Simultaneous Detection of Five
Fluorescent Proteins:
Customization of a Beckman Coulter
Cyan Flow Cytometer
E. Michael Meyer1, Kristen Butela2, Bryan Goddard2,
Jeffrey G. Lawrence2, Albert D. Donnenberg1,3
University of Pittsburgh Cancer Center Flow Cytometry Facility, University of
Pittsburgh, Pittsburgh, PA1 University of Pittsburgh Department of Biological
Sciences, Pittsburgh PA2 University of Pittsburgh School of Medicine,
Department of Medicine, Pittsburgh PA3
Abstract
Green Fluorescent Protein (GFP) tagged vectors have been of great
utility for selection and analysis of cells on the basis of expression of
specific genes. The development of GFP derivatives and other
fluorescent proteins now offers a palate of markers with varying
degrees of spectral separation. Although flow cytometric analysis
and sorting of cells expressing GFP is widespread, the use of
multiple fluorescent proteins (FPs) has lagged behind. Multi-laser
flow cytometers initially evolved in the direction of detecting multiple
fluorochrome-conjugated antibodies.
Detection of multiple
fluorescent proteins requires a reconfiguration of optical filters to
match their unique spectral properties.
The Beckman Coulter Cyan was among the first cytometers
equipped with 405nm, 488nm, and 643nm solid state lasers. The
standard filter sets are configured for detection of near u.v. excitable
dyes (Cascade, Pacific), FITC, PE and APC, their tandems.
Although this standard configuration is not well suited for the
analysis of multiple FPs, the modular design of the CyAn filter block
is well suited to custom configuration.
.
Abstract
In order to implement experiments involving the fluorescent barcoding of live Salmonella typhimurium LT2, we reconfigured the filter
set on our CyAn for detection of five fluorescent proteins (mKalama,
CFP, GFP, YFP, and dsRed). Filters were chosen in consultation
with Brian Manning (Chroma Technology, Bellows Falls, VT) based
on the emission spectra of the FPs. mKalama and CFP are excited
by the 405 nm laser. Emissions are resolved with 440/20 and
470/20 BP filter respectively, separated with a 455 LP dichroic. GFP,
YFP and dsRed are excited by the 488nm laser. Emissions are
resolved with 510/20, 550/30 and 610/20 BP filters and separated by
530 and 575 LP filters.
Bacterial events expressing single FPs were triggered on log
forward scatter. Initial PMT voltage settings were determined
empirically and optimized using an algorithm described in a
separate abstract submitted to this meeting (A. D. Donnenberg).
The algorithm determines a set of PMT voltages which minimize
spectral overlap within the limits of user determined constraints on
maximal and minimal gain. This combined approach of optimizing
optical filters and PMT settings permits the resolution of organisms
expressing 31 distinct combinations of FPs.
The Project
The Lawrence Lab proposed a project to examine the bacterial spectrum
that presents in the lower intestine of the common Anole. They wished to
pre-label different strains of Salmonella typhimurium with fluorescent
proteins (FPs) and observe the distribution various strains. They wished
to utilize several FPs and wanted to know the limitations of the Cyan flow
cytometer. Could we visualize the bacteria and how many different FPs
could we see at one time?
Can We Do This Experiment?
Earlier the previous year, we had completed
another project where a researcher wished
to detect GFP infected Mycobacterium
tuberculosis. We knew from experience that
the Cyan could detect small particles and do
so in the context of a GFP analysis.
Electron Micrograph by Christophe Demay
from the Institut Pasteur de Guadeloupe
Piuri M, Jacobs WR Jr, Hatfull GF (2009)
Fluoromycobacteriophages for Rapid, Specific, and
Sensitive Antibiotic Susceptibility Testing of
Mycobacterium tuberculosis. PLoS ONE 4(3): e4870.
doi:10.1371/journal.pone.0004870
Common Fluorescent Proteins
Hawley, Hawley, and Telford have
provided comprehensive guidelines for
the analysis of multiple Fluorescent
Proteins. In 2001 several invaluable
references describe the simultaneous
analysis of 4 FPs.
Hawley, T. S., Herbert, D. J., Eaker, S. S., and Hawley, R. G. 2004. Multiparameterfl ow cytometry of fluorescent protein reporters.
Flow Cytometry Protocols, Second Edition, Methods Mol. Biol. 263. T.S. Hawley and R.G. Hawley, eds. (Humana Press
Inc., Totowa, NJ) 219-238.
Pruitt, S. C., Mielnicki, L. M., and Stewart, C. C. 2004. Analysis of fluorescent protein expressing cells by flow cytometry. Flow
Cytometry Protocols, Second Edition,Methods Mol. Biol. 263. T.S. Hawley and R.G. Hawley, eds. (Humana Press Inc.,
Totowa, NJ) 239-258.
Hawley, T.S. ,Telford, W.G., Ramezani, A., and Hawley, R.G. 2001. Four-Color Flow Cytometric Detection of Retrovirally Expressed
Red, Yellow, Green, and Cyan Fluorescent Proteins Biotechniques 30:1028-1034.
Hawley, T.S. ,Telford, W.G., and Hawley, R.G. 2001. “Rainbow” Reporters for Multispectral Marking and Lineage Analysis of
Hematopoietic Stem Cells. Stem Cells 19:118-124.
Common Fluorescent
Proteins
The excitation and emission
spectra of five FPs are shown
in this poster from Chroma
Technology.
Through the implementation of
“tight”
band
pass
and
appropriate dichroic filters , it
seemed possible that we could
increase
the
number
of
available fluorescent proteins
that could be used at one time.
