Transcript Single Cell Informatics - The George S. Wise Faculty of

end
MII
MI
Single Cell Informatics
entry
Motivation
Some phenomena can only be seen when filmed at
single cell level! (Here: excitability)
Outline
• Motivation: Similar cells respond differently
– most methods don’t see that: uarrays, gels, blots
• Possible reasons:
– The cells are actually not similar
– molecular “noise”
• How can we tell? Look at single cells!
– Imaging
– Image analysis
– Statistical analysis/model fitting
• Examples
– Yeast meiosis
– Apoptosis
– Competence in bacteria
Decision making in cells:
switching from one state to another
signal
apoptosis
filamentation
cell state change
sporulation
differentiation
Similar cells respond differently to the same signal
What can lead to variable responses?
1.The cells differ in some aspects (type, size, …)
2.Molecular “noise”
4
How can we study this?
Most methods average over
cells
Need to follow many single cells over time
along the process
end
MII
entry
MI
meiosis marker
But how do we track molecular levels in living cells?
The GFP revolution
• Allows tagging and
monitoring a specific protein
in vivo
• Different variants/colors allow
multiple tagging in the same
cell.
7
Example: Yeast entry into meiosis
starvation
meiosis
meiosis & sporulation
Difference between cells: time of decision
8
Yeast have a decision point
replication
MI
MII
meiosis
commitment
cell
cycle
When do cells commit?
What controls this timing and variability?
end
Regulation of entry into meiosis
signals
acetate
early genes
middle genes
We can fluorescently
tag different levels along
this pathway!
9
glucose
Ime1
master regulator
transcriptional program
nitrogen
late
genes
10
Approach: live cell imaging
early gene
YFP
30-50 positions, every 5-10 min
(1000-4000 cells/experiment)
t
rich medium
poor
medium
DIC images
YFP images
Custom image analysis
•Controlled temperature, flow
Annotation of events+more
Image analysis steps
• Cell segmentation
• Cell tracking
• Fluorescent signal measurement
These have to be tailored to cell type, motility,
signal location, etc.
Example: Image analysis for yeast
nuclear signals
1) Identify Cells
3) Identify *FP “blobs”
2) Map cells between time points
4) Map blobs to cells
mapped
t
12
cell
# cells
identified
t
Results of image analysis
 Intensities
 Num of signals
 Distance
YFP level
 Cell Size
Time
MI MII
• Large number of single cells over time
• Automated experiment + post-process
13
14
Data extraction: timing distributions
early
genes↑
tearly
“wait”
tMI
progress
tMII
Time
tearly = onset time of
early meiosis genes
15
Two-color use for event annotation
Adding another fluorescent
marker allows annotating
more events.
Hypothesis: meiosis entry
is determined by last
mitosis
6.3±2.3hr
nutrient
shift
last
mitosis
11.1±2.2hr
Conclusion: Countdown
to meiosis occurs in
parallel to the cell cycle
Htb2-mCherry ▄▄
Dmc1-YFP ▄▄
tearly
t
16
Two colors: level vs. timing
regulator
Regulator promoter activity
early genes
t early
tearly
promoter activity
Regulator promoter activity affects entry time
Molecular “noise” → spread in decision times
17
Model of causative effects
cell cycle
cell size
40%
nutrient signals
pIME1 activity
35%
onset time of early genes
80%
decision time
Large number of single cell measurements let us build a model of causative links
between molecular levels, phenotypes, event timings.
Comparing two promoter activities
The time tracks verify the circuit model:
The red and green genes are anti-correlated
Summary
• Similar cells behave differently
– molecular noise, non-molecular factors
• Quantitative fluorescent time lapse
microscopy
– Follow single cells over time
– Track protein levels/promoter activities in them
• Test dynamics of circuits (network motifs)
• Test dependencies between molecular levels,
event times, morphological properties