Cell Biology and Toxicology. 2002; 18: 157^173.

# 2002 Kluwer Academic Publishers. Printed in the Netherlands

An improved resazurin-based cytotoxicity assay for hepatic cells

M.K. McMillian1, L. Li1,4, J.B. Parker1, L. Patel2, Z. Zhong2,5, J.W. Gunnett3, W.J. Powers1
and M.D. Johnson1
Departments of 1Molecular Toxicology and 2Exploratory Technology and 3Endocrine Group, The R.W.
Johnson Pharmaceutical Research Institute, Raritan, NJ, USA; 4Present address: Schering-Plough
Research Institute, Lafayette, NJ, USA; 5Present address: Cell & Molecular Technologies, Inc.,
Phillipsburg, NJ, USA

Received 16 August 2001; accepted 12 December 2001

Keywords: Alamar Blue, cytotoxicity assay, dicumarol, HepG2, hepatocytes, resazurin, resoru¢n

Abstract

A simple resazurin-based cytotoxicity assay is presented for screening of cytotoxicity in hepatocytes

and liver cell lines. Human hepatoma (HepG2) cells in 96-well culture plates were exposed to
known toxic (cisplatin, 5-£uorouracil, ethionine, £ufenamic acid, and di£unisal) and control
(transplatin, 5-chlorouracil, methionine, and acetylsalicylic acid) compounds for 1^3 days, and
resazurin (5 mmol/L) was added. A conventional short-term (1 h) assay was ¢rst performed, where
cytotoxicity is indicated by decreased reduction of resazurin to its £uorescent product resoru¢n.
Our improved assay consists of additionally measuring £uorescence 2^4 days later, when
cytotoxicity is indicated by a striking increase in the concentration of resoru¢n, resulting from two
distinct processes. First, viable liver-derived cells slowly convert resoru¢n to non£uorescent
metabolites. Fluorescence of control cell wells decreased to background during a 2- to 4-day
exposure to resazurin. This metabolism of resoru¢n was largely blocked by dicumarol and to lesser
extents by disul¢ram and SKF525a. Second, dead or dying cells slowly convert resazurin to
resoru¢n but do not further metabolize resoru¢n; thus this £uorescent metabolite accumulates to
high levels in wells with dead cells by 2 to 4 days. A similar increase in £uorescence associated with
cytotoxicity was observed in primary cultures of rat hepatocytes using the long-term resazurin-
based assay. In addition to an improved signal relative to the short-term assay, the inversion of the
£uorescent signal from high = alive short-term to high = dead long-term allows determination of
two independent cytotoxicity endpoints after addition of one innocuous vital dye.

Abbreviations: FLU, relative £uorescence units; 5FU, 5-£uorouracil; MTT, 3-(4,5-dimethylthiazol-

2-yl)-2,5-diphenyltetrazolium bromide; XTT, 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetra-
zolium-5-carboxanilide

Introduction compounds (reviewed by Harbell et al., 1997;

