BMS 631 Flow Cytometry: Theory 2 CR Course

1
BMS 631 Flow Cytometry Theory (2 CR Course)
Purdue University Cytometry Laboratories April
14, 2008
Cell Sorting
Guest Lecturer
James F. Leary, Ph.D. SVM Professor of
Nanomedicine Professor of Basic Medical Sciences
and Biomedical Engineering Member Purdue Cancer
Center Oncological Sciences Center Bindley
Biosciences Center Birck Nanotechnology
Center Email jfleary_at_purdue.edu
2
Why is cell sorting important ?
Once cell biologists and immunologists had found
the existence of many different cell
subpopulations with distinct properties and
functions, the need to purify these cell
subpopulations became important. Purified cell
subpopulations are used for

• Confirm the identity of cells identified through
  flow cytometric analysis
• Single cell cloning
• Homogeneous cell sorting for cell biology and
  immunology research
• Purified cell subpopulations for transplantation
  (e.g. stem cell subsets)
• Homogeneous cell subpopulations for gene array
  and proteomic analyses

3
Alternative cell separation technologies

• density gradient cell separation
• magnetic bead cell separation
• affinity column cell separation
• centrifugal elutriation cell separation
• aqueous polymer phase cell separation
• cell electrophoresis
• selective lysis of undesired cells

N.B. These are semi-quantitative cell separation
methods used for enrichment of large numbers of
specific cell subpopulations that work fine IF
these cell subpopulations can be separated on the
basis of a single parameter.
4
GENERALIZED FLOW CYTOMETER/CELL SORTER
cell
temporary
sheath
suspension
data storage
fluid
computer
CRT
system
for user
piezo-electric
permanent
interaction
transducer
data storage
beam
light
signal
flow
light
light
analog-
shaping
collection
detectors
processing
to-digital
source
optics
optics
conversion
electronics
cell
droplet
sort
charging
control
electronics
grounded
electrostatic
deflection
charging
collar
deflection
plates
sorted cells
5
Flow Cytometer Just another form of a
fluorescence microscope!
(just tip the picture (not the flow cytometer!)
on its side)
fluorescence
PMT 3
collection
optics
flow cell
sheath
light scatter
PMT 2
PMT 1
virtual slide
laser beam
virtual fluorescence microscope
The second half of a flow cytometer/cell sorter
is an exquisitely sensitive and rapid single cell
micromanipulator!
6
TYPES OF FLOW CELLS
high efficiency appendage
cylindrical
stream- in air
square cross-section
square cross-section appendage
laser
laser
lens
laser
fluorescent light
laser
reflector
laser
Microscope arc-lamp based systems
rectangular cross section with optical gel
interface
epi-illumination
on slides
syringe
objective
slide
laser
objective
N.B. The type of flow cell can have a big effect
on the cell sorting (or lack thereof!)
7
Hydrodynamic Focusing
(the first step of cell sorting is the
hydrodynamic micromanipulation of cells by
concentric laminar flow!)
Sample fluid
Sheath fluid
Initially the sample stream has a parabolic
velocity profile (Hagen-Poiseuille flow)
Cells and chromosomes will tend to align their
long axes in the direction of flow, especially if
the sample stream is made smaller in diameter
Cell
Laser
Eventually the sample stream is in fully
developed flow
N.B. The primary reasons for hydrodynamic flow
are to (1) control the position of the cells so
that it flows through the central part of the
excitation source (where the variation in
excitation intensity is a minimum), (2) to insure
that the cell flows through the focal point of
the fluorescence collection optics and (3) to
orient the cell or chromosome
8
cell sample
charge pulse
fluorescence
PMT 3
collection
optics
flow cell
sort
electronics
logic
sheath
flow cytometer
PMT 2
light scatter
PMT 1
laser beam
droplet formation
computer and software
flow cytometer/
deflection plates
cell sorter

