots of WEI vs. CBV due to repeated Mannitol boluses (bolus 1,
bolus 2, and bolus 3) or CO 2 inhalation. . ................
5-6
102
T2-weighted images of a wild type mouse (a) and eNOS-/- mouse (b)
after SPION injection ......
5-7
........
..........
..
112
Comparison of CBV at cortical and sub-cortical levels between eNOS
KO and WT mice.
5-9
111
Comparison of VSI and MVD at cortical and sub-cortical levels between eNOS KO and WT mice. ...................
5-8
99
Changes of CBV and WEI in response to CO 2 challenge: CBV (a) and
WEI (b) .................
5-4
..
..................
...
......
113
H&E (magnification: 400x) staining of the cerebral cortex of wild
type (a, c) and eNOS knockout (b, d) mice. Morphologically, eNOS
knockout mouse is characterized by higher number of blood vessels
with smaller vessel size in the cerebral cortex when compared with
wild type mouse. ...................
6-1
..........
114
MRI angiography showing partial reperfusion of previously occluded
MCA (a, orange arrow), and representative diffusion-weighted images
(DWI) and apparent diffusion coefficient (ADC) maps of mouse subjected to transient lhr middle cerebral artery (MCA) occlusion (b) are
shown in this figure.
...................
........
121
6-2
Changes of CBV and WEI due to transient 1hr MCAo. Animal group
with poor CBV restoration showed significantly higher WEI, indicating
a dramatically increased BBB water permeability, thus more severe
BBB damage than the group with relatively good CBV restoration. . 122
6-3
in vivo T 2-weighted images of mice underwent 1hr transient MCAo.
Images were acquired 24 hours after occlusion (a. mouse 1, b. mouse 2).129
6-4
3D ex vivo images of mouse underwent 1hr transient MCAo (mouse 2).
Images were acquired 24 hours after occlusion. Image acquired before
Gd-DTPA staining (a) revealed little contrast; while Gd-DTPA staining not only enhanced overall image contrast but also clearly revealed
the stroke affected area (b). . ........
6-5
.......
.
. . . . .
130
Axial, horizontal, and sagital view of a stroke mouse brain initially
stained with 5 mM Gd-DTPA. The brain was collected 24 hours after
one-hour MCA occlusion. ......
6-6
......
.
..... .
. .... .
.
131
Axial, horizontal, and sagital view of a stroke mouse brain stained with
0.36 mM MnC12 after the initial staining with 5 mM Gd-DTPA. . . . 132
6-7
Axial, horizontal, and sagital view of a stroke mouse brain initial
stained with 0.36 mM MnC12 . The brain was collected 24 hours after one-hour MCA occlusion .........
6-8
.........
.....
133
Changes of ex vivo brain tissue R 1 and R2 before contrast agent staining, after staining with 5 mM Gd-DTPA, during washout, and after
the second staining. ................
6-9
. . . . . .
. . . . . 135
Changes of ex vivo brain tissue R 1 and R 2 before contrast agent staining, after staining with 0.36 mM MnC12 , and during washout.
.....
136
6-10 Comparison of MRM staining (5 mM Gd-DTPA, a) with the conventional Nissl stain (b). .
. . .
...
. ... . .
.
137
List of Tables
1.1
Ratio R1 over R 2 for dipolar and scalar relaxation contributions in
.
different motional regimes (table is adapted from chapter 3 in [14])
2.1
28
MRI properties of Gd-DTPA and collagen I gel. Measured longitudinal
relaxivity of Gd-DTPA at 14 T, corresponding RI changes of collagen I
gel after adding 10 mM Gd-DTPA, and the resulting SNR improvement. 36
2.2
To document proliferative activity, we report here the percentage of
MIB-I positive cells for the 14T specimen versus control. The invasion
data result from measuring the straight-line distance between an invasive cell and a circle drawn around the spheroid. The measurements
are done on a print of a 20x magnification, and represent [arbitrary
units]. (See text for more details) ...................
3.1
.
41
Concentrations of Gd-DTPA and MnC12 staining buffer and ex vivo
brain. Total buffer volume is 5 ml.
. ...................
54
3.2
Relaxivities of Gd-DTPA and MnC12 in water and ex vivo brain at 14T 54
3.3
Relaxivities of MnC12 in ex vivo brain measured at different field strength:
0.47T, 1.4T, and 14T. The NMRD profile strongly suggests macromolecule binding of Mn 2 + in exercised brain tissue. . ..........
54
18
Chapter 1
Introduction
Magnetic resonance imaging (MRI) has been widely used in clinical diagnosis and
basic research. It offers excellent tissue contrasts based on intrinsic tissue T 1, T2 , or
water diffusion, which allows visualization of various anatomical structures as well as
pathological abnormalities. In addition, with the use of MRI contrast agents, we can
not only greatly improve MRI's ability for tissue delineation and disease detection,
but also open new approaches for understanding the biophysical and pathological
mechanisms in different diseases. This thesis focuses on two major applications of
MRI contrast agents: the improvement of tissue differentiation ability of ex vivo
magnetic resonance microscopy (MRM) and the development of an MRI technique
for in vivo detection of blood-brain barrier (BBB) change.
1.1
MR contrast agents
Based on magnetic susceptibility properties, materials can be classified as diamagnetic, paramagnetic, superparamagnetic, and ferromagnetic. Atoms in diamagnetic
materials have no unpaired electrons, and therefore, no permanent magnetic dipole
moments. When placed in a strong magnetic field, diamagnetic materials are repelled
by the externally applied magnetic fields, hence they have a weak negative magnetic
susceptibility. Unlike diamagnetic material, atoms in paramagnetic materials have
permanent magnetic dipole moments due to unpaired electrons. When placed in a
magnet, they will align with the external magnetic field, generating a net magnetic
field. However, this net magnetic field can not sustain itself when the external field
is removed. Ferromagnetic materials have high magnetic susceptibility and magnetic
memory properties. They will remain magnetized even after the external magnetic
field is removed. Superparamagnetic materials are composed of very small ferromagnetic crystallites (1-10 nm). Each small crystallite has a net magnetic moment, which
is influenced independently by the external magnetic field. These small crystallites
function as individual units and behave like atoms in paramagnetic materials. Superparamagnetic materials, therefore, are similar to paramagnetic materials in this
regard, but their magnetic susceptibilities can be significantly higher than those of
paramagnetic materials.
Currently, the majority of MR contrast agents are based on paramagnetic or superparamagnetic materials.
Most clinically approved contrast agents are different
forms of paramagnetic gadolium chelates. Other paramagnetic contrast agents (e.g.
MnCl 2 ) and superparamagnetic contrast agents (e.g. superparamagnetic iron oxide
nano particles (SPION)) are currently limited to animal studies. Based on tissue distributions, MRI contrast agents can be divided into different categories. Intravascular
(blood pool) contrast agents stay exclusively in the intravascular space, and have a
molecular weight of 70,000 and above; extracellular agents are distributed in the extracellular space, and have a relatively short blood half-life; intracellular agents can
enter into the intracellular space through different cellular uptake mechanisms. Furthermore, contrast agents can also be functionalized to become target specific, which
allows in vivo visualization of bio-markers for understanding the disease processes.
These MR contrast agents can affect both T and T2 relaxation times of the tissue.
However, because of their inherent magnetic properties, their abilities to affect tissue relaxation times are different; theses properties can be utilized for different MRI
studies in both clinics and research labs. For example, superparamagnetic MR contrast agents can produce strong and locally varying magnetic fields, leading to strong
T 2 and T2 * relaxation enhancements because of their high MRI susceptibility. This
susceptibility effect usually dominates T relaxation enhancement, resulting in sig-
nificant signal reduction in structures containing superparamagnetic contrast agents.
This class of contrast agents has now been actively used in molecular imaging to
visualize target-specific processes at the molecular level [1]. The superparamagnetic
MR contrast agents have also been extensively used in fMRI studies as a blood-pool
agents and in various animal models of human diseases [2, 3]. Paramagnetic contrast
agents, on the other hand, can cause a predominant reduction in tissue T relaxation
times, producing increased MR signal intensity on Ti-weighted images. Paramagnetic MR contrast agents (CA) usually contains metal ions (e.g. Mn 2 + and Gd 3 + )
with unpaired electrons. These unpaired electrons can interact with tissue protons,
causing the reduction in both T 1 and T 2 relaxation times. Depending on the nature
of these interactions and the strength of the external magnetic field, paramagnetic
contrast agents have different relaxation abilities in tissue that can also be affected
by their biophysical and biochemical properties in biological samples. Concenquently,
paramagnetic contrast agents are widely used in various in vivo studies to help delineate pathological changes and evaluate disease treatments. They are also commonly
used in the ex vivo studies to the increase signal-to-noise ratio (SNR) and enhance the
image contrast for anatomical phenotyping of various transgenic and mutant mouse
models [4, 5].
Paramagnetic and superparamagnetic MR contrast agents can affect tissue T
and T2 relaxations through the dipole-dipole interaction, scalar coupling relaxation,
and Curie relaxation.
For superparamagnetic contrast agents, the high magnetic
susceptibility effect can also cause the T 2 relaxation enhancement, and even larger T 2 *
effects. In general, their contributions to tissue T and T 2 are usually concentration
dependent, and can be expressed using Equation 1.1, in which ri is the relaxivity of
the contrast agent of interest and is normally expressed in the unit of mM-' - S - 1 .
In next section, we will discuss in detail mechanisms governing contrast agent T and
T2 relaxation in solution. Understanding the relaxation mechanisms in solution will
help us to interpret the relaxation effect of different contrast agents in tissue.
Ti
=
obs
+ ri [CA] , i = 1, 2
Ti
intrinsic
(1.1)
Relaxation mechanisms of paramagnetic MR
1.2
contrast agents in solution
1.2.1
Introduction
Paramagnetic relaxation enhancement comes from two major components: the innersphere contribution and the outer-sphere contribution [6, 7]. Inner-sphere relaxation
arises from the strong interaction between the coordinative bound-water molecules
and the magnetic moment of unpaired electrons in the contrast agent. Water protons that diffuse close to paramagnetic contrast agent can also experience relaxation
enhancement without undergoing coordinative binding. This so called outer-sphere
relaxation is modulated through translational diffusion, electron spin dynamics, and
molecular tumbling. In general, it is less efficient and more difficult to describe because of the stochastic nature of the distances between protons and electrons.
The theory of inner-sphere relaxation was first formulated by Solomon, Bloembergen, and Morgan in the 1950s and 1960s [8, 9, 10]. In the SBM approach, the
hyperfine interactions (dipole-dipole and scalar coupling) between electrons and protons are modulated by molecular reorientation, chemical exchange between the coordinated water and the bulk water, and electron relaxation. In addition, for metal
ions with high S values, the interaction between proton spins and the time-averaged
electron magnetic moment can also cause significant relaxation enhancement, which
is called Curie relaxation. For the three relaxation mechanisms (dipole-dipole, scalar
coupling, and Curie), the time dependency of the proton-electron interactions is characterized by correlation time (-T), which can be influenced by time constants such as
the molecular tumbling (r), the chemical exchange (FTm),
(T7).
and the electron relaxation
Next three sections are brief summaries of these different paramagnetic relax-
ation mechanisms. Detailed descriptions can be found in the following references by
Slicter, Solomon, Bloembergen, Morgan, Lauffer, and Review of coordinate chemistry
[9, 11, 8, 10, 6, 12, 13].
1.2.2
Proton paramagnetic relaxation due to dipole-dipole
interaction
Dipole-dipole interactions can be modulated by molecular tumbling, chemical exchange, and electron relaxation. Correlation time (%T)for dipole-dipole interaction
can thus be expressed as:
1
-T =
c
11
+ - +--
Tr
Ts
1
(1.2)
Tm
where, the time constants were defined in the above section. Equation 1.3 and 1.4 are
the Solomon-Bloembergen-Morgan equations governing the dipole-dipole relaxation
[11, 8]: for longitudinal relaxation,
2 0
R=
2
(S + 1)
ggp2
r+6
15 4~
3,
1 + (wi - ws)2
1
2F
S 67
1 + (wI + Ws)272
(1.3)
for transverse relaxation,
R2
=-
1
(o
15
4L
2Bgp
S(S
2
e
B
WS)2T c
1 + (wI +1±
++
+ 1)
4F +
6T,
W2 _2
1
(i
73
)
WS)27c2
+
f
1 +WiTc2
(1.4)
where -7 is the proton gyromagnetic ratio, S is the electron spin quantum number,
AB
is Bohr magneton, ge is the electronic Lande factor for the free electron, r is electronspin distance, w, and ws are proton and electron Larmor frequencies respectively.
1.2.3
Proton paramagnetic relaxation due to scalar (contact)
coupling
The contact or scalar coupling relaxation is through the flip-flop mechanism. By its
nature, it is not modulated by the reorientation of the molecule. The correlation time
for contact relaxation is modulated by electron relaxation times as well as chemical
change, which can be calculated using the following equation:
1
T
c
1
1
Ts
Tm
- --
(1.5)
The equation for longitudinal relaxation enhancement is given by Equation 1.6, and
Equation 1.7 describes the transverse relaxation enhancement due to contact coupling.
RI
=
R2
=
S(S + 1)
IS(S
+ 1)
(1.6)
e
1A)
2T2
Tie +
2eT
(1.7)
where :A is the electron-proton hyperfine coupling constant, T 1e and T2e are electron
relaxation times, w1 and ws are proton and electron Larmor frequencies respectively.
For metal ions with long electron relaxation times, contact coupling to R 2 relaxation can be sizable leading to enhanced proton transverse relaxation.
1.2.4
Proton paramagnetic relaxation due to Curie relaxation
In the presence of an external magnetic field, there exists a small population difference between electron states. This time-averaged electron magnetic moment creates
a fluctuating local magnetic field that is independent of electron spin relaxation. This
local magnetic field, however, is still modulated by molecular motion and chemical
exchange.
The proton relaxation due to this local magnetic field is called Curie
relaxation, or magnetic susceptibility relaxation. The correlation time for Curie relaxation will be affected by the molecular rotation and chemical exchange, and can
be expressed as:
1
7-c
1
1
- + r
24
7Tm
(1.8)
The contribution of Curie relaxation to R 1 and R 2 can be expressed using the following
two equations:
R1
RI
2 P0
2
3 rr
2 2
2
(Sz)2
geB
6
5 4
r
(1.9)
1 + w,
3Tr7
2Wg44 (S + 1)2
/
Tr
e
kieB
1 + W27)1
(3kT)2r6
(1.10)
52
5 47
R2
1~(Io
-(Sz)
7Ig2B2 2
4, + + 37,T1
47
r6
S
1o
47r
22
+ 1)-2
47,
e B(
(3kT)2r6
g4(S
1+
+ +
2
37-,
(1.12)
r
Here, 7,is the proton gyromagnetic ratio, AUB is Bohr magneton, ge is the electronic
g factor, r is electron-spin distance, wu is proton Larmor frequencies.
Curie relaxation can be significant if electron relaxation times dominate the correlation time in dipolar coupling relaxation.
1.2.5
Comparison of the three relaxation mechanisms
Paramagnetic relaxation enhancement has a strong field dependency. Due to the
nature of different relaxation mechanisms, the field-dependent relaxation profiles
(NMRD: nuclear magnetic relaxation dispersion) exhibit different characteristics. For
dipole-dipole relaxation, the NMRD profile (Figure 1-la) of the longitudinal relaxation rate, R 1, has a plateau at low field. It starts to decrease, then reaches another
plateau, and starts to decrease again as the magnetic field increases. The two inflection points correspond to wus" TdiP , 1 and w,
7cdip
1, respectively, where Tcip
is the correlation time governing the dipole relaxation. The NMRD profile of R 2 is
similar to that of R 1 except that it reaches another plateau at a high field due to
the field-independent term, 4r ip in equation 1.4. The NMRD profiles of R1 and R 2
due to scalar (contact) relaxation also plateau at a low field, and decrease with increasing magnetic field (Figure 1-1b). However, they have only one inflection point at
1, in which " is the correlation time for the scalar coupling relaxation.
ws - 7Co"
°
The two curves diverge after ws . c" > 1 because of the field-independent term, Tie,
in equation 1.7 (Figure 1-1b). For Curie relaxation, the magnitude of the induced
magnetic moment is proportional to the external magnetic field (Figure 1-1c). Therefore, the contribution due to Curie relaxation is field dependent, and is proportional
to the square of the external magnetic field strength. Moreover, Curie relaxation is
usually observed at the high magnetic field (> 100 MHz), and its contribution to R1
is often negligible. It is, therefore, more relevant for the T2 relaxation leading to the
line broadening.
In the motion limited regime where wI --~ < 1 (usually true for the field strength
up to 100 MHz), R 1 equals R 2 . There are no field-dependent terms. R 1 is maximum at r7 = w;, then starts to decrease as r continues to increase. On the other
hand, R 2 continues to increase due to the non-dispersion term. In this regime, for
most paramagnetic contrast agents, the dominant relaxation mechanism is the dipoledipole relaxation. However, when the electron relaxation time (r7) is longer than the
rotational correlation time (r), the contribution of the scalar (contact) coupling to
paramagnetic relaxation, especially T2 , can be sizable. The contribution of the scalar
coupling term and the dipole term to R 2 relaxation is proportional to
s/Fr,.
For
example, if r, is on the order of 10- 10 to 10-11 second and Tr is on the order of 10to 10-
9
(e.g., Mn2+), the contribution of the scalar coupling to R 2 can be 102 to 103
times higher than that of the dipole-dipole interaction. Curie relaxation, on the other
hand, is always negligible under this condition.
Curie contribution, in general, becomes higher with increasing field, S, and Tr/Ts
ratio. Outside the motion limited regime, its contribution to R 2 can be appreciable.
The contribution of Curie relaxation to R 1, on the other hand, is leveled off, and
usually is small compared to the dipole relaxation, due to the fact that wl -
is much
greater than one in this regime. The relative contributions of dipole-dipole interaction
and contact (scalar) coupling to relaxations outside the motion limited regime are
dependent on the magnetic field strength as well as the correlation times governing
the two processes. In the range where IWsl Ton > 1 > IWi1di
p
,
the contribution from
contact coupling is small (equation 1.6). The dominant relaxation mechanism for R 1
is the dipolar interaction. However, under this condition, the relative contribution of
7.5
5
79
0.01
0.1
10
1
100
1000
Proton Larmor Frequency (MHz)
17.5
5 -
2.5
0.01
0.1
10
1
100
1000
Proton Larmor Frequency (MHz)
0
I.10'
7.5"10'
510'
2.5*10'
(MHI)
squared
Proton Lartor Frequency
(c)
Figure 1-1: Effect of different relaxation mechanisms on proton paramagnetic relaxation (figures are adapted from chapter 3 in [14]).
these two mechanisms to R 2 relaxation is highly dependent on the correlation times.
For paramagnetic contrast agents with long electron relaxation times (i.e. large scalar
coupling correlation times), scalar (contract) coupling relaxation can be big, leading
to a high R 2 /R
1
the ratio of R 2/R
ratio. On the other hand, if the dipole-dipole coupling is dominant,
1
is around one. In the region where wIwlc > 1, the R 1 relaxation
is not very effective, leading to a ratio of R 2 /R
1
much larger than one. Table 1.1
summaries the relative contributions of dipole-dipole and scalar coupling relaxation
to R 1 and R 2 Table 1.1: Ratio R 1 over R 2 for dipolar and scalar relaxation contributions in different
motional regimes (table is adapted from chapter 3 in [14])
WSTc < 1
R 1/R 2 (dipolar dominant)
R 1/R 2 (scalar dominant)
1.3
1
1
WSTc > 1 >
6/7
<1
WIT
WIT c > 1
c
< 1
1<
MR contrast agents in magnetic resonance microscopy
With advances in MRI techniques, MR image resolution has been greatly improved
over the past decades, approaching a microscopic level.
Magnetic resonance mi-
croscopy (MRM) has now become an important tool for anatomical phenotying
and morphologic screening of different mouse strains and various transgenic/mutant
mouse models [15, 16, 5, 17, 18].
Comprehensive mouse brain atlas of C57BL/6J
and 129S1/SvImJ strains [19, 20, 21] have also been established based on MRM images. Despite the great success of MRM in anatomical studies, MRM images based
on intrinsic tissue contrast lacks the flexibility and target-specificity offered by conventional histological staining. One the other hand, there exist multiple relaxation
mechanisms and varying tissue-dependent affinities for different contrast agents. This
may potentially help improve tissue differentiation and provide tissue staining flexibility resembling that found in conventional histology. We hypothesize that the use
of extrinsic contrast agents in MRM studies can both improve overall signal-to-noise
ration (SNR) and help to achieve tissue/cellular specificity in MRM. Using an in
vitro system of Glioma spheroid suspended in a collagen I matrix, we first demonstrated that the use of MR contrast agent can reduce acquisition time, improve SNR,
and allow us to monitor dynamic tumor growth. We then explored the changes of
image contrast in mouse brans stained with different MR contrast agents at a 14T
MRI system. Our results showed that the distinct tissue relaxation properties and
tissue-dependent affinities of different MR contrast agents may be utilized in MRM
brain staining to manipulate image contrast for better tissue differentiation and structure visualization. Finally, we applied the protocol developed using normal mouse
brain to the Kif2la knockout mouse model for anatomical phenotyping and to a
transient middle cerebral artery occlusion (MCAo) mouse stroke model for potential
histopathological MRM staining.
1.4
Detection of BBB damage due to stroke using
MR contrast agent
Stroke, the most common cerebrovascular disease, is the third leading cause of adult
mortality and morbidity in the United States and developed countries. Of various
strokes, ischemic stroke accounts for about 85% of total incidents. After an ischemic
insult, the affected brain area first undergoes cytotoxic edema, which is later followed
by vasogenic edema. Brain swelling and increased intracranial pressure as a result of
cerebral edema are two major causes of mortality in stroke patients. Among various
factors, the impairment of blood-brain barrier (BBB) integrity is mainly responsible
for interstitial fluid accumulation or vasogenic edema. In previous studies, the quantification of BBB permeability was performed using the extravasation of extrinsic
markers (e.g., Evans Blue, radio labeled isotopes, Gd-DTPA) [22, 23, 24]. However,
these extrinsic markers for measuring vascular permeability are usually not capable
of crossing the BBB at acute stages of ischemia, thus, limiting their ability to ob-
serve the progressive weakening of BBB. It is therefore desirable to find biomarkers
that allows the detection of early or subtle changes in BBB integrity and permits
qualitative and quantitative monitoring of BBB alteration. Thus, the second focus of the thesis is the development of MRI technique that uses intrinsic water as
a qualitative and quantitative biomarker to evaluate the overall BBB integrity by
utilizing effect of transvascular water exchange on Ti-weighted MRI signal intensity.
First, the theoretical background was presented. Parameters affecting transvascular
water exchange measurement were determined using computer simulations based on
two- and three-compartment models. This new MRI method was characterized using
two well-known physiological and pharmacological challenges: graded hypercapnia
and systemic administration of Mannitol. Furthermore, we applied this technique to
study the baseline BBB water permeability and CBV changes in the eNOS knockout
mouse model and the BBB damage resulting from ischemic stroke using a mouse
model of transient stroke.
Chapter 2
MR Microscopy of glioma spheroid
in collagen I matrix
2.1
Introduction
High-grade malignant gliomas are characterized by rapid volumetric growth and extensive local tissue infiltration. Despite all efforts to improve diagnostics and therapy,
the outcome remains dismal with a five-year survival rate below 3.3% in the main
age group of 45-years and older [25]. Since the surrounding tissue is thought to impact, and perhaps even guide the tumors invasive patterns, much weight is currently
being put on better understanding the dynamic interaction of an expanding brain
tumor with its microenvironment. As such, following an interdisciplinary approach,
collaborators of this project have previously employed an in vitro glioma multicellular tumor spheroid (MTS) model to investigate the spatial and temporal dynamics of
MTS expansion within extracellular matrix environments [26, 27, 28], and started to
model them in silico [29, 30, 31]. In these studies, various aspects of the interaction
between a growing spheroid system and its environment were studied and mathematically modeled, including extracellular matrix concentration, mechanical forces, and
invasion directionality.
In the 1980s, several groups independently acquired microscopic images of different biological samples using magnetic resonance microscopy [32, 33, 34]. These
early studies demonstrated the feasibility of magnetic resonance microscopic imaging
with its unique sensitivity to tissue water environment. Since then, with continued
advances in magnetic resonance imaging (MRI) techniques (incl. stronger gradient
coils, improved RF coil design, and more powerful computers), imaging resolution
has steadily increased. We are now able to acquire three-dimensional (3D) in vitro
images at a high SNR with a resolution approaching the single cell level. This allows
us to employ MRI, the imaging modality of choice in vivo, to study these microscopic
tumor models also in vitro.
We have previously reported that magnetic resonance microscopy (MRM) enables
us to distinguish an MTS of roughly 250 ypm in diameter in a collagen I gel and
provides us with a true 3D view of the MTS within the matrix [35]. However, the
time required to obtain high-resolution 3D images was considerably long, typically
10 hours.