Skehan, 1999; Wilson, 2000). Since the liver is
Cytotoxicity assays provide convenient in vitro a common and serious target of most orally
screens for comparing relative toxicities of administered drug toxicities, primary liver cells
158
and human hepatic cell lines are frequently
used as targets in these screening assays. Cyto-
toxicity assays are based on cellular functions
that are particularly sensitive to and irreversi-
bly inhibited by toxins. The best cytotoxicity
assays are those that are easiest to perform. A
battery of dissimilar cytotoxicity assays is
usually run, giving complementary data that
are easier to interpret. Among the more com-
monly used are cellular permeability assays;
some measure live cell function (for example,
the ability to accumulate acetoxymethyl ester
dyes), others re£ect loss of the ability to
exclude dyes such as trypan blue and propi-
dium iodine. Assay of the release of cytosolic
lactate dehydrogenase into the media is time-
honored but cumbersome. Cytotoxicity assays
dependent on the metabolic state of treated
cells provide a di¡erent approach. Cellular
ATP levels can be easily measured in a lumi-
nescent assay. Reductase-based assays (such as
MTT, XTT, WST-1, and Alamar Blue assays)
require live cells or at least functioning mito-
chondria to convert a precursor dye into a
measurable £uorescent or colored product.
Reductase-based assays are heavily used as
they are sensitive, easy to use, and inexpensive.
Resazurin, a redox-sensitive dye long used
for detecting bacteria and yeast, is the primary
reporter dye in Alamar Blue, a proprietary
mixture with other compounds (poising
agents) added to optimize mitochondrial
reduction and inhibit nonspeci¢c reduction of
resazurin (Lancaster and Fields, 1996).
According to the package insert, Alamar Blue
(resazurin) is reduced by respiring mitochon-
dria, and the reduced form (resoru¢n) is a
more sensitive reporter than other commonly
used mitochondrial reductase dyes such as
MTT and XTT. MTT and XTT are reduced
by components early in the respiratory chain
and ultimately block electron £ow and respira-
tion; these indicator dyes are thus inherently
toxic. Alamar Blue substitutes for molecular
oxygen as an electron acceptor for oxidoreduc-
tases, such as cytochrome oxidase, the last
cytochrome in the respiratory chain; accepting
electrons at this last step does not interfere
with respiration, and thus resazurin is non-
toxic. Recent reports have questioned the
mitochondrial location of Alamar Blue (as well
as MTT and XTT) reduction, as well as the
need for poising agents (Rasmussen, 1999).
Although the details of its cellular metabolism
remain unclear, resazurin is a useful short-term
general indicator of cellular metabolic activity;
in particular, healthy cells reduce resazurin
more e¡ectively than do dead or dying cells.
Resazurin reduction leads to a loss of oxygen
and a gain of hydrogen in the molecule, and its
reduced product, resoru¢n, can be detected
both colorimetrically and £uorometrically.
Resoru¢n is a highly £uorescent molecule,
and also serves as an indicator product of
several cytochrome P450 isoform-sensitive
substrates. Unlike other reductase indicator
dyes, resazurin is nonmutagenic and relatively
nontoxic, and can be washed free from the cells
so that other assays can be performed. The
resazurin-based cytotoxicity assay is also rela-
tively insensitive to interference from drugs,
serum, and phenol red (Page et al., 1993;
Mershon et al., 1994; Ramu et al., 1996;
Gazzano-Santoro et al., 1997; Pitt et al., 1997;
Zhi-Jun et al., 1997; Nociari et al., 1998;
Mathew et al., 1999).
Since resazurin is relatively nontoxic, it is
possible to add resazurin and follow resoru¢n
production over a period of days to continually
assess e¡ects of agonists and growth factors on
cellular proliferation (Ahmed et al., 1994). This
approach may work well for many types of
cells, and is more convenient and easier to
interpret than performing a series of short-
term viability/proliferation assays on similarly
treated ``sister'' plates over a period of days.
Fortuitously, we found that this long-term
resazurin exposure approach does not work
well for liver-derived cells, but rather yields an
improved cytotoxicity assay. After examining a
 159

number of prototypical cytotoxins ^ some Alamar Blue (resazurin) assay. 100 ml of

