Multicolor Flow Cytometry Workshop
Holden Maecker, PhD
Learning Objectives
•
Explain the critical aspects of digital and multicolor flow cytometry that make it different from traditional analog flow cytometry with 1–4 colors
•
Describe the role of instrument configuration in the performance of multicolor flow cytometry
•
Perform instrument QC using BD
TMCytometer Setup and Tracking beads, and understand the use of baseline and application settings in BD FACSDiva
TMsoftware
•
Design a robust multicolor reagent panel, understanding the role of spillover, tandem dyes, and antibody titration
•
Create appropriate controls for a multicolor experiment, and be able
to find and correct potential problems in multicolor data
Schedule: Day 1
9:00–10:30 I. Introduction, review of basic concepts 10:30–10:45 Break
10:45–11:45 II. Digital and multicolor flow cytometry Exercise 1: Adjusting biexponential displays
12:00–1:00 Lunch
1:00–2:00 III. Instrument setup, optimization, and QC Exercise 2: Determining stain index and spill index 2:00–4:00 Acquisition of data:
• Instrument characterization using CS&T
• Baseline and application settings determination
• Compensation using BD™ CompBeads
• 8-color stained PBMCs
Schedule: Day 2
9:00–10:00 IV. Design and optimization of multicolor panels:
• Selection of fluorochromes
• Matching fluorochromes with antibody specificities
• Determining application-specific settings
Demonstration: Visualizing data on a virtual cytometer 10:00–12:00 Data analysis in BD FACSDiva software
12:00–1:00 Lunch
1:00–2:30 V. Controls and Data QC
Exercise 3: Finding and correcting a spillover problem 2:30–2:45 Break
2:45–4:00 Review and summary, discussion of participant issues
Multicolor Flow Cytometry Workshop:
I. Review of Flow Cytometry Basics
Outline
• Definitions, what can be measured by flow cytometry
• Fluidics: Sheath and sample streams, flow cells, sorting
• Optics: Lasers, filters
• Electronics: PMTs, signal processing
• Fluorochromes: Spectra, spillover
• Data analysis: FCS files, gating, statistics
Definitions
• Flow cytometry: The study of cells as they move in fluid suspension, allowing multiple
measurements to be made for each cell.
• FACS
TM: Fluorescence-activated cell sorting
What Measurements Can Be Made?
• Forward scatter (FSC): Proportional to cell size
• Side scatter (SSC): Proportional to cell granularity
• Fluorescence:
- Binding of fluorescent-labeled antibodies - Ca
++-sensitive dyes within cells
- Fluorescent proteins expressed by cells
- Binding of DNA dyes
Scatter Profile of Lysed Whole Blood
Side Scatter
Forward Scatter
0 200 400 600 800 1000 02004006008001000
Largest and most granular population
Smallest and least
granular population Lymphocytes
Monocytes Granulocytes
Major Components of a Flow Cytometer
• Sample injection port
• Sheath and waste reservoirs
• Flow cell
• Laser(s)
• Optical filters
• Photomultiplier tubes ( PMTs ) or photodiodes
• Signal processor
Cytometer Fluidics Create Laminar Flow
Sample stream
Sheath stream
Cell Laser beam
Flow cell
Cell Sorting
Multicolor Experiment Cytometer Configuration
PMT
Longpass filter Bandpass
filter
Background and Autofluorescence
All cells have a certain level of background fluorescence due to:
• Autofluorescence from pigments and fluorescent moieties on cellular proteins
• Non-specifically bound antibodies and free antibody in the sample stream
The level of autofluorescence varies with the wavelength of excitation and collection:
• Highest in FITC, PE detectors
• Lowest in far red (APC, Cy™7) detectors
Fluorescence Sensitivity
Detection Efficiency (Q): The number of photoelectrons generated per molecule of fluorophore
• Dependent upon fluorophore, laser, filters, PMT sensitivity, voltage gain setting, etc.
Background (B): Non-specific signal intrinsic to the system
• Dependent upon autofluorescence, unbound fluorophore, ambient
light, etc.