To dynamically monitor MTS growth, we would have to increase the
temporal resolution while maintaining a comparable signal to noise ratio (SNR). For
this very purpose, we applied Gd-DTPA, the contrast agent widely used in clinics, to
the collagen I matrix, achieving not only a significantly shortened MR imaging time
down to 3 hours, but also increasing the SNR from 20 to over 40. Therefore, we are
able to dynamically monitor both global and local changes of the MTS over a 12-hour
period at a temporal resolution of 3 hours and an isotropic spatial resolution of 24
ym. Using conventional immunohistochemistry techniques we were able to confirm
cell viability in the MRM sample post imaging, as well as to find patterns of cell
proliferation and invasion that seem distinctively different from the controls and thus
warrant further analyses.
In summary, we argue that MRM enables us to examine MTS growth dynamically
with a true 3D view. The nature of MRM data allows us to segment the spheroids
contour out of its microenvironment, reconstruct and visualize its surface, and subsequently analyze the MTS distinct 3D dynamics from any selected angle. Tracking
tumor expansion non-invasively down to the single-cell scale by using a clinically relevant imaging modality should facilitate data integration, and, in combination with
in silico modeling, will yield valuable insights into the critical interaction between the
tumor and its microenvironment.
2.1.1
Methods
2.1.2
Multicellular tumor spheroid (MTS) and extracellular
matrix
Cell culture:
The U87mEGFR [35, 36, 37] cell line is cultured in 10 mL of high-glucose Dulbecco's
Modified Essential Medium (DMEM; Invitrogen, Carlsbad, CA), supplemented with
1% Penicillin-Streptomycin (PS; Invitrogen), 10% Fetal Bovine Serum (FBS; JRH
Biosciences, Lenexa, KS), 0.5 mg/mL Geneticin (Invitrogen) and 20 mM hepes buffer
(Invitrogen) in 10 cm diameter Petri dishes (Corning, Corning, NY). Once cells are
confluent, they are twice rinsed with Phosphate Buffered Saline (Invitrogen) and
detach after the addition of 1 mL Trypsin-EDTA (Invitrogen).
After 5-10 min, 9
mL of fresh cell media is added to neutralize the Trypsin. The cell solution is then
transferred into a 15 mL centrifuge tube and centrifuged for 5 min at 1200 RPM ( 2 x
103 m/s 2 ). After aspirating the supernatant, the cells are resuspended in fresh media
at a concentration of 2 x 105/mL.
Multicellular spheroids:
Spheroids are then generated using the hanging droplet method [38], with a drop
size of 20 pL, which seeds
4 x 103 cells in each spheroid to yield a diameter of
approximately 400 pm. Briefly, multiple droplets of 20 pL of cell solution are pipetted
onto the inside of a 100 mm-diameter Petri dish cover. After placing the cover back
on a culture medium-filled dish, the dish itself is placed in the incubator (5% CO 2,
37 'C). Surface tension maintains droplet integrity, while gravity pulls cells together
at the bottom of each droplet. Cells are left to form spheroids in the incubator for 3
days before they are collected.
Extracellular matrix:
The extracellular matrix (ECM) model is a 1.5 mg/mL bovine collagen type I (Inamed
Biomaterials, Fremont, CA) matrix supplemented with 10% FBS, 10% Minimum Essential Medium (MEM; Invitrogen), 1% PS, 50 mM sodium bicarbonate (NaHCO 3;
Sigma, St. Louis, MO) buffer. For image enhancement, 10 mM Gd-DTPA (Magnevist; Berlex, Wayne, NJ, U.S.A.) is added to the collagen solution.
To induce
polymerization, a few pL of 1 M sodium hydroxide (NaOH; Sigma) are added until a neutral pH is reached.
The choice of 1.5 mg/mL concentration used here is
based on work by Kaufman et al. [26] and represents a reasonable compromise between achieving a sufficiently high viscoelastic modulus while avoiding blocking cell
motility altogether.
Sample preparation:
For each sample, we pipette 280 pLs of collagen solution into a pPCR tube (VWR,
West Chester, PA). After 15-30 min in the incubator to allow the gel to initiate
polymerization, a spheroid is pipetted into this collagen gel, which is then placed back
in the incubator for several hours to allow cells to attach properly to the ECM before
imaging. To assess the impact of MRM on cell viability, we compared the histology
of the MR-imaged specimen with a control sample that was treated identically but
for one difference: instead of being imaged overnight, it was kept in the incubator.
As soon as the experiment was halted, both the imaged specimen and the control
samples were fixed, sectioned and stained.
2.1.3
Magnetic resonance microscopy
We first measured Gd-DTPA (Magnevist) relaxivity in water at 14T (Magnex, 89mm
vertical bore, gradient strength 100 gauss/cm, Bruker Biospin System) using the
conventional inversion recovery spin echo sequence at three different concentrations
(0.075, 2.5, 5 mM). The Gd-DTPA concentration (10 mM) used here was then determined based on its dosage (0.2 - 0.5 mmol/kg) in MRI mouse model studies [39, 40]
and its relaxivity at 14T.
The MTS in collagen with 10 mM Gd-DTPA was imaged using MRM at 14T.
Specifically, we first used a modified FLASH pulse sequence to acquire multi-slice
multi-echo (TR/TE = 400/3 8 13 18 ms; FOV = 0.95 cm, Matrix size = 256 x 256;
slice thickness = 400 Am; Flip angle= 300) images with an in-plane resolution of 37
x 37 Am2 to localize the MTS in the collagen I gel. A quick Ti measurement was
then performed using a modified IR_RARE sequence (TR/TE: 1500/7.86 ms; rare
factor: 2; matrix size: 128x128; 10 slices with 0.4 mm slice thickness; FOV: 0.6 cm;
TI: 5.621 55.620 105.620 205.620 505.620 1005.620 2005.620 ms). A three-dimension
spoiled gradient echo (FLASH) sequence (TR/TE = 20/5.5 ms; FOV = 1.2 x 0.6
x 0.6; matrix size: 512 x 256 x 256; signal was optimized at the Ernst angle) was
then used to acquire high-resolution (24 ,m isotropic) microscopic images every three
hours for a total of 12 hours. After imaging, the sample was fixed for histopathological
analysis as described below. MR images were displayed employing an ImageJ software
package [41] and then segmented using the 3D Slicer software package [42]. T1 of the
sample and Gd-DTPA relaxivity were fitted using Matlab (Mathwork, Inc., Natick,
MA). After the MTS was manually segmented for each time point, its corresponding
volume and surface area were calculated.
2.1.4
Histology
The spheroids were harvested and processed as described previously [28], fixed in
10% formalin and embedded in paraffin.
Blocks were serially sectioned in 7 Am
thick sections and stained with hematoxylin and eosin (H&E). In order to assess
the proliferative activity, immunohistochemistry was performed by using cell cycleunspecific MIBI antibody (which detects the nuclear Ki-67 antibody so that quiescent
cells (GO) remain unstained) (DAKO, M7240, USA; 1 : 50 dilution). The so called
MIB-1 labeling index was then calculated as the fraction of MIB-1 positive cells
from the total number of cells. In addition, images of the test specimen and control
were printed. The number of invading cells was counted and the distance of each
cell was calculated from a circle drawn at the circumference of the spheroid. The
distances were measured by drawing the shortest straight line from the circle around
the spheroid to the invading cells. The measurements are in cm, but do not represent
true distance of invasion (due to magnification of the image printed), but rather
represent an arbitrary unit by which distances of the cells from the two MTS, i.e.
14T specimen versus control, can be compared.
2.2
2.2.1
Results
Magnetic resonance microscopy of MTS in collagen I
gel
Prior to imaging the MTS, we characterized the rl relaxivity (longitudinal relaxivity)
of Gd-DTPA at 14T and the SNR improvement after adding 10 mM Gd-DTPA into
the collagen I gel that did not contain any tumor cells. Table 2.1 shows that with
10 mM Gd-DTPA, we were able to reduce the T 1 relaxation time of the collagen gel
from around 3000 ms to about 24 ms. The imaging time was reduced from 10 hours
to 3 hours with a two-fold increase of SNR (from
-
20 to over 40). The original
Table 2.1: MRI properties of Gd-DTPA and collagen I gel. Measured longitudinal
relaxivity of Gd-DTPA at 14 T, corresponding R 1 changes of collagen I gel after
adding 10 mM Gd-DTPA, and the resulting SNR improvement.
ri (mM- 1 S - 1)
R 1 (S- 1)
SNR
Gd-DTA
3.88 ± 0.12
Collagen I Gel
without 10 mM Gd-DTPA
Collagen I Gel
with 10 mM Gd-DTPA
0.003
16
42.32
42
MRM images of the MTS specimen are shown in Figure 2-1. The time series clearly
documents the growth of the spheroid from different angles and at different planes.
Even with the 2D view, MRM showed different patterns of cellular growth. Images on
the left in Figure 2-1 show cells growing attached to the main body (red circle); while
those on the right show some cells seemingly moving away from the main spheroid.
It becomes apparent that with reduced imaging time and increased SNR, MRM can
indeed monitor the spatio-temporal expansion of an MTS system.
2.2.2
Segmentation and 3D visualization of MTS
The 3D reconstruction shows rather heterogeneous expansion patterns with rough
surface growth areas throughout over the course of the observation period (Figure
2-2a). Intriguingly, once started (orange droplets at the proximal apex of the MTS
at time point: 3 hours) glioma cell expansion into the gel seems to give rise to an
'imprinting' process which confirms the 'trailblazer' concept that has been described
previously in [28, 26]. Finally, corresponding to the MTS preserved structural viability
(compare with Fig. 3) Figure 2-2b also confirms its functionality in that both MTS
volume and surface area increase throughout the observation period.
2.2.3
Histopathology
To semi-quantitatively evaluate the viability of the MTS that has been imaged over
12 hours with 14T MRM, standard immunohistochemistry was performed and the
histopathology results were compared with the control (as described above).
As
documented in Figure 2-3, the imaged MTS not only remains viable overall but,
compared to the control tumors the 14T specimen even contained a substantially
larger proliferative fraction (see Table 2.2).
More specifically, the MIB-1 labeling
index for the control is 12.5% (30 positive cells out of 239 counted) as compared to
the MIB-1 labeling index for the specimen which is 21.1% (50 positive cells out of
236 cells counted) almost doubled. Furthermore, reviewing comparable sections, in
the 14T specimen only 5 cells could be found to invade the gel; conversely, a total of
32 cells invaded the collagen matrix in the control experiment. While the min-max
values expectedly (given the difference in n) showed considerably more heterogeneity
in the control group, intriguingly, the overall mean invasion distance was with 2.10
versus 2.11 almost identical between 14T specimen and control. We note that a 2nd
control experiment confirmed the results.
3 hours
6 hours
9 hours
12 hours
Figure 2-1: The images to the left depict the axial view, whereas those to the right
show an arbitrary angle. These 2D images confirm the surface heterogeneity that
emerges already 12 hrs post placement of the MTS into the gel, and depict a small
group of invasive cells (images on the right) that can also be seen in Figures 2-2a and
2-3. On the other hand, images on the left demonstrate that certain parts of the solid
spheroid actually grew into the gel (red circle).
1.9
0.15
1.85
0.145
1.8
0.14
1.75
0.135
-4- volurre
surface
..
1.7
0.13
3
9
6
12
time [hour]
(b)
Figure 2-2: Volumetric rendering and growth dynamics of the MTS in the collagen I
matrix at four consecutive time points. (a) The segmented MTS is reconstructed in
3D and the color-coding indicates the growth increase at each time point. The MTS
appears to grow anistropic, which may indicate regional heterogeneities in composition of either MTS or microenvironment, or both, and/or point towards a heterogeneous local interaction between cells and gel. (b) MTS volume and surface area are
calculated and plotted over time.
(a)
(b)
(c)
(d)
Figure 2-3: Histopathology findings, comparing 14T MTS with control. (a). H&E
staining of 14T specimen (original magnification x1OO) showing only a small number of cells infiltrating the gel. Insert shows infiltrating cells in higher magnification
(original magnification x400). (b). MIB-1 immunostaining of 14T specimen (original magnification x200), highlighting cells that are not in GO as dark brown nuclei.
(c). H&E staining of control (original magnification x100) showing a large number of
infiltrating cells in adjacent gel. Insert shows the infiltrating cells in higher magnification (original magnification x400). (e). MIB-1 immunostaining of control (original
magnification x200), highlighting cells that are not in GO as dark brown nuclei.
Table 2.2: To document proliferative activity, we report here the percentage of MIB1 positive cells for the 14T specimen versus control. The invasion data result from
measuring the straight-line distance between an invasive cell and a circle drawn around
the spheroid. The measurements are done on a print of a 20x magnification, and
represent [arbitrary units]. (See text for more details)
14T
Control
2.3
Proliferation
50 pos/236 21.1%
30 pos/239 12.5%
n: 5
n: 32
Mean: 2.10
Mean: 2.11
Invasion
SD: 2.07 min-max: 0.50 - 4.70
SD: 1.6 min-max: 0.30 - 6.50
Discussion
Our data provide clear evidence that magnetic resonance microscopy can be used
to study the dynamic growth of multi-cellular tumor spheroid embedded in a collagen I matrix, at a resolution close to the single cell level (Figure 2-1).
Adding 10
mM Gd-DTPA to the gel dramatically reduced the image acquisition time without
any detrimental impact on cell viability (Figure 2-3). Intriguingly, the glioma cells
showed distinct spatio-temporal expansion patterns into the gel with spotty surface
expansion across the spheroid and the notion of a 'trailblazer' mechanism that guides
cell motility in that single cells follow each other along preformed pathways (Figure
2-2a). The histological results showed surprisingly not only more proliferative activity yet concomitantly also less cells pursuing invasion in the MRM specimen than
in the control. Intriguingly, the mean invasive distance was virtually identical which
supports the notion that the phenotype itself remained unaffected. This preliminary
data support the notion of a "dichotomy" between proliferation and migration, which
was experimentally shown by Giese et al. [43] and modeled with a molecular switching mechanism by Athale et al [44] and Zhang et al [45].
The impact of high field
MR (14 Tesla in our study) on cancer cells is somewhat controversial. For instance,
Santini et al [46] showed that a sinusoidal 50 Hz magnetic field of ImT significantly
increased spheroid invasive properties, but had no damage on growth. Short et al.
[47] showed that for the two cell lines exposed to a 4.7T field up to 72 hours, there
was no change in cell growth rate. On the contrary, Raylamn et al [48] demonstrated
that human cancer cells exposed to 11.7T for 64 hours had a slower growth rate. In
our study, the spheroid also experienced a high frequency small magnetic field (B1
field, RF pulse) in addition to the 14T main magnetic field. Although the results
from a single case study must be interpreted with proper caution, the increase in
proliferation and concomitant reduction in the number of invasive cells observed in
our study may indicate that different exposure time and magnetic field strength can
impact tumor expansion patterns. Hence, our results warrant a more extended study
to investigate in detail the impact of MRM on normal and cancer cell viability and
phenotypes.
Kaufman et al [26] showed different patterns of MTS growth and mobility in collagen I matrices of varying concentrations. They observed a more effective invasion
in high concentration (1.5 and 2.0 mg/mL) gels at early time. The choice of 1.5
mg/mL concentration used here was a compromise between having a high modulus
to maintain sample integrity and allow cells to move around quickly, and low optical
density to allow us to locate the spheroid easily by eye and enable, perhaps, comparisons with optical imaging in the future. Previously, Brandl et al [49] used an
11.7T spectrometer to study multicellular tumor spheroids that were generated from
human malignant MV3 melanoma. In their study, the spheroid was allowed to grow
for seven to fourteen days in agar to a diameter of 400 1000 pm with a starting dense
population of 5-106 cells. This may lead to central necrosis due to limited nutrition
and oxygen in the core region. They indeed observed a necrotic core in their sample.
The fact that the authors did not observe any morphologic change over the four to
five-hour imaging time is less surprising since their agar environment is far more rigid
than the collagen I gel used in our study here.
The advantage of using MRM in this context lies in its non-invasiveness, and its
true three-dimensional full-scale (global and local) view of MTS dynamic growth. It
is very difficult, if not impossible to acquire whole three-dimension images of MTS
in collagen I gel using most modern optical imaging methods. Confocal microscopy
allows three-dimensional imaging; however, traditional single photon reflectance or
fluorescence methods are limited by the working distance of objectives, typically a
few hundred microns. Two-photon confocal microscopy and second harmonic generation can go beyond this limit, but still remain within 1 mm of the surface [50, 51, 52].
Conventional light microscopy can surpass this limitation as well, but this is accompanied by a loss in axial resolution. Moreover, a general limitation of light microscopy
is that it is constrained by the opacity of tissues. To overcome the loss of signal due
to increasing depth, one has two choices - with clear drawbacks: i) increasing light
power may decrease cell viability very quickly, together with the photobleaching seen
in typical fluorescence imaging; ii) fixing and staining the cells prior to imaging usually yields a better signal-to-noise ratio than live cells, but the mere fact of fixing the
system is an obvious loss in evaluating the time evolution of a living sample. That
is not to say of course that optical methods are incapable of generating useful and
quantitative information. Interesting studies by e.g. Nygaard et al [53] and recent
work by Kaufman et al [26] yield useful information about MTS growth, though with
the limitations described above.
With the presently achieved MRM resolution already, we observe spotty volumetric changes. Future studies of MTS growth using MRM will reveal even more detailed
structural changes in a true 3D sense as the image resolution further improves. MRM
is beginning to bridge the imaging scale that, until recently, has been accessible to
optical techniques only, with the clinical MRI data level which is still plagued by
a rather poor spatial and temporal resolution. Such three-dimensional volumetric
studies of the morphologic growth of MTS using MRM will provide complimentary
information about the growing tumor and the dynamic interaction with its microenvironment. With the help of integrative in silico modeling insights gained with MRM
in vitro can help us understand better the complex cancer growth patterns seen in
patients.
2.4
Conclusions
We have shown that magnetic resonance microscopy can be used to study dynamic
tumor growth in vitro with a resolution that approaches that of light microscopy.
Based on this microscopic MRI pilot data, we argue that one may be able to use this
clinically relevant imaging modality also in vitro, particularly once combined with
molecular imaging techniques. Being able to rely on one imaging modality only, or
even primarily, would greatly enhance our ability to successfully integrate in vitro
with in vivo data, using innovative in silico models that bridge between these scales.
Chapter 3
MRM staining of ex vivo mouse
brain at 14T
In Chapter 2, using an in vitro system of Glioma spheroid suspended in a collagen I
matrix, we demonstrated that the use of Gd-DTPA can greatly increase the temporal
resolution and SNR of MR microscopy (MRM), which allows to dynamically monitor
the tumor growth. However, the application of MR contrast agents in MRM studies
is not limited to the reduction of imaging time and the increase of SNR. The multiple
relaxation mechanisms and varying biophysical/biochemical properties of different
contrast agents can potentially provide us an MRM "staining" flexibility resembling
that found in conventional histology. Therefore, we intended to extend the utility of
contrast agents for MRM applications. In this chapter, we exploited the potential
use of two commonly used MR contrast agents, Gd-DPTA and MnC12 , to manipulate image contrast. An MRM staining protocol was developed by drawing upon
the multiple relaxation mechanisms and tissue-dependent staining properties characteristic of each contrast agent. Our results show that differential MRM staining
can be achieved using multiple MR contrast agents, revealing detailed cytoarchitecture, and may ultimately offer a window for investigating new techniques by which
to understand biophysical MR relaxation mechanisms and perhaps to visualize tissue
anomalies even at the molecular level.
3.1
The development of MRM staining using GdDTPA and MnC12 in the wild type C57BL/6
mice
3.1.1
Introduction
Magnetic resonance microscopy (MRM) is a specialized magnetic resonance imaging
(MRI) technique for imaging ex vivo specimens at microscopic resolution in tens of
microns (or < _ 104 pm 3 ). Since the first MRM images were obtained in the 1980s
[32, 54], MRM has been extensively used in various anatomical studies of normal and
transgenic/mutant mice [17, 15, 55, 18, 4, 16], and developing mouse embryos [56, 57].
Unlike conventional histology, which requires tissue dissection, MRM provides capabilities to capture large amounts of anatomical information without destroying 3D
tissue structures, and thus allows geometrically accurate visualization of anatomical
structures. As greater numbers of transgenic or mutant mice have become available
for translational studies of human diseases, MRM has become a highly important tool
for anatomical phenotyping and morphologic screening. For instance, several groups
have successfully completed comprehensive mouse brain atlasing of C57BL/6J and
129S1/SvImJ strains [20, 19, 21] based on MRM images. MRM has also demonstrated
success for phenotyping of several mutant mice strains such as the Reeler mouse [16]
and the CACNA1A knock-in migraine mouse [4].
Despite the great success of MRM for anatomical studies of various mouse strains
and transgenic/mutant mice, MRM imaging that relies on intrinsic tissue contrast
lacks the flexibility and target specificity offered by conventional histological staining. Although many of the MRM studies mentioned above have used MRI contrast
agents (mainly gadolinium chelates), the main purpose of their use to date has been
limited to reducing tissue T, relaxation times, so as to achieve improved overall signalto-noise ratio (SNR) and shorter image acquisition time. These studies have therefore
not explored the full capabilities of these agents for MRM; MR contrast agents can
regionally change T as well as T 2 and T 2* relaxation times. Moreover, multiple
relaxation mechanisms and their field dependency can lead to change in relaxation
enhancement effects of different contrast agents at different field strengths. For example, MnC12 has a scalar coupling relaxation mechanism, which exerts a strong effect
on T 2 relaxation especially at high field [9, 8]. In addition, when applied to ex vivo tissue specimens, commonly used MR contrast agents may exert distribution and tissue
binding effects as a result of intrinsic chemical characteristics, and therefore produce
differentiable MRI contrast between heterogeneous biological structures. Moreover,
the differences in microscopic distributions of various contrast agents may illustrate
different selective enhancement on T2 * relaxation due to differential sensitivity to
local tissue compartmentalization [58, 59]. Therefore, it is quite possible that we
can achieve better tissue differentiation and henceforth develop detailed tissue/cell
specific MRM staining methods by employing various MR contrast agents.
Gd-DTPA and MnC12 are two MR contrast agents commonly used for in vivo
studies. These agents are also known to have distinct in vivo bio-distributions when
exogenously administered. In vivo, Gd-DTPA, an extracellular agent, is incapable of
entering viable cells [60], whereas MnC12 can move across the neuronal cell membrane
through calcium channels [61, 62] or across synapses [63]. Gd-DTPA typically exerts
its influence on tissue MR relaxation through dipole-dipole interaction [6].
Mn 2+ ,
besides having dipolar relaxation capabilities, has an additional scalar coupling relaxation effect, which especially affects T 2 relaxation at high field, [13]. Although
relaxation mechanisms are well known in solution and to some extent in vivo, the
ex vivo distribution of these compounds and their effect on MR relaxation rates in
fixed specimens have not been precisely documented. Based on the different tissue
distribution and biophysical properties demonstrated in various in vivo studies, we
hypothesized that Gd-DTPA and MnCl 2 will would manifest different relaxation rates
and distribution properties in ex vivo brain tissue, leading to spatially distinct image
contrasts, and thus provide potential staining flexibility in MRM. To test our hypothesis, we administered these two contrast agents to widely available C57BL/6 mouse
brains and acquired ex vivo images using our 14T MRI system. We first characterized the tissue relaxation properties of each contrast agent in different brain regions.
Then, we examined the dependence of tissue relaxation rates on staining duration
and determined contrast agent concentrations and imaging parameters to achieve the
desired image contrast for each MR contrast agent. Our specific goals in this study
were (1) to determine staining and imaging parameters for high-resolution ex vivo
magnetic resonance microscopy, and (2) to examine the relaxation mechanisms, biophysical properties, and tissue affinities of these two MR contrast agents in mouse
brains. Our results demonstrated the potential use of these two staining agents to improve ex vivo image contrast in the brain, and furthermore the possibility of achieving
tissue- or cell-specific contrast using such MRM staining methods.
3.1.2
Materials and methods
A total of thirty adult male C57BL/6 mice (8 to 12 weeks old, Taconic Farm, Germantown, NY) were used in this study. Extracted brains were divided into 4 groups:
(1) five animals were used as controls and received no contrast agent; (2) eight were
used to optimize contrast agent concentration and measure relaxivity; (3) four were
used for a contrast agent penetration study; and (4) thirteen (four for Gd-DTPA and
nine for MnC12 ) were used to optimize MRM image acquisition.
Mouse brain preparation:
All brains were fixed through transcardial perfusion. Specifically, after they were
anesthetized with 1.5% isoflurane in 30% 02 balanced by N2 0, control mice were
transcardially perfused with 30-cc heparinized saline at room temperature, then with
10-cc 4% paraformaldehyde (PFA) in phosphate-buffered solution (PB). The brain
was removed from the skull, post-fixed in 4% PFA for 12 hours, and transferred to 5
ml phosphate-buffered saline (PBS) solution afterwards. For those animals used for
the MRI staining procedure, we prepared the brains as described above except for
that we mixed contrast agents in 4% PFA used in transcardial perfusion and also in
PBS buffer for extensive staining after fixation. In order to see whether the presence
of Mn 2+ in transcardial perfusion buffer makes any difference on image contrast due
2
to active uptake of Mn 2 + by living cells, three mice in the Mn + group were fixed
without Mn 2 + in 4% PFA during transcardial perfusion.
In the time course study of contrast agent penetration, mouse brains were incubated in a mixture of PBS and contrast agent for one, three, five, and seven days
before T1 measurement. For all other experiments, the brains were stored in the PBS
buffer containing contrast agent of desired concentration for five to seven days before
MRI acquisition.