selective for dividing cells, others known to be Alamar Blue (diluted 1:50 in Hanks bu¡er)
hepatotoxic ^ using the short-term (hours) was added under sterile conditions to treated
resazurin-based reductase assay, we discovered 96-well plates at indicated times. Immediately
that long-term exposure to this dye for several after adding Alamar Blue, zero-time-point
days produced a second, more pronounced £uorescence readings of the 96-well plate were
index of cytotoxicity. This cytotoxicity assay recorded using a Fluorometric Imaging Plate
shows increased £uorescence associated with Reader (FLIPR, Molecular Devices, Sunny-
cell death, and importantly is not dependent vale, CA, USA). FLIPR uses an argon laser
on viable cell number. (488 nm 0.5 W) to excite the plate while
collecting images from each well with a cooled
CCD camera (emission ¢lter bandpass 510^
Methods 570 nm). The emission intensity is expressed as
relative £uorescence units (FLU). The initial
Materials. Alamar Blue was purchased from
zero-time-point readings (which averaged
Biosource International (Carlsbad, CA, USA);
about 6000 FLU and were essentially equal to
resazurin and resoru¢n were from Molecular
readings from an empty culture plate) were
Probes (Eugene, OR, USA) and Electron
subtracted from later readings. Readings of
Coupler (phenazine, methosulfate, PMS) was
60 000 FLU saturated the CCD camera, but
from Boehringer Mannheim (Indianapolis, IN,
maximal levels seldom exceeded 45 000 FLU.
USA). Cisplatin, transplatin, 5-£uorouracil, 5-
In three experiments, puri¢ed resazurin (10
chlorouracil, ethionine, methionine, £ufenamic
mmol/L in Hanks bu¡er) was added for
acid, di£unisal, acetylsalicylic acid, dicumarol,
comparison with Alamar Blue. Fluorometer
disul¢ram, SKF525a, and Triton X-100 were
readings also were taken from pooled control
from Sigma Chemical Co. (St. Louis, MO,
or ethionine-treated samples to determine
USA). Drugs were diluted in 0.3% DMSO
wavelength ranges for excitation and emission
(¢nal) prior to each experiment.
of reduced Alamar Blue. Setting emission at
600 nm, the £uorescence increased with excita-
Cells. HepG2 human hepatoma cells were
tion from about 510 nm to a shoulder at 530
plated in 96-well black clear-bottom plates
nm to a peak at 572 nm. When excited at 530
(Poly¢ltronics, NUNC) at 90% con£uency
nm, a pronounced emission peak was observed
(unless stated otherwise) in minimal essential
for the ethionine sample at about 585 nm, in
medium (MEM) containing 10% fetal bovine
agreement with reported values for reduced
serum (Gibco). Quadruplicate wells were
Alamar Blue and resazurin. Using excitation
treated 24^72 h after plating with test
at 530 nm and emission at 585 nm gave
compounds in 100 ml ¢nal volume. HepG2
ethionine:control signals almost identical to
cells were kept at 378C in a humidi¢ed cell
those obtained using the FLIPR.
culture incubator in 5% CO2/95% air.
Con£uent primary rat hepatocytes were
Statistics. Unless otherwise stated, data were
obtained from In Vitro Technologies (Balti-
analyzed by ANOVA followed by Dunnett's
more, MD, USA) and were delivered plated in
test (comparing to controls) using a JMP2
collagen-coated Costar 96-well black clear-
statistics package (SAS, Research Triangle
bottom plates (Corning). Cell media was
Park, NC, USA).
changed to serum-free phenol red-free MEM
plus bovine serum albumin and compounds
added as described above.
160