Common Fluorophores for Ab Conjugation
FLUOROCHROME Type of molecule Typical excitation laser Approximate emission peak
Fluorescein
isothyocyanate (FITC) Small organic 488 nm 518 nm
Alexa Fluor® 488 Small organic 488 nm 518 nm
Phycoerythrin (PE) Protein 488 or 532 nm 574 nm
PE-Texas Red® Protein tandem 488 or 532 nm 615 nm
PE-Cy™5 Protein tandem 488 or 532 nm 665 nm
Peridinin chlorophyll
protein (PerCP) Protein 488 or 532 nm 676 nm
PerCP-Cy™ 5.5 Protein tandem 488 or 532 nm 695 nm
PE-Cy™ 7 Protein tandem 488 or 532 nm 776 nm
Allophycocyanin (APC) Protein 633 nm 659 nm
Alexa Fluor® 647 Small organic 633 nm 667 nm
Alexa Fluor® 700 Small organic 633 nm 718 nm
APC-Cy™ 7 or APC-H7 Protein tandem 633 nm 784 nm
Pacific BlueTM Small organic 405 nm 454 nm
AmCyan Protein 405 nm 487 nm
Fluorescence Spillover
Emission of FITC in the PE channel
Compensating for Spillover
Uncompensated Compensated
FITC mean fluorescence PE mean fluorescence --- --- Negative Positive Negative Positive --- --- --- ---
Uncompensated 125 3,540 185 1,650
Compensated 125 3,560 135 135
1,650 – 185 3,540 – 125
% Spillover = X 100
FCS Files
• FCS 2.0 and FCS 3.0 conventions
• Often referred to as list-mode files
• Contain all of the measurements (FSC-H, FSC-A,
SSC-H, SSC-A, FL1-H…) for each individual cell
processed in a given sample
Hierarchical Gating
Web Reference Tools
• BD Spectrum Viewer:
www.bdbiosciences.com/spectra
• Maecker lab weblog:
http://maeckerlab.typepad.com
(protocols, manuscripts, literature updates)
Multicolor Flow Cytometry Workshop:
II. Digital & Multicolor Flow Cytometry
Differences From Analog Instruments
• Optics: Fiber optics and octagons/trigons
• Fluidics: Optimized for high flow rates
• Electronics: Digital signal processing
PMT Octagon and Trigon
PMT
Longpass filter
Bandpass
filter
Filter Nomenclature Conventions
Longpass (LP) filter: Allows light above a certain wavelength to pass, reflects shorter wavelengths
• Example: 505 LP = 505 nm longpass
Bandpass (BP) filter: Allows light within a certain range of wavelengths to pass (above and below a specified
midpoint)
• Example: 530/30 BP = 515–545 nm bandpass
Effect of Flow Rate
• Higher flow rates mean a broader sample stream ( less precise focusing)
• Less precise focusing means less accurate fluorescence measurement of dim populations ( population spreading)
• Higher flow rates also increase the frequency of
coincident events (can be gated out based on FSC area vs height)
• In practice, flow rates of 8,000–12,000 events per second are acceptable on the BD™ LSR II (vs 2,000–3,000
events per second on a BD FACSCalibur™ flow
cytometer)
Digital Signal Processing
Generates high resolution fluorescence values that can include negative numbers
• No compression of populations at the low end of the fluorescence scale
• More accurate representation of dim populations
Allows compensation to be performed in the software at any time
• Uncompensated data and any associated compensation matrix are both stored
• Compensation can be changed at any time
Peak area and peak height can both be recorded for all
parameters
Biexponential Display of Digital Data
0 103 104 105
0 103 104 105
101 102 103 104 105
101 102 103 104 105
101 102 103 104 105 101
102 103 104 105
Uncompensated Compensated
Log
Biexp
0 103 104 105
0 103 104 105
A B
C D
APC Area
APC-Cy7 Area
Antibody capture beads stained with 3 levels of an APC reagent
The transformed display shows aligned populations In the APC-Cy7 dimension
All populations align correctly
Spillover Affects Resolution Sensitivity
Without CD45 AmCyan With CD45 AmCyan
CD19 FITC
Conclusions
• Optical platforms using octagons and trigons result in more efficient light collection and flexibility in the use of detectors and filters
• BD LSR II fluidics allow running at higher flow rates with minimal compromise to the data
• Digital signal processing provides more accurate
representation of dim populations, and more accurate and flexible compensation—but logarithmic data display may not be appropriate
• More colors mean more spillover issues, with loss of
resolution sensitivity in affected detectors
Exercise 1
Adjusting biexponential displays:
1. Open the FCS file “exercise1.fcs”
2. Gate on small lymphocytes, then on double-positive events for CD45 AmCyan vs CD3 Pacific Blue
3. From this gate, create a plot of CD4 FITC vs CD8 APC-H7 4. Turn on biexponential scaling for the x- and y-axes, and
note the changes to the plots
5. Turn on manual biexponential scaling and experiment with various scaling factors for FITC and APC-H7, noting how the plots change
Questions:
1. Would gating be affected by biexponential scaling?
2. Is it important to use the same scaling for all samples in an
experiment?
Multicolor Flow Cytometry Workshop:
III. Instrument Setup and QC
Outline
1. Configure your instrument
2. Characterize your instrument 3. Design your panel
4. Optimize settings for your panel 5. Run appropriate controls
6. QC your data
Outline
1. Configure your instrument
• Number and type of lasers
• Number of PMTs per laser
• Choice of filters and dichroic mirrors
These choices will determine:
• What fluorochromes you can use effectively
• How well certain fluorochrome combinations will perform
How Do We Measure Performance?
W2
W1
D
Where D = difference between positive and negative peak medians W = 2 x rSD (robust standard deviation)
Stain Index = D / W
Resolution Sensitivity
CD127 PE
300 400 500 600 700 0
5 10 15 20 25 30 35 40
25 mW green laser (532 nm) 100 mW blue laser (488 nm) 25 mW blue laser (488 nm)
PMT voltage
Stain index
An Example: Green vs Blue Lasers
• Green laser is more efficient for PE and PE tandems
• Blue laser is more efficient for FITC, PerCP, and GFP
Second Example: Filters and Spillover
Outline
2. Characterize your instrument
• Obtain minimum baseline PMT settings
• Track performance over time
This allows you to:
• Run the instrument where it is most sensitive
• Be alert to changes in the instrument that might affect
performance
SDEN = 20
Baseline PMTV is set by placing the dim bead MFI to equal 10X SD Baseline PMTV is set by placing the dim bead MFI to equal 10X SDENEN
460 V
Automated Baseline PMT Voltage Determination Using BD CS&T
MFI= 200
Performance Tracking
A variety of parameters can be tracked:
• Linearity, CVs, laser alignment
• PMT voltages must hit target values
Data can be visualized in Levey-Jennings plots:
400 425 450 475 500 525 550
10/22/04 11/11/04 12/01/04 12/21/04 01/10/05 01/30/05 02/19/05 03/11/05
Time
PMT Voltage
FITC Channel (Blue laser) FITC Channel (Blue laser)
Exercise 2
Calculating stain index and spill index:
1. Open the FCS file “exercise2.fcs” (AmCyan Compbeads) 2. Calculate the stain index in the primary detector (AmCyan)
by determining:
[Median (positive peak)] - [Median (neg peak)]
2 x rSD (neg peak)
3. Calculate the spill index in FITC by determining the FITC stain index as above, then calculating:
[Stain index (FITC) / Stain index (AmCyan)]
Questions:
1. What is an acceptable stain index?
2. How high can the spill index be before it is problematic?
Stain Index for Various Fluorochromes
Reagent Clone Filter Stain Index
PE RPA-T4 585/40 356.3
Alexa Fluor® 647 RPA-T4 660/20 313.1
APC RPA-T4 660/20 279.2
PE-Cy7 RPA-T4 780/60 278.5
PE-Cy5 RPA-T4 695/40 222.1
PerCP-Cy5.5 Leu-3a 695/40 92.7
PE-Alexa Fluor® 610 RPA-T4 610/20 80.4 Alexa Fluor® 488 RPA-T4 530/30 75.4
FITC RPA-T4 530/30 68.9
PerCP Leu-3a 695/40 64.4
APC-Cy7 RPA-T4 780/60 42.2
Alexa Fluor® 700 RPA-T4 720/45 39.9
Pacific Blue™ RPA-T4 440/40 22.5
AmCyan RPA-T4 525/50 20.2
Antibody Cocktail for Data Acquisition
• CD4 FITC
• CD127 PE
• HLA-DR PerCP-Cy™5.5
• CD45RA PE-Cy7
• CD25 APC
• CD8 APC-H7
• CD3 V450
• CD45 AmCyan
Schedule: Day 2
9:00–10:00 IV. Design and optimization of multicolor panels:
• Selection of fluorochromes
• Matching fluorochromes with antibody specificities
• Determining application-specific settings
Demonstration: Visualizing data on a virtual cytometer 10:00–12:00 Data analysis in BD FACSDiva software
12:00–1:00 Lunch
1:00–2:30 V. Controls and Data QC
Exercise 3: Finding and correcting a spillover problem 2:30–2:45 Break
2:45–4:00 Review and summary, discussion of participant issues
Multicolor Flow Cytometry Workshop:
IV. Panel Design & Application Settings
Outline
3. Design your panel
• Reserve the brightest fluorochromes for the dimmest markers and vice versa
• Avoid spillover from bright populations into detectors requiring high sensitivity
• Beware of tandem dye issues
• Titrate antibodies for best separation
This allows you to:
• Maintain resolution sensitivity where you need it most
• Avoid artifacts of tandem dye degradation
Various Fluorochromes—Stain Index
Reagent Clone Filter Stain Index
PE RPA-T4 585/40 356.3
Alexa Fluor®647 RPA-T4 660/20 313.1
APC RPA-T4 660/20 279.2
PE-Cy7 RPA-T4 780/60 278.5
PE-Cy5 RPA-T4 695/40 222.1
PerCP-Cy5.5 Leu-3a 695/40 92.7
PE-Alexa Fluor® 610 RPA-T4 610/20 80.4 Alexa Fluor® 488 RPA-T4 530/30 75.4
FITC RPA-T4 530/30 68.9
PerCP Leu-3a 695/40 64.4
APC-Cy7 RPA-T4 780/60 42.2
Alexa Fluor® 700 RPA-T4 720/45 39.9
Pacific Blue™ RPA-T4 440/40 22.5
AmCyan RPA-T4 525/50 20.2
Spillover Affects Resolution Sensitivity
Without CD45 AmCyan With CD45 AmCyan
CD19 FITC
Note that this is only an issue when the two markers (CD45
and CD19) are co-expressed on the same cell population.
Special Requirements for Tandem Dyes
Compensation requirements for tandem dye conjugates can vary, even between two experiments with the same antibody
• Degrade with exposure to light, temperature, and fixation
• Stained cells are most vulnerable
Solutions:
• Minimize exposure to above agents
• Use BD stabilizing fixative if a final fix is necessary
• Run label-specific compensation
False Positives Due to Tandem Degradation
A.
False positives in APC channel reduced in absence of APC-Cy7
False positives in PE channel remain
Gating scheme CD8 APC-Cy7+ cells CD4 PE-Cy7+ cells
B.