Magnetic Resonance Imaging and Microcopy:
All images were acquired with a 14T magnet (Magnex Scientisfic, England) connected
to a ParaVision console (Bruker, Biospin, MA) with a gradient of 100 Gauss/cm
(Bruker Biospin, MA). The brains prepared for ex vivo imaging were placed inside a
glass tube filled with a proton-free susceptibility-matching agent (Perfluoro-compound
FC-40, Fisher Scientific) to eliminate background signal. We used a Bruker 10 mm
volume coil for all image acquisitions.
For two-dimensional T1 , T2 , and T2* measurements, we used an inversion-recovery
prepared RARE (IR _ RARE) sequence, a multi-slice multi-echo (MSME) sequence,
and a multi-echo spoiled gradient echo (MGE) sequence. All images were acquired
with an FOV of 0.95x0.95 cm, matrix size of 128x128 (in plane resolution of 74x74
pm 2), and slice thickness of 0.5 mm. Imaging parameters for MSME were: TR =
1200 ms, 8 echoes, at times 8.5, 17.1, 25.6, 34.1, 42.6, 51.1, 59.7, and 68.2 ms; for
MGE were: TR=1200 ms, 8 echoes, first TE = 3.4 ms, echo spacing 2.45 ms; for
IR-RARE were: TR/TE = 5000/7.9 ms, TI = 5.62, 55.62, 105.62, 205.62, 505.62,
1005.62, and 2005.62 ms (for brain stained with high concentration contrast agent),
and TR/TE = 10000/7.9 ms, TI = 5.62, 105.62, 305.62, 505.62, 1005.62, 2005.62,
and 3005.62 ms (for brains without contrast agent or in low concentration contrast
agent).
For high-resolution magnetic resonance microscopy, we used a 3D spoiled gradient echo sequence (3D SPGR). Two sets of high-resolution images were acquired
with TR/TE = 35/10 ms, FOV: 1.3x1.3x0.95 cm, Matrix:
512x512x256 (resolu-
tion: 25x25x37 ?m3), 8 averages (about 11 hours), and TR/TE = 35/8 ms, FOV:
1.45x0.95x0.95 cm, Matrix: 400x256x256 (resolution: 37x37x37 pm 3 ), 16 averages
(about 11 hours), respectively. The flip angle was chosen to be the Ernst angle to
maximize the signal.
Inductively coupled plasma (ICP) mass spectrometry:
For each brain used for relaxivity calculation, one hemisphere was used for ICP analysis, and the other was used for T1 and T2 measurement to calculate contrast agent
relaxivities at 0.47T, 1.4T and 14T. Each brain sample was first dissolved in 2 ml
70% HNO 3 at 37°C overnight. Samples were then centrifuged at 5000 rpm and 4oC
for 15 min. For brains stained with MnC12 , we diluted 0.75 ml of the completely
dissolved brain sample with 2 ml diluents (each 2L diluents containing 1.857 L water,
0.143 L 70% HNO 3 , 2 ml Triton, 20 ppb final concentration Strontium, 20 ppb final
concentration Lutetium, and 40 ppb final concentration gold). For the corresponding
Mn 2+ staining buffer, we mixed 0.5 ml staining buffer with the 1.5 ml diluent. For
those brains stained in Gd-DTPA, 50 pl of the dissolved brain sample was diluted
with 2 ml diluents. For the corresponding Gd-DTPA staining buffer, we mixed 50 pl
staining buffer with 5 ml diluent. At each step, samples were carefully weighed for
calculating the final Gd and Mn concentrations. We used Agilent 7500a ICP-MS to
measure the amount of Gd and Mn in brain samples and the corresponding staining
buffers.
Relaxivity measurement, choice of concentration and imaging parameters:
We used four different concentrations to calculate the relaxivities of each contrast
agent: 0 mM, 2.5 mM, 5 mM, and 7.5 mM for Gd-DTPA, and 0 mM, 0.12 mM, 0.24
mM and 0.36 mM for MnCl 2 . For aqueous solution and brain samples, T, T2 , and
T2 were measured using 2D IR-RARE, MSME, and MGE sequences described above.
In addition, we performed inductively coupled plasma (ICP) mass spectrometry to
determine accurate contrast agent concentrations in the brain. We also measured the
ex vivo relaxivities of Mn 2+ at two other field strengths (0.47T and 1.4T) using the
same brain samples that were analyzed via ICP. The relaxivity of each contrast agent
was then obtained using linear regression analysis.
To determine the concentration and imaging parameters for high-resolution MRM,
we first quickly scanned the brains stained with contrast agent of different concentrations using the 2D MGE sequence described previously. We then calculated the SNR
and contrast-to-noise ratio (CNR) for multiple regions from brains stained with different concentrations of Gd-DTPA or MnC12 , scanned at different TEs. Those brains
that exhibited the best CNR were imaged again using a 3D gradient echo sequence
,
(resolution: 50x74x74 pm 3 ) involving different flip angles (100, 250, 550 700, and
900) and TR/TE of 35/4 and 8 ms. The concentration and imaging parameters were
selected based on the combined results from 2D and 3D measurements.
Time course of contrast agent penetration:
Mouse brains stained with 5 mM Gd-DTPA and 0.24 mM MnCl 2 were imaged at one
day, three days, five days, and one week after soaking in PBS buffer with contrast
agents of desired concentrations. We measured Ti using IR-RARE sequences at each
time point. We then calculated Ti at each time point in order to characterize tissue
T1 change over the time.
Data Analysis:
Region of interest (ROI) analysis was used to investigate changes of tissue TI, T2 , and
T2* after MRI contrast agent staining, and the relaxivities of Gd-DTPA and MnCl 2
in ex vivo mouse brain. ROIs pertaining signal from the whole brain, cerebral cortex, striatum, thalamus, hippocampus, and cerebellum were manually outlined based
on Paxinos mouse brain atlas [64] using AFNI [65]. We used Matlab (Mathworks,
Natick, MA) to analyze MRM data. TI, T2 , and T2 in each ROI were fitted using
non-linear least square or first-order polynomial algorithms based on averaged MR
signal intensity. In tissue relaxivity calculations, concentrations measured from ICP
analysis were used. We used Mouse BIRN Atlasing Toolkit (MBAT) [66] for brain
segmentation and structure labeling. All the structures were identified and labeled
based on the Allen Brain Atlas [67] and Paxinos mouse brain atlas. ImageJ software package [41] was used for image visualization. Unless otherwise specified, all
statistical analyses in the study were performed using a two-tailed t-test. Statistical
significance was accepted at a confidence level of 0.95. All of the numerical data were
presented as averages ± standard deviations.
3.1.3
Results
Relaxivity measurement:
We first determined contrast agent concentrations in ex vivo brain using ICP analysis.
Table 3.1 shows contrast agent concentrations in the PBS staining buffer (5 ml) before
and after brain staining and in the corresponding ex vivo brains. Buffer Gd-DTPA
concentration was not significantly affected by brain staining over time. There was
only slight change in Gd-DTPA concentration, going from 5 mM to 4.9 mM and
from 2.5 mM to 2.4 mM after brain incubation (Table 3.1).
Mn
On the other hand,
2+
concentrations in the buffer dropped significantly after brain staining; while,
its concentration in the brain was much higher than that of the buffer. In addition,
there was no obvious linear relationship between resultant brain Mn 2+ concentration
and initial buffer concentration (see Table 3.1).
Table 3.2 summarizes the relaxivities of the two contrast agents in both aqueous
solution and ex vivo mouse brain at 14T. The molar relaxivities (rl and Tr2 ) of GdDTPA and MnCl 2 were analyzed using linear regression based on R1 and R 2 measured
at five different concentrations in both solution and tissue. The rl of Gd-DTPA in
aqueous solution and ex vivo brain were measured to be 3.88 ± 0.12 mM - 1 -S - and
3.58 ± 0.01 mM
-
(13.84 ± 0.79 mM
.S- 1 , respectively. Meanwhile, the r 2 of Gd-DTPA in ex vivo brain
-1
- S - ) was significantly higher than in aqueous solution (5.434
± 0.11 mM - 1 - S - ) (see Table 3.2). The MR relaxivity of MnCl 2 trended differently
than Gd-DTPA. The rl of MnC12 in solution was 6.3 ± 0.1 mM the rl of 8.57 ± 0.55 mM
-1
1
- S - 1, lower than
-S - 1 in ex vivo tissue. MnC12 also exhibited a relatively
high transverse relaxation enhancement effect, with greater r 2 in aqueous solution
(162.44 ± 3.74 mM - 1' S - 1 ) than in ex vivo brain (124.9 + 15.44 mM - ' - S-).
Table 3.3 shows the relaxivities of Mn 2+ in ex vivo brain tissues at different field
strengths, with ri the highest at 0.47T (69.89 ± 6.67 mM-' - S-'). The rl drops
dramatically as the field strength increases, whereas r 2 increases as magnetic field
increases.
We further evaluated the Gd-DTPA tissue distribution based on the relaxivities
and ICP results. Gd-DTPA has a similar rl in ex vivo tissue and in aqueous solution
(3.2), indicating that the T 1 relaxation enhancement of Gd-DTPA in ex vivo tissue
is also caused by the same dipole relaxation mechanism that governs paramagnetic
proton relaxation in solution. The similar rl of Gd-DTPA in tissue and solution
also imply that there is no protein or macromolecule binding of Gd-DTPA in brain
tissue. We, therefore, can use the concentrations of Gd-DTPA in tissue and staining
buffer to estimate the tissue space that is occupied by Gd-DTPA after brain staining.
From the ICP results (3.1), we found that the final brain Gd-DTPA concentration
was 1.97 mM if 2.5 mM staining buffer was used or 3.81 mM if 5 mM staining buffer
was used. Assuming that the total brain volume could be accessible by Gd-DTPA,
the final Gd-DTPA tissue concentration would be equal to that of staining buffer
since we are using a large volume of buffer to stain the brain. By taking the ratio
of Gd-DTPA concentration in tissue to that in staining buffer, we estimated that
Gd-DTPA occupied about 77% of the total tissue space. This result indicates that
Gd-DTPA is not uniformly distributed in ex vivo brain tissue. In addition, as Fisel
and Weisskoff pointed out, the microscopic distribution of contrast agent can generate
local magnetic gradient that influences proton relaxation over a longer range than the
dipole relaxation [58, 59]. Together with electron-proton dipole-dipole interactions,
this can cause higher T2 relaxation enhancement in tissue than in solution. Indeed,
this is what we observed in our study: higher r 2 in ex vivo brain tissue than in
aqueous solution (3.2: 13.84±0.79 mM
-
- S - 1 in tissue vs. 3.58±0.01 mM - 1 - S - 1 in
solution). This further suggests that Gd-DTPA is not uniformly distributed in tissue,
and is probably confined in local microscopic space.
Table 3.1: Concentrations of Gd-DTPA and MnC12 staining buffer and ex vivo brain.
Total buffer volume is 5 ml.
Mn 2+
Concentration (mM)
(in buffer, initial)
0
0.12
0.24
0.36
Concentration (mM)
(in buffer, after staining)
0.00014
0.082
0.19
0.23
Concentration (mM)
(in the brain)
0.002
0.29
0.59
0.71
2.5
5.0
2.36
4.92
1.97
3.81
Gd-DTPA
Table 3.2: Relaxivities of Gd-DTPA and MnC12 in water and ex vivo brain at 14T
r, (mM- 1 -S- 1)
r 2 (mM- 1'S - 1)
Ratio: r 2 /rl
Gd-DTPA
In water
In tissue
3.88 + 0.12
3.58 ± 0.01
5.434 ± 0.11 13.84 + 0.79
1.4
3.86
MnC12
In water
In tissue
6.3 ± 0.1
8.57 ± 0.55
162.4 ± 3.74 124.9 ± 15.44
25.79
14.57
Table 3.3: Relaxivities of MnC12 in ex vivo brain measured at different field strength:
0.47T, 1.4T, and 14T. The NMRD profile strongly suggests macromolecule binding
of Mn 2 + in exercised brain tissue.
Bo (T)
0.47
1.4
14
Larmor freq (MHz)
20
60
600
ri (mM-1.-S- 1) r 2 (mM-.S - 1)
69.89 ± 6.67
98.01 ± 8.82
56.78 ± 4.35
105.92 ± 8.89
8.57 ± 0.55
124.9 ± 15.44
Ratio: r 2 /rl
1.40
1.87
14.7
Choice of concentration, staining time, and imaging parameters:
To determine the contrast agent concentration to use, we characterized image contrast between three brain regions - cortex, thalamus, and striatum - using a 2D MGE
sequence. Figure 3-1 shows the CNR (cortex vs. thalamus and cortex vs. striatum)
as a function of TE for brains immersed in the two contrast agents at different concentration. For brains stained with Gd-DTPA, we found that the best CNR for the
selected regions (cortex vs. thalamus and cortex vs. striatum) was achieved with a
concentration of 5 mM (Figures 3-1c and 3-1d). Brains stained with Mn 2 + showed
different optimal concentrations in the same selected regions (Figures 3-1a and 3-1b).
The CNR between cortex and thalamus (Figure 3-1a) peaked around TE=8 ms, at a
concentration of 0.24 mM. However, the CNR between cortex and striatum decreased
as TE increased, and reached its maximum at a concentration of 0.36 mM. These
results suggest that the optimal concentration for these experiments for Gd-DTPA is
5 mM, and for Mn 2+, 0.24
0.36 mM. Based on these data, we chose to use 5 mM
Gd-DTPA and 0.24 or 0.36 mM MnC12 in our subsequent MRM studies.
We determined image parameters using a low resolution 3D FLASH (i.e. 3D
SPRG) sequence with different flip angles and TEs at the TR of 35 ms. Images in
Figure 3-2 show how contrast changes in mouse brains stained with 5 mM Gd-DTPA
(Figure 3-2a) and 0.24 mM MnC12 (Figure 3-2b) at different flip angels (100, 250, 550 ,
700, and 90') and with TEs of 4 ms and 8 ms. As shown in the top panel of Figures
3-2a and 3-2b, we observed no significant change in image contrast with increased
T 1 weighting, i.e., increased flip angle, which indicated a loss of T 1 contrast in fixed
brain tissue [68]. However, the overall image contrast was greatly enhanced as the
echo time was increased from 4 ms to 8 ms. Accordingly, we chose to image the brain
samples with TE of 8 ms at the Ernst angle, to optimize image contrast and SNR.
To understand the dependency of tissue relaxation times on the duration of contrast agent staining, T 1 of mouse brains stained with 5 mM Gd-DTPA or 0.24 mM
mM MnC12 were measured at one day, three days, five days, and seven days after
incubation in the PBS buffers doped with corresponding contrast agent. As shown in
-- 1 mM
-I-2.5 mM
*-SmM.
7.5 mM
8
_
7
E
_U 6
E
8
-- 1 mM
-- 2.5 mM
EM 7
E
-*-S5mM
-
7.5 mM
6
,,.,
5
5
d
e--t----
d/
4
44
X
(
0)3
.43
-
0
0
U
2_
u
Vz
0
0
10
8
6
TE (ms)
4
2
10
8
6
TE (ms)
4
2
12
12
(b)
(a)
2.5
-1-0.12 mM
S-0-0.24 mM
--0.36 mM
2.5
-U-0.12 mM
-- 0.24 mM
--- 0.36 mM
E
0
0
I.S 2
4
..
o
p 0.5
Z
1 0
P 0.5
Z
U
2
1
0
4
6
8
TE (ms)
(c)
10
12
14
2
4
6
8
10
12
14
TE (ms)
(d)
Figure 3-1: CNR vs. TE of mouse brains stained with Gd-DTPA or MnC12 at different
concentrations based on 2D MGE sequence. Top panel: brains in 1, 2.5, 5, or 7.5
mM Gd-DTPA, left: cortex v.s. thalamus, right: cortex v.s. striatum; bottom panel:
brains in 0.12, 0.24, or 0.36 mM MnC12 showing image contrast between cortex and
thalamus on the left, cortex and striatum on the right.
Figure 3-2: Coronal sections of mouse brains stained with 5 mM Gd-DTPA (a) or
0.24 mM MnCl 2 (b). Images were acquired using 3D FLASH sequence with different
flip angles (from left to right, 10', 250, 550, 700, and 900) and TEs (top: TE = 4 ms,
bottom: TE = 8 ms).
I- --
800
--~--~111-----
700
S5 mM Gd-DTPA
1 0.24 mM MnCI2
600
500
E 400
300
200
J
100
0
I
'xZ
Day 1
Day 3
/
It
VwA
'Ifg
Day 5
Day 7
Figure 3-3: Changes in tissue Ti relaxation time over time for mouse brains stained
with 5 mM Gd-DTPA and 0.24 mM MnCl 2 . These data demonstrate that at least
three days are needed for the contrast agent to fully penetrate brain tissue. To ensure
good tissue staining, approximately five days are needed before imaging.
Figure 3-3, we plotted the changes in T 1 in different brain regions over time, which
clearly demonstrated that after a five-day incubation period, the contrast agent concentration reaches equilibrium across the whole brain. To ensure full penetration and
equilibrium of contrast agent, all of the brains used for subsequent MRM studies were
incubated for at least five to seven days before imaging.
Structure and image contrast of ex vivo brain in Gd-DTPA and MnC1 :
2
Figure 3-4 shows coronal, sagittal, and axial views of mouse brains in 0.24 mM
MnC12
(Figure 3-4a) and 5 mM Gd-DTPA (Figure 3-4b). Gd-DTA and MnC1 greatly en2
hanced overall image contrast to allow visualization of detailed anatomical structures
of mouse brain. Subcortical structures such as hippocampus, caudate putamen, thalamus, globus pallidus, and external capsule can be clearly identified from these highresolution images. Furthermore, our data also demonstrate that the two contrast
agents stain the brains differently, in multiple brain regions, as shown in Figure 3-4.
For example, in the cerebellar cortex, the image contrast between granular layer and
molecular layer of brains stained by these two agents is reversed. The granular layer
was stained dark by MnCl 2 but appeared bright for brains in Gd-DTPA. Furthermore,
the distinct layer (possibly Purkinje layer in the cerebellar cortex) between granular
and molecular layers that is stained dark in Gd-DTPA brains is not apparent in the
MnC12-stained brain (Figure 3-4). In the cerebral cortex, Mn 2+ seems to stain a particular cortical layer dark (Figure 3-4a sagital view). On the other hand, Gd-DTPA
appears to stain the cortex uniformly. The lamina structure of hippocampus is more
obvious in the Mn2+-stained brains than in those stained by Gd-DTPA. Interestingly,
the dentate gyrus was stained bright by both contrast agents, but appears better
resolved by Gd-DTPA staining.
3.1.4
Discussion
This study clearly demonstrates that Gd-DTPA and MnC12 affect T 1 , T 2 , and T 2 *
relaxation times through different relaxation mechanisms in solution and ex vivo brain
samples. Moreover, these contrast agents exhibited different biological properties and
tissue affinities. As a result, MR images of ex vivo mouse brains immersed in GdDTPA or Mn 2+ solutions revealed different local image contrasts. The distinct image
contrast indicated independent tissue differentiation abilities of each contrast agent,
suggesting that tissue differentiation can be achieved through the use of different
MRI contrast agents, providing flexibility analogous to that inherent in conventional
histology.
Relaxation mechanisms and tissue affinity of Gd-DTPA and MnC12:
Paramagnetic contrast agents affect tissue relaxation primarily through electronproton dipole-dipole interaction [6]. In addition, for paramagnetic ions like Mn 2 +,
which have long electron spin relaxation times, the contribution of scalar coupling to
transverse relaxation becomes prominent, especially at high magnetic field [13]. This
Figure 3-4: high-resolution MRM of ex vivo mouse brain stained with 5 mM GdDTPA2 (a) or 0.36 mM MnCl (b) reveals detailed anatomical structure with different
image contrast in certain regions brain, e.g., cerebellum, cortex, and hippocampus.
cerebral cortex
medial habenula
Shippocampus
m lateral habenula
Shypothalamus
optic tract
thalamus
Caudate Putamen
globus pallidus
cirngulum bundle
n
m
m
m
m
amrygda la
substantia innominata
corpus callosum
anterior commissure
fimbria
cerebellum
ntdsrna capstriatu
brain stem
fundus of striaturn
midbrain
Figure 3-5: high-resolution MRM of ex vivo mouse brain stained with 0.24 mM MnC12
(a) or 5 mM Gd- DTPA (b) reveals detailed anatomical structure with different image
contrast in certain regions brain, e.g., cerebellum, cortex, and hippocampus.
may lead to the strong T 2 relaxation enhancement effect of Mn 2 + , both in solution and
in tissue, observed in our study at 14T. However, this scalar coupling relaxation has
negligible effect on T 1 relaxation at relatively high magnetic fields [8], as T 1 relaxation
enhancement for Mn 2 + ion occurs mainly through dipole-dipole interaction. Dipole
relaxation can be significantly affected by a change in the molecular tumbling rate, a
reflection of binding to macromolecules, and has strong field dependency. Previously,
the manganese ion has been shown to provide highly enhanced T 1 relaxation effect
(rl = 56 mnM
-
' - S - 1 at 20 MHz) in the rat myocardium due to intracellular protein
binding [69]. In our study, the relaxivity of Mn 2+ in ez vivo mouse brains, measured
at the same field strength, agrees with results published in the literature, implying
protein or macromolecule binding of Mn 2+ ion. This potential binding to proteins
or macromolecules is further supported by field-dependent relaxivity measurements
(NMRD) of Mn 2 + in ex vivo brain (Table 3.3). Moreover, ratios of r 2 /ri at relatively
high field strengths further demonstrate the significant contribution of scalar relaxation mechanism to transverse relaxation at high field. Perhaps still more important,
ICP analysis has revealed that manganese ion concentration is much higher in the
brain than in the corresponding staining buffer. Indeed, the buffer Mn 2+ concentration dropped significantly after brain incubation, which suggests that Mn 2 + has a
high tissue affinity even in excised tissue. These findings together would seem to suggest tissue Mn 2 + binding mechanisms that are independent of active cellular uptake
described in MEMRI experiments [70, 71].
Gd-DTPA, unlike Mn 2 + , influences proton relaxation mainly through dipole-dipole
interaction. The relaxivity values measured in aqueous solution agree well with literature results. Using the tissue Gd-DTPA concentration measured from ICP, the rl
of Gd-DTPA was found to be 3.58±0.01 mM - 1 - S - 1. This value is rather close to
that measured in aqueous solution, indicating the same R 1 paramagnetic relaxation
mechanism in ex vivo tissue. In addition, the lower Gd-DTPA concentration (Table
3.1) in ex vivo samples as revealed by ICP analysis suggests that Gd-DTPA is compartmentalized, as it is in vivo. Such compartmentalization creates microscopic field
gradients, which contribute to transverse relaxation [72] and give rise to increased
r 2 in ex vivo tissue, and thus higher r 2 /rl ratio in ex vivo brain compared to that
measured in solution (Table 3.2). Based on the ratio of Gd-DTPA concentration in
tissue to that in solution, we estimated that Gd-DTPA was distributed to a tissue
compartment occupying approximately 77% of the total brain volume. This somewhat surprising percentage, larger than one would expect from the extracellular space
alone, suggests at least part of the original intracellular space becomes accessible to
Gd-DTPA following tissue fixation. Moreover, the lower tissue Gd-DTPA concentration (3.81 mM), virtually unchanged buffer concentration after brain staining, and
similar rl in aqueous solution and ex vivo tissue indicate that Gd-DTPA has relatively low tissue affinity compared to Mn 2+ , and that there is little or perhaps no
tissue binding or active staining process for Gd-DTPA in excised brain tissue.
Bio-distribution and its effect on image contrast:
Gd-DTPA, as mention earlier, demonstrated a selective T 2 and T 2 * enhancement in
ex vivo brain tissue due to the overall compartmentalization. MnC12 , on the other
hand, has a more complicated bio-distribution than Gd-DTPA. Brains stained with
MnC12 showed distinct tissue- or region-specific relaxation time differences. For example, in cerebellar cortex, the change of R 1 is statistically higher in granular layer
than in molecular layer (Figure 3-6a). Mn 2 +, a calcium analog, is known to be able
to enter neurons via calcium channels [61, 62], and has high affinity to membrane and
intracellular proteins [73]. Unlike molecular layer that consists of neuronal dendrites,
fiber tracks, and small interneurons, granular layer is a densely populated neuronal
cell layer. It is therefore likely that relatively more Mn 2 + is drawn to this layer, accumulates, and is compartmentalized in the neuronal cell bodies that reside there. This
Mn 2+ tissue compartmentalization further creates a local microscopic gradient that
leads to the small but significant increase in R 2* relaxation observed in the granular
layer (Figure 3-6b). Enhanced T 1 and T2 * relaxation associated with accumulation
and compartmentalization of Mn 2+ in the granular layer, in turn, lead to the unique
contrast we observed in the T 2 *-weighted image (Figures 3-8b): darker granular layer
and brighter molecular layer. The differentiation of the cerebellar lamina that we
observed using Mn 2 + in ex vivo mouse brain is also consistent with previous in vivo
MRI studies in rats and mice using MnC12 [74, 75]. Moreover, the presence of Mn 2+
in the buffer used for transcardial perfusion was not a factor for the overall image
contrast enhancement. The image contrast remains similar between Mn 2+ stained
brains that were initially fixed with (Figure 3-7b) or without (Figure 3-7a) Mn 2+ in
the transcardial perfusion buffer. Together, these results suggest that the presence
of activity-independent Mn 2+ tissue binding or uptake mechanisms, which do not require cells or tissue to be alive. Such biochemical mechanisms appear to dominate the
tissue contrast observed in selective brain regions in our ex vivo study, and further
suggest that some of the observed MEMRI contrast (e.g. the cerebellar lamina) may
be from activity-independent uptake or binding.