Results with toxicity (Figure 2). The apparent metabo-

A conventional Alamar Blue assay (1 h, 1:100
dilution, 1%) gave strikingly di¡erent results
for some toxic compounds depending on their
duration of exposure to HepG2 cells (Figure
1a). When the duration of exposure was for
only 24 h, many compounds appeared less
toxic, i.e., the cells reduced more Alamar Blue)
than when exposed for 3 days or longer.
Toxicity was observed after short-term expo-
sure to £ufenamic acid and cisplatin, but
several days' exposure was required to see
pronounced concentration-dependent e¡ects
of 5FU and ethionine (Figure 1a).
Following Alamar Blue £uorescence
changes over several days (using the same
plates read for Figure 1a) yielded a very
di¡erent set of results (Figure 1b). In contrast
to the decreased £uorescence observed with
the traditional 1 h assay, cytotoxicity was
indicated predominantly by a pronounced
increase in £uorescence after several days with
Alamar Blue (Figure 1b). No similar
pronounced increases in resoru¢n £uorescence
were observed with compounds in the absence
of cells (data not shown).
Resoru¢n metabolism by control cells
The £uorescence associated with nontoxic or
untreated wells decreased by the second day of
exposure to Alamar Blue, while the £uores-
cence increased and persisted in the toxic
compound-treated wells. These di¡erential
changes greatly increased the signal associated
lism of resoru¢n from control wells was not
observed in preliminary experiments with
other nonhepatocyte-derived cell lines (from
human lymphoblasts and monocytes and
mouse neuroblasts). The increase in long-term
Alamar Blue £uorescence was observed with
cytotoxins in these nonhepatic cell lines, but
since there was no corresponding decrease in
£uorescence in control wells, our long-term
cytotoxicity assay was of little use (data not
shown).
In control HepG2 cell wells, three inhibitors
of ``liver-speci¢c'' enzymes ^ SKF525a,
disul¢ram, and dicumarol ^ blocked the
decrease in Alamar Blue £uorescence observed
over time (Figure 3a). Dicumarol was the most
e¡ective and least toxic of these inhibitors
(Figure 3a,b). These inhibitors were similarly
e¡ective at decreasing metabolism of resoru¢n
added to HepG2 cells (Figure 4).
The increases in long-term Alamar Blue
£uorescence associated with cytotoxicity were
reproducible both in time and in magnitude for
selected compounds. Highest £uorescence
(425 000 FLU) was observed with cytotoxins
that acted rapidly: high concentrations (30
mmol/L) of cisplatin, the hepatotoxic non-
steroidal anti-in£ammatory drugs £ufenamic
acid and di£unisal, and the detergent Triton.
However, even relatively small increases in
£uorescence (45000 FLU with 5FU, for
example) with prolonged Alamar Blue incuba-
tion were easily distinguished from control
wells or relatively nontoxic analogs (typically
500^2000 FLU).

Figure 1 (opposite). E¡ects of short-term (1 h) and long-term (3 h) Alamar Blue monitoring times in HepG2 cells exposed to a variety of
model toxicants for 1 or 3 days. Fluorescence responses (in relative £uorescence units, FLU) from cells exposed for 1 day (white bars)
or 3 days (black bars) to cisplatin (cis-P), transplatin (trans-P), 5-£uorouracil (5FU), ethionine (ethio), methionine (methio), £ufenamic
acid (£uf), or di£unisal (di£u) were determined after Alamar Blue addition. Concentrations are mmol/L. Control data were pooled
from wells treated with media or 0.3% DMSO (vehicle for £uf and di£u). (a) Alamar Blue added 1 h prior to £uorescence
determination. Note the increased sensitivity of the assay with 3 days exposure to drugs. (b) Alamar Blue present for 3 days. Note the
pronounced £uorescence signal obtained with higher concentrations of toxic compounds. Data were pooled from three experiments.
*p50.05 from control wells by Dunnett's test.
161
162

Figure 2. Time course for toxicity-induced changes in Alamar Blue £uorescence. HepG2 cells were treated with toxic compounds for 3
days before Alamar Blue addition. Alamar Blue £uorescence was then measured repeatedly after 1 day (gray bars), 2 days (black bars),
and 3 days (white bars). Note the decreasing £uorescence over time in control wells and with low concentrations of drugs. Data were
pooled from two experiments (eight wells for each condition). Abbreviations are as given in Figure 1.

While experiments were usually performed
in a cell culture incubator at 378C, plates kept
at room temperature showed essentially
identical short-term Alamar Blue reduction.
Long-term loss of control well resoru¢n
£uorescence occurred more slowly, but essen-
tially identical changes in £uorescence were
observed at 7 days at room temperature and
at 3 days at 378C (data not shown).
Is resazurin the active component of Alamar
Blue?
In comparisons of Alamar Blue (1%) with
resazurin (5 mmol/L, a major component of
Alamar Blue), there was little di¡erence
between the dyes in both traditional short-term
exposure experiments (Figure 5a) and in our
long-term cytotoxicity assay (Figure 5b).

