multiplexed cellular analysis - aacc
TRANSCRIPT
Sean Bendall Department of Pathology
Stanford University
AACC ChicagoOct. 2, 2015
MassivelyMultiplexed Cellular
Analysis
Learning Objectives
• Understand the challenges and opportunities for single cell analysis in human pathobiology
• The use of elemental mass spectrometry in highly multiplexed cellular measurements
• Single cell mass spectrometry can be ‘routinely’ applied in the hematopoietic immune system
• ‘Single cell signatures’ can be used to predict disease outcome
• Multi-parameter subcellular imaging – watch out!
• Accessing most interesting cell types is costly (Parameters)
Parameters are the currency of single cell analysis:
• Leaves little room for the measurement of “cellular behavior”
• Major consideration for precious (clinical) samples and rare event analysis
How do we get more parameters?
Adding more parameters to bioanalysis
Fluorescence
• Up to 12 colors can be “routine”• 17 colors have been reported• High background
Elemental Isotopes
• > 100 non-biological elemental mass channels
• No compensation required• Zero background
140 145 150 155 160 165 170 175Isotopic mass (Da)
100 -
90 -
80 -
70 -
60 -
50 -
40 -
30 -
20 -
10 -
0 -
How do you get > 50 parameters?
Europium “152” is actually a 50/50 mix of 151Eu and 153Eu
Each vertical bar in these elements is a different isotope that can be separately measured
Measureby TOF
Mass cytometry enables single-cell proteomics
Stimulatecells in vitro
Crosslinkproteins
Stain with isotope tagged Abs
Nebulize single-celldroplets
Ionize(7500K)
Permeabilizecell membrane
Isotopically enriched
lanthanide ions (+3)
x 6 polymers= 180 atoms per antibody
30-site chelating polymer
• Metal tags instead of fluorescence
• 38 commercial Abparameters
• Up to 42 Ab parameters with optimization
A Rh or Ir based metalointercalator is a cellular indicator signal and yields DNA content information
Mass Cytometry ReagentsHow do we track cellular events?
Ornatsky O, et al. Anal Chem. 2008
Fresh PBMc stained with 27 markers (mix I):
Lymp B
Lymp CD4+T CD2 CD3 CD4CD45
CD45RA; CD20; CD45; CD38; CD19; CD40 CD49dCD71
CD2 175LuCD3 152SmCD4 142NdCD7 139LaCD8 146NdCD10 168ErCD11b 158Gd
CD13 166ErCD15 170ErCD19 171YbCD20 156GdCD31 144NdCD33 141PrCD34 169Tm
CD36 150NdCD38 165HoCD40 172YbCD44 151EuCD45 159TbCD45RA 153EuCD49d 145Nd
CD56 176YbCD64 148NdCD71 167ErCD90 174YbCD117 147SmHLA-DR 160Gd
Be careful what you wish for…
“seeing” single cell information in 30+ dimensions?
Biaxial plots are not a scalable solution
Parameters: 32Plots: 496
Can we create 2D maps representing higher dimensional data?
Deep Profiling of the Human
Hematopoietic Immune System
Bendall & Simonds et al., Science 2011.
Experimental design: Human bone marrow signaling
CD45CD3CD4CD8CD34CD19CD20CD33CD11bCD123
SPADE projects bone marrow as a continuum of phenotypes
0 100Intensity (%max)
Bendall & Simonds et al., Science 2011.
pSTAT5
0 100Intensity (%max)
Rediscovery of canonical signaling pathwaysBasal IL-7
Bendall & Simonds et al., Science 2011.
Clinical recovery from surgery
correlates with single-cell immune
signatures
Gaudilliere and Fragiadakis et al., Sci Transl Med, 2014
CyTOF enables a systems-level view with single-cell resolution
4. Normalization
Before normalization
After normalization
Sig
nal I
nten
sity
Time
BL 1h 24h72h1mo
1. Barcoding 3. CyTOF
Mass
Tim
e
21 Surface Markers
12 Functional Markers
2. Antibody Staining
Tim
e
5. De-barcoding
Minimum separation
Cel
l cou
nt
6. Analysis
Gaudilliere and Fragiadakis et al., Sci Transl Med, 2014
Trauma induces a redistribution of cell subsets in the blood
Gaudilliere and Fragiadakis et. al, Sci Transl Med, 2014
Surgery induces time-dependent and cell-type specific activation of immune signaling networks
1h 24h 72h 1mo -1 10
arcsinh ratio over baseline
Gaudilliere and Fragiadakis et al., Sci Transl Med, 2014
Patient n
...