With CD8 APC-Cy7 and CD4 PE-Cy7:
Without CD8 APC-Cy7:
Antibody Titration Basics
For most purposes, the main objective is to maximize the signal-to-noise ratio (pos/neg separation)
• This may occur at less than saturating antibody concentrations
• This may or may not be the manufacturer’s recommended titer, depending on the application
Titer is affected by:
• Staining volume (eg, 100 L)
• Number of cells (not critical up to ~5 x 10
6)
• Staining time and temperature (eg, 30 min at RT)
• Type of sample (whole blood, PBMCs, etc)
Antibody Titration Example
1 10
100 10001
10 100 1000
10000
signal
noise S:N
ng antibody
Outline
4. Optimize settings for your panel
• Derive experiment-specific PMT settings
• Run compensation controls for each experiment
This allows you to:
• Use the most appropriate settings for your panel
• Avoid gross errors of compensation
Application Settings for a New Panel (I)
Balancing detectors and checking spillover:
1. Start with the current baseline CS&T settings
2. Run single-stained BD
TMCompBeads to see if all populations are on scale
• Decrease voltage if positives are off-scale
• Increase voltage if the negative mean is below zero
3. Verify that each positive bead is at least 2x brighter in its primary detector vs other detectors (use the unstained control worksheet)
• If not, increase voltage in the primary detector
• Spill indexes for all combinations should be <0.8
Application Settings for a New Panel (II)
Optimizing voltages for cells of interest:
1. Run fully-stained cells and:
• Decrease voltages for any detectors where events are off- scale
• Increase voltages for any detectors where low-end resolution is poor (SD
NEG PEAKshould be 5–10x SD
EN)
2. Save application settings
3. Run single-stained BD CompBeads and calculate compensation
4. Run samples
Application Settings for an Existing Panel
1. Start with the current CS&T settings
2. Apply previously saved application settings
3. Run single-stained BD CompBeads and calculate compensation
4. Run samples
Demonstration
Visualizing data using a virtual cytometer:
1. Demonstration of data display as PMT voltages change
2. Note the percentage of variance due to electronic noise at different voltages
Questions:
1. What percentage of the variance contributed by electronic noise is acceptable?
2. Do you need to calculate this for all detectors and all panels?
3. Is there such a thing as too high a voltage?
Multicolor Flow Cytometry Workshop:
V. Controls and Data QC
Outline
5. Run appropriate controls
• Instrument setup controls (eg, voltage and compensation determination)
• Gating controls (eg, FMO)
• Biological controls (eg, unstimulated samples, healthy donors)
This allows you to:
• Obtain consistent setup and compensation
• Gate problem markers reproducibly
• Make appropriate biological comparisons and conclusions
BD CompBeads as Single-Color Controls
BD CompBeads provide a
convenient way to create single- color compensation controls:
• Using the same antibodies as in the experimental samples
• Creating a (usually) bright and uniform positive fluorescent peak
• Without using additional cells
Frequent Compensation Questions
Do I need to use the same antibody for compensation as I use in the experiment?
• Yes, for certain tandem dyes (eg, PE-Cy7, APC-Cy7) Are capture beads better than cells for compensation?
• Usually, as long as the antibody binds to the bead and is as bright or brighter than stained cells
Should compensation controls be treated the same as experimental samples (eg, fixed and permeabilized)?
• Yes, although with optimal fix/perm protocols this may
not make much difference
Gating Controls
Isotype control: Non-specific antibody of same isotype as the test antibody. For example :
• IgG
1FITC + IgG
2aPE + IgG
1APC
Fluorescence-minus-one (FMO) control: All test antibodies except the one of interest. For example :
• CD3 FITC + CD4 APC (no PE)
Combined control: All test antibodies except the one of interest, which is replaced by an isotype control. For example :
• CD3 FITC + IgG
2aPE + CD4 PE
Biological controls can sometimes be used as gating controls.
Gating Controls (continued)
• Isotype controls don’t take spillover into account
• FMO controls don’t take background staining into account
• Combined controls take both into account, but still may
not accurately represent the background staining of the
test antibody
Comparison of Gating Controls
Consider Using Lyophilized Reagents
• Lyophilization provides increased stability, even at room temperature or 37°C
• One batch of reagents can be used for an entire longitudinal study
• Pre-configured plates (BD Lyoplate™ plates) can avoid errors of reagent addition
• Complex experiments (multiple stimuli, multiple polychromatic staining cocktails) become easier
• Lyophilized cell controls can provide run-to-run
standardization
Outline
6. QC your data
• Visually inspect compensation
• Visually inspect gating
• Set sample acceptance criteria
This allows you to:
• Avoid classification errors and false conclusions due to
improper compensation and/or gating, or sample artifacts
Visually Inspect Compensation
• Create a template containing dot plots of each color combination in your experiment, then examine a fully stained sample for possible compensation problems
• Yikes!
Compensation Problems Can Have Cascading Effects
Compensation at 110% Compensation at 15%
Visually Inspect Gating
• Check gating across all samples in the experiment.
• Gates may need to be adjusted across donors and/or experimental runs. Dynamic (eg, snap-to) gates may help in some cases.
IL-2 PE
IFN FITC