Figure 3 (opposite). E¡ects of enzyme inhibitors on long-term decrease in resoru¢n £uorescence in HepG2 cell wells. (a) Dicumarol,
disul¢ram, and SKF525a were added together with Alamar Blue and £uorescence was determined at 1 h (gray bars), 1 day (black gars),
and 2 days (white bars). (b) Short-term (1 h) Alamar Blue assay of cytotoxicity after 2 days exposure to the enzyme inhibitors (using a
sister plate to that in (a)). Note that nontoxic concentrations of dicumarol (30 mmol/L and higher) and disul¢ram (300 mmol/L)
prevented clearance of resoru¢n. Results shown are representative of three experiments.
163
164

Figure 4. E¡ects of enzyme inhibitors on removal of exogenous resoru¢n by HepG2 cells. Dicumarol, disul¢ram, and SKF525a were
added together with resoru¢n and £uorescence determined at 2 h, 14 h, 2 days, and 3 days (increasingly dark bars). Note the similarity
to Figure 3, where resazurin-derived resoru¢n was metabolized. Results shown are representative of three experiments.

Although the bulk of these experiments
employed Alamar Blue, no di¡erences were
observed when using resazurin instead.
Resazurin reduction appears to account for
the increased £uorescence observed with long-
term Alamar Blue exposure.
Addition of an electron coupler
Phenazine methosulfate (PMS), an electron
coupler that greatly increased the signal of
other mitochondrial function (reductase)
indicators such as XTT and WST-1, had a
small e¡ect on the short-term Alamar Blue
assay (Figure 6a). However, PMS dramatically
increased the basal (but not toxicant-induced)
£uorescence after long-term Alamar Blue
exposure, and thus eliminated the signal
associated with toxicity (from over 20-fold to
less than 2-fold, Figure 6b). Prolonged
exposure to PMS did not decrease viability of
the HepG2 cells when measured with the
short-term Alamar Blue assay (data not
shown).

Figure 5 (opposite). Comparisons of £uorescence changes of Alamar Blue (white bars) and resazurin (black bars) in (a) short-term (1 h)
assay and (b) long-term (3-day) assay. Treatment of HepG2 cells with drugs was for 3 days prior to addition of Alamar Blue (1:100 ¢nal
dilution) or resazurin (5 mmol/L). Data were pooled from two experiments. Abbreviations and units are as given in Figure 1.
165
166

Figure 6. E¡ects of an electron coupler (phenazine methosulfate, PMS, 20 mmol/L, black bars) on Alamar Blue reduction (£uorescence
increase) at (a) 1 h and (b) 3 days. HepG2 cells were exposed for 3 days to cisplatin (cis-P) and transplatin (trans-P) at indicated
concentrations (mmol/L), then PMS and Alamar Blue were added. Note that PMS only slightly increased Alamar Blue reduction and
did not substantially a¡ect cytotoxicity assessment at 1 h, but markedly increased Alamar Blue reduction in cells with viable cells and
essentially blocked the signal in the long-term Alamar Blue assay. Data were pooled from two experiments.