Patient 2
% Pop. A
Patient 1
% Pop. B
Group 1Group 2
Group 1
Group 2
Pop. Characterization Endpoint Association
Cell 1Cell 2 .. ..Cell n
Patient 1
Raw Data
A
B
Pop. Identification
Citrus: A Data-Driven Method for the Identification of Stratifying Cellular Subsets
Prop p...
Patient 2
Prop 1
Patient 1
...
Patient n
Prop 2
Subset CharacterizationSubset Identification
Endpoint
Regularized Model
Group 1
Group 2
Group 2
...
Regularized Regression
G1 G2
Cluster Phenotype
Predictive Features
Stratifying Subsets
Bruggner et. al, PNAS, 2014
• Automatically identify cell subsets
• Calculate properties of clusters on per-sample basis
• Create regularized regression model of experimental endpoint
• Predictive features in model correspond to stratifying clusters and properties
Signaling responses in CD14+ monocyte subsets correlate with surgical recovery
Gaudilliere and Fragiadakis et al., Sci Transl Med, 2014
What about phenotype vs.
function in disease?
i.e. AML
Single Cell Proteomic AML Analysis Pipeline
Levine, Simonds & Bendall et al. Cell 2015
Phenograph – a graph-based single cell method for identifying high dimensional ‘neighborhoods’
Levine, Simonds & Bendall et al. Cell 2015
In ‘Healthy’ – Conserved Phenotypes Have Conserved Signaling Behavior
Levine, Simonds & Bendall et al. Cell In Press
Partition Cells by Function Instead of Phenotype
In AML – A Disconnect Between Cell Phenotype & Signaling Behavior Exists
Levine, Simonds & Bendall et al. Cell 2015
The Frequency of Inferred Functionally Primitive Cells (IFPC) Can Identify a Corresponding Gene
Expression Signature
Levine, Simonds & Bendall et al. Cell In Press
The IFPC Gene Signature Predicts AML Survival in Two Independent Cohorts
Better than previously identified signatures in these and related studies
Levine, Simonds & Bendall et al. Cell 2015
Mapping pluripotent stem cell differentiation and de-differentiation• Zunder E et al. Cell Stem Cell 2015• Lujan E et al. Nature 2015
Massive Immune Monitoring and Discovery –• Spitzer MH et al. , Science 2015• Brodin P et al. Cell 2015
Single Cell Clinical Biomarkers and predictive medicine• Gaudillière B et al. Science Trans. Med. 2014
Antigen specific T cell Responses• Newell et al . Nature Biotech 2014• Newell et al. Immunity 2011
Computational Challenges and Opportunities in Single Cell Biology• Shekhar K. et al. PNAS 2014• Qiu P et al. Nature Biotech 2012• Other papers above
Immune Cell Diversity• Strauss-Albee DM et al. Science Trans. Med. 2015• Horowitz A et al. Science Trans. Med. 2013
Other Notable Mass Cytometry Applications
in Cell Biology
There is a lot of room left in the Mass Cytometry analysis space….
103 198
Now 38
6 x PdBarcoding I = IdU
S‐Phase Pt = CisplatingViability
203,05,09
In‐113/15
“Single-cell suspensions not
required”
Viewing the future of high dimensional imaging?Using lanthanide ‘mass reporters’
Giesen Nature Methods, 2014
Angelo Nature Medicine, 2014
Viewing the future of high dimensional imaging?Using lanthanide ‘mass reporters’
MIBI – multiplexed ionbeam imaging
Viewing the future of high dimensional imaging?Using lanthanide ‘mass reporters’
dsDNAH&E
10-D Simultaneous
Imaging of Ductal Breast
Carcinoma
ER+, PR+, Her2+ ER+, PR+, Her2-Clinical Path:
Marker
Angelo Nature Medicine, 2014
Viewing the future of high dimensional imaging?Using lanthanide ‘mass reporters’
Angelo Nature Medicine, 2014
Viewing the future of high dimensional imaging?Using lanthanide ‘mass reporters’
Angelo Nature Medicine, 2014
END