-
cell sorter
sorted droplets


containing cell


-
of interest

-
-

-
-
-
9
GENERALIZED FLOW CYTOMETER/CELL SORTER
cell
temporary
sheath
suspension
data storage
fluid
computer
CRT
system
for user
piezo-electric
permanent
interaction
transducer
data storage
beam
light
signal
flow
light
light
analog-
shaping
detectors
collection
processing
to-digital
source
optics
optics
conversion
electronics
cell
droplet
sort
charging
control
electronics
Grounded Charging collar
electrostatic
deflection
Deflection plates
sorted cells in variety of capture devices
10
Types of Cell Sorting
B. single-cell sorting
A. bulk sorting
C. fluidic switching
D. indexed sorting
E. zapping
single-cell sorting where listmode data are
stored on each cell cloned
detect cells not of interest in beam 1
destroy cells not of interest in beam 2
zap !
4
1
2
3
11
Types of Cell Sorting

• Fluidic switching sorting (initially thought to
  be too slow) in 1960s
• Droplet (ink-jet printing) sorting (1970
  present)
• Aerosol generation problems in sterility and
  biohazards
• Rise of alternative cell sorting technologies
  (e.g. magnetic)
• Back to fluidic switching sorting? (lab-on-a-chip
  sorters)
• (technology changes!)

12
Cell Sorting is Really Real-Time Data Analysis
Cell sorting is really just another form of data
analysis except that it needs to process the
information and then make an attempt to classify
the data (and associated cell!) in real-time
(defined as the period between the laser
intersection point and the sort decision point at
the droplet break-off position). The difference
is that, unlike off-line data analysis, the sort
classification algorithms need to be very fast
(less than 1 millisecond) and need to summon the
rest of the sorting device to physically separate
cells classified as of interest or possibly of
interest within a very carefully defined time
interval to insure that the cell of interest is
captured. Since data can travel and be computed
much faster than the physical cells are moving,
this allows for the possibility of extremely
sophisticated and multi-step algorithms to be
used (e.g. complicated multivariate statistical
classifier functions).
13
Types of Sort Regions
Rectilinear Bit-mapped
cell subpopulations A B
A
A
Y
Y
B
B
X
X

• Advantages good boundaries
• Disadvantages
• more difficult to implement,
• high memory requirements, especially for more
  than 2 dimensions,
• slower than rectilinear regions to implement

• Advantages
• simple to implement,
• low memory requirements
• very fast to implement
• Disadvantages crude boundaries

14
Simple Boolean Logic for Flow Cytometry Sort
Gates Combining Different Regions
OR
Most common sort logic
AND
NAND
NOR
15
VENN DIAGRAMS SHOWING PERCENT POSITIVES OF FOUR
COLOR IMMUNOFLUORESCENCE
B
A
ABC-D-
AB-CD-
C
partialABCD-
A-B-CD
D
16
Formation of droplets inkjet printing
Sample injection needle
piezo-electric crystal
The goal of ink-jet printing is to form a stable
instability of oscillations (typically at a
resonant frequency) from a jet exiting a nozzle
orifice to produce droplets breaking off at a
stable and predictable distance (break-off
point) from the nozzle. Then we can wait the
amount of time that it takes for a cell to go
from the laser intersection point to the droplet
break-off position and issue a sort command
through the sort control logic.
nozzle
orifice
cell and laser intersection point
jet
oscillations
break-off point
satellite droplet
droplets
17
Sort timing delays - droplet charging
To sample injection needle at top of flow cell
laser excitation of cell X
cell
sort injection needle is sent a radio-frequency
pulse causing ions in sheath fluid to move up or
down the sheath column
cell X at time t
-
-
-
-
flow cytometric measurements
sort delay time dt to be set by cell sorter
operator





Droplet break-off position



where cell X will be at time t dt
sort decision







electrostatic deflection of charged droplet
containing cell of interest

-



free sorted droplet




18
Sort Delay Settings (adapted from Lindmo et al.,
1990)
13
12
14
15
11
12
13
14
11
12
14
13
15
11
15
Breakoff distance increased by 1/4 droplet
recovery
recovery
recovery
14
13
12
13
15
12
14
13
12
14
Sort delay setting (in droplets, and in fractions
of a droplet)
19
Relationship between droplet frequency, pressure,
and orifice diameter
The highly nonlinear equations describing jet
formation, pressure relationships, and droplet
break-off position are complicated and based
mostly on experimentally derived modeling
equations. But there are at least two important
take-home messages

• Droplet frequency increases as the square root
  of the pressure
• (so you need to increase the system pressures
  by a factor of four to
• double the number of droplets/sec to try to
  increasing sorting speed)
• Droplet frequency decreases with increasing
  orifice diameter
• (so if you want to sort much larger cells and
  need to increase the orifice
• diameter to prevent clogging and keep the
  sorting stable, the droplet
• frequency and rate will go down)

