3D Image Cytometry: from tissue physiology, pathology, to...
Transcript of 3D Image Cytometry: from tissue physiology, pathology, to...
3D Image Cytometry: from tissue physiology, pathology, to pharmacology
Peter T. C. So
Department of Mechanical Engineering & Biological Engineering, MIT
FemtoFab Inc.
Outline
• Cytometry
– Biomedical needs for 3D cytometry
• Frontiers in 3D cytometry
– In vivo flow cytometry
– Cell based 3D cytometry
– Tissue based 3D cytometry
• Physiology and Pathophysiology
– Neurobiology and memory plasticity
– Cancinogenesis, cardiac hypertrophy, and liver fibrosis
• System biology and pharmaceutical discovery
– Introduction to system biology & image informatics
– Pharmaceutical discovery via tissue cytometry: fibrogenesis
Flow cytometry
Cytometry: the characterization and measurement of cells and cellular
constituents
Flow cytometry: a technique for counting cells suspended in fluid as they flow
one at a time past a focus of exciting lighthttp://medical-dictionary.thefreedictionary.com/
Cytometry:
(1) Mark cells based on QUANTIFIABLE parameters
(2) Measure large population (~ 106) to achieve STATISTICAL power
(3) High throughput computation for population classification
scq.ubc.ca hmds.org.uk
Applying flow cytometry in carcionogenesis
*Spontaneous mutation is a primary cause of cancer
*A rapid mutation assay will allow the quantification of
the toxicity of new drugs and chemicals
Hendricks et al., PNAS 2003
From flow to image cytometry
DNA repair involves complex protein machinery
Tanaka, Cytometry A ,2007
Mistrik, Cell Cycle,2009
Flow Cytometry (~106) Image Cytometry (< 103)
Frontiers of Image Cytometry
Image cytometry = high throughput microscopy + fast image processing
Image cytometer capable of detecting
1 rare cell in 107 cell background based
on automated high throughput imaging
Image cytometer is conceptually
simple but can be complex in
implementation. One example is the
need for autofocus
Baja, Cytometry ,2000
Shen, Cytometry A ,2008
Why extend cytometry to 3D?
Bennett et al., JBC, 1996
Pancreas islet b-cell exhibit synchronous glucose response in 3D vs
asynchronous response in 2D
Endothelial Cell Survival in 2D vs 3D Liver Culture
Primary rat liver sinusoid endothelial cell (+hepatocyte) survival rate differs in 2D vs 3D culture
Presents of other cell types, i.e. Kupffer
and stellate cells are also important
Hwa et al., FASEB, 2007
Differential Gene Expression in 3D/2D Liver Culture
Differential gene expression is observed 500 genes examined. 14 identifiable genes are in the
TGFb pathway important for stellate cell activation
Hwa et al., FASEB, 2007
Classes of Cytometry
Cytometry
Flow cytometry Image cytometry
In vivo
flow
cytometry
Optofluidic
microscopy
2D image
cytometry
3D image
cytometry
Cell based
3D cytometry
Tissue based
3D cytometry
Structured light
3D cytometry
Light sheet
3D cytometry
Multiphoton
3D cytometry
High Throughput 3D Imaging Cytometer
Light sheet based
cytometer
Outline
• Cytometry
– Biomedical needs for 3D cytometry
• Frontiers in 3D cytometry
– In vivo flow cytometry/Optofluidic microscopy
– Cell based 3D cytometry
– Tissue based 3D cytometry
• Physiology and Pathophysiology
– Neurobiology and memory plasticity
– Cancinogenesis, cardiac hypertrophy, and liver fibrosis
• System biology and pharmaceutical discovery
– Introduction to system biology & image informatics
– Pharmaceutical discovery via tissue cytometry: fibrogenesis
In Vivo Flow Cytometry
Flow cytometry in vascular system of living animals
Geogakoudi, Cancer Research ,2004
Novak, Opt. Letters, 2004
Flow/Image cytometry – optofluidic microscope
Cui et al., PNAS 2004
Cell based 3D cytometry – structured light
Rapid 3D Cell Based Cytometry based on HiLo microscopy & Fourier transform spectroscopy
Target: 1000 cells (in 3D)/second
Cell/Tissue based 3D cytometry –light sheet
Similar light sheet based approach led by Sharpe and Stelzer groups
Connection with system biology in mapping cell lineages
Even better light sheet?