The increase in basal Alamar Blue reduction
after addition of the electron coupler suggested
that prolonged exposure to Alamar Blue, or at
least to the product resoru¢n, might inhibit
further production of reduced Alamar Blue
(resoru¢n) in control cells. If this was the case,
the e¡ect appears to be largely reversible.
Removal of Alamar Blue after 7 days'
exposure, and replacement of fresh Alamar
Blue and media (but with no replacement of
toxic compounds) resulted in less sensitivity
but similar short-term (Figure 7a) and long-
term £uorescent changes to those observed
with Alamar Blue-naive cells (Figure 7b).
Comparisons of toxicity in fast and slow-growing
HepG2 cells
As expected, fast-growing HepG2 cultures
(treated at 50% con£uency) were more a¡ected
by 5FU, ethionine, and low concentrations of
cisplatin than were slow-growing cells (treated
at 490% con£uency, Figure 8). These di¡er-
ences were most notable when Alamar Blue
was added a day after treatment (Figure 8a).
More prolonged exposure to ethionine and
5FU produced much larger increases in
Alamar Blue £uorescence, but the con£uent
cells also showed increased responses (Figure
8). When Alamar Blue was added a week after
drug treatment, cytotoxic responses to low
concentrations of drugs were increased and
responses to high concentrations of drugs were
diminished; thus the relationship between
concentration and response was not obvious
(Figure 8c). Alamar Blue reduction also
increased in the control con£uent cell wells
with prolonged incubation times and this
increase may re£ect death of cells (Figure 8).
Flufenamic acid and high concentrations of
cisplatin produced pronounced responses
regardless of growth stage, and the e¡ects of
the higher concentrations diminished over time
regardless of con£uency (Figure 8). The con-
trol compounds transplatin and 5-chlorouracil
167
produced minimal evidence of toxicity even
when Alamar Blue was added after one week
of treatment (Figure 8c).
Long-term Alamar Blue incubation with
primary rat hepatocytes
Similar to HepG2 cells (Figure 1), primary rat
hepatocyte cultures reduced Alamar Blue in
the long-term assay after exposure to toxic
compounds. A time course for a typical
response is shown in Figure 9. There were
s i m i l ar iti e s a s wel l a s d i ¡e re n c e s i n
compounds' e¡ects on primary rat hepatocytes
and HepG2 cells. The nonsteroidal anti-in£am-
matory drugs di£unisal and £ufenamic acid
produced cytotoxicity at similar concentra-
tions in both cell types. In contrast, aspirin
appeared cytotoxic to rat hepatocytes, but not
to HepG2 cells even at concentrations as high
as 300 mmol/L (data not shown). Cisplatin was
surprisingly potent in primary hepatocytes,
showing e¡ects at 3 mmol/L (data not shown)
despite the nonproliferative state of these cells.
Transplatin was minimally toxic at the same
concentrations. Ethionine (3^30 mmol/L) was
not toxic initially, but toxicity became apparent
at later times as was observed in HepG2 cells;
the control amino acid methionine (10 and 30
mmol/L) was not toxic.
Discussion
There are two major advantages of this
prolonged-incubation resazurin-based cyto-
toxicity assay. First, the signal is superior to
that generally observed in the conventional
s h o r t- t e r m A l a m a r B l u e a s s ay. T h e
pronounced increase in £uorescence associated
with toxicity is easy to detect in screening of
pharmaceuticals and may be particularly use-
ful for the 384-well plate format. Second, the
mechanistic basis for the prolonged-incubation
assay is clearly di¡erent from that for the
168

Figure 7. E¡ect of exposure to Alamar Blue on subsequent responses to Alamar Blue. HepG2 cells were exposed to drugs for 7 days
with (white bars) or without (black bars) Alamar Blue in the ¢nal 4 days of drug exposure. Medium with and without Alamar Blue was
then replaced with medium plus Alamar Blue and £uorescence was determined at (a) 1 h and (b) 4 days. Note that cells that were
preexposed to Alamar Blue appeared slightly less sensitive to some compounds such as ethionine (ethio), but more sensitive to
di£unisal (di£u). Concentrations are mmol/L. Data were pooled from two experiments.
 169

Figure 8. Comparison of Alamar Blue reduction (increased £uorescence) in growing (50% con£uent, white bars) and con£uent (black
bars) HepG2 cells. Compounds were added at the indicated concentrations (mmol/L) for (a) 1 day, (b) 3 days, and (c) 7 days prior to
addition of Alamar Blue. Fluorescence was determined after 4 days exposure to Alamar Blue. Note that the growing cells were more
sensitive than con£uent cells to most drugs, particularly when Alamar Blue was added 1 or 3 days after drugs. Note also that the cells
(both growing and con£uent) became more sensitive to compounds with prolonged exposure prior to Alamar Blue addition. Data were
pooled from two experiments at each time point.
170

Figure 9. Long-term Alamar Blue reduction in primary rat hepatocyte culture. Cells were exposed to compounds for 3 days prior to
addition of Alamar Blue, and £uorescence readings were taken 2 days (gray bars), 4 days (black bars), and 6 days (white bars) after
Alamar Blue addition. Abbreviations are as given in Figure 1, except for aspirin (acetylsalicylic acid, ASA). Data are from one
representative experiment; other compounds also produced increases in Alamar Blue £uorescence in primary rat hepatocytes similar
to those obtained with HepG2 cells (all previous ¢gures).