Although both contrast agents can significantly shorten T 1, T 2 , and T 2* relaxation
times in ex vivo brains imaged at 14T, images (Figures 3-2a and 3-2b, top panel)
acquired with the TI-weighted sequence revealed relatively poor contrast compared
to those acquired with the T 2 *-weighted sequence (Figures 3-2a and 3-2b, bottom
panel). This difference suggests that image contrast in ex vivo mouse brain tissue
originates mainly from T 2 or T 2 * contrast, probably due to the loss of TI contrast
in fixed tissue [68]. We, therefore, used T 2*-weighted images in all our ex vivo MRM
studies. This is different from the imaging protocols used in Johnsons group, in which
Ti-weighted images were used primarily for high resolution MRM [5]. Difference in
specimen preparation may account for the different choice of imaging protocols. In
the active staining method developed by Johnsons [15] group, brains were perfusefixed with 50 mM ProHance and subsequently fixed in 10% formalin without contrast
agent. However, in our protocol, 5 mM Gd-DTPA 4% PFA solution was used for
both transcardial perfusion and subsequent incubation. As a result, contrast agent
distribution and final concentration in the brain are possibly different, leading to
different tissue T 1 , T 2 and T 2 * relaxation times and differently optimized MR imaging
parameters.
Images of mouse brains stained with Gd-DTPA or MnC12 revealed different staining patterns and varying contrast in multiple regions of the brain (Figure 3-4). This
7.5
I
I
Molecular
granular
7
6.5
6
inn
14
5.5
5
4.5
4
3.5
3
(a)
150
140
*
I
I
130
120
S110
100
90
01
j
Molecular
granular
(b)
Figure 3-6: Changes in R 1 , and R 2 * values in cerebellar molecular and granular layers
of mouse brain after stained with 0.36 mM MnC12 . The higher AR1 and AR 2* in
granular layer indicate that Mn 2+ is possibly compartmentalized into neuronal cells
in this layer.
i
(a)
(b)
2+
Figure 3-7: Images of ex vivo mouse brains fixed without (a) and with (b) Mn in
4% PFA during transcardial perfusion. Image contrast is not significantly affected by
the different staining protocols.
tissue differentiation and potentially flexible staining capability offered by MR contrast agents suggests a path toward future studies that might detail the principles behind the biophysical and biochemical distributions of these agents that may facilitate
the development of new agents with more selective staining properties. When looking
at the gene expression maps from the Allen Brain Atlas (http://www. brain-map.
org), we found that the staining patterns of both MnC12 and Gd-DTPA showed
positive correlation with certain gene expression profiles. For example, the spatial
expression of calretinin or calbindin 2 (calb2, the main intracellular Ca 2+ binding
protein expressed in cerebella granular cells) showed striking correspondence with
the staining patterns of MnC12 and Gd-DTPA (Figures 3-8a, 3-8b, and 3-8c, respectively). Moreover, the image contrast observed between Gd-DTPA and MnC12
stains was reversed in these regions. In another example, the distinctive dark line
between the two layers in the cerebellar cortex of our Gd-DTA-stained brains shows
remarkable correlation with lecithin cholesterol acyltransferase (Lcat, an important
protein in lipid metabolism expressed in cerebellum) expression (Figures 3-8c and
3-8d). However, it is also important to note that the correlations observed here are
not specific to Calb2 or Lcat. For example, other genes such as Zinc1 (zinc finger
protein of the cerebellum 1) and CAMKK2 (calcium/calmodulin-dependent protein
kinase kinase 2) also showed correlation pattern similar to that observed for Calb2.
Although correlations between the staining patterns (using Mn 2 or Gd-DTPA) and
the gene expression maps from the Allen Brain Atlas are not specific to particular
genes, these interesting correlations suggest that MRM staining may have a great potential in high-resolution molecular imaging with the development of highly selective
or target specific smart MR contrast agents. As a result, MRM is not only an important tool for anatomical screening and phenotyping, but it may also play a significant
role in studies of gene expression [76] and pathological diagnosis [77].
r 2aKI4-2(X Alien .
r 2Wa4-Z4fVAAlen
iruefor Brain Sicnce
for Brain Science
ihurinrw
ieo
(a)
2%W-240 Allen &riorf
in
(b)
(c)
for
&Brai
fience
(d)
Figure 3-8: Regions in the cerebellum of brains stained with MnCl12 or Gd-DTPA showed correlation with gene expression
patterns of Ca 2 + binding protein calretinin (calb2) in cerebellum and lectithin cholesterol acyltransferase (Lcat). Calb2 an
Lcat expression images were obtained from online Allen Brain Atlas (http://www.brain-map.org).
3.2
3.2.1
MRM of kif2la knockout new born mouse brains
introduction
Transgenic mouse models are invaluable in understanding the genetic effects on the
development and progression of various human diseases. As more and more genetically
mutated mouse models become available, there is an increasing demand for characterizing normal brain variations, mapping genotype to phenotype, and investigating
neurological abnormalities. Magnetic resonance microscopy (MRM) has shown great
utility in anatomical and phenotypic screening, allowing the visualization of brain
structures with different image contrasts in a true 3D nature [5, 78]. Furthermore,
specimens after MRM can also be used for conventional histological analysis, allowing
co-registration, direction comparison, and cross validation of images obtained from
different modalities using the same animal.
Kif2la encodes a member of the kinesin family of molecular motor proteins that
are involved in anterograde axonal transport of materials along microtubules. It is
primarily expressed in the central nervous system of the mouse, and is implicated
in human congenital fibrosis of the extraocular muscles type 1 (CFEOM1) disease,
an autosomal-dominate disorder caused by the mal-development of brain stem nuclei
and brain stem pathways. In order to understand the phenotypic effect of Kif2la
on brain development, we performed MRM study of the new-born Kif2la knockout
mouse brains, and compared the resultant MRM images of Kif2la knockout mice
with those from age-matched wild type mice.
3.2.2
Material and methods
Animal preparation:
New born Kif2la knockout and wild type mice were fixed with 10% formalin right
after birth. One week before MRM imaging, formalin fixed brains were transfered
to phosphate buffered saline (PBS) solution containing 5 mM Gd-DTPA for MRM
staining.
MR Microscopy:
High resolution MR imaging were performed at 14T (Magnex Scientisfic, England)
connected to a ParaVision console (Bruker, Biospin, MA) using a gradient of 100
Gauss/cm (Bruker Biospin, MA). A Bruker 10 mm volume coil was used in all image
acquisition. We used a 3D FLASH (i.e. 3D spoiled gradient echo sequence) for high
resolution MR microscopy. Specifically, the imaging acquisition parameters were:
TR/TE = 35/8 ms, resolution: 30 pm isotropic acquisition (reconstructed to yield
images with an isotropic resolution of 18 ym).
Data analysis:
We used ImageJ software package [41] for image visualization.
3.2.3
Results and discussion
Representative high resolution (18 pum isotropic) images of the new born wild type
(WT) and Kif2la knockout (KO) mouse brains are shown in axial and sagital views
(Figure 3-9, top - WT, bottom - KO ). Distinctive anatomical difference in multiple
brain regions can be found between the Kif2la KO mouse and the WT mouse by
visual inspection. For example, the layered cortical structure seen in the WT mouse
is completely missing in the Kif21a KO mouse at birth (Figure 3-9). The septum
appears dark in the WT brain, however is bright in the Kif21a KO brain. Striatum
also appears abnormal with scattered hyper-intensive dots in the Kif2la KO mouse.
The image contrast in the white better, e.g. the corpus callosum, is also different
between the Kif2la KO mouse and the WT mouse. In addition, the Kif2la KO mouse
brain looks smaller compared to the WT by gross examination. For future studies, it
will be important to perform volumetric analysis of the whole brain and different brain
structures, to acquire diffusion tensor images (DTI) or diffusion spectrum magnetic
resonance imaging (DSI) to further characterize changes of white matter in these
knockout animals, and to compare MRI results with those from histological analysis.
Moreover, brains at different developmental stages should also be imaged to look
(a)
(c)
(b)
(d)
and kif2la KO
Figure 3-9: High resolution MR Images of wild type (top panel)
regions.
(bottom panel) mice revealed anatomical differences in multiple brain
into the genetic effect of Kif2la on the embryonic development.
In summary, the
preliminary results from this study demonstrated that MRM staining can greatly
increase the tissue differentiation ability of MRM, and may help to better understand
the phenotypic effects of genes of interest when combined with conventional histology.
Chapter 4
MRI detection of BBB water
exchange: theoretical background
4.1
Introduction
The integrity of the blood-brain barrier (BBB) is crucial to normal brain function.
The BBB serves to maintain ionic homeostasis, to deliver oxygen and nutrient to
neuronal cells in the central nervous system (CNS), and to protect the brain from
neurotoxic or adversely neuroactive blood-borne agents. Many diseases affecting the
CNS can cause alteration or failure of the BBB. The compromised BBB, in turn,
may contribute to the development and progression of these diseases, causing further
damages to the brain. Therefore, early detection and monitoring of the alteration
of BBB integrity are very important in the diagnosis, prevention, and treatment of
cerebral and neurological disorders.
The BBB damage is usually detected through the spatial leakage profiles of extrinsically administrated markers such as staining dyes [22, 79], fluoresceins [80], radiolabeled compounds [23, 81], or gadolinium based compound [24]. However, these
methods are only possible when BBB is compromised to the extent that allows extravasation of these markers. It is therefore desirable to develop a technique that
allows the early detection of BBB damage. Previously, Donahue et al. showed that
absolute cerebral blood volume (CBV) could be quantified through the use of intravas-
cular MRI contrast agents, the accuracy of which was highly affected by the water
exchange rate across the BBB [82].
This observation led to several quantification
strategies not only for the cerebral blood volume (CBV) but also for transvascular
water exchange rate (WER). In particular, for the CBV quantification, MRI techniques based on either direct T 1 quantification or signal intensity have been used,
usually assuming a two-compartment tissue model consisting of intra- and extravascular spaces. Either infinitely fast or slow (relative to the intravascular and the
extravascular T 1 relaxation times) transvascular WER assumption is applied in order to quantify CBV [60, 83, 83]. Such assumptions, therefore, give rise to errors in
absolute CBV estimation based on changes of T 1 relaxation time or signal intensity
caused by intravascular MR contrast agent. Although not desirable in CBV quantification, these errors can actually be minimized for accurate CBV measurement or
amplified to reflect the degree of transvascular water exchange by carefully selecting
pulse sequences and calculation methods.
Characterization of CBV and WER based on changes in tissue and vascular T 1
relaxation time as proposed by Schwarzbauer et al. [83] requires accurate T 1 measurements before and after MR contrast agent administration. Fast imaging acquisition
and complicated calculation algorithms are usually needed for quantitative WER estimation. On the other hand, CBV quantification based on signal intensity can be easily
achieved by using a conventional 3D spoiled gradient echo sequence and intravascular
long circulating MRI contrast agent Gd-PGC. Furthermore, information regarding
transvascular water exchange can also be obtained by utilizing the flip angle dependency of the CBV measurements. This MRI technique for simultaneous CBV and
WER measurement is based on steady state measurement, thus, allowing the detection of spatiotemporal alterations of BBB integrity under different pathophysiological
conditions. In addition, high-resolution anatomical information and angiograms are
inherent to the 3D gradient echo sequence used for data acquisition.
In this chapter, we first present the theoretical background of measuring transvascular water movement across the BBB using MRI based on a two-compartment tissue
model. Computer simulations were then conducted to characterize parameters that
Figure 4-1: Diagram of a cerebral capillary enclosed in astrocyte end-feet. Characteristics of the blood-brain barrier are indicated: (1) tight junctions that seal the
pathway between the capillary (endothelial) cells; (2) the lipid nature of the cell
membranes of the capillary wall which makes it a barrier to water-soluble molecules;
(3), (4), and (5) represent some of the carriers and ion channels; (6) the "enzymatic
barrier" that removes molecules from the blood; (7) the efflux pumps which extrude
fat-soluble molecules that have crossed into the cells.
may affect the BBB water exchange measurement using two- and three-compartment
models.
4.2
Two-compartment water exchange model and
BBB water exchange measurement
4.2.1
Two-compartment tissue model
The blood-brain-barrier (BBB) is a protective barrier formed by capillary endothelia, pericytes enclosed within the basal lamina, and the surrounding cuff of astrocytic
endfeet (Figure 4-1, http: //www. answers. com). It greatly restricts the passive move-
O
•
e
O
rtravscular
compadment
Figure 4-2: Two-compartment tissue model: assuming immediate and complete mixing of protons within each of the compartments
ment of water and most water-soluble materials across the BBB and protects the brain
from neurotoxic or adversely neuroactive blood-borne agents. Lipid soluble molecules
(e.g. 02 and steroid hormones) are able to diffuse through the barrier, while water
soluble molecules such as sodium ion and glucose use specific transport systems. In
general, substances that have a molecular weight of more than 500 daltons cannot
cross the BBB.
Due to the existence of this biological barrier, the brain is usually modeled as
a two-compartment system consisting of the BBB-enclosed capillary space as the
intravascular compartment and the brain tissue outside the BBB as the extravascular
compartment (figure 4-2). In this model, fast water exchange is assumed between
plasma and blood cells. Thus, the intravascular space is a well mixed homogeneous
compartment. In the extravascular compartment, water molecules are also assumed
to be well mixed between intracellular and extracellular spaces. We further assume
that water exchange only occurs between capillaries and the surrounding tissue. No
exchange happens in arteries and veins. Incorporating water exchange into the Bloch
equations, we can obtain two coupled differentiation equations describing the spin
dynamics of this two-compartment system as following (equations 4.1 and 4.2, [84]):
_v=_
-v
dt
T1,iv
Mev - MevTI,ev
dMe
dt
- Mvkie + Mevkei
(4.1)
Mekei + Mikie
(4.2)
where, Mli, Me are the magnetizations in the intravascular and the extravascular
spaces, respectively. kie, kei are the exchange rate constants of water moving from
the intravascular space to the extravascular space and vice versa. In addition, these
rate constants are not affected by the blood flow. Ti,ijandTI,ev are the T relaxation
times in the intravascular and the extravascular spaces, respectively. Under detailed
balance condition, we have: Mivkl 2 = Mevk21 or vivkl 2 =
Vevk21,
with viv and vev as
the volume fractions of the intravascular and extravascular spaces, respectively. In
addition, at any time, vi, + Vev = 1 and the total magnetization M = Miv + Mev.
Equations (4.1) and (4.2) can be further rewritten as:
d(Miv,o - Mv)
=d__
-
dt
d(MevO - Mev)
M o+ (Mev,o - Mev)kei
(4.3)
- Mev,o ,- Mev + (MAI,o - Miv)kei
(4.4)
T'1
dt
where
iV
T1,
-
+
iv
T',ev
kiv and
-
T1,
+ key. By solving these two differentiation
T
equations 4.3 and 4.4, we can obtain a normalized analytical solution with the initial
condition: Mo = Mz(t = 0)
-t
H(t) =
vle
-t
Tiv,app f+ Ve
Tlev,app
where
i"
2
1
4
ev
1 --
iv=
1
liv,app
[(Vev - viv)(
C1 + C2
+ kie + kei]
1-
C2
(4.5)
1
=
C1 -
C2
Tev,app
11
C,
=
C2
=
1
+ T
1
-[(
1
+ ki
I
Tiev
+ kei)
k +
+ kie - kei) +
4
kiekei
Without the water exchange, MRI signal is simply the weighted sum of the signals
from the two compartments (i.e., SI = viMiv + v,Mev). However, as revealed in
Equation 4.5, MRI signal intensity is actually affected by the bidirectional water
movement across the BBB and the T difference between the two compartments. We
therefore can take advantage of this dependency of MRI signal intensity on transvascular water exchange to obtain water exchange information across the BBB by adjusting
MRI imaging parameters and changing the T relaxation time of one compartment.
4.2.2
Method for water exchange measurement
Quantification of CBV usually involves in the use of T contrast agents. Typically, the
exchange rate between compartments is assumed to be either very high (fast exchange)
[83] or insignificant (no exchange) [82, 85, 86]. CBV is then calculated based either
on tissue R 1 change (equation 4.7) or on signal intensity change (equation 4.6) before
and after contrast agent administration. Since most living tissues do not exhibit such
extreme exchange characteristics, neither of these assumptions accurately describes
the biological system [60] leading to CBV over or under estimation. Donahue et.
al. [82] demonstrated that the accuracy of CBV measurement is highly dependent
on water exchange rate, choice MRI pulse sequences, and imaging parameters. In
particular, she showed that accurate CBV can be quantified from MRI signal intensity
measured using a gradient echo sequence with large flip angle (the exchange insensitive
regime), while CBV calculated from MRI signal intensity measured at small flip angle
is highly affected by transvascular water exchange (the exchange sensitive regime)
[82]. Although undesirable in CBV quantification, the flip angle dependent exchangesensitive Ti-weighted MRI signal intensity can actually be utilized to measure water
exchange across the BBB.
Spre
Q
STpost
BVno-exchange
-
tissue
S' blood
ppost
S1,tissue
BVfast-exchage
-
RO
1l,blood
-
S
tissue
blood
(4.6)
pre
Al,tissue
R(prr
l,blood
(4.7)
)
Based on the above observations, we propose to use a widely available 3D spoiled
gradient echo (3D SPGR) pulse sequence and a long-circulating strictly intravascular
MRI contrast agent (Gd-PGC: 500,000 Da) [87] for absolute CBV and transvascular
water exchange measurement. Specifically, quantitative MRI measurement will be
performed in the exchange sensitive and insensitive regimes (i.e., at small flip angles
and large flip angles). As shown in Figure 4-4, the apparent cerebral blood volume
(BVapp, Equation 4.6) measured using relatively small flip angles is highly dependent
on the water exchange across the BBB based on computer simulation assuming a 2%
cerebral blood volume. Lowering the flip angle generally produces an overestimation
of the CBV. Such overestimation is dependent on the water exchange rate (WER),
and deviates further away from the true CBV as WER increases (Figure 4-4). However, the absolute CBV, which is quantified using MRI signal intensity measured
at relatively large flip angle (Equation 4.9), is insensitive to the bidirectional water
movement across the BBB (Figure 4-4). We therefore defined a secondary marker,
water exchange index (WEI), to evaluate transvascular water exchange (i.e., BBB
integrity). This secondary MRI-derived parameter, WEI, is defined as the ratio of
the BV app to the absolute CBV (Equation 4.8). Furthermore, WEI is approximately
linearly proportional to WER over exchange rate between 1 - 10 Hz, and only begins
to asymptotically depend on increasing WER beyond 10 Hz (Figure 4-5) based on
computer simulation using two-compartment model. Thus, using the proposed MRI
technique, we can achieve simultaneous measurement on the changes in the BBB
integrity and CBV.
The contrast agent, Gd-PGC, used in our study is strictly intravascular, and
has a long blood half-life (Figure 4-3) [87]. It, therefore, permits the monitoring of
temporal changes of CBV and BBB integrity. This method is advantageous over
[ Blood Signal after Gd-PGC
0/
Gd-PGC
1.5
2.5
2
3
3.5
4
Time after reperfusion (hr)
Figure 4-3: Time course of intravascular MR signal after intravenous administration
of Gd-PGC
dynamic contrast enhancement MRI technique in that water exchange sensitive MRI
signal intensity won't be affected by the change of blood T1 due to the clearance of
contrast agent from the blood stream. In addition, a 3D SPGR sequence is used in
data acquisition. Therefore the blood in-flow effect is minimized.
B Vapp
WE = B200
Q TP0
st
aCBV = -tissue,900
S Tpost
S blood,9 0 0
4.3
(4.8)
re
S rP
- Spre
'tissue,O90o
_
S
(4.9)
blood,900
Parameters affecting WEI
To understand the effect of imaging related parameters and contrast agent dose on
WEI measurement, we performed computer simulations using the two-comparment
model describe above. Specifically, we assumed a tissue T 1 of 1.8 sec, and blood T 1
of 2.3 sec before contrast agent administration. The choice of T1 values are based on
0.16
0.14
E
---------
------------------
1
-- ------
--------- ---------
S 0.12
--
'-
- E-3 Hz
A 5 Hz
10 Hz
---
..
.
-+-1 Hz
H
00
.4
0.04
0.06
------.------ ()
--- ------.--- ----------------- -- "--.- ...---------.---.- ---------.-----------------. . .- 60
. . . . .80
.
-... 100
40
20
0
Flip Angle (degree)
0-8
I 0.14 ------
1 Hz
----
--
--
- - - - - - - - - - - ~-----------------------------0.12
~~0.082
0
40
20
0
---------
. 0.06 -------
g
0.04 ------- -- ---
4
0.02
m
- - - - 10 Hz
- 5Z-
- -
---------------------------- ----------
-- -
am-4
me-
-
true CBV
0
20
40
60
80
100
Flip Angle (degree)
(b)
rate. MeasureFigure 4-4: Dependency of Vapp on flip angle, TR, and water exchange
ment at larger flip angle reflects true cerebral blood volume; while those at smaller
flip angles are sensitive to transvascular water exchange rate.
81
the T 1 measurements done at 9.4T [88, 89].
4.3.1
Effect of imaging related parameters
Measurement of WEI can be affected by imaging parameters other than flip angle.
For example, varying TR can result in different WEI values. In general, WEI increases as TR becomes longer, which in particular is the case for transvascular water
exchange rate (WER) ranging from 1 - 10 Hz (Figure 4-5a). Increasing TR elevates
the sensitivity of WEI measurement at low WER values on one hand, but reduces the
linear relationship between WEI and WER on the other hand. In addition, longer
TR usually preduces a longer imaging time.
Another imaging parameter that needs to consider is the echo time, TE. Although
Gd-PGC is used as a Ti agent, it also has transverse relaxation enhancement effect,
leading to reduced T 2 relaxation time. Moreover, Gd-PGC will be confined in the intravascular space once administrated into the blood stream. This tissue compartmentalization of Gd-PGC may generate local magnetic field gradient, leading to reduced
MRI signal intensity [58, 59]. If not taken into account, this susceptibility effect after
contrast agent administration might introduce error into the WEI calculation or even
cause experiment failure.
In the proposed MRI method, the quantification of CBV and WEI is based on flip
angle dependent water-exchange sensitive Ti-weighted MRI signal intensity. Errors
in flip angle may affect the MRI signal intensity, and ultimately introduce errors in
the CBV and WEI calculation. We therefore conducted error analysis to characterize
the effect of inaccurate flip angle on the CBV and WEI quantification. Computer
simulation was based on the two-comparmtent model described previously. Parameters used in simulation were: TR=30ms, tissue T 1 = 1.8 sec, blood T 1 = 2.3 sec
and 100 ms before and after contrast agent administration. We assumed that the
error in flip angle has a Gaussian distribution. The actual flip angles were randomly
sampled from the Gaussian distribution assuming either 10% or 20% error. As shown
in Figure 4-6a, good CBV measurement can still be achieved with an error of less
than 5% regardless of the extent of the error in the actually flip angle as long as a
10 mns
-A-30 nms
-
4
---
20 ms
40 ms
-I- 50nu -s4- omn
5
0
10
15
20
25
WER (Hz)
4.5
4
3.5
2.5
0
5
10
15
20
25
WER (Hz)
Figure 4-5: Dependency of calculated WEI on contrast agent dose (i.e. TR (a) and
intravascular RPost) (b) at different WER. Increasing contrast agent dose and TR can
increase the sensitivity of WEI. However, the linearity relationship between WEI and
WEI is reduced as TR increases.
0.022
*
--
C
0.0215
-
No erreor
10% error
- 20% error
0.021
OC
0.
m 0.0205
l
N
0.02
0.0195
0.019
1
I
5
0
10
15
20
25
Exchange Rate (Hz)
3
2.5
t
1.5
0
1±
no error
S10%error in flip angle
-20% error in flip angle
A
5
10
15
20
25
Exchange Rate
in flip angles
Figure 4-6: Error in CBV and WEI quantification due to the inaccuracy
at different transvascular water exchange rates.
large flip angle is used for imaging acquisition. There is only a slightly overestimation
of the true CBV when water exchange rate (WER) is high. On the other hand, WEI
is more sensitive to flip angle errors compared to CBV (Figure 4-6)b. Although the
average WEI measured with 10% and 20% errors in the actual flip angles is only
slightly deviated from the "true" WEI (i.e. no errors in flip angle), the standard
deviation becomes bigger as the error gets larger. Nevertheless, we can still perform
good WEI measurement with an error less than 10% over a large WER range if the
flip angle has an error of less than 10%.
4.3.2
Effect of contrast agent dose
The change of intravascular (blood) T 1, i.e., contrast agent concentration, can also
affect WEI quanitfication. Much like increasing TR affects WEI, increasing the contrast agent dose (i.e., decreasing intravascular longitudinal relaxation time, Tf/t)
increases the overall WEI values (Figure 4-5b), thus increasing the sensitivity of WEI
to WER. Decreasing T1, t also reduces the asymptotic dependence of WEI on WER
as shown in Figure 4-5b, thereby increasing the linearity between WEI and WER.
