0 1995 Wiley-Liss, Inc.
Cytometry 20:181-184 (1995)
Measurement of Tumor Necrosis Factor Activity by
Flow Cytometry
A n n Levesque, Alain Paquet, and M i c h e l Page.'
Department of Biochemistry, Faculty of Medicine, Universite Laval Sainte-Foy, Quebec, Canada
Received for publication July 11, 1994; accepted December 6, 1994
Tumor necrosis factor-a (TNF-a) is a monokine of
17 kDa produced by activated macrophages and var-
ious cells involved in the immune system. We pro-
pose a new method for the measurement of TNF
activity using flow cytometry. After an incubation
with TNF,L929 cells were harvested and treated with
a calcein-AM and ethidium homodimer-1 solution.
Nonfluorescent calcein-AM is hydrolyzed by intra-
cellular esterases to yield fluorescent calcein. The
ethidium homodimer-1 is a high-af€inity red fluo-
rescent DNA dye that is internalized only through
altered cell membranes. A very good correlation
was observed between the calcein fluorescence in-
Tumor necrosis factor was originally defined as a fac-
tor produced in animals primed with Bacillus Calmette-
Guerin (BCG) in response to endotoxin (4). Tumor ne-
crosis factor-a (TNF-a)plays a central role in the host
immunological response against various bacteria and par-
asitic infections (2). It was found to be responsible for
the necrosis of various tumors and to be a cytotoxic fac-
tor produced by T cells (6). Cells of the mononuclear
phagocyte lineages are the primary source of TNF-a, but
it is now known that many other cell types produce this
cytokine, including lymphocytes, mast cells, and even
fibroblasts ( 15). TNF is involved in various pathological
and physiological situations, and it is implicated alone or
in association with other cytokines such as interleukin
(1L)- I , IL-6, and interferons in the pathogenesis of differ-
ent diseases such as septic shock and cerebral malaria and
autoimmune diseases such as rheumatoid arthritis
(17,10,7).
Various assays have been reported for the measure-
ment of T N F cytotoxicity (1,3,5,8,9,11-13,16). These as-
says use different cell lines, various dyes, different incu-
bation periods, or other test parameters. Many of these
assays evaluate only the remaining adherent cells after
TNF treatment without a strict control on the number of
cells seeded. This causes an inherent error, because the
activity of TNF is related to the percentage of cells killed.
We report herein a sensitive method for the measure-
ment of T N F activity using calcein-AM and ethidium ho-
modimer-1 as fluorescent probes for a simultaneous mea-
surement of living and dead cells.
tensity and the number of viable cells as well as the
ethidium fluorescence and the number of cells with
altered membranes. The assay is sensitive, inexpen-
sive, and correlates with the already reported crystal
violet assay while measuring membrane alteration
by TNF.It allows the simultaneous measurement of
total living and dead cells. There is no interference
with culture medium components. This method is
rapid and may be used for routine measurement of
TNk: activity. 0 1995 Wiley-Llss, Inc.
Key terms: Bioassay, cytotoxicity, ethidium ho-
modimer-1, calcein-AM, calcein
MATERIALS AND METHODS
Cell Lines
L929 cells are a murine fibroblastoid cell line obtained
from the American Type Culture Collection (ATCC,
Rockville, MD). Cells were maintained in culture using
RPMI 1640 medium (Gibco, Grand Island, NY) contain-
ing 10% (v/v) supplemented calf serum (HyClone Labo-
ratories, Logan, UT).
Recombinant TNF
The recombinant human TNF-aused for these assays
was produced in our laboratory (14). This purified
TNF-ahad a specific activity of 10' units per milligram as
determined by the standard T N F assay of Aggarwal and
Kohr ( 1). One unit is defined by the quantity of TNF that
produces 50% lysis of L929 fibroblasts.
Optimization of Dye Concentrations and
Incubation Time
Mouse L929 fibroblasts were trypsinized and washed
three times; 5 X lo4 cells were seeded in a 96-well flat-
bottom microtiter plate in 100 pl phosphate-buffered sa-
line (PBS), p H 7.2. Various concentrations of either cal-
Address reprint requests to Michel Pa& Department of Biochemistry,
Faculty of Medicine, Universite Lavd Sainte-Foy, Quebec, Canada G l K
7P4.
182 LEVESQUE ET AL.

cein-AM or ethidium homodimer- 1 (Molecular Probes,

Eugene, OR) were added to each well in 50 pl PBS. Final
concentrations varied from 0.1 p M to 2.0 p M for cal-
cein-AM and from 0.1 p M to 5.0 p M for ethidium ho-
modimer-1. The plates were incubated at room temper-
ature, and fluorescence readings were taken every 10
min with an excitation at 485 nm and an emission at 645
nm using a Cytofluor 2300 (Millipore Instruments, Mis-
sissauga, Ontario, Canada). The same procedure was per-
formed with heat-killed L929 cells.
0 0
Flow Cytometry 0 10000 20000 30000 40000 50000