conventional short-term Alamar Blue assay
and other reductase-based proliferation assays
(XTT, MTT, and WST-1). A concern with all
cytotoxicity assays based on healthy cell
function is that they are fundamentally
proliferation assays, and proliferation can be
inhibited independently of frank cytotoxicity.
For example, di¡erentiation of cell lines can be
interpreted as toxicity with such assays. To
avoid confounding results from reductase-
based cytotoxicity assays, assays are needed
that re£ect cell death per se rather than a
decrease in healthy cell number.
In the prolonged-incubation assay, the
balance between resazurin reduction (resoru¢n
production) by dead or dying cells and
resoru¢n metabolism by healthy HepG2 cells
determines the £uorescence. Only when this
balance favors cell death is there a clear-cut
signal. Given the dependence of the long-term
assay on cell death, it is somewhat surprising
that the sensitivity to cytotoxins is comparable
or better than that obtained in the short-term
Alamar Blue assay. Other advantages of the
long-term assay are the same as noted for the
traditional short-term Alamar Blue assay;
resazurin is an ``add-and-read'' substrate that
is relatively nontoxic and chemically insensitive
to serum, phenol red, and most drugs and
antioxidants. Optimally, both short- and long-
term resazurin assays should be run on the
same cell culture plates to assess cytotoxicity
from two di¡erent vantages. The obvious
disadvantage of our assay is that prolonged