Multiphoton light sheet by Fraser group Bessel Beam -- Planchon
Status: Transforming embryology now and beyond
Tissue based cytometry – multiphoton
PMT
Ti-Sa laser
Galvanometer-drivenmirror
(Y-axis scanner)
Polygonal mirror(X-axis scanner)
Objectivelens
Computer-controlled
Specimen stage
Dichroic
mirror
Laser diode
Photo diode
Piezoelectrictranslator
Inside of microscope
PMT
Ti-Sa laser
Galvanometer-drivenmirror
(Y-axis scanner)
Polygonal mirror(X-axis scanner)
Objectivelens
Computer-controlled
Specimen stage
Dichroic
mirror
Laser diode
Photo diode
Piezoelectrictranslator
Inside of microscope
Single point scanning with high speed scanner
Kim et al., AO, 1999
Padera et al., Mol Imaging, 2001
In collaboration with Ki Hean Kim (MIT)
In collaboration with Rakesh Jain, MGH
Bewersdorf et al., OL, 1998
Buist et al., JM, 1998
y: 6
4 m
m
x: 64 mm
y: 6
4 m
m
x: 64 mm
White light
800 images at 640 frames/sec : 1,25 sec sequence
Calcium wave propagating through a cardiac myocyte labeled with flo3
Bahlmann et al., OE, 2007
Tissue based cytometry – multiphoton
Single point scanning with high speed
scanner
Challenges in tissue imaging: aberrations
Rueckel et al., PNAS, 2007
0
10
20
30
40
50
60
70
80
20 40 60 80
Depth (um)
Perc
enta
ge Im
prov
emen
t
Cha et al., JBO, 2010
0
20
40
60
80
100
120
140
160
50 100 150
Per
cen
tag
e Im
pro
vem
ent
Depth (um)
Challenges in tissue imaging: scattering 2
Emission PSF can significantly reduce signal to background ratio in confocal/imaging geometry
PMT
TPM
CCD
CCD-based MMM
Kim et al, OE 2007
Design of MMM based on Multi-anode PMT
Scanner
Microlens array
Eyepiece
L2
Tube lens
L3 Objective
Specimen
Multianode PMT (8x8)
Dichroic
L1
L4
Challenge & solution of MMM deep imaging
Kim et al, OE 2007
Maximum likelihood reassignment of scattered photons
• CCD
– blurred image
• Multianode PMT +
descanning method
– The focus the center
of each pixel of the PMT
– Still leakage of the
scattered emission
photons to neighboring
pixels ghost images
• Larger separation between
foci
• Image post processing
*Kim, K.H., et al., Multifocal multiphoton microscopy based on multianode photomultiplier tubes. Optics Express, 2007. 15(18): p. 11658-11678.
CCD MAPMT
MAPMT
w/ ML
Photon
Reassignment
Outline
• Cytometry
– Biomedical needs for 3D cytometry
• Frontiers in 3D cytometry
– In vivo flow cytometry
– Cell based 3D cytometry
– Tissue based 3D cytometry
• Physiology and Pathophysiology
– Neurobiology and memory plasticity
– Cancinogenesis, cardiac hypertrophy, and liver fibrosis
• System biology and pharmaceutical discovery
– Introduction to system biology & image informatics
– Pharmaceutical discovery via tissue cytometry: fibrogenesis
Z-Stack, Individual Slices
Reconstructed 3-D View
Computational Model of Dendrite Branches
Application of Multiphoton Microscopy in Neurobiology
Structural Basis of Memory Plasticity
Lee et al., PloS Biology 2006Lee et al., PloS Biology 2006
Background on memory plasticity
Hubel and Wiesel’s pioneering work in 1960s defined the “critical period” during
developmental when significant neuronal structural reorganization occurs. After this
period, adult memory plasticity is mostly attributed to “masking” & “unmasking” of
existing connections
This model is partly modified by advent of two-photon
microscopy & the work of Denk, Svoboda, and Yurst on
calcium signaling and dynamic remodeling of spines on
excitatory pyramidal neurons.
Does neuron-neuron connections modify on the arbor level?
Svoboda, Science 1996, Trachtenberg, Nature 2002
Discovery of neuron remodel on the dendrite level
As potential of memory plasticity
Lee et al., PloS Biology 2006
The arbor of excitatory
pyramidal neurons is static
The arbor of
inhibitory
interneurons are
dynamics and they
are GABAergic
Lee et al., PNAS 2009
Apply high throughput image cytometry to characterize neurons
capable of remodeling on the arbor level
The arbor remodeling is layer dependent
Lee et al., PNAS 2009
Apply high throughput image cytometry to characterize neurons
capable of remodeling on the arbor level
Remodeling interneurons exhibit at least three morphological
sub-types as determined by PCA and cluster analysis
Interneuron remodeling is experience dependent
Monocular visual deprivation
results in remodeling in the
binocular visual cortex with up to
16% of branch tips remodeled.
Chen et al., Nat. Neuro. 2011
Interneuron remodeling is related to dissociation and formation
of synapses
Chen et al., Nat. Neuro. 2011
Quantifying genetic mutation probability on the organism level
*Spontaneous mutation is a primary cause of cancer
*A rapid mutation assay will allow the quantification of
the toxicity of new drugs and chemicals
Hendricks et al., PNAS 2003
Recombination rate is 1 in 106 to 105.