The FACScan flow cytometer (Becton Dickinson Can- Cells

ada, Saint-Laurent,Quebec, Canada), with a single excita-
tion wavelength at 488 nm, was used for all experiments.
The green fluorescence produced by calcein was ob-
served by emission at 530 nm, and the fluorescence of
ethidium homodimer-1 was observed by emission at 650
FIG. 1 . Standard curves of L 9 2 9 fibroblasts with calcein-AM (corre-
lation coefficient for living cells = 0.995; correlation coefficient for
dead cells = 0.999). Cells were seeded and incubated for 20 min in the
presence of calcein-AM (1.0 pM). The data represent the mean of qua-
druplicate determinations.

nm.
4000
Kinetics of TNF Cytotoxicity in Flow Cytometry
L929 fibroblasts (5 X 104hell) were seeded in a 24-
well plate in 200 pI RPMI containing 10%supplemented
calf serum. After a 4 h adherence period, the medium was
changed for 200 p1 of a TNF solution (2 ng/well) in RPMI w 2000-
1640 containing 100 ng actinomycin D (13). After vari-
ous incubation periods with TNF at 37°C in a 5% CO,
incubator, 100 p1calcein-AM solution ( 3 p M ) and ethid-
ium homodimer-1 (7.5 p M ) were added to the tested
wells, and the cells were incubated for 20 min. After this
period, supernatant was removed, and cells were 0 20000 40000 60000 80000 100000
trypsinized, washed, and added to their respective super- Cells
natants. These cells were then analyzed on the flow cy-
tometer as described above. FIG. 2. Standard curves of L929 fibroblasts with ethidium ho-
modimer-1 (correlation coefficient for living cells = 0.992; correlation
coefficient for dead cells = 0.999). Cells were seeded and incubated for
TNF Bioassay 20 min in the presence of ethidium homodimer-1 (2.5 pM). The data
L929 cells (5 X 104/well) were seeded in a 24-well represent the mean of quadruplicate determinations.
plate in 200 p1 RPMI 1640 containing 10% supplemented
calf serum. Cells were incubated for 4 h at 37°C in a 5%
CO, incubator. Various TNF concentrations were added
per well in the presence of actinomycin D (1 pg/ml) for
a 12 h incubation period. One hundred microliters of
calcein-AM(3 pM) and ethidium homodimer-1 (7.5 p M )
- 6 0
solution were added to each well. After a 20 min incuba-
tion period, the cells were collected and analyzed as de-
E
3
scribed above. s 4-