The sensitivity of WEI drastically increased with decreasing TIJ t particularly in the
range of relatively high WER.
4.3.3
Summary
Increasing TR or contrast agent dose can potentially result in better WEI estimation
based on the above computer simulations. However, the effect of imaging parameters
and contrast agent dose on the linear relationship between WEI and WER and the
sensitivity of WEI to WER must be considered when choosing these parameters for
CBV and WEI measurement. The benefit of increased sensitivity of WEI measurement by increasing TR is offset by the reduced linearity of WEI on WER (Figure
4-5a) and prolonged imaging time. In addition, the sensitivity is only amplified over
the low WER range, leading to a reduced detectable WER range (Figure 4-5a). On
the other hand, it is always beneficial to increase the contrast agent dose. Higher dose
Figure 4-7: Three-compartment tissue model assuming no water exchange between
intravascular and intracellular spaces
improves the linearity between WEI and WER, and thus providing a more proportional representation of WER (Figure 4-5a). Moreover, measurement sensitivity can
also be increased over a larger range of WER when contrast agent dose is increased
(Figure 4-5a).
4.4
Three-compartment model: effect of transcellular water exchange
4.4.1
Three-compartment tissue model
In the two-compartment tissue model, the extravascular extracellular space (interstitial space) and the intracellular space are modeled as one compartment assuming
fast transcellular exchange. However, Quirk et al [60] showed that water exchange
across the cell membrane was actually slow, an order of magnitude slower than the
water exchange across the erythrocyte membrane. We therefore incorporated a third
compartment, intracellular space, into previously described two-compartment model
consisting of the intra- and extra-vascular spaces (Figure 4-7). We further assumed
that there is no direct water exchange between the intravascular and the intracellular
spaces. The modified Bloch equations incorporating water exchange phenomena for
the three-compartment model can thus be described as:
dMb(t)
dt
dIev(t)
dMe(t)
dt
dMc()
dt
-
A-b,O
Mbkb,ev + Mevkev,b
(4.10)
Tl,b
iev,0
T ,ev
Me- M
-Io
Mekev,b - Mekev,ic + Mbkb,ev + Mickic,ev (4.11)
M - M
T1,ic
ickic,ev +
Mevkev,ic
(4.12)
where, Mb(t), Mev(t), and Mic(t) are the time-dependent magnetizations of the intravascular, the interstitial (extravascular extracellular), and the intracellular spaces,
respectively. Tl,b, T,ev, and T1,ij are the tissue longitudinal relaxation times in the intravascular, the interstitial, and the intracellular compartments. kb,ev and kev,b are the
rate constants of the water exchange from the intravascular space to the interstitial
space, and the interstitial space to the intravascular space, respectively. Parameters
kev,ic and kic,e, are the rate constants of the water exchange from the interstitial com-
partment to the intracellular compartment and from the interstitial compartment to
the intracellular compartment, respectively. At any moment, the total magnetization
is equal to the sum of the magnetizations from the three compartments:
M(t) =
i(t)
+ Mev (t) + Mic(t)
(4.13)
Under the condition of detailed balance, at equilibrium:
Mevkev,b
(4.14)
Mevkev,ic = Mickic,ev
(4.15)
Mbkb,ev
Previously, we explored various imaging parameters and contrast agent dose on
CBV and WEI quantification using two-compartment computer simulation. However, the two-compartment model cannot account for the effect of transcellular water
exchange on the CBV and WEI measurement. In addition, as the definition of WEI
indicates (Equation 4.8), any change in CBV can potentially affect WEI measurement
even though there is no change in BBB water exchange rate. Therefore, in order to
use WEl to estimate the changes of the BBB integrity, it is necessary to characterize
factors affecting WEl quantification and investigate the sensitivity of WEI to the
transvascular water exchange rate (WER) in relation to other parameters, in particular the transcellular water exchange and the CBV. Therefore, we performed computer
simulation using the three-compartment model described above to characterize the
effect of transcellular water exchange on the measurement of CBV and WEI, and to
look into the influence of CBV on WEI quantification.
4.4.2
Effect of transvascular and transcellular water exchange
on CBV and WEI measurements
The bidirectional water movement across cell membrane may affect the accuracy and
sensitivity of CBV and WEl measurement, especially under pathophysiological conditions. However, because of the existence of the two barriers, i.e., the BBB and
the cellular membrane, between the intravascular and intracellular spaces, the influence of the intravascular contrast agent on the intracellular water signal is probably
limited. Indeed, BV app measured at different flip angles is not significantly affected
by the transcellular water exchange as shown in Figure 4-8. In particular, accurate
CBV calculation (measurement at large flip angle) is not affected whether the tissue
is treated as two- or three-compartment model.
Plotted in Figure 4-9 is the simulation result of the dependency of WEI on water
exchanges across the BBB and cellular membrane using the three-compartment model
assuming a 2% intravascular volume. As revealed in this figure, very small differences
can be found when the water exchange rates across the cellular membranes are varied
from slow to fast, indicating a minimal effect of transcellular water exchange on WEI
measurement. It is also shown in Figure 4-9 that WEI is approximately linear to the
transvascular water exchange rate (WER) over the range between 0 - 10 Hz regardless
of the transcellular exchange rate, suggesting WEI is a not significantly affected by
the transcellular water exchange.
0.045
-
-
-
-
.
--
. .-..
.
- - .--..
.
-..
. .-
0
intra/ext
racellular
-------- -----
E o.o4
:10 Hz
-- -Intra/ext
. ---------- -----
> 0.035
go
0
.
racellular --
&
1Hz
------
0.03
---
----
transvascular:.
1
H
10
g 0.025
------- ------------------------------
I.
0.
CL
----------- -
0.02
0.015
0
20
40
60
80
100
Flip Angle
(a)
0.08
0.07
-4-intra/extra
,
--------- -------------cellular: 10"
0.06
-- ----- ---------
Hz
0.05
--
intra/extra
cellular: 1
------------------
Hz
--
transvascular:
----------
0.04
0.03
0.02
0.01
-- -- -
r
----.........-
--- -
3 Hz
r ........
-
r .......
',
,----,
,...--- ------r----.-- _-..
.......
I
I
I
I
I
I
I
I
I
0
20
40
60
Flip Angle
100
(b)
Figure 4-8: Effect of water exchange between interstitial and extracelllular spaces on
CBV.
89
2.4
2.2
2
1.8
transcellular water
exchange rate
1,
---
1 Hz
-0--3 Hz
1.4 -
r 5 Hz
-X- 7 Hz
-- 9 Hz
e---Two compartment
1.2
0
9
6
3
Transvascular Exchange Rate (Hz)
12
Figure 4-9: Effect of transcellular water exchange on WEI quantification.
4.4.3
Effect of intravascular volume on WEI measurement
usIn the proposed MRI technique, the transvascular water exchange is evaluated
BV'"
ing WEI, which is derived from the flip angle dependent exchange sensitive
measurement. As a result, changes in CBV may potentially affect WEI quantification, hence, the evaluation of the BBB integrity. Under normal, stable physiological
probconditions, the fractional volume ratio of intravascular and interstitial spaces
cellular
ably is not an influential factor on WEI measurement. However, dynamic
timeand hemodynamic changes under pathophysiological conditions can be highly
dependent and provide complex factors that might introduce errors in WEI measureWEI
ment. To better understand the effect of CBV (i.e. intravascular volume) on
the efquantification, a three-compartment simulation was conducted. We simulated
different
fect of different intravascular volume fractions on WEI measurement under
change, an
transvascular water exchange rates. Indeed, even without BBB integrity
(Figincrease in the intravascular volume fraction can lead to WEI underestimation
ure 4-10a). In addition, the underestimation of WEI gets greater as the intravascular
volume fraction becomes bigger or the transvascular water exchange rates gets higher
(Figure 4-10a). However, the influence of CBV on WEI calculation is marginal when
compared to the changes of WEI due to the underlying WER fluctuations, especially within the pathophysiological CBV range (shaded region, 4-10). This is further
demonstrated in Figure 4-10b, where calculated WEIs at different intravascular volume fractions start to deviate from one anther only when the BBB water exchange
becomes very fast.
4.5
Summary
Using computer simulation based on the two- and three-compartment models, we
demonstrated that the MRI-derived parameter, WEI, is approximately linearly proportional to the BBB water exchange rate (WER), and is sensitive to the WER
changes. This technique, therefore, allows us to evaluated the BBB damage by monitoring the changes in the BBB water exchange. By using a widely available MRI
pulse sequence (3D spoiled gradient echo), no complicated calculation algorithms or
fast MRI acquisition techniques are needed. In addition, the use of the long circulating intravascular MRI contrast agent allows us to perform steady-state measurement
and enables the dynamic monitoring of the CBV and BBB integrity changes under
different conditions. Furthermore, cerebral blood volume, high-resolution anatomical
information, and MRI angiograms can also be obtained in one data set, therefore,
providing a fast MRI technique to acquire multiple MRI-derived cerebrovascular parameters.
transvascular water
exchange rate (Hz)
2.4
-
0-0 -*-1
-*-3
-i-2
-- 4 -- 5
-+--6 -8
2
1.6
1.2
0.8
0
0.5
0.4
0.3
0.2
Voume Ratio
(intravascular/interstitial)
0.1
0.6
2.6
2.4
2.2
2
1.8
1.6
1.4
1.2
1
3
6
WER (Hz)
9
different BBB water
Figure 4-10: Effect of intravascular blood volume on WEI under
permeabilities.
Chapter 5
The in vivo characterization of the
proposed MRI technique and its
application in the eNOS knockout
mouse model
5.1
In vivo characterization using Mannitol administration and graded hypercapnia
5.1.1
Introduction
In Chapter 4, we described the theoretical background of an MRI technique for simultaneous BBB integrity and CBV detection, and explored the effect of imaging
parameters, contrast agent dose, and water exchange across the BBB and the cellular membrane on such measurements using computer simulation based on the twoand three-compartment tissue models. In this chapter, we performed in vivo experiments using two well-known physiological and pharmacological challenges (graded
hypercapnia and Mannitol) to further investigate and characterize the the ability
and sensitivity of this MRI method to independently detect and dynamically monitor
changes of CBV and BBB integrity.
Graded hypercapnia and Mannitol are known to influence both CBV and BBB
permeability, however, to relatively different extents. The elevation of the partial
pressure of arterial CO 2 (pCO 2 ) has a strong vasodilation effect, leading to an increase in CBV that is dependent on the level of arterial pCO [90, 91]. In addition,
2
hypercapnia has also been shown to cause an increase in BBB permeability [92, 93],
possibly due to vasodilation [22, 94]. Mannitol, an osmotic agent, is widely used to
reduce acutely raised intracranial pressure and to open the BBB for drug delivery to
the CNS [95, 96]. It can cause dramatic opening of the BBB, resulting in increased
BBB permeability [97]. However, its effect on CBV is still a subject of controversy.
Recently, an increasing number of reports have revealed that CBV increases in response to acute exposure to hyper-osmotic agent like Mannitol [98, 99, 100]. Thus,
the ability of the proposed MRI technique to detect BBB water exchange and CBV
changes independently can be explored using these two models. Moreover, we further
modified the existing protocol for Mannitol experiment by repeatedly administrating
Mannitol to retain the opening of the BBB, which allows us to detest the capability
of this MRI technique in monitoring the change of both BBB water exchange and
CBV.
Therefore, the goals of this study are to test the ability and sensitivity of this MRI
method to detect the BBB integrity and CBV changes, to examine its ability to detect
such changes independently using graded hypercapnia and Mannitol- induced BBBopening experiments, and to investigate its ability to monitor the dynamic changes
of BBB water exchange and CBV using repeated Mannitol bolus injections.
5.1.2
Materials and method
Intravascular MR Contrast Agent:
Protected graft copolymer bearing covalently linked Gd-DTPA residues (Gd-PGC)
was synthesized and purified as described previously [87]. The batches of Gd-PGC
were characterized on size-exclusion HPLC column (TSK3000SW, Supelco, Bellefonte
PA) and gadolinium concentrations were determined by elemental analysis.
The
polymer was dissolved at 50 mg/ml (13 mM Gd) in DPBS, at pH 7 and sterilefiltered.
Animal preparation:
A total of nine male C57BL/6 mice (8-12 wks old, around 21g each) were used in
the study. Six of them were prepared with femoral vein cannulation in each leg:
one for constant Mannitol infusion and the other for contrast agent (Gd-PGC) administration and Mannitol bolus injection. The other three mice were used for the
graded hypercapnia experiment. The right femoral vein cannulation was prepared for
contrast agent (Gd-PGC) injection. Continuous inhalation of gas mixture (30% 02
and 70% N20 2) with 1.5 % isoflurane was used for anesthesia during surgical preparations. Immediately following surgery, mice were relocated to a head coil for MRI
data acquisition. During MRI, the mice were anesthetized with 1.5 % isoflurane in
a 1:1 gas mixture of medical air and oxygen while body temperature ( 36.5°C) was
monitored.
MRI acquisition:
All MRI data were acquired at 9.4T (Magnex Scientific Ltd, Oxford, UK) interfaced
with the ParaVision system (Bruker Biospin, Billerica, MA) using a gradient strength
of 150 Gauss/cm. Mouse volume head coil was used for all image acquisition. To
eliminate the in-flow effect in 2D acquisition, a 3D-SPGR (spoiled gradient echo) pulse
sequence was used for imaging. MRI images were collected with TR/TE = 30/3 ms,
matrix size = 64 x 64 x 128, FOV = 1.6 x 1.6 x 3.2 cm, number of average = 1, and
varying flip angles at 30' and 90'. Additional 3D-SPGR scans were performed with
TR/TE = 30/[5, 7] ms at the fixed flip angle of 300 for calculating the T2 change due
to the injection of the intravascular contrast agent.
The experiment paradigm for the animal group injected with Mannitol is shown in
figure 5-1a. Specifically, after two sets of baseline data collection, we started to infuse
the mouse with 20% Mannitol at a rate of 100 pl/hr. In addition, three Mannitol
boluses (20%, 100 rpl) were given every 30 minutes. Two sets of 3D volumetric images
as described above were acquired after each bolus administration.
For the animal group subject to the CO 2 challenge, two different CO 2 doses (10%
and 15%) were given. After two sets of baseline image collection, two serial CO 2 levels
(10% and 15% respectively) were performed (5-1b).
During each CO 2 challenge,
MRI data were collected five minutes after the mice were given 10% or 15% CO .
2
Following CO 2 challenges, the gas mixture was switched back to 1:1 medical air
and oxygen. Another baseline CBV and WEI measurement was then performed 10
minutes afterwards.
Data analysis:
Region of interest (ROI) analysis was used to investigate the changes of absolute
cerebral blood volume (CBV) and WEI under these two physiological and pharmacological challenges. The ROI in the blood compartment was drawn within the inferior
cerebral vein (http://www.informatics.jax. org/).
The area of cerebral cortex
supplied by the middle cerebral artery (MCA) was manually outlined based on the
Paxono's mouse brain atlas [64]. The CBV under baseline, Mannitol injection or CO
2
inhalation was calculated using Equation 4.9 based on the signal intensity obtained at
a flip angle of 900, and WEI was evaluated using Equation 4.8. Because the intravascular contrast agent can also affect the transverse (T 2 or T 2 *) relaxation, regional
T2* was calculated based on the data acquired using the 3D SPGR sequence with
variable TE (i.e., 3.0, 5.0, and 7.0 ms). The value of calculated T 2 * was then used
for signal correction to obtain the purely Ti-weighted MRI signal for accurate CBV
and WEI estimation.
Statistical test:
Unless otherwise specified, all statistical analyses for the ROI presented in the study
were performed using a paired two-tailed t-test. For correlation analyses, we applied
the Pearson product-moment correlation test (and linear regression). All of the numerical data were presented as averages ± standard error of the mean. Statistical
scan
scan
wall"',
SC
2
->[scan 3
scan 4
scan7
scan8
administration (a) and CO 2 challenge (b)
Figure 5-1: Experiment Paradigm: intravenous Mannitol
significance was accepted at a confidence level of 0.95.
5.1.3
Results
Representative images from different brain regions before and after contrast agent
injection were shown in Figure 5-2a. In these Ti-weighted images, large blood vessels
and venous sinuses are clearly visible after intravenous Gd-PGC injection. The MRI
signal in the blood compartment remained time-independent after Gd-PGC administration, indicating no-leakage and long blood half-live of Gd-PGC (Figure 5-2b).
Figure 5-2c is the MRI angiogram of a typical mouse brain, which is readily available
with 3D gradient echo sequence. The baseline CBV in mouse cerebral cortex was
measured to be 2.48% ± 0.18%, and WEI was 1.52 + 0.12.
The responses of CBV and WEI to CO 2 challenge were shown in Figure 5-3a
and 5-3b, respectively. Significant increase of CBV over baseline after CO 2 challenge
were observed. CBV increased by - 50% due to the increased amount of CO 2 in the
breathing gas mixture. However, no significant increase in CBV was observed during
15% CO 2 challenge when compared to that during 10% CO 2. On average, WEI was
higher compared with baseline values. However, no statistical significant difference
was detected.
The temporal changes of fractional cerebral blood volume (CBV) and water exchange index (WEI) induced by the serial Mannitol bolus injections are shown in Figure 5-4a and 5-4b, respectively. Both CBV and WEI in the cerebral cortex increased
significantly following an initial intravenous Mannitol injection. CBV continued to
increase following subsequent Mannitol injections. On average, WEI also increased
after the second and the third bolus injections when compared to the WEI after the
first bolus. However, no statistical significance was reached. On the other hand, the
WEI at all time points remained statistically higher than the baseline WEI.
Plotted in Figure 5-5 is the relationship between WEI and CBV for Mannitol
(excluding the baseline) and graded CO 2 challenges. The slopes of the WEI vs. CBV
curves due to repeated Mannitol injection (i.e. bolus 1, 2 and 3) and CO 2 inhalation
are very similar to each other.
4
3.5
3
u
i 2.5
2
0
N
S1.5
E
o 0.5
0
(b)
(c)
and
mouse brain acquired before (top panel)
Figure 5-2: Representative images of
of
administration (a), and temporal evolution
after (bottom panel) contrast agent
contrast
MRI
half life of this intravascular
vascular MRI signal showing the long
structure (c).
agent (b), 3D angiography revealing vascular
0.05
0.04
0.03
0.02
ooo"
A
2.1
2
1.9
1.8
1.7
1.6
1.5
1.4
Figure 5-3: Changes of CBV and WEI in response to CO 2 challenge: CBV (a) and
WEI (b).
100
0.05
I
*
I
I
I
I
0.04 -
0.03
0.02
(a)
2.1
*
2
*
1.9
S1.8
1.7
1.6
1.5
1.4
(b)
Figure 5-4: Changes of CBV and WEI after intravenous Mannitol administration:
CBV (a) and WEI (b).
101
_ __
~~_~~~~~
2.1
y = 10.96x + 1.4385
2
R2 = 0.8535
1.9
1.8
1.7
L3.175x + 1.2641
1.6
R2 = 0.9965
1.5
0.02
0.025
0.03
0.035
0.04
0.045
0.05
CBV
*
-
*
C02
-
- Linear (C02)
Mannitol
- Linear (Mannitol)
Figure 5-5: Plots of WEI vs. CBV due to repeated Mannitol boluses (bolus 1, bolus
2, and bolus 3) or CO 2 inhalation.
102
5.1.4
Discussion
The blood-brain barrier is crucial to normal brain function. Even subtle functional
changes without morphological alterations can cause long lasting effect on disease development and progression. For example, in ischemia and traumatic brain injury, the
opening of the BBB can lead to vasogenic edema, resulting in increased intracranial
pressure, severe neurological damage, and even death. It is therefore necessary to understand the pathological changes of the BBB, which may help to better understand
disease evolution and provide effective treatment options and methods for evaluating
the efficacy of pharmacological intervention. Currently, the commonly used vascular
permeability measurements are based on the extrinsically administrated molecules
(such as staining dyes and Gd-DTPA). BBB disruption is evaluated by examining
either the leakage profile of these staining dyes on dissected tissue slices or hyperintensity regions in MRI images. Since either the brain is destroyed or time is needed
for the MRI contrast agent to clear out of the brain, these methods are not readily
adaptable to the dynamic monitoring of BBB alteration. In this study, we demonstrated that the proposed MRI technique can be used not only to detect the acute
change of CBV and WEI, but also to monitor their temporal evolution by employing water as an intrinsic biomarker. Moreover, the ability to detect the different
CBV and WEI responses under graded hypercapnia and Mannitol challenges further
demonstrated that sensitive and independent measurement of BBB water exchange
and CBV can be achieved using this MRI technique. In addition, this MRI technique
does not require accurate tissue and blood T 1 measurement as in the method proposed by Schwarzbauer et al. [83], and thus allows fast 3D imaging acquisition and
dynamic CBV and WEI monitoring.
Changes of CBV and WEI in response to CO 2 inhalation
CO 2 is a potent vasodilator and has been shown to cause CBV increase that is dependent on partial pressure of blood CO 2 (pCO 2 ) [91]. Hypercapnia, thus, is often
used as vascular challenge in clinical applications and basic research. In this study,
103
we used 10% and 15% CO 2 in the breathing gas mixture.
Significant increase of
CBV over baseline was observed during the two CO 2 challenges. However, no significant increase in CBV was observed during 15% CO 2 challenge when compared
to that during 10% CO 2 . Since the total blood volume of mouse is very small (6 - 8 ml/100g, [101]), it is hard to withdraw blood to measure blood gases during
experiment. We therefore had no knowledge about the blood pCO 2 level during the
experiment. Furthermore, the mice were not ventilated during the experiment. It was
therefore difficult to control the blood CO 2 level by changing the percentage of CO
2
in breathing gas mixture. The small but not significant increase in CBV probably reflects slightly different pCO 2 level during the two CO 2 experiments. Nevertheless, all
animals showed strong response to CO 2 challenge. CBV was significantly increased
from 2.7% to around 4.0%, to 4.3%, and then went back to 2.8% (Figure 5-3), suggesting good sensitivity of CBV measurement. Previous studies using labeled protein
[94] or radio isotope labeled sucrose [102] demonstrated increased BBB permeability
under hypercapnia condition. Different mechanisms have been suggested to explain
the observed change in BBB permeability. Johansson et al. [94] suggested that BBB
dysfunction was the combined effect of cerebral vasodilation and high blood pressure.
On the other hand, Mollgard & Sorensen [22] revealed that the cerebrovascular tight
junction became more permeable in hypercapnia probably due to vasodialtion. In
current study, we didnt detect significant WEI increase probably due to the substantial inter-subject variability. However, there is a clear trend that the averaged WEI
increased during the two CO 2 challenge periods (Figure 5-3b), indicating small but
appreciable increase of BBB water exchange to water. This increase of BBB water
exchange is probably due to hypercapnia-induced vasodilation
Changes of CBV and WEI in response to Mannitol injection
Mannitol is widely used to lower elevated intracranial pressure (ICP) [99, 103] and to
temporally open the blood-brain-barrier (BBB) for drug delivery [95, 96]. Although
the exact mechanism of Mannitol on tight junction structure and function is unknown,
the opening of BBB is thought to be the result of the disengagement of tight junctions
104
between adjacent endothelial cells due to endothelial cell shrinkage [97]. This leads to
the disruption of BBB integrity resulting in increased BBB water exchange . On the
other hand, studies regarding cerebral blood volume change in response to Mannitol
administration have been ambiguous; some revealed a decrease in CBV [104], while
others showed significant CBV increase [98, 100]. In our study, we showed that there
was a significant acute increase of CBV over the baseline at the cortical level after
an initial Mannitol injection, which is in good agreement with previous studies using
MRI [100] and other methods [99] (Figure 5-4a).
The BBB water exchange rate
as reflected by WEI also increased acutely and significantly after the first Mannitol
injection (Figure 5-4b), demonstrating the ability and sensitivity of this technique to
detect increased BBB water exchange rate associated with Mannitol induced BBB
opening. The responses of CBV and WEI to subsequence Mannitol bolus injections
were different. The CBV continued to increase significantly (Figure 5-4a). However,
the measured WEI did not change significantly compared to that in response to the
first Mannitol bolus (Figure 5-4b), suggesting that water exchange across the BBB is
probably a tightly controlled process.
Sensitivity of WEI to WER and the effect of CBV changes on WEI
Measurement of CBV showed robust and significant response to Mannitol induced
BBB opening and graded CO 2 challenges. However, WEI responded differently under these two conditions. There was a dramatic increase of WEI over the baseline
due to initial Mannitol injection; while, no significant change of WEI was observed
during CO 2 challenge. Interestingly, when the WEI vs. CBV curves obtained under
the two experiment conditions were compared, we found that the slope of the WEI vs.
CBV curve due to repeated Mannitol injections (excluding the baseline) was similar
to that obtained during CO 2 challenge (Figure 5-5), implying that the WEI increases
due to the 2nd and 3rd Mannitol injections are probably caused by the same mechanism that is also responsible for the WEI increase during CO 2 inhalation (i.e., CBV
increase). On the other hand, the large increase in WEI due to the 1st Mannitol bolus is more likely induced by the sudden change in osmotic pressure across the BBB,
105
demonstrating that WEI is highly sensitive to the underlying BBB integrity change.