These take-home messages have important
implications with respect to whether cells of a
given size can be sorted, how fast, and how that
process scales.
20
Cell viability and explosive decompression

• Contrary to popular opinion, the greatest damage
  to cells comes not from shear stresses or
  sudden deceleration but rather to explosive
  decompression whereby the pressure surrounding
  the cell suddenly (in milliseconds) goes from
  cell sorter pressures (15 45 psi) to ambient
  (13.7 psi 1 atm) pressures.
• Except for a few fragile cell types, shear
  pressure is not a significant factor for
  viability of most cell types
• During capture in a test tube or microtiter
  well, the cell does undergo several gs of
  deceleration. To minimize this stress on the
  cells we typically sort into fluid in the
  collection container. The cell ultrastructure can
  be poor after sorting but can rapidly recover.

21
How sort decisions are made and implemented

• Sort decisions are made on the basis of the
  upstream flow cytometric measurements
• Typically an aliquot of cell sample is run and
  analyzed by more or less traditional flow
  cytometric data analysis.
• The cells of interest are chosen and sort
  algorithms are chosen to best capture that
  real-time data stream so that real-time sort
  decisions can be made on the subsequent cells
  that are processed.
• Data can be acquired during sorting so that the
  flow cytometric characteristics of the sorted
  cells are known and recorded.

22
Cell-cell coincidence in sorting and
anti-coincidence

• The probability of coincidence of two or more
  cells in a single sort unit (e.g. one drop, two
  drops, three drops) is much higher than the
  probability of coincidence of cells in the laser
  beam (which only occurs at very high cell
  processing rates).
• The probability of both in-beam and in sort unit
  coincidence are governed by queuing theory,
  frequently approximately by Poisson arrival
  statistics.
• If we desire high sort purity we can reject all
  cases where there are two or more cells in the
  sort unit (so called anticoincidence)
• More sophisticated anti-coincidence strategies
  have been developed to allow for improved purity
  with less sacrifice of yield.
• High-speed sorting of cells under conditions
  where anti-coincidence is partially or completely
  shut off is known as enrichment sorting.

23
Effects on cell yield and purity
Sort yield and purity are always in trade-off
mode. So it is important to determine BEFORE a
sort experiment how many cells (of what frequency
in the total cell population) are required and
whether it is practical, or even feasible, to
obtain the desired number (yield) of cells at the
desired purity.
24
Computing Cell Sorting Purities
Leary, J.F. "Strategies for Rare Cell Detection
and Isolation" In Methods in Cell Biology Flow
Cytometry (Edited by Z. Darzynkiewicz, J.P.
Robinson, H.A. Crissman), vol. 42 pp. 331-358,
1994.
25
Actual example computing sort purity
26
(No Transcript)
27
High-speed enrichment sorting
high-speed analysis (gt 100,000 cells/sec)
real-time data classification
non-rare cell, not -of-interest (e.g. original
frequency 99. 999 )
rare cell, of interest (e.g. original
frequency 0.001 )

high-speed sorting ( gt 100,000 cells/sec)



droplets containing cells not-of-interest are
charged, sorted and discarded




-


droplets containing rare cells of interest are
uncharged and sorted straight-ahead for high
efficiency recovery and localization

-



-





-



-





-





highly enriched rare cell subpopulation (e.g.
final sorted frequency 33.33 )
to waste
28
Conclusions and Take-Home Messages

• To calculate the purity of the total sort one
  must calculate the purity of each sort unit (1, 2
  or 3 droplets/sorting unit depending on choice of
  user and capability of the instrument).
• It should be obvious that the longer the queue
  length the greater the probability that two or
  more cells will be found in the queue length
  defined by the sorting unit.
• While 3-droplet sorting is the easiest to set up
  and maintain, it leads to the greatest amount of
  contamination. In contract, 2-droplet sorting is
  still fairly easy to set up and maintain yet
  leads to a considerable drop in contamination due
  to cell-coincidence in the sorting unit.
• For very high speed sorting, 1-droplet sorting
  can be performed, but it is more difficult to set
  up and is less stable in terms of sort fluidic
  stability.