Blue Fluorescent Protein would
prove to be inadequate and
until the development of BFP
variants the assay would be
limited to four.
Improvement of Blue Fluorescent Proteins (BFPs)
Improvement of BFPs
“Despite the fact that it has neither the
highest intrinsic brightness nor the best
photostability of the new variants
mKalama1 is the brightest blue FP
when expressed in bacteria This result
suggests that the intensive directed
evolution
effort
that
produced
mKalama1 selected for efficient protein
folding and chromophore maturation in
bacteria in addition to high intrinsic
brightness.”
The development of mKalama as a robust 405nm excitable Fluorescent
Protein is a welcome addition. It allows for the expansion of a well
characterized 4 color panel into the lower wavelength domain. 5 colors
are now available without the costly addition of a green of yellow laser.
Cyan Native Optical Configuration
The Beckman Coulter Cyan is
designed to detect 9 colors.
Typical configuration allows for
the detection of 2 colors excited
by the 405nm laser, 5 colors
excited by the 488nm laser and
2 colors excited by the 640nm
laser. For the development of
this assay, the cyan would be
limited to the original 3 laser.
We could not add a green or
yellow laser. Optical filters
(dichroics and bandpass filters)
are
mounted
in
easily
accessible filter holders. New
filters can easily be mounted
into spare holders and replaced
in a matter of seconds.
Cyan Modified Optical Configuration
405 nm excitation of mKalama and CFP
488nm excitation of GFP, YFP and dsRed
mKalama, CFP, GFP, YFP,
and dsRed were studied and
spectral characteristics were
taken
into
account.
Consultation with Chroma
Technology made it easy to
select new optical filters.
Utilizing the existing 405nm
and 488nm laser lines,
replacing 5 bandpass filters
and 3 dichroic filter now
allows for the simultaneous
detection of all 5 FPs while at
the same time minimizing
spectral overlap.
Optical Filter Comparison
Original Filters
Bandpass Filter
450/50
530/30
530/40
575/35
613/20
Pacific Blue
Cascade Yellow
FITC
PE
ECD
Dichroic Filter
485 LP
545 LP
595 LP
Modifided Filters
Bandpass Filter
420/20
470/20
510/20
550/30
610/20
mKalama
CFP
GFP
YFP
dsRed express
Dichroic Filter
455 LP
530 LP
575 LP
Individual Fluorescent Protein Profiles
PMT Voltage Sweep
Scatter Plot
Scatter Plot
Scatter Plot
2
2
4
LMFI
3
mKalama
1
CFP
0
-1
300
2
dsRed
LMFI
LMFI
1
0
500
700
-1
300
900
VOLTS
Scatter Plot
500
700
900
600
800
VOLTS
Scatter Plot
3
3
2
2
600
800
VOLTS
YFP
LMFI
0
400
LMFI
1
1
1
GFP
1000
0
-1
200
0
400
600
VOLTS
800
-1
200
400
VOLTS
The first step in understanding how voltage changes effect spillover
coefficients (compensation) is to observe the change of measured Mean
Fluorescence Intensity (MFI) of a particular FP as voltage changes.
Individual FPs are run at various voltages (in this case, 20 volt increments)
Regression analysis of results are shown above.
Spillover Calculator
PBMC
Min
Capture 3rd Peak 3rd Peak PBMC
MIN
VOLT
MAX
VOLT
Min
target
Max
target
ORIG
%OFF
AXIS
VOLTS
VOLTS
Mkalama
400
900
Mkalama
Mkalama
822
CFP
400
900
CFP
CFP
GFP
400
900
GFP
GFP
YFP
400
900
YFP
DSRed
400
900
DSRed
INTO
THAT
Regression
Coefficients
ADDED NEW VOLTS
MACRO
B
C
886
AMkalama
0.01
-4E-06
592
-2
590
BCFP
0.01
-4E-06
470
10
480
CGFP
0.01
-7E-06
YFP
508
5
513
DYFP
0.01
-7E-06
DSRed
530
-13
517
EDSRed
0.01
-4E-06
PREDICTED SPILLOVER MATRIX
THIS
THIS
GFP
Mkalama
Mkalama
2.70%
CFP
CFP
GFP
GFP
0.34%
2.56%
YFP
YFP
0.24%
1.89% #####
DSRed
DSRed
0.20%
1.62%
#####
1.71%
1.37%
YFP
DSRed
INTO THAT
Mkalama
CFP
GFP
0.19%
0.39%
Mkalama
0.17%
0.30%
CFP
11.70%
1.05%
GFP
0.12%
3.65%
8.98%
YFP
0.07%
2.32% #####
DSRed
0.04%
1.24%
#####
8.65% #####
4494
64
OBSERVED SPILLOVER MATRIX
MkalamaCFP
PREVIOU
S SS
INCREMENT
#####
YFP
4.82%
0.96%
DSRed
0.62%
2.03%
0.14%
0.39%
40.83%
4.68%
1
1.95%
#####
14.04%
GFP and YFP emission have a great deal of spectral overlap. Optimal PMT
gain settings minimize spectral compensation. Having the ability to predict the
relationship between voltage change and MFI change, one can observe the
relative effect of voltage change on spectral spillover. Spillover is minimized
when detector voltages are “balanced”
In search of Optimal PMT Gain Albert D Donnenberg PhD CYTO 2010
Parrallel session 4-19 8:30 am Wednesday May 12
First Run Results
First Run Results
Screen capture from the first run with modified filter set
Conclusion: It is possible to resolve five fluorescent proteins
using only conventional blue and red lasers.
Spectrum and filters