Computer simulation using three-compartment water exchange model showed that
WEI was approximately linearly proportional to WER (Figure 4-9). It also demonstrated that changes in CBV could affect WEI measurement (Figure 4-10a). However,
compared to the changes WER on WEI, the effect of CBV change on WEI measurement is small, and probably is not an influencing factor (Figure 4-10b). Indeed, in
vivo experiments using repeated intravenous Mannitol bolus injections and graded hypercapnia revealed that different degree of BBB inegrity change can be detected even
though the underlying CBV may be similar (Figure 5-5), indicating that independent
measurement of CBV and WEI can be made using this MRI technique.
In summary, MRI experiments using the two well known physiological and pharmacological challenges, graded hypercapnia and Mannitol, demonstrated that the
proposed MRI technique can detect changes of BBB water exchange and CBV independently and sensitively. More importantly, since many diseases affecting the central
nervous system are associated with damaged or disrupted BBB, this technique provides means to simultaneously monitor the alteration of BBB integrity and the change
of cerebral blood volume, which may help to understand the pathological mechanisms
of disease progression, to discover therapeutic treatment options, and to investigate
the efficacy of such treatments.
5.2
The change of baseline MRI-derived vascular
parameters in the eNOS knockout mouse model
5.2.1
Introduction
The Nitric Oxide (NO) produced by endothelial nitric oxide synthase (eNOS) plays
an important role in regulating the vascular tone and maintaining the blood pressure [105]. It also has antithrombotic, anti-inflammatory, and antiproliferative effect
by inhibiting vascular smooth muscle proliferation, platelet aggregation, and leukocyte adhesion [106]. eNOS knockout (eNOS-/-) mice were therefore generated to
106
study and understand the role of eNOS in vascular function and cerebrovascular diseases. Mice lacking endothelial derived NO have systematic hypertension [107, 108],
and develop larger cerebral infarction [109]. However, despite the extensive studies
performed using eNOS-/- mice, the baseline vascular parameters of the eNOS-/mouse model have not been fully documented yet.
Magnetic resonance imaging (MRI) has become an important tool in the study
and understanding of the pathological mechanisms and development of various human diseases using mouse models. It is advantageous over other imaging methods
by providing both anatomical and physiological information. Previously, Donahue et
al. showed that accurate cerebral blood volume (CBV) can be measured using the
widely available gradient echo sequence and an intravascular MRI T contrast agent
[82], but the accuracy of the CBV measurement was highly affected by the water exchange rate across the BBB. She further demonstrated that the source of error comes
from the sensitivity of the MRI signal to transvascular water exchange at different flip
angles. This lead to our idea of extracting the transvascular water exchange (WE)
information by using the flip-angle dependent, water exchange sensitive Ti-weighted
MR signal intensity [110] (details in Chapter 4 and Section 5.1). In addition, information regarding the vessel size and the mean vessel density can also be obtained
if MRI susceptibility contrast agent is used [59, 111, 112, 113]. These MRI derived
physiological parameters, i.e., CBV, transvascular WE, vessel size, and mean vessel
density are all affected by cerebrovascular pathophysiological changes under different
disease conditions. Thus, characterizing these vascular parameters in eNOS-/- mice
is not only important in the understanding of the phenotypic effect of eNOS gene
but also crucial in the understanding of eNOS functions under different pathological
conditions.
In current study, we aim to establish the baseline vascular parameters, i.e., CBV,
- /transvascular water exchange, vessel size, and mean vessel density in the eNOS
and the wild type (WT) mice using MRI, and to understand whether there is cerebral
- / - mice. Our results demonvascular change other than hypertension in the eNOS
strated that there are significant differences in average vessel size, mean vessel density,
107
and BBB water exchange between eNOS-/- and WT mice under normal (healthy)
condition. However, no significant difference in CBV was detected. These findings
suggest that there are basal morphological and functional changes of the cerebral
vasculature as a result of the knockout of eNOS gene. Establishing these baseline
changes using MRI-derived parameters will help interpret and understand the effect
of endothelial NO in various disease states.
5.2.2
Materials and methods
Animal preparation:
A total of eight male WT C57BL/6 mice (-12 wks old and about 24g each) and
five eNOS-/- mice ( 12 wks old and 24g each) were used in this study. An intravenous catheter was placed in the right femoral vein in each animal for contrast agent
injection. Continuous inhalation of gas mixture (30% 02 and 70% N20) with 1.5
% isoflurane was used for anesthesia during the surgical preparations. Immediately
following surgery, mice were relocated to a mouse head coil for MRI acquisition. During MRI acquisitions, the mice were anesthetized with 1.5 % isoflurane in a 1:1 gas
mixture of medical air and oxygen while body temperature (- 36.50C) was monitored.
MRI acquisition:
All MRI experiments were performed at 9.4T (Magnex Scientific Ltd, Oxford, UK) interfaced with the ParaVision system (Bruker Biospin, Billerica, MA) with a gradient
strength of 150 Gauss/cm. A volume head coil was used for all image acquisition. To
eliminate the in-flow effects in 2D acquisition, 3D SPGR (spoiled gradient echo) pulse
sequence was used for imaging acquisition. 3D images were acquired with TR/TE =
30/3 ms, matrix size = 64 x 64 x 128, FOV = 1.6 x 1.6 x 3.2 cm, one average, and
varying flip angles at 30' and 90'
.
Additional 3D SPGR scans were performed with
TR/TE = 30/[5.0, 7.0] ms at the fixed flip angle of 30' for calculating the T2 * change
due to the intra-vascular contrast agent. Prior to and following contrast agent administration (Gd-PGC: 17 ymol Gd/kg), two sets of 3D images were acquired for CBV
108
and WEI evaluation. Following the CBV and WEI measurement, superparamagnetic
iron oxide nanoparticles (SPION: 20mg/kg) was injected for the vessel size and mean
vessel density measurement. Multi-slice multi-echo spin echo sequence (TR/TE =
3000/[10 20 30 40 50 60 70 80 90 100 110 12] ms, FOV = 1.5 x 1.5 x 0.5 cm, matrix
size = 96 x 96) was used for the T2 relaxation times before and after SPION administration. 2D spoiled gradient echo images with four different TEs (TR/TE: 1000/[3
5 7 10] ms, FOV = 1.5 x 1.5 x 0.5 cm, matrix size = 96 x 96) were acquired for the
tissue T 2* relaxation times before and after SPION injection.
Data analysis:
Region of interest (ROI) analysis was used to investigate the changes of absolute
cerebral blood volume (CBV), water exchange index (WEI), vessel size index (VSI),
and mean vessel density (MVD) in the eNOS - / - mice and WT mice.
An ROI
in the blood compartment was drawn within inferior cerebral vein (http://www.
informatics . j ax. org/). The area of cerebral cortex supplied by the middle cerebral
artery (MCA) and the sub cortical areas (striatum and thalamus) were manually
outlined based on the Paxonos' mouse brain atlas [64]. CBV was calculated using
equation based on the MRI signal at a flip angle of 90', and WEI was evaluated
with Equation 5.1. Since intravascular contrast agent can affect transverse (T2 or
T2*) relaxation as well, regional T2 * was also calculated based on the 3D SPGR MRI
signal acquired at varying TE (i.e., 3.0, 5.0, and 7.0 ms). The values of the T2 * before
and after contrast agent injection were then used for signal correction to obtain the
purely TI-weighted signal intensity for accurate CBV and WEI estimation. For the
vessel size and mean vessle density characterization, the R 2 and R 2 * in the cerebral
cortex and sub cortical areas before and after SPION administration were calculated
based on least square fitting using MATLAB. Vessel size index was determined from
the ratio of change of R 2 * to that of R 2 (Equation 5.2), and mean vessel density index
(MVD) was calculated using Equation 5.3.
WEI
= CBV 3 0ICBV 90
109
(5.1)
VS
= AR/AR2
MVD = AR 2 /(AR*)
(5.2)
2
(5.3)
Statistical test:
Unless specified otherwise, all statistical analyses for the ROI presented in this study
were performed using a two-tailed t-test. All data were presented as the average ±
standard error of the mean. Statistical significance was accepted at a confidence level
of 0.95.
5.2.3
Results
Representative T2-weighted images of the wild type mouse and the eNOS-/- mouse
are shown in Figure 5-6a and b, respectively. No apparent anatomical difference can
be observed in the eNOS-/- mice. However, as shown in Figure 5-7a, the vessel
size index (VSI) defined as the ratio of AR 2* over AR 2 is significantly smaller at the
cortical level in the eNOS-/- mice compared to that in the wild type (WT) mice.
At the sub-cortical level, the VSI is also smaller in the eNOS-/- mice than that in
the WT mice. However, no statistical significance was reached. In addition, average
vessel sizes in the sub-cortical area are higher than that in the cerebral cortex for
both eNOS-/- and WT animals.
Figure 5-7b summaries the results of another MRI derived parameter: mean vessel density index (MVD) through the use of susceptibility contrast agent (SPION).
Cerebral cortical MVD is significantly higher in the eNOS-/- mice than that in WT
mice. No statistical difference of MVD in sub-cortical area was detected between the
eNOS KO and the WT mice. In addition, we observed a significantly higher MVD
in the cerebral cortex than the MVD in the sub-cortical area for both eNOS KO and
WT mice.
Quantitative comparisons of CBV and WEI between the eNOS-/- and the WT
mice are shown in Figure 5-8a and 5-8b, respectively. There is no statistically significant difference of CBV at both cerebral and sub-cortical levels between the eNOS-/110
Figure 5-6: T2-weighted images of a wild type mouse (a) and eNOS-/- mouse (b)
after SPION injection
111
35 r*eNOS
KO
F WT
I
~T
cortex
sub cortical
0.35
0.33
0.31
0.29
0.27
0.25
0.23
0.21
0.19
0.17
0.15
cortex
sub cortical
(b)
Figure 5-7: Comparison of VSI and MVD at cortical and sub-cortical levels between
eNOS KO and WT mice.
112
0.028
SeNOS KO
WT
0.026
0.024
0.022
0.02
0.018
0.016
0.014
0.012
0.01
sub cortical
cortex
1.9
N eNOSKO
1.8
E1WT
T
1.7
T
1.6
1.5
1.4
1.3
1.2
1.1
1
sub cortical
cortex
(b)
Figure 5-8: Comparison of CBV at cortical and sub-cortical levels between eNOS KO
and WT mice.
113
(a)
(b)
(c)
(d)
Figure 5-9: H&E (magnification: 400x) staining of the cerebral cortex of wild type
(a, c) and eNOS knockout (b, d) mice. Morphologically, eNOS knockout mouse is
characterized by higher number of blood vessels with smaller vessel size in the cerebral
cortex when compared with wild type mouse.
and WT mice (Figure 5-8a). The water exchange index (WEI) in the eNOS-/- mice
is significantly higher than that in the WT mice at both cortical and sub-cortical
levels (Figure 5-8b).
Figure 5-9 shows the standard H&E staining of two cerebral cortex areas (medial
and lateral) supplied by the middle cerebral artery in the eNOS-/- and the wild
type mice, respectively. On average, the eNOS-/- mice (Figure 5-9b and d) seem to
have a larger number of vessels with reduced vessels size compared to the wild type
mice (Figure 5-9a and c). Further quantitative analysis based on H&E staining also
showed that the average vessel is smaller in the eNOS-/- mice.
114
5.2.4
Discussion
The eNOS-/- mice are normal in terms of appearance and general behavior [107,
108].
From the T2-weighted images (Figure 5-6a - wild type and b - eNOS-/-),
no apparent anatomical difference can be found in the eNOS - / - mice (Figure 56a), indicating that there is no gross brain anatomy change in the eNOS knockout
mice. Measurement of the whole brain volume based on the Ti-weighted images also
revealed no significant difference in total brain volume between the eNOS-/- mice
and the wild type mice. However, MRI measurement showed significant differences
in cerebral vasculature between the eNOS knockout and the wild type mice. The
average vessel size as reflected by VSI is significantly reduced in the eNOS - / - mice
at the cortical level (Figure 5-7a), likely due to the reduced basal vasodilation as a
result of the loss of endothelial NO production. On the contrary to what we expected,
no significant difference in the cerebral blood volume (CBV) at the cortical level was
reached between the eNOS-/- and the WT mice (Figure 5-8a), suggesting a possible
increase in the vessel density in these eNOS-/- mice. Indeed, the MRI measurement
of the average vessel density revealed an increase in the mean vessel density in the
eNOS-/- mouse(Figure Chap5:fig:VSI:MVD)b, probably as a compensation for the
reduced vessel size in order to maintain similar basal CBV. The H&E staining of the
cerebral cortex revealed that the eNOS-/- mouse tends to have thinner vessels and
larger number of vessels in the cortex than the wild type mouse (Figure 5-9). Together
these results demonstrate that there is baseline cerebral vasculature change in the
eNOS - / - mice, indicating the importance of eNOS in normal vascular morphology.
More importantly, we observed a significant increase in the calculated water exchange index (WEI) in the eNOS knockout mouse (Figure 5-8b), suggesting an increased BBB water exchange in the eNOS-/- mouse under normal physiological condition. It has been shown previously that there is intrinsic vascular vulnerability and
perhaps increased BBB permeability because of the chronic hypertension [114, 115].
The eNOS-/- mouse is known to have systematic hypertension as a result of the
loss of endothelial derived NO production [107, 108]. Consequently, the observed in115
crease of BBB water exchange is very likely caused by this systematic hypertension in
eNOS knockout mouse, thus reflecting the alteration of BBB integrity. This baseline
BBB water exchange increase in the eNOS-/- mouse may have great implications
in the understanding of the biophysical and pathological effects of eNOS on stroke
progression and provide a possible preventive or treatment option in stroke management. For future studies, it will be important to further investigate the biochemical
or physiological basis of this basal BBB water exchange change, for example, through
immunohistological assays (e.g. changes of water channels expression). In addition,
the degree of these vascular changes on vascular function and physiology should be
tested using various vascular stimuli such as gas exposure to CO and 022
In summary, we measured the baseline cerebral blood volume, average vessel size,
mean vessel density, and BBB water exchange at both cortical and sub-cortical levels
in the eNOS-/- and the wild type mice. Our results demonstrate that even though
there is no gross anatomical or behavior changes in the eNOS knockout mouse, there
exist baseline cerebral vasculature alterations, in both morphology and function, in
these knockout animals. Understanding these basal vascular changes may help to
further understand the role of eNOS in various cardiovascular diseases and to identify
potential therapeutic interventions.
116
Chapter 6
In vivo BBB water exchange
measurement and ex vivo MRM
staining of a mouse model of
transient ischemic stroke
6.1
The detection of BBB water permeability change
in a mouse model of transient ischemic stroke
6.1.1
Introduction
Following an ischemic insult, the affected brain tissue undergoes a series of changes
associated with initially reversible damages that eventually evolve into irreversible
tissue damage [116, 117]. Biophysical alterations due to ischemic injury include cell
swelling, reduced water diffusion, disruption of the blood-brain-barrier (BBB), and
loss of cell membrane integrity. Early restoration of the blood supply, therefore,
is crucial in stroke recovery and favorable patient outcomes. Despite the probable
efficacy, reperfusion therapies are often hindered by the limited time window and by
the risk of hemorrhagic transformation. Among various factors, the weakening of
117
the blood-brain-barrier (BBB) plays an important role in stroke progression and the
effectiveness of treatments.
Recent studies using immunohistochemical assays revealed that perivascular water
channel proteins (aquaporins), mediators of bidirectional water flow across BBB, are
impaired and/or dysfunctional as early as one hour after ischemic onset [118]. Such
early alterations of the intra/extra vascular water exchange across the blood-brain
barrier provide a measurement target of interest for monitoring the acute weakening
and breakdown of BBB. This motivated us to use intrinsic water molecules as a
qualitative and quantitative biomarker to investigate the early BBB impairment.
In previous chapter, we presented a novel MRI technique, which can be used to
quantitatively monitor the bidirectional water movement across the BBB (i.e., BBB
water exchagne). Using computer simulation and in vivo graded hypercapnia and
Mannitol induced BBB opening experiments, we demonstrated that transvascular water movement, or intra-/extra-vascular water exchange can be quantified for assessing
the integrity of the BBB using a traditional 3D spoiled gradient echo sequence and a
long-circulating intravascular contrast agent (Gd-PGC). The goal of this study was to
investigate the acute changes of BBB integrity and CBV due to ischemic insult in the
transient 1 hr filament middle cerebral artery occlusion (MCAo) mouse model using
this new MRI technique by employing the trans-vascular movement of the intrinsic
water molecules as the measurement marker.
6.1.2
Material and Method
Animal preparation:
A total of eight male WT C57BL/6 mice (-10 wks old, around 24g each) were used
in the study. Continuous inhalation of a gas mixture (30% 02 and 70% N 20) with
1.5% isoflurane was used for anesthesia during the surgical preparation. The standard
intraluminal middle cerebral artery occlusion method was used. Briefly, an incision
was made in the external carotid artery, and a silicon-covered 8-0 nylon monofilament
was advanced through the ICA to the origin of the anterior cerebral artery to occlude
118
the MCA and posterior communicating artery. In addition, an intravenous catheter
was placed in the right femoral vein in each animal for contrast agent injection. One
hour after occlusion, the filament was withdrawn, and external carotid artery was
tied off.
Immediately following reperfusion, mice were relocated to a mouse head coil for
MRI acquisition. During MRI acquisitions, the mice were anesthetized with 1.5 %
isoflurane in a 1:1 gas mixture of medical air and oxygen while body temperature
( 36.5 'C) was monitored.
MRI acquisition:
All MRI experiments were performed at 9.4T (Magnex Scientific Ltd, Oxford, UK)
interfaced with the ParaVision system (Bruker Biospin, Billerica, MA) using a gradient strength of 150 Gauss/cm. Mouse volume head coil was used for all image
acquisition. To eliminate the in-flow effect in 2D acquisition, 3D SPGR (spoiled gradient echo) pulse sequence was used for imaging acquisition. 3D images were acquired
with TR/TE = 30/3 ms, matrix size = 64 x 64 x 128, FOV = 1.6 x 1.6 x 3.2 cm,
number of average = 1, and varying flip angles at 300 and 900. Additionally, the 3D
SPGR scan was performed with TR/TE = 30/[5.0, 7.0] ms at the fixed flip angle of
30' for calculating the T 2* change due to the intra-vascular contrast agent. Prior to
and following contrast agent administration (Gd-PGC: 17 ,pmol Gd/kg), two sets of
3D images were acquired to evaluate the the CBV and WEI as a result of stroke.
Data analysis:
Region of interest (ROI) analysis was used to investigate changes of absolute cerebral blood volume (CBV), water exchange rate (WER). ROI pertaining to signal from
blood compartment was drawn within inferior cerebral vein (http: //www. inf ormatics.
jax.org/). Area of cerebral cortex supplied by middle cerebral artery (MCA) was
manually outlined based on Paxon mouse brain atlas [64]. CBV was calculated using
equation based on images acquired flip angle of 900, and WEI was evaluated with
Equation 5.1. Since intravascular contrast agent can affect transverse (T2 or T2 *)
119
relaxation as well, regional T2 * was calculated based on 3D SPGR with varying TE
(i.e., 3.0, 5.0, and 7.0 ms), the value of which was used for signal correction to obtain
the purely Ti-weighted signal intensity for accurate CBV and WEI estimation.
Statistical test:
Unless specified otherwise, all statistical analyses for the ROI presented in the study
were performed using a two-tailed t-test. Data were presented as average ± standard
error of the mean. Statistical significance was accepted at a confidence level of 0.95.
6.1.3
Results
MRI angiography of a transient one hour MCAo mouse is shown in Figure 6-1b.
Previously occluded left middle cerebral artery (Figure 6-la, orange arrow) was partially reperfused after filament withdrawn. Representative diffusion-weighted images
(DWI, top panel) and apparent coefficient (ADC) maps (bottom panel) are shown
in Figure 6-1b. Ischemic lesion is clearly visible in DWI and ADC maps around two
hours after reperfusion (three hours after the onset of occlusion). Interestingly, in
these relatively early stage after reperfusion, spatial heterogeneity of ischemic region
is clearly visible in both DWI and ADC maps. In addition, ischemic border is characterized as hyperintensity in DWI or lower ADC values in ADC maps (Figure 6-1b).
Measurements of the absolute CBV and transvascular water exchange index (WEI)
were conducted approximately two hours after the start of reperfusion. Of the total 8
mice studied, two distinct groups of animals can be classified based on the CBV values
after reperfusion. Group one has poor ipsilateral CBV restoration; while, group two
has significantly better ipsilateral CBV restoration that is also significantly different
from contralateral CBV (Figure 6-2a).
Interestingly, animals with low ipsilateral
CBV showed a dramatic increase of WEI, indicating the possible severe BBB damage
(Figure 6-2b, group 1). On the other hand, animals with the restored CBV (Figure
6-2a, group 2), had an ipsilateral transvascular WEI similar to the contralateral WEI
120
occluded MCA (a, orange arrow), and representative
Figure 6-1: MRI angiography showing partial reperfusion of previously
maps of mouse subjected to transient 1hr middle
diffusion-weighted images (DWI) and apparent diffusion coefficient (ADC)
cerebral artery (MCA) occlusion (b) are shown in this figure.
0.035
..........
-......
.
.- -
contra
ipsi
0.03
IT
0.025
I
T
T
0.02
0.015
0.01
0.005
0
group 1
group 2
. contra M ipsi
*
*
group 1
group 2
Figure 6-2: Changes of CBV and WEI due to transient 1hr MCAo. Animal group
with poor CBV restoration showed significantly higher WEI, indicating a dramatically
increased BBB water permeability, thus more severe BBB damage than the group with
relatively good CBV restoration.
122
(Figure 6-2b, group 2).
6.1.4
Discussion
Disruption of the blood-brain barrier (BBB) plays an important role in the development of vasogenic edema and leads to increased risk of hemorrhagic transformation
after recirculation [116, 117]. Understanding BBB alteration at acute stage thus is
not only important in the study of stroke pathology but also in the diagnosis, prognosis, and therapeutic treatment of stroke. In the current study, using a novel MRI
technique, we demonstrated that the BBB can be highly compromised at acute stages
of transient stroke after poor reperfusion that is characterized by CBV deficiency. In
addition, poor CBV restoration after reperfusion seems to strongly correlate with
the increased BBB water exchange. These results demonstrate the importance of effective reperfusion on cerebral vascular changes and have important implications for
understanding the efficacy of reperfusion treatment and drug development.
After an ischemic insult, the affected area of the brain first undergoes cytotoxic
edema, which is later followed by vasogenic edema. The disrupted BBB after ischemic stroke is often associated with brain edema, especially vasogenic edema. In
this respect, MR contrast agent has been commonly used to study the dysfunctional
blood-brain-barrier (BBB) at relatively late stages of stroke progression [119, 24, 120].
However, the evaluation of the BBB permeability change is only possible when the
compromised BBB allows the extravasation of extrinsic makers (e.g. Gd-DTPA).
Therefore, it is not facile to observe the weakening of the BBB at early stages of
stroke since extrinsic markers usually cannot cross the BBB. On the other hand, the
MRI technique used in this study employs intrinsic water as a biomarker to assess
BBB integrity. Water is a small molecule, and can diffuse through the BBB. Therefore, it is more sensitive to the alterations of BBB, allowing the detection of acute or
even hyper-acute BBB damages after ischemic attack. In addition, our measurement
strategy is to utilize steady-state MRI detection instead of using dynamic enhanced
MRI technique, which allows quantitative monitoring of progressive BBB damage
with a better temporal resolution after reperfusion.
123
In this study, animals were be divided into two distinct groups based on ipsilateral CBV after reperfusion (Figure 6-2a). The group with poor CBV restoration after
recirculation (group 1) has a dramatically increased transvascular water exchange in
the stroke affected brain region, implying possible severe BBB damage acutely after
reperfusion (Figure 6-2b: group 1). Animals in the second group with good CBV recovery have ipsilateral WEI values similar to those in the contralateral side, implying
unchanged transvascular water exchange in stroke affected region. Together, the results from CBV and WEI measurements indicate that the extent of reperfusion must
be taken into account when evaluating the outcome of reperfusion and the efficacy
of therapeutic treatment. Even though the major artery is successfully reperfused
(e.g. MCA), micro-vasculature may not be effectively recirculated. As a result, the
progression and evolution of ischemic injury may not be the same and need to be
considered separately.
In summary, using the novel MRI technique fully described in Chapter 4, we
demonstrated that the intrinsic water molecule can be used as a biomarker for BBB
integrity evaluation in the study of a transient MCAo mouse model of stroke. Our results indicate the importance of good CBV restoration in BBB integrity at acute stage
after reperfusion. Successful restoration of CBV may also affect the stroke outcome.
Understanding the acute changes of CBV and BBB integrity after stroke, therefore,
may help to investigate the mechanisms of reperfusion injury, the efficacy of drug
intervention, and the prevention of hemorrhagic transformation at later stages. For
future studies, it will be important to monitor the dynamic changes of CBV and WEI
at acute stages of reperfusion to investigate the temporal evolution of BBB damage, to
characterize the biochemical and physiological basis of changes in transvascular water
movement using immunohistochemical assays (e.g., AQP4) and other relevant measurements (e.g., glucose consumption). In addition, histological correlation of MRI
results is also important in understanding the effect of these early vascular changes
on stroke outcome. The attempt to characterize acute vascular damages by transvascular WER is largely a work in progress, but may provide a basis for understanding
early vascular impairments at acute phases of stroke progression.