It is a “needles in a haystack problem”
Young Mouse Pancreas Wide-field Imaging
Wide field image is not quantitative in deep tissue
Wikor-Brown, PNAS, 2008,
Kwon, Roy Soc Interface, 2009
3D cell cluster image reveal clonal ex
Wikor-Brown, PNAS, 2008,
Kwon, Roy Soc Interface, 2009
Distribution of number of cells in each foci
4 weeks vs 70+ weeks old
Recombination still occurs at old age (same rate as animal age)
Majority of recombinant cells are originated from clonal expansion (over 90%)
Identification of stem cells?
Wikor-Brown, PNAS, 2008,
Kwon, Roy Soc Interface, 2009
Genetic Basis for Cardiac hypertropy
http://www.cincinnatichildrens.org
2-D cross section of myocytes
(collagen stained with Texas Red-
Malemide)
3-D reconstructed myocyte morphology
Multi-Scale Functional modeling of Whole Mouse Heart
Ragan et al., JBO, 2007; Huang, JBO 2009
Genetic changesArchitectual changesFunctional biomechanical changes
Effect of Desmin Mutation in Mouse Heart
Huang, JBO 2009
Fibro-C, a new in vivo index for liver fibrogenesis after bile duct ligation:
a path toward less invasive fibrosis staging
Tai et al., Journal of Biomedical Optics, 2009
Improving the clinical utility of fibrosis assay using liver surface imaging
Ting et al., Journal of Biomedical Optics, 2010
Outline
• Cytometry
– Biomedical needs for 3D cytometry
• Frontiers in 3D cytometry
– In vivo flow cytometry
– Cell based 3D cytometry
– Tissue based 3D cytometry
• Physiology and Pathophysiology
– Neurobiology and memory plasticity
– Cancinogenesis, cardiac hypertrophy, and liver fibrosis
• System biology and pharmaceutical discovery
– Introduction to system biology & image informatics
– Pharmaceutical discovery via tissue cytometry: fibrogenesis
Perlman, Science 2003
Emergence of system biology
Conventional single factor analysis
System biology – Multiple factors analysisMultiple factor analysis using microarray
Perlman, Science 2003
Emergence of image informatics
Perlman et al., Science 2004
Perturbation:
Known &
unknown
“drug” at
different
dosages
Readout: morphology & protein distribution
Heat map
Most drugs that target the same pathway has similar heat map
independent of their chemical structure
Perlman et al., Science 2004
Molecular pathway signature of a drug
100 x 13 x 11 x20,000 x2 = 0.5B
compounds dilution readout cells control
Classifying unknown drug & drug interaction
Blue:
unknown drug
Perlman et al.,
Science 2004
Perlman, Science 2003
Image informatics = System biology + Structure
Bakal, Science, 2007 (Started with Perlman et al., Science 2004,)
Bakal – Gene regulation of cellular morphologyD
ow
n/u
p r
egula
tion o
f 200+
genes
Degree of similarity to
classified phenotypes vs control
Results:
*Clustering of interacting genes
*Linking gene cluster with their
functions
Future?
Genes/proteins Structure
Pattern formation
Structure Function
Force, adhesion, migration
Starting studies:
(1) Liver fibrosis (in collaboration with Hanry Yu, Colin Sheppard, NUS)
(2) Nerve regeneration (in collaboration with Ioannas Yannas, MIT)
Tissue Image Informatics: Image informatics in 3D
Fibroblast – myofibroblast
transformation and their reorganization
is a key step in tissue regeneration
Collagen Scaffold properties
affects regeneration efficacy
Regeneration:
Fewer &
disorganized
myofibroblast
Scar:
Nurmerous &
organized
myofibroblast
Yannas, JRS-Interface, 2005Meyer-ter-vehn et al., IVOS, 2006
Using image informatics to elucidate molecular mechanisms of
fibrogenesis disease processes in liver fibrosis & nerve regeneration
Chia et al., J. Hepatology, Submitted
Tzeranis et al., Roy. Soc. Interface, In press
With collaborators: H. Yu (NUS) , I. Yannas (MIT),
B. Roysam (RPI), X.Wei (Fudan University)
HSC activation is a key
step toward fibrosis
High content study of HSC pharmacological responses
Typical HSC drug response data
Correlate ex vivo HC imaging responses to drug in vivo responses
Drug efficacy is based on
in vivo results reported in
literature
Correlation between in
vivo drug efficacy and ex
vivo drug induced
morphological changes
provides information on
the mechanisms of drug
action
Most effective drugs
targets apoptosis,
contraction, and
proliferation
Image Informatics:
Quantitative model of pathophysiology
Re
gu
late
pro
tein
exp
ressio
n
Re
gu
late
ma
trix
pro
pe
rty
High throughput
Image segmentation/classification
Data clustering & interpretation
Quantitative Multiscale Model
of pathophysiologyHigh throughput,
high content
Image cytometry
Targeted perturbation
of signaling network
Ph
arm
ace
utica
l in
terv
en
tio
n
Acknowledgement