TNF Cytotoxicity by the Crystal Violet Method

The method described by Aggarwal and Kohr (1) was
used as a reference. After staining with crystal violet and 0 2 4 6 8 10 12 14
rinsing, the absorbance was measured with a Thermomax Time (hours)
ELISA reader (Molecular Devices, Menlo Park, CA).
FIG. 3. Kinetics of TNF cytotoxicity on 1.929 fibroblasts. Cells (5 X
RESULTS AND DISCUSSION 104/well) were seeded and cultured for various pcriods in the presence
Measurement of Intracellular Calcein of 2 ng TNF and 100 ng actinomycin D. Fluorescence was measured after
a 20 min incubation with calcein-AM (1.0 pM) and ethidium ho-
Calcein-AM hydrolysis increased steadily in the pres- modimer-1 ( 2 . 5 KM). The data represent the mean of duplicate drter-
ence of viable cells for about 20 min (Fig. 1). For prac- minations.
 TUMOR NECROSIS FACTOR ACTIVITY
100
,001 ,o1 ,1 1 10
TNF (ng/well)
FIG. 4. TNF cytotoxicity on L929 fibroblasts after 12 h (correlation
coefficient for living cells = 0.977;correlation coefficient for dead cells
= 0.968). Cells ( 5 x lO'/well) were cultured in the presence of various
concentrations of tumor necrosis factor (TNF) and in the presence of
100 ng actinomycin D. After 12 h, calcein-AM and ethidium ho-
mtdimer-1 were added, and fluorescence was measured. The data rep-
resent the mean of duplicate determinations.
tical reasons, a final concentration of 1.0 p M calcein-AM
was chosen for the TNF bioassay. This concentration
yielded a significant signal, with a minimal background
for dead cells. Increasing the concentration enhanced the
background without any substantial increase in the fluo-
rescence produced by living cells.
Measurement of Intracellular Ethidium
Homodimer-1
When ethidium homodimer was tested under the same
conditions with living and dead cells, we could observe
an increase in fluorescence in the presence of dead cells
for the first 20 min (Fig. 2). An optimal concentration of
2.5 p M ethidium homodimer was chosen.
Kinetics of TNF Cytotoxicity
The cytotoxicity of TNF on L929 cells was measured
over a period of 12 h at a fixed concentration of 2 ng/
well. Figure 3 shows that the percentage of living cells is
equal to the percentage of dead cells after about 7.5 h of
incubation. TNF cytotoxicity reaches a plateau after
about 14 h of incubation.
TNF Bioassay
Figure 4 shows the standard curves of TNF cytotoxicity
obtained after 12 h of incubation. A correlation coeffi-
cient of 0.977 was obtained between the TNF concentra-
tion and the number of living cells, and a correlation of
0.968 was obtained for the number of dead cells.
An inhibiting concentration at 50% (ICs0) of 108 pg
TNF/well (540 pg TNF/ml) was obtained with the ethid-
ium homodimer- 1 staining, and an IC,, of 6 1.6 pg TNFI
well (308 pg TNF/ml) was obtained with the calcein-AM
staining. By extrapolation, the lower limits of detection
were 1.03 pg TNF/well(5.2 pg TNF/ml) for the ethidium
183
Table 1
Comparison of Detection Methods for Tumor Necrosis Factor
Cytotoxicity Measurement
Ethidiurn
hornodimer-1 Calcein-AM
Methods w i t h FACS w i t h FACS Crystal violet
Incubation t i m e 12 12 18
(hours)
Lower limit o f 5.2 3.2 56 1
detection
(Pg/mQ
IG.0 ( p d m l ) 540 308 2,240
TNF activity 1.9X107 3.2X lo7 4.3X lo6
(units/me)
homodimer- 1 and 0.64 pg TNF/well(3.2 pg TNF/ml) for
the calcein-AM.
When the results obtained using this method are com-
pared with the standard crystal violet bioassay results in
Table 1, we observe that the lower limit of detection of
TNF activity was more than 100-fold lower than the limit
that could be measured by the crystal violet assay.
The fluorescence assay described above may be per-
formed directly in culture medium. In addition, the assay
allows the measurement of both living and dead cells
simultaneously, unlike other assays described, which
measure either the remaining attached cells or the non-
attached dead cells. Thus, inherent error due to varia-
tions in seeding density can be avoided.
LITERATURE CITED
1. Aggarwal BB, Kohr WJ: Human necrosis factor. Methods Enzymol
116:448-456, 1985.
2. Beutler B, Cerami A The biology of cachectidTNF-a primary me-
diator of the host response. Annu Rev Immunol 7:625-655, 1989.
3. Branch DR, Shah A, Guilbert LJ: A specific and reliable bioassay for
the detection of femtomolar levels o f human and murine tumor
necrosis factors. J Immunol Methods 145251-261, 1991.
4. Carswell EA, Old JL, Kassel RL, Green S, Fiore N, Williamson B: An
endotoxin-induced serum factor that causes necrosis of tumors. Proc
Natl Acad Sci USA 72:3666-3670, 1975.
5. Flick DA, Gifford GE: Comparison of in vitro cell cytotoxic assays for
tumor necrosis factor. J Immunol Methods 68:167-175, 1984.
6. Granger GA. Williams TW:Lymphocyte cytotoxicity in vitro. Acti-
vation and release of a cytotoxic factor. Nature 218:1253-1254,
1968.
7. Keffer J, Probert L,Cazlaris H, Georgopoulos S, Kaslaris E, Kioussis D,
Kollias G Transgenic mice expressing human tumor necrosis factor:
A predictive genetic model of arthritis. EMBO J 13:4025-4031,
1991.
8. Klostergaard J: A rapid extremely sensitive, quantitative microassay
for cyto-toxic cytokines. Lymphokine Res 4309-31 7, 1985.
9. Kramcr SM, Carver ME: Serumfree in vitro bioassay for the detec-
tion of tumor necrosis factor. J lmmunol Methods 93:201-206,
1986.
0. Manncl DN, Northoff H, Bauss F, Fak W: Tumor necrosis factor
alpha: A cytokine involved in toxic effects of endotoxin. Rev Infect
Dis (Suppl 5 ) 9:602-606, 1987.
1 . Meager A, Leung A, Woolley J: Assays for tumor necrosis factor and
related cytokines. J Immunol Methods 116:1-17, 1989.
2. Morgan CD, Mills KC, Letkowitz DL, k k o w i t z SS: An improved
colorimetric assay for tumor necrosis factor using WEHI 164 cells
cultured on novel microtiter plates. J lmmunol Methods 145:259-
262. 1991.
3. Ostrove JM, Gifford GE: Stimulation of RNA synthesis in L-929 cells
184 LkVESQUE ET AL.

by rabbit tumor necrosis factor. Proc Soc Exp Biol Med 160:354-
358,1979.
14. Paquet A, Levesque A, Page M: Rapid purification method for human
recombinant tumor necrosis factor alpha. J Chromatogr 667:125-
130, 1994.
IS. Semenzato G: Tumor necrosis factor: A cytokine with multiple bio-
logical activities. Br J Cancer 61354-361, 1990.
16. Suyama K, Goldstein J, Green S: Effects of murine tumor necrosis
factor on friend erythroleukemic cells. Exp Cell Bid 53:85-92,
1985.
17. Tracey KJ, Fong Y,Hesse DG, Manogue KR, Lee AT, Kuo GC, Lowry
SF, Cerami A AnticachectidTNF monoclonal antibodies prevent
septic shock during lethal bactcrdemia. Nature 330:662-664, 1987.