(days) incubation with resazurin is required to
fully assess toxic potential of drugs. This was
particularly evident for ethionine and 5FU, but
for these compounds the delayed response
partly is due to the long exposure time required
for development of toxicity. Similar cyto-
toxicity ¢ndings with HepG2 cells and primary
rat hepatocytes suggested that the long-term
resazurin-based assay might be useful for a
variety of cells; however, preliminary results
with other cell lines indicate that this cyto-
toxicity assay is liver-speci¢c.
A major component of the signal observed
in our long-term assay results from the
decrease in resoru¢n £uorescence over time in
control wells. Although resoru¢n derivatives
have long been used as substrates for cyto-
chrome P450s in hepatocytes (Donato et al.,
1993), a recent paper indicates that measure-
ment of resoru¢n may be problematic due to
metabolism of this product by liver cells
(Evans, 1999). Dicumarol, an inhibitor of
quinone oxidoreductase, was added to liver cell
suspensions to inhibit resoru¢n metabolism,
but at best only partially inhibits (Evans,
1999; Lubet et al., 1985). Dicumarol is also an
inhibitor of glucuronidation (Segura-Aguilar et
al., 1986), and disul¢ram is an inhibitor of
glutathione conjugation reactions (Manyike et
al., 2000), which may be major clearance path-
ways of resoru¢n. Dicumarol was the most
e¡ective inhibitor of resoru¢n metabolism we
identi¢ed in our assay with HepG2 cells.
Resoru¢n interactions with aldehyde dehydro-
genase (Kitson, 1999) and with cytochrome
P450 might also contribute to the observed
decrease in £uorescence over time in control
wells; however, the respective inhibitors of
these two enzymes, disul¢ram and SKF525a,
had only weak inhibitory e¡ects at nontoxic
concentrations. While the metabolism of resor-
u¢n by control cell wells appeared hepatocyte-
speci¢c in preliminary experiments, other
metabolically highly active cells ^ perhaps
intestinal cell lines and kidney proximal
171
tubular cells (Hammond and Strobel, 1992;
Evans, 1999) ^ that remove resoru¢n are also
candidates for our long-term resazurin-based
assay.
As noted above, the basis for the enhanced
£uoresc enc e observed after prolonged
resazurin exposure to dying cells is not estab-
lished. Clearly, the signal results from both
increased, presumably ongoing, reduction of
resazurin in wells with cytotoxins, and a
pronounced decrease in £uorescence, due to
resoru¢n metabolism, in control wells.
Resazurin is essentially an electron scavenger
and for several hours after addition to cells its
reduction appears tightly coupled to mitochon-
drial respiration and cellular reductase activ-
ities, so reduction is high in control healthy cell
wells and low in dead cells (but highest in
stressed cells). A steady-state accumulation of
£uorescence would be expected with continual,
constant-rate reduction of the dye, but the
reduced resazurin product resoru¢n is not
stable in HepG2 cultures. Instead, there is a
pronounced (and fortuitous) loss of £uores-
cence over the time course of days in control
wells. Resoru¢n or its decomposition product
may compete with resazurin for mitochondrial
cytochrome oxidase and other important
reductases, since there appeared to be
abundant substrate (blue color) remaining in
the control wells after prolonged exposure;
washing and replacing the media with fresh
dye resulted in pronounced ``recoupled'' short-
term reduction of resazurin. Alternatively, an
electron coupler, such as phenazine metho-
sulfate, or a poising agent may be lost due to
decomposition or metabolism with time from
the Alamar Blue mixture or from the media.
Our data show that an electron coupler can
restore resazurin reduction in control wells
even with prolonged exposure times; however,
PMS may also couple resazurin reduction to
cellular glutathione or NADH levels rather
than respiration (Candeias et al., 1998;
Rasmussen and Nicolaisen, 1999).
172
Dying cells product abundant free and
unpaired electrons during the explosive lipid
peroxidation associated with necrosis, and a
slight increase in short-term resazurin reduc-
tion was observed after treatment with certain
cytotoxins, notably £ufenamic acid and
di£unisal. If there are other components of
Alamar Blue that suppress acute reduction of
resazurin by electrons and free radicals from
dying cells, these also must be lost within a
day, since pronounced £uorescence increases
were observed by this time. Electron produc-
tion rather than cell death per se is required to
reduce resazurin. When cells were dead and
presumably no longer generating electrons via
lipid p eroxidation, for example, after
prolonged treatment with high concentrations
of cisplatin and £ufenamic acid, resazurin
reduction was considerably less than in
concurrent wells containing dying cells treated
with lower concentrations of cisplatin and
£ufenamic acid. Surprisingly, we were unable
to block cytotoxin-induced resoru¢n £uores-
cence with antioxidants (data not shown). It is
likely that lipid peroxidation and electron
generation were too far advanced in our
experiments to be e¡ectively blocked by anti-
oxidants added just prior to Alamar Blue. It
will be di¤cult to block electron production by
dying cells while still allowing expression of
cytotoxicity, which could establish this
proposed mechanism of the assay.
It is likely that the prolonged resazurin-
based assay detects only necrotic cell death.
Although we repeatedly saw slight increases in
£uorescence in control wells after weeks of
continuous exposure, no pronounced increase
in resoru¢n £uorescence was ever observed,
although visual inspection revealed crenulated
dead cells in these wells at the end of these very
long-term experiments. Similarly, some of the
less e¡ective cytotoxins in this assay, such as
5FU and low concentrations (3 mg/ml or less)
of cisplatin, may predominantly kill cells by
inducing apoptosis. There is apparently much
less free electron generation and lipid perox-
idation associated with apoptosis than with
necrosis (Copin et al., 1998; Finstad et al.,
1998; Mizuoka et al., 1999; Gumpricht et al.,
2000).
Results obtained with the cytotoxins chosen
for this study validated the long-term assay,
and largely con¢rmed reported results with the
cytotoxins. Cisplatin at concentrations greater
than 10 mmol/L (3 mg/ml) produced equal
death in dividing and con£uent cells, but
targeted dividing cells at concentrations less
than 1 mmol/L; transplatin was toxic only at
300 mmol/L (100 mg/ml). The hepatotoxic
n o n s teroi d al anti- i n £a m m ato r y d r ug s
£ufenamic acid and di£unisal were very toxic
at 300 mmol/L, but toxicity dropped sharply at
lower concentrations, con¢rming reported
¢ndings (Masabuchi et al., 1998); aspirin (300
mmol/L) was nontoxic in HepG2 cells (data
not shown), but was toxic in rat hepatocytes.
The nonphysiological amino acid ethionine
was toxic at high concentrations (3 mmol/L
and higher), but the toxicity required pro-
longed exposure; the amino acid methionine
was nontoxic at the same high concentrations.
Similarly, 5-£