124
Ex vivo MRM of ischemic injured mouse brains
6.2
- toward MRM histopathological staining
6.2.1
Introduction
The mouse models of stroke are widely used in stroke research to understand the
mechanisms of neuronal damage and repair, and to test the efficacy of neuroprotective therapies. In addition, the availability of various transgenic or mutant mouse
models allows researchers to study the genetic effect on ischemic brain injury and to
find the therapeutic targets. Common histological methods to visualize and quantify ischemic brain infarction are through tissue staining such as hematoxylin-eosin
(H&E), triphenylte-trazolium chloride (TTC), and cresyl violet (CV). These methods
usually require tissue sectioning and dehydration, thus destroy the three-dimensional
tissue structure. Infarction volume is therefore usually quantified based on 2D images
obtained using light microscope. Therefore, it is desirable to develop imaging technique that allows visualization and reliable quantification of ischemic infarct volume
without destroying 3D tissue structures.
Among different clinical imaging techniques (e.g. CT, SPECT, PET, MRI), MRI
has been increasingly used in the diagnosis and management of acute ischemic stroke
because of its excellent soft tissue contrast and sensitivity and relative specificity in
detecting ischemic insult. It is also a widely used imaging tool in various in vivo
stroke studies using different animal models. However, in vivo MRI stroke studies
using animal models, especial mouse models, are usually limited by the imaging resolution and SNR within a given scan time. To quantify and visualize stroke affected
brain areas, high resolution MR images are usually required. Moreover, images with
histological comparable contrast or histological correlation are also needed.
In Section 3.1, using two commonly used MR contrast agents, Gd-DTPA and
MnC12 , we demonstrated that image contrast can not only be greatly enhanced but
also be manipulated by utilizing the properties of different MR contrast agents. In
addition, further tissue preparation for histologyical and pathological purpose is also
125
possible, as the brain stays intact during MRM acquisition. Therefore, direction comparison and across validation of images obtained using MRM or histological methods
can be done in the same animal. The goals of this study are thus to investigate the
changes of image contrast of ex vivo stroke mouse brains by using Gd-DTPA and
MnC12 as the MRM staining agents, to examine the ability of MRM staining in the
visualization and quantification of the ischemic infarc volume, and to compare the
MRM staining with that of conventional histological methods.
6.2.2
Material and Methods
Animal preparation:
Two male C57BL/6J mice (- 25g) were anesthetized with 2% halthane in 70% 02
balanced by 30% N 20. The standard intraluminal middle cerebral artery occlusion
method was used. Briefly, unilateral middle cerebral artery (MCA) occlusion was
performed by inserting a silicon-covered 8-0 nylon filament into the external carotid
artery (ECA) and advancing it to the origin of the anterior cerebral artery. One
hour after occlusion, the filament was withdrawn, and external carotid artery was
tied off. Twenty-four hours after the surgery and in vivo MR imaging, the mice were
transcardial fixed. The brain was first perfused with heparinized saline, the with 4%
paraformaldehyde (PFA). The brains were then taken out of the skull and post fixed
in 4% PFA overnight.
Brain MRM staining
Either 5 mM Gd-DTPA or 0.36 mM MnC12 was used to stain the brain. In general,
the brains were imaged one week after MRM staining to ensure the full penetration of
contrast agent into the brain. For MRM double staining, two staining protocols were
explored. In protocol one, the brain was initially stained with 5 mM Gd-DTPA. After
the high-resolution imaging acquisition and the T 1 , T 2 measurements, the brain was
washed twice with PBS buffer approximately one and a half weeks in between. Before
changing the PBS buffer (i.e., the washing buffer), the T 1 and T 2 were measured to
126
look into the effectiveness of washout. The brain was then stained with the second
contrast agent: 0.36 mM MnC12 . Then high resolution images were acquires, and
the brain T 1 and T 2 were measrued. In protocol two, the brain was initially stained
with 0.36 mM MnC12 . After the high resolution MR imaging and the T 1 and T 2
measurements, the brain was washed twice with PBS buffer approximately one and
a half weeks in between. We didn't further stained the brain with Gd-DTPA because
of the residual MnC12 in the brain.
Magnetic resonance imaging:
The in vivo data acquisition was performed at 9.4T (Bruker Biospin), and the in vivo
imaging was obtained using a 14T system (Bruker Biospin). For in vivo measurement,
the multi-slice multi-echo spin echo (MSME, TR/TE: 3000/10, 20, 30, 40, 50, 60, 70,
2
80, 90, 100, 110, 120 ms, Matrix size: 96x96, FOV: 1.5x1.5 cm , slice thickness: 0.5
mm) and the 3D spoiled gradient echo (FLASH, TR/TE: 35/3.2 ms, FOV: 2x1.5x1.5,
3
Matrix size: 164x128x128 cm 3 and reconstructed to 256x128x128 cm ) sequences
were used for the T 2-weighted and 3D images, respectively. For the in vivo study
at 14T, T 1 , T 2 , and T 2 * of the brain were measured using inversion recovery RARE
(IR_RARE), MSME, and multi-gradient echo (MGE) sequences, respectively. Highresolution 3D images were acquired using a spoiled gradient echo sequence (FLASH).
The imaging parameters for high resolution data acquisition are as following: TR/TE:
3
35/10 ms, Matrix: 512x512x256, resolution: 18.5 x 18.5 x 25 IMm, average: 12. The
flip angle was chosen to maximize the signal intensity.
Data anlysis:
We performed region of interest (ROI) analysis to characterize the T 1 and T 2 in
the cerebral cortex, striatum, thalamus, and hippocampus in the ipsilateral and the
contralateral hemispheres. ROIs were drawn using the AFNI software package [65].
We used Matlab (Mathworks, Natick, MA) to analyze MRM data. T and T2 in each
ROI were fitted using non-linear least square or first-order polynomial algorithms
based on averaged MR signal intensity. ImageJ software package [41] was used for
127
image visualization.
6.2.3
Results
In vivo T 2-weighted images acquired 24 hours after transient 1hr MCAo were shown
in Figure 6-3. Stroke affected areas are characterized by hyper-intensity in these T 2
weighted images because of the cerebral edema. As shown in the images, there is
a widespread damage in striatum, thalamus, hippocampus, and cerebral cortex. In
fact, much of the ipsilateral hemisphere is affected.
We stained one of the MCAo brains (mouse 2) with 5 mM Gd-DTPA first. MRM
images (50 pm isotropic) were acquired before and after Gd-DTPA staining as shown
in Figure 6-4a and 6-4b, respectively. Clear midline shift can be visualized in images
acquired before staining. However, the stroke affected areas are not readily identifiable in the pre-contrast images perhaps except in the ipsilateral hippocampus (Figure
6-4a). After 5 mM Gd-DTPA staining, the stroke affected regions can be easily identified (Figure 6-4b) in addition to midline shift. Interestingly, the dentate gyrus and
CA region in ipsilateral hippocampus appear dark in both pre and post contrast agent
stained images. In addition, ipsilateral hippocampus seems to be smaller compared
to the contralateral hippocampus.
To better understand the image contrast observed in the brain stained with GdDTPA, we performed higher resolution MRM study. Figure 6-5 shows the MRM
images of the brain stained with 5 mM Gd-DTPA at a resolution of 18.5 x 18.5
x 25 pm3 . Images of the brain areas affected by the MCAo are stained dark and
spatially heterogeneous. The boundaries of the affected regions are more distinct in
both striatum and cortex. It is also interesting to note that these boundaries are
stained darker by Gd-DTPA than the rest part of the region affected by the stroke.
In Section 3.1, we demonstrated that Gd-DTPA and MnC12 have different tissue
affinities and brain staining properties. We, therefore, further explored the possibility
of brain double staining using these two contrast agents. Following high resolution
imaging of the brain stained 5 mM Gd-DTPA, we first tried to wash out Gd-DTPA by
putting the brain in PBS buffer. We then stained the brain with 0.36 mM MnC1 . In
2
128
Figure 6-3: in vivo T 2 -weighted images of mice underwent 1hr transient MCAo.
Images were acquired 24 hours after occlusion (a. mouse 1, b. mouse 2).
129
(a)
(b)
Figure 6-4: 3D ex vivo images of mouse underwent 1hr transient MCAo (mouse
2). Images were acquired 24 hours after occlusion. Image acquired before Gd-DTPA
staining (a) revealed little contrast; while Gd-DTPA staining not only enhanced overall image contrast but also clearly revealed the stroke affected area (b).
130
stained
Figure 6-5: Axial, horizontal, and sagital view of a stroke mouse brain initially
occluMCA
one-hour
after
hours
with 5 mM Gd-DTPA. The brain was collected 24
sion.
131
Figure 6-6: Axial, horizontal, and sagital view of a stroke mouse brain stained
with
0.36 mM MnC12 after the initial staining with 5 mM Gd-DTPA.
132
Figure 6-7: Axial, horizontal, and sagital view of a stroke mouse brain initial stained
with 0.36 mM MnC12 . The brain was collected 24 hours after one-hour MCA occlusion.
133
Figure 6-6, MRM images of the same brain stained with 0.36 mM MnC12 are shown.
Instead of being stained dark, the stroke affected areas in the Mn 2 + stain appear
bright. In addition, within these regions, some areas are much brighter than the rest
of the region, probably indicating brain damage to different extent.
To exclude the possible effect of residual Gd-DTPA on the image contrast in MnC12
staining (Figure 6-6), we stained the other brain directly with 0.36 mM MnC12 . The
resultant images are shown in Figure 6-7. Comparing Figure 6-6 and Figure 6-7, we
find that the overall image contrast is rather similar.
We also measured brain T 1 and T 2 to understand the contrast difference between
the ipsi- and contra-lateral hemispheres and to evaluate the effectiveness of contrast
agent washout. Changes of tissue R 1 and R 2 in the brain initially stained with GdDTPA and MnC12 are plotted in Figure 6-8 and Figure 6-9, respectively. In general,
the R1 and R 2 are lower in stroke affected regions than in the contralateral part
(Figure 6-9) in the brain without any contrast agent. After Gd-DTPA staining, the
R 1 and R 2 become higher in the ipsilateral hemisphere than those in the contralateral
hemisphere (6-8). However, in the MnCl 2 staining, the R 1 and R 2 in the ipsilateral
hemisphere is lower than those in the contralateral hemisphere (6-9). For the brain
initially stained with Gd-DTPA then with MnC12 , the R 1 and R2 in the ipsilateral
hemisphere becomes lower in the ipsilateral hemisphere than those in the contralateral
side (6-8), which is in agreement with the results from brain initially stained with
MnCl 2 . We then compared the washout patterns of these two contrast agents. For
Gd-DTPA, tissue R 1 and R 2 dramatically dropped in both hemispheres after the first
washout, and the second wash seems to have little effect on the R 1 and R 2 changes
(Figure 6-8). For MnC12 , changing PBS buffer seems to help Mn 2+ washout as both
R 1 and R 2 dropped after each time the buffer is changed. In addition, after washing
the brain initially stained with MnC12 twice with PBS, the R 1 still remained higher
than the pre-contrast value although R2 almost went back to baseline.
Figure 6-10 shows the results of MRM staining (a) and Nissl staining (b) of the
same axial slice of a stroke mouse brain. The infarcted area can be visualized in both
MRM staining and Nissl staining. In comparison with the Nissl staining, the MRM
134
20
18
16
14
S12
10
8
6
4
2
0
ma
'U
u
a-
C
m
C
a
0
5 mM GdDTPA
1
lm
0
washout
time 1
0
washout
time 2
0.36 mM
MnCI2
E cortex U striatum 0 thalamus D hippocampus
120
100
80
,h
60
% 40
N
20
0
hippocampus
U cortex U striatum E thalamus OE
Figure 6-8: Changes of ex vivo brain tissue R 1 and R 2 before contrast agent staining,
after staining with 5 mM Gd-DTPA, during washout, and after the second staining.
135
OOM
74l
19
contra
ipsi
contra
pre-contrast
ipsi contra
0.36 mM
MnCI2
U cortex U striatum
ipsi contra
wash 1
ipsi
wash 2
thalamus E hippocampus
(a)
120
100
80
60
"4e
N
40
20
0
contra
ipsi
pre-contrast
contra
ipsi
0.36 mM
MnCI2
Scortex • striatum
contra
ipsi contra
wash 1
ipsi
wash 2
thalamus F hippocampus
Figure 6-9: Changes of ex vivo brain tissue R 1 and R before contrast agent
staining,
2
after staining with 0.36 mM MnC12 , and during washout.
136
(b)
Figure 6-10: Comparison of MRM staining (5 mM Gd-DTPA, a) with the conventional Nissl stain (b).
137
staining showed similar infarct area, however, with better visualization ability.
6.2.4
Discussion
The idea of magnetic resonance histology (MRH) was first proposed by Dr. Johnson
at Duke University [15]. Since then, many MRI studies have been done mainly in the
anatomical phenotyping and morphological screening of different mouse strains and
transgenic/mutant mouse models [18, 72, 21, 4, 5]. Few studies, however, have focused
on the development of MRM staining methods that can provide histopathological
information. In the current study, we developed a new protocol for MR tissue staining
based on the ex vivo mouse brain MRM study described in Chapter 3, and applied it
to a transient mouse model of stroke. The preliminary results showed that both GdDTPA and MnC12 can help in delineating stroke affected brain regions, however, with
different tissue contrast and differentiation abilities. By staining the brain first with
Gd-DTPA then with MnCl 2 , we further demonstrated that MRM double staining can
be achieved in the same brain sample with the staining pattern in the infarcted area
complimentary to each other. This double staining protocol can potentially increase
the ability of MRM staining to detect regions of the brain with different fate (e.g.
infarct core vs. penumbra) as a result of ischemic insult.
Image contrast of Gd-DTPA or MnC12 stained stroke brains
The image contrast in the ipsilateral side is reversed between the Gd-DTPA staining and the MnC12 staining (Figures 6-5 and 6-6). Compared to the contralateral
hemisphere, the ischemic affected region stained by Gd-DTPA is hypointensity on
the T 2 *-weighted image, while, that stained by Mn 2 + is hyperintensity. Further R 1
analysis revealed that the R 1 and R 2 values in the stroke affected areas were higher
than the contralateral part in the Gd-DTPA staining, however were lower in the Mn 2 +
staining (Figures 6-8a and 6-9a). This is very likely due to the disrupted normal tissue structure in the infarted area, and the differences in the tissue distribution and
relaxation enhancement effect of these two agents.
138
In Chapter 3, we demonstrated that Gd-DTPA stains the brain mainly through
passive diffusion and affects the longitudinal relaxation mainly through the electron
proton dipole-dipole interaction. The higher R1 values observed in the ischemic area
compared to that in the contralateral hemisphere probably reflects that more GdDTPA is accumulated in this area (Figure 6-8a). Since the normal tissue structure is
compromised in this region, more tissue space may become accessible to contrast agent
leading to higher amount of Gd-DTPA in ischemic area compared to the contralateral
side. As a result, ischemic region appears dark in the T 2*-weighted image (Figure 65). On the other hand, Mn 2+ can bind to proteins or macromolecules in ex vivo brain
in addition to passive diffusion into the tissue. Binding to protein or macromolecules
can leads to enhanced TI relaxation in tissue. Consequently, the destruction of normal tissue/cell structure and cellular content in ischemic area leads to less Mn 2+
accumulation or protein/macromolecule binding in the ipsilateral hemisphere, thus
lower RI and R 2 values (Figure 6-9a and b). As a result, in the T2 *-weighted image
of Mn 2+ stained brain, stroke affected region appears bright (Figures 6-6 and 6-7).
Moreover, the slow washout pattern of the MnC12 staining further suggests that the
ex vivo Mn 2 + binding to proteins or macromolecules is more likely not specific.
Histological correlation
The Nissl stain and MRM stain revealed similar infarct area, demonstrating the ability of MRM staining to identify the overall brain regions affected by the ischemic
insult. In addition, MRM staining seems to have a better infarct delineation ability
in terms of gross examination. The rim of the ischemic areas that showed different
image contrast, however, is not clearly visible in the Nissl staining. These preliminary results warrant further investigation to look into the neuronal correlation of the
staining patterns observed in the MRM ex vivo staining.
In summary, we have shown that both contrast agents can help enhance image contrast and delineate ischemic areas, however with different infarct staining properties.
These two staining methods seem to complement each other. Furthermore, the com139
parison of the infarct area using the same brain stained with the MR contrast "staining" agents and Nissl showed good agreement, indicating that MRM staining can be
used for infarct volume study. Moreover, we also developed an MRM double staining
method to stain the same brain with these two contrast agents sequentially, which
enables better ischemic infarct differentiation. Future work will include refinement of
the staining protocol, detailed comparison of ischemic infarct region between these
two staining, segmentation and visualization of ischemic affected regions, and detailed
histology or immunohistochemistry study to understand features observed in MRM
staining.
140
Bibliography
[1] Alan Jasanoff. Mri contrast agents for functional molecular imaging of brain
activity. Current Opinion in Neurobiology, 17(5):593-600, 2007.
[2] Joseph B. Mandeville, John J. A. Marota, C. Ayata, Michael A. Moskowitz,
Robert M. Weisskoff, and Bruce R. Rosen. Mri measurement of the temporal
evolution of relative cmro2 during rat forepaw stimulation. Magnetic Resonance
in Medicine, 42(5):944-951, 1999.
[3] L. M. Hamberg, R. Macfarlane, E. Tasdemiroglu, P. Boccalini, G. J. Hunter,
J. W. Belliveau, M. A. Moskowitz, and B. R. Rosen. Measurement of cerebrovascular changes in cats after transient ischemia using dynamic magnetic
resonance imaging. Stroke, 24(3):444-450, 1993.
[4] R van der Landen, R.C.G. van de Ven, L. van der Weerd, M. Fodor, R.R.
Frants, R.E. Poelmann, A.M.J.M. van den Maagdenberg, J.H.C. Reiber, M.A.
van Buchem, and F. Admiraal-Behloul. Automatic morphological assessment in
a transgenic cacnala knockin migraine mouse model in in-vivo high-resolution
mrm. Biomedical Imaging: Nano to Macro, 2006. 3rd IEEE InternationalSymposium on, (0-7803-9577-8/06), 2006.
[5] Alexandra Badea, Peter J. Nicholls, G. Allan Johnson, and William C. Wetsel.
Neuroanatomical phenotypes in the reeler mouse. Neurolmage, 34(4):13631374, 2007.
[6] Randall B. Lauffer. Paramagnetic metal complexes as water proton relaxation
agents for nmr imaging: theory and design. Chem. Rev., 87(5):901-927, 1987.
[7] P. Caravan, J. J. Ellison, T. J. McMurry, and R. B. Lauffer. Gadolinium(iii)
chelates as mri contrast agents: Structure, dynamics, and applications. Chem.
Rev., 99(9):2293-2352, 1999.
[8] N. Bloembergen. Proton relaxation times in paramagnetic solutions. The Journal of Chemical Physics, 27(2):572-573, 1957.
[9] I. Solomon. Relaxation processes in a system of two spins. Physical Review,
99(2):559, 1955.
[10] N. Bloembergen and L. 0. Morgan. Proton relaxation times in paramagnetic
solutions, effects of electron spin relaxation. The Journal of Chemical Physics,
34(3):842-850, 1961.
[11] I. Solomon and N. Bloembergen. Nuclear magnetic interactions in the hf
molecule. The Journal of Chemical Physics, 25(2):261-266, 1956.
[12] Charles P. Slichter. LinkPrinciples of magnetic resonance. Springer, 1990.
[13] I Bertini, C Luchinat, and S. Aime. Nmr of paramagnetic substances. Coordination Chemistry Reviews, 150:1-293, 1996.
[14] Chapter 3 relaxation. Coordination Chemistry Reviews, 150:77-110, 1996.
[15] G. Allan Johnson, Gary P. Cofer, Boma Fubara, Sally L. Gewalt, Laurence W.
Hedlund, and Robert R. Maronpot. Magnetic resonance histology for morphologic phenotyping. Journal of Magnetic Resonance Imaging, 16(4):423-429,
2002.
[16] A. Badea, A. A. Ali-Sharief, and G. A. Johnson. Morphometric analysis of the
c57bl/6j mouse brain. Neurolmage, 37(3):683-693, 2007.
[17] Helene Benveniste, Katie Kim, Lin Zhang, and G. Allan Johnson. Magnetic
resonance microscopy of the c57bl mouse brain. Neurolmage, 11(6):601-611,
2000.
[18] Michel Cyr, Marc G. Caron, G. Allan Johnson, and Aki Laakso. Magnetic resonance imaging at microscopic resolution reveals subtle morphological changes
in a mouse model of dopaminergic hyperfunction. Neurolmage, 26(1):83-90,
2005.
[19] Anjum A. Ali, Anders M. Dale, Alexandra Badea, and G. Allan Johnson. Automated segmentation of neuroanatomical structures in multispectral mr microscopy of the mouse brain. Neurolmage, 27(2):425-435, 2005.
[20] Allan MacKenzie-Graham, Erh-Fang Lee, Ivo D. Dinov, Mihail Bota, David W.
Shattuck, Seth Ruffins, Heng Yuan, Fotios Konstantinidis, Alain Pitiot,
Yi Ding, Guogang Hu, Russell E. Jacobs, and Arthur W. Toga. A multimodal, multidimensional atlas of the c57bl/6j mouse brain. Journal of Anatomy,
204(2):93-102, 2004.
[21] Y. Ma, P. R. Hof, S. C. Grant, S. J. Blackband, R. Bennett, L. Slatest, M. D.
McGuigan, and H. Benveniste. A three-dimensional digital atlas database of the
adult c57bl/6j mouse brain by magnetic resonance microscopy. Neuroscience,
135(4):1203-1215, 2005.
[22] K. Mollgard and S. C. Sorensen. The permeability of cerebral capillaries to a
tracer molecule, alcian blue, with a molecular weight of 1390. In The Pathology
of Cerebral Microcirculation.De Gruyter and Co., Berlin, 1974.
142
[23] Oak Za Chi, Hwu Meei Wei, Xiaowen Lu, and Harvey R. Weiss. Increased
blood-brain permeability with hyperosmolar mannitol increases cerebral 02 consumption and o2 supply/consumption heterogeneity. J Cereb Blood Flow Metab,
16(2):327-333, 1996.
[24] Quan Jiang, James R. Ewing, Guang Liang Ding, Li Zhang, Zheng Gang Zhang,
Lian Li, Polly Whitton, Mei Lu, Jiani Hu, Qing Jiang Li, Robert A. Knight,
and Michael Chopp. Quantitative evaluation of bbb permeability after embolic
stroke in rat using mri. J Cereb Blood Flow Metab, 25(5):583-592, 2005.
[25] CBTRUS. Statistical report: Primary brain tumors in the united states statistical report, 1998-2002 (years data collected). Technical report, The Central
Brain Tumor Registry of the United States (CBTRUS), 2005-2006.
[26] L. J. Kaufman, C. P. Brangwynne, K. E. Kasza, E. Filippidi, V. D. Gordon,
T. S. Deisboeck, and D. A. Weitz. Glioma expansion in collagen i matrices: analyzing collagen concentration-dependent growth and motility patterns. Biophys.
J., 89(1):635-650, 2005.
[27] V. D. Gordon, M. T. Valentine, M. L. Gardel, D. Andor-Ardo, S. Dennison,
A. A. Bogdanov, D. A. Weitz, and T. S. Deisboeck. Measuring the mechanical stress induced by an expanding multicellular tumor system: a case study.
Experimental Cell Research, 289(1):58-66, 2003.
[28] T. S. Deisboeck, M. E. Berens, A. R. Kansal, S. Torquato, A. O. StemmerRachamimov, and E. A. Chiocca. Pattern of self-organization in tumour systems: complex growth dynamics in a novel brain tumour spheroid model. Cell
Proliferation,34(2):115-134, 2001.
[29] Leonard M. Sander and Thomas S. Deisboeck. Growth patterns of microscopic
brain tumors. Physical Review E, 66(5):051901, 2002.
[30] Pier Paolo Delsanto, Caterina Guiot, Piero Giorgio Degiorgis, Carlos A. Condat,
Yuri Mansury, and Thomas S. Deisboeck. Growth model for multicellular tumor
spheroids. 85(18):4425- 4227, 2004.
[31] A. Mojsilovic, B. Rogowitz, J. Gomez, and Deisboeck T.S. Analysis, reconstruction and visualization of malignant brain tumors: a case study in data
synthesis. Analyt. Quant. Cytol. Histol., 24:125-133, 2002.
[32] James B. Aguayo, Stephen J. Blackband, Joseph Schoeniger, Mark A. Mattingly, and Markus Hintermann. Nuclear magnetic resonance imaging of a
single cell. Nature, 322(6075):190-191, 1986.
[33] Z. H. Cho, C. B. Ahn, S. C. Juh, H. K. Lee, R. E. Jacobs, S. Lee, J. H. Yi,
and J. M. Jo. Nuclear magnetic resonance microscopy with 4-um resolution:
Theoretical study and experimental results. Medical Physics, 15(6):824, 1988.
143
[34] G. A. Johnson, M. B.Thompson, S. L. Gewalt, and C. E. Hayes. Nuclear
magnetic resonance imaging at microscopic resolution. Journal of Magnetic
Resonance, 68(1):129-137, 1986.
[35] Shuning Huang, David Vader, David A. Weitz, Gangping Dai, Bruce R. Rosen,
and Thomas S. Deisboeck. Magnetic resonance microscopy of a glioma spheriod
in a collagen i matrix. In Proceeding of ISMRM 14th SCIENTIFIC MEETING
& EXHIBITION, Seattle, Washington, USA, 2006.
[36] R. Nishikawa, X. Ji, R. C. Harmon, C. S. Lazar, G. N. Gill, W. K. Cavenee, and
H. S. Huang. A mutant epidermal growth factor receptor common in human
glioma confers enhanced tumorigenicity. PNAS, 91(16):7727-7731, 1994.
[37] Motoo Nagane, Frank Coufal, Hong Lin, Oliver Bogler, Webster K. Cavenee,
and H. J. Su Huang. A common mutant epidermal growth factor receptor confers enhanced tumorigenicity on human glioblastoma cells by increasing proliferation and reducing apoptosis. Cancer Res, 56(21):5079-5086, 1996.
[38] Jens M. Kelm, Nicholas E. Timmins, Catherine J. Brown, Martin Fussenegger, and Lars K. Nielsen. Method for generation of homogeneous multicellular
tumor spheroids applicable to a wide variety of cell types. Biotechnology and
Bioengineering, 83(2):173-180, 2003.
[39] D. Revel, R. C. Brasch, H. Paajanen, W. Rosenau, W. Grodd, B. Engelstad,
P. Fox, and J. Winkelhake. Gd-dtpa contrast enhancement and tissue differentiation in mr imaging of experimental breast carcinoma. Radiology, 158(2):319323, 1986.
[40] Gang Chen, Sune Norhoj Jespersen, Michael Pedersen, Qi Pang, Michael R.
Horsman, and Hans Stodkilde-Jorgensen. Intravenous administration of gddtpa prior to dwi does not affect the apparent diffusion constant. Magnetic
Resonance Imaging, 23(5):685-689, 2005.
[41] M.D. Abramoff, P.J. Magelhaes, and S.J. Ram. Image processing with imagej.
Biophotonics International,, 11(7):36-42, 2004.
[42] 3d slicer.
[43] A Giese, MA Loo, N Tran, D Haskett, SW Coons, and ME. Berens. Dichotomy
of astrocytoma migration and proliferation. Int J Cancer., 67(2):275-82, 1996.
[44] Chaitanya A. Athale and Thomas S. Deisboeck. The effects of egf-receptor
density on multiscale tumor growth patterns. Journal of Theoretical Biology,
238(4):771-779, 2006.
[45] Le Zhang, Chaitanya A. Athale, and Thomas S. Deisboeck. Development of
a three-dimensional multiscale agent-based tumor model: Simulating geneprotein interaction profiles, cell phenotypes and multicellular patterns in brain
cancer. Journal of Theoretical Biology, 244(1):96-107, 2007.
144
[46] Maria Teresa Santini, Gabriella Rainaldi, Antonella Ferrante, Paola Indovina,
Gianfranco Donelli, and Pietro Luigi Indovina. A 50 hz sinusoidal magnetic
field does not damage mg-63 three-dimensional tumor spheroids but induces
changes in their invasive properties. Bioelectromagnetics, 27(2):132-141, 2006.
[47] WO Short, L Goodwill, and CW Taylor. Alteration of human tumor cell adhesion by high-strength static magnetic fields. Investigative Radiology, 27:836-840,
1992.
[48] Anaira C. Raylman, Raymond R. andClavo and Richard L. Wahl. Exposure
to strong static magnetic field slows the growth of human cancer cells in vitro.
Bioelectromagnetics, 17(5):358-363, 1996.
[49] M Brandl, JC Tonn, K Kotitschke, R Goldbrunner, S Kerkau, and A. Haase.
Quantitative nmr microscopy of multicellular tumor spheroids and confrontation cultures. Magnetic Resonance in Medicine, 34(4):596-603, 1995.
[50] Fritjof Helmchen and Winfried Denk. Deep tissue two-photon microscopy. Nat
Meth, 2(12):932-940, 2005.
[51] G. McConnell. Improving the penetration depth in multiphoton excitation laser
scanning microscopy. Journal of Biomedical Optics, 11(5):054020, 2006.
[52] Christopher B. Raub, Vinod Suresh, Tatiana Krasieva, Julia Lyubovitsky,
Justin D. Mih, Andrew J. Putnam, Bruce J. Tromberg, and Steven C. George.
Noninvasive assessment of collagen gel microstructure and mechanics using multiphoton microscopy. Biophys. J., page biophysj.106.097998, 2006.
[53] Svein J. T. Nygaard and Ole-Bjorn Tysnes. Quantification of glioma cell invasion by confocal laser scanning microscopy in an in vitro co-culture system.
Cancer Letters, 105(1):45-49, 1996.
[54] G. A. Johnson, M.B. Thompson, S.L. Gewalt, and C.E. Hayes. Nuclear magnetic resonance imaging at microscopic resolution. J Magn Reson, 68(1):129137, 1986.
[55] 0. Natt, T. Watanabe, S. Boretius, J. Radulovic, J. Frahm, and T. Michaelis.
High-resolution 3d mri of mouse brain reveals small cerebral structures in vivo.
Journal of Neuroscience Methods, 120(2):203-209, 2002.
[56] Jrgen E. Schneider, Simon D. Bamforth, Stuart M. Grieve, Kieran Clarke,
Shoumo Bhattacharya, and Stefan Neubauer. High-resolution, high-throughput
magnetic resonance imaging of mouse embryonic anatomy using a fast gradientecho sequence. Magnetic Resonance Materials in Physics, Biology and Medicine,
16(1):43-51, 2003.
[57] Jiangyang Zhang, Linda J. Richards, Paul Yarowsky, Hao Huang, Peter C. M.
van Zijl, and Susumu Mori. Three-dimensional anatomical characterization of
145
the developing mouse brain by diffusion tensor microimaging.
20(3):1639-1648, 2003.
Neurolmage,
[58] C. Richard Fisel, Jerome L. Ackerman, Richard B. Buxton, Leoncio Garrido,
John W. Belliveau, Bruce R. Rosen, and Thomas J. Brady. Mr contrast due
to microscopically heterogeneous magnetic susceptibility: Numerical simulations and applications to cerebral physiology. Magnetic Resonance in Medicine,
17(2):336-347, 1991.
[59] Robert Weisskoff, Chun S. Zuo, Jerrold L. Boxerman, and Bruce R. Rosen.
Microscopic susceptibility variation and transverse relaxation: Theory and experiment. Magnetic Resonance in Medicine, 31(6):601-610, 1994.
[60] James D. Quirk, G. Larry Bretthorst, Timothy Q. Duong, Avi Z. Snyder,
Charles S. Springer Jr, Joseph J. H. Ackerman, and Jeffrey J. Neil. Equilibrium water exchange between the intra- and extracellular spaces of mammalian
brain. Magnetic Resonance in Medicine, 50(3):493-499, 2003.
[61] Kazuhiko Narita, Fumiko Kawasaki, and Hiroshi Kita. Mn and mg influxes
through ca channels of motor nerve terminals are prevented by verapamil in
frogs. Brain Research, 510(2):289-295, 1990.
[62] Michael Aschner and Judy L. Aschner. Manganese neurotoxicity: Cellular effects and blood-brain barrier transport. Neuroscience and Biobehavioral Reviews, 15(3):333-340, 1991.
[63] Atsushi Takeda, Yuka Kodama, Shioji Ishiwatari, and Shoji Okada. Manganese
transport in the neural circuit of rat cns. Brain Research Bulletin, 45(2):149152, 1998.
[64] Keith B.J. Franklin and George Paxinos. The Mouse Brain in Stereotaxic Coordinates. Academic Press, third edition edition, 2007.
[65] RW Cox. Afni: Software for analysis and visualization of functional magnetic
resonance neuroimages. Computers and Biomedical Research, 29:162-173, 1996.
[66] Boline JK, MacKenzie-Graham A, Yuan H Shattuck D, Anderson S, Sforza DM,
Wang J, Williams RW, Wong W, Martone ME, Zaslavsky I, and Toga AW. A
digital atlas and neuroinformatics framework for query and display of disparate
data. In Society for Neuroscience, page 100.12, 2006.
[67] Ed S. Lein, Michael J. Hawrylycz, Nancy Ao, Mikael Ayres, Amy Bensinger,
Amy Bernard, Andrew F. Boe, Mark S. Boguski, Kevin S. Brockway, Emi J.
Byrnes, Lin Chen, Li Chen, Tsuey-Ming Chen, Mei Chi Chin, Jimmy Chong,
Brian E. Crook, Aneta Czaplinska, Chinh N. Dang, Suvro Datta, Nick R.
Dee, Aimee L. Desaki, Tsega Desta, Ellen Diep, Tim A. Dolbeare, Matthew J.
Donelan, Hong-Wei Dong, Jennifer G. Dougherty, Ben J. Duncan, Amanda J.
Ebbert, Gregor Eichele, Lili K. Estin, Casey Faber, Benjamin A. Facer, Rick
146
Fields, Shanna R. Fischer, Tim P. Fliss, Cliff Frensley, Sabrina N. Gates,
Katie J. Glattfelder, Kevin R. Halverson, Matthew R. Hart, John G. Hohmann,
Maureen P. Howell, Darren P. Jeung, Rebecca A. Johnson, Patrick T. Karr,
Reena Kawal, Jolene M. Kidney, Rachel H. Knapik, Chihchau L. Kuan,
James H. Lake, Annabel R. Laramee, Kirk D. Larsen, Christopher Lau,
Tracy A. Lemon, Agnes J. Liang, Ying Liu, Lon T. Luong, Jesse Michaels, Judith J. Morgan, Rebecca J. Morgan, Marty T. Mortrud, Nerick F. Mosqueda,
Lydia L. Ng, Randy Ng, Geralyn J. Orta, Caroline C. Overly, Tu H. Pak,
Sheana E. Parry, Sayan D. Pathak, Owen C. Pearson, Ralph B. Puchalski,
Zackery L. Riley, Hannah R. Rockett, Stephen A. Rowland, Joshua J. Royall,
Marcos J. Ruiz, Nadia R. Sarno, Katherine Schaffnit, Nadiya V. Shapovalova,
Taz Sivisay, Clifford R. Slaughterbeck, Simon C. Smith, Kimberly A. Smith,
Bryan I. Smith, Andy J. Sodt, Nick N. Stewart, Kenda-Ruth Stumpf, Susan M.
Sunkin, Madhavi Sutram, Angelene Tam, Carey D. Teemer, Christina Thaller,
Carol L. Thompson, Lee R. Varnam, Axel Visel, Ray M. Whitlock, Paul E.
Wohnoutka, Crissa K. Wolkey, Victoria Y. Wong, et al. Genome-wide atlas of
gene expression in the adult mouse brain. Nature, 445(7124):168-176, 2007.
[68] Cynthia Mills Schumann, Michael H. Buonocore, and David G. Amaral. Magnetic resonance imaging of the post-mortem autistic brain. Journal of Autism
and Developmental Disorders,31(6):561-568, 2001.
[69] Wibeke Nordhoy, Henrik W. Anthonsen, Morten Bruvold, Heidi Brurok, Sissel Skarra, Jostein Krane, and Per Jynge. Intracellular manganese ions provide strong t1 relaxation in rat myocardium. Magnetic Resonance in Medicine,
52(3):506-514, 2004.
[70] Yi-Jen Lin and Alan P. Koretsky. Manganese ion enhances tl-weighted mri during brain activation: An approach to direct imaging of brain function. Magnetic
Resonance in Medicine, 38(3):378-388, 1997.
[71] Robia G. Pautler and Alan P. Koretsky. Tracing odor-induced activation in the
olfactory bulbs of mice using manganese-enhanced magnetic resonance imaging.
NeuroImage, 16(2):441-448, 2002.
[72] N. Kovacevic, J. T. Henderson, E. Chan, N. Lifshitz, J. Bishop, A. C. Evans,
R. M. Henkelman, and X. J. Chen. A three-dimensional mri atlas of the mouse
brain with estimates of the average and variability. Cereb. Cortez, 15(5):639645, 2005.
[73] Wibeke Nordhoy, Henrik W. Anthonsen, Bruvold Morten, Jynge Per, Krane
Jostein, and BrurokHeidi. Manganese ions as intracellular contrast agents:
proton relaxation and calcium interactions in rat myocardium. NMR in
Biomedicine, 16(2):82-95, 2003.
[74] Ichio Aoki, Yi-Jen Lin Wu, Afonso C. Silva, Ronald M. Lynch, and Alan P.
Koretsky. In vivo detection of neuroarchitecture in the rodent brain using
manganese-enhanced mri. Neurolmage, 22(3):1046-1059, 2004.
147
[75] Afonso C. Silva, Jung Hee Lee, Carolyn W. H. Wu, Jason Tucciarone, Galit
Pelled, Ichio Aoki, and Alan P. Koretsky. Detection of cortical laminar architecture using manganese-enhanced mri. Journal of Neuroscience Methods,
167(2):246-257, 2008.
[76] Christina H. Liu, Young R. Kim, Jia Q. Ren, Florian Eichler, Bruce R. Rosen,
and Philip K. Liu. Imaging cerebral gene transcripts in live animals. J. Neurosci., 27(3):713-722, 2007.
[77] Megan L Blackwell, WC. Hubbard, M Selig, Bruce R Rosen, Ann B. Moser, and
Florian Eichler. Contrast-enhanced ex vivo mr reveals inflammatory zone in xlinked adrenoleukodystrohpy. In Proceeding of ISMRM 16th Scientific Meeting
& Exhibition, page 2067, Toronto, Ontario, Canada, 2008.
[78] Alexandra E. Petiet, Matthew H. Kaufman, Matthew M. Goddeeris, Jeffrey
Brandenburg, Susan A. Elmore, and G. Allan Johnson. High-resolution magnetic resonance histology of the embryonic and neonatal mouse: A 4d atlas
and morphologic database. Proceedings of the National Academy of Sciences,
105(34):12331-12336, 2008.
[79] R. Suzuki, T. Yamaguchi, T. Kirino, F. Orzi, and I. Klatzo. The effects of
5-minute ischemia in mongolian gerbils: I. blood-brain barrier, cerebral blood
flow, and local cerebral glucose utilization changes. Acta Neuropathol Berl,
60:207-16, 1983.
[80] Ludmila Belayev, Raul Busto, Weizhao Zhao, and Myron D. Ginsberg. Quantitative evaluation of blood-brain barrier permeability following middle cerebral
artery occlusion in rats. Brain Research, 739(1-2):88-96, 1996.
[81] E. Preston and DO. Foster. Evidence for pore-like opening of the bloodbrain
barrier following forebrain ischemia in rats. Brain Research, 761:4-10, 1997.
[82] Kathleen M. Donahue, Robert M. Weisskoff, David A. Chesler, Kenneth K.
Kwong, Alexei A. Bogdanov Jr, Joseph B. Mandeville, and Bruce R. Rosen.
Improving mr quantification of regional blood volume with intravascular tl
contrast agents: Accuracy, precision, and water exchange. Magnetic Resonance
in Medicine, 36(6):858-867, 1996.
[83] Schwarzbauer C, Syha J, and Haase A. Quantification of regional blood volumes
by rapid tl mapping. Magnetic Resonance in Medicine, 29(5):709-712, 1993.
[84] Harden M. McConnell. Reaction rates by nuclear magnetic resonance.
Journal of Chemical Physics, 28(3):430-431, 1958.
The
[85] Y.R. Kim, M. D. Savellano, D. H. Savellano, R. Weissleder, and A. Bogdanov Jr.
Measurement of tumor interstitial volume fraction: Method and implication for
drug delivery. Magnetic Resonance in Medicine, 52(3):485-494, 2004.
148
[86] Demsar F, Roberts TP, Schwickert HC, Shames DM, Mann JS van Dijke CF,
and Brasch RC Saeed M. A mri spatial mapping technique for microvascular
permeability and tissue blood volume based on macromolecular contrast agent
distribution. Magnetic Resonance in Medicine, 37(2):236-242, 1997.
[87] Bogdanov AA Jr, Weissleder R, Frank HW, Bogdanova AV, Nossif N, Schaffer
BK, Tsai E, Papisov MI, and Brady TJ. A new macromolecule as a contrast
agent for mr angiography: preparation, properties, and animal studies. Radiology, 187:701-706, 1993.
[88] Rob C.G. van de Ven, Bianca Hogers, Arn M. J. M. van den Maagdenberg,
Huub J. M. de Groot, Michel D. Ferrari, Rune R. Frants, Robert E. Poelmann,
Louise van der Weerd, and Suzanne R. Kiihne. T1 relaxation in in vivo mouse
brain at ultra-high field. Magnetic Resonance in Medicine, 58(2):390-395, 2007.
[89] Yu-Ting Kuo, Amy H. Herlihy, Po-Wah So, Kishore K. Bhakoo, and Jimmy D.
Bell. In vivo measurements of t1 relaxation times in mouse brain associated
with different modes of systemic administration of manganese chloride. Journal
of Magnetic Resonance Imaging, 21(4):334-339, 2005.
[90] Martin Reivich. Arterial pco2 and cerebral hemodynamics.
206(1):25-35, 1964.
Am J Physiol,
[91] John W. Belliveau, Bruce R. Rosen, Howard L. Kantor, Richard R. Rzedzian,
David N. Kennedy, Robert C. McKinstry, James M. Vevea, Mark S. Cohen, Ian L. Pykett, and Thomas J. Brady. Functional cerebral imaging by
susceptibility-contrast nmr. Magnetic Resonance in Medicine, 14(3):538-546,
1990.
[92] C.J. Clemedson, H. Hartelius, and G. Holmberg. The influence of carbon dioxide
inhalation on the cerebral vascular permeability to trypan blue ("the bloodbrain barrier"). Acta path. microbiol. scand., 42:137-149, 1958.
[93] R. W. P. Cutler and C. F. Barlow. The effect of hypercapnia on brain permeability to protein. Arch. Neurol. (Chic.), 14:54-63, 1966.
[94] Barbro Johansson and Bengt Nilsson. The pathophysiology of the blood-brain
barrier dysfunction induced by severe hypercapnia and by epileptic brain activity. Acta Neuropathologica,38(2):153-158, 1977.
[95] Mary Kay Gumerlock, Barbara Drew Belshe, Richard Madsen, and Clark
Watts. Osmotic blood-brain barrier disruption and chemotherapy in the treatment of high grade malignant glioma: patient series and literature review. Journal of Neuro-Oncology, 12(1):33-46, 1992.
[96] David Fortin, Cathy Gendron, Marie Boudrias, and Marie-Pierre Garant. Enhanced chemotherapy delivery by intraarterial infusion and blood-brain barrier
disruption in the treatment of cerebral metastasis. Cancer, 109(4):751-760,
2007.
149
[97] Rachel C. Brown, Richard D. Egleton, and Thomas P. Davis. Mannitol opening
of the blood-brain barrier: regional variation in the permeability of sucrose, but
not 86rb+ or albumin. Brain Research, 1014(1-2):221-227, 2004.
[98] P. Ravussin, D.P. Archer, J.L. Tyler, E. Meyer, M. Abou-Madi, M. Diksic,
Yamamoto. L., and D. Trop. Effects of rapid mannitol infusion on cerebral
blood volume, a positron emission tomographic study in dogs and man. J.
Neurosurg., 64(1):104-13, 1986.
[99] P. Ravussin, D. P. Archer, E. Meyer, M. Abou-Madi, L. Yamamoto, and
D. Trop. The effects of rapid infusions of saline and mannitol on cerebral
blood volume and intracranial pressure in dogs. Can J Anesth, 32(5):506-515,
1985.
[100] Weili Lin, Richard P. Paczynski, Karthikeyan Kuppusamy, Chung Y. Hsu, and
E. Mark Haacke. Quantitative measurements of regional cerebral blood volume using mri in rats: Effects of arterial carbon dioxide tension and mannitol.
Magnetic Resonance in Medicine, 38(3):420-428, 1997.
[101] Janet Hoff. Methods of blood collection in mouse. Technique, 29(10):47-53,
2000.
[102] C. A. Evans, J. M. Reynolds, M. L. Reynolds, and N. R. Saunders. The effect of
hypercapnia on a blood-brain barrier mechanism in foetal and new-born sheep.
J Physiol, 255(3):701-714, 1976.
[103] LF Marshall, RW Smith, LA Rauscher, and HM Shapiro. Mannitol dose requirements in brain-injured patients. J Neurosurgery, 48:169-172, 1978.
[104] JP Muizelaar, EP Wei, HA Kontos, and DP Becker. Mannitol causes compensatory cerebral vasoconstriction and vasodilation in response to blood viscosity
changes. J Neurosurgery, 59:822-828, 1983.
[105] Robert F. Furchgott and John V. Zawadzki. The obligatory role of endothelial cells in the relaxation of arterial smooth muscle by acetylcholine. Nature,
188(5789):373-376, 1980.
[106] Matthias Endres, Ulrich Laufs, James K. Liao, and Michael A. Moskowitz.
Targeting enos for stroke protection. Trends in Neurosciences, 27(5):283-289,
2004.
[107] Paul L. Huang, Zhihong Huang, Hiroshi Mashimo, Kenneth D. Bloch,
Michael A. Moskowitz, John A. Bevan, and Mark C. Fishman. Hypertension in mice lacking the gene for endothelial nitric oxide synthase. Nature,
377(6546):239-242, 1995.
[108] Edward Shesely, Nobuyo Maeda, H. S. Kim, Kaushik Desai, John Krege, Victor
Laubach, Paula Sherman, William Sessa, and Oliver Smithies. Elevated blood
150
pressures in mice lacking endothelial nitricoxidesynthase. Proceedings of the
National Academy of Sciences of the United States of America, 93(23):1317613181, 1996.
[109] Zhihong Huang, Paul L. Huang, Jianya Ma, Wei Meng, Cenk Ayata, Mark C.
Fishman, and Michael A. Moskowitz. Enlarged infarcts in endothelial nitric
oxide synthase knockout mice are attenuated by nitro-l-arginine. J Cereb Blood
Flow Metab, 16(5):981-987, 1996.
[110] Y.R. Kim, E. Tejima, S. Huang, D. N. Atochin, G. Dai, E. H. Lo, P. L. Huang,
A. Bogdanov Jr, and B. R. Rosen. In vivo quantification of transvascular water
exchange during the acute phase of permanent stroke. Magnetic Resonance in
Medicine, 60(4):813-821, 2008.
[111] Jerrold L. Boxerman, Leena M. Hamberg, Bruce R. Rosen, and Robert M Weisskoff. Mr contrast due to intravascular magnetic susceptibility perturbations.
Magnetic Resonance in Medicine, 34(4):555-566, 1995.
[112] Joelle Dennie, Joseph B. Mandeville, Jerrold L. Boxerman, Scott D. Packard,
Bruce R. Rosen, and Robert M. Weisskoff. Nmr imaging of changes in vascular morphology due to tumor angiogenesis. Magnetic Resonance in Medicine,
40(6):793-799, 1998.
[113] Irene Tropres, Stephan Grimault, Albert Vaeth, Emmanuelle Grillon, Cecile
Julien, Jean-Francois Payen, Laurent Lamalle, and Michel Decorps. Vessel size
imaging. Magnetic Resonance in Medicine, 45(3):397-408, 2001.
[114] Masaki Ueno, Haruhiko Sakamoto, Hidekazu Tomimoto, Ichiro Akiguchi,
Masayuki Onodera, Cheng-Long Huang, and Kenji Kanenishi. Blood-brain barrier is impaired in the hippocampus of young adult spontaneously hypertensive
rats. Acta Neuropathologica,107(6):532-538, 2004.
[115] Gerald A. Grant, N. Joan Abbott, and Damir Janigro. Understanding the
physiology of the blood-brain barrier: In vitro models. J Cereb Blood Flow
Metab, 13(6):287-293, 1998.
[116] Riidiger von Kummer and Tobias Back. Magnetic Resonance Imaging in Ischemic Stroke. Springer, 2006.
[117] H. S. Markus. Cerebral perfusion and stroke. J Neurol Neurosurg Psychiatry,
75(3):353-361, 2004.
[118] Marios C. Papadopoulos, Geoffrey T. Manley, Sanjeev Krishna, and A. S. Verkman. Aquaporin-4 facilitates reabsorption of excess fluid in vasogenic brain
edema. FASEB J., pages 04-1723fje, 2004.
[119] James R. Ewing, Robert A. Knight, Tavarekere N. Nagaraja, John S. Y Yee,
Vijaya Nagesh, Polly A. Whitton, Lian Li, and Joseph D. Fenstermacher. Patlak
151
plots of gd-dtpa mri data yield blood-brain transfer constants concordant with
those of 14 c-sucrose in areas of blood-brain opening. Magnetic Resonance in
Medicine, 50(2):283-292, 2003.
[120] Chien-Yuan Lin, Chen Chang, Wai-Mui Cheung, Ming-Huang Lin, Jean-Ju
Chen, Chung Y. Hsu, Jyh-Horng Chen, and Teng-Nan Lin. Dynamic changes
in vascular permeability, cerebral blood volume, vascular density, and size after
transient focal cerebral ischemia in rats: evaluation with contrast-enhanced
magnetic resonance imaging. J Cereb Blood Flow Metab, 28(8):1491-1501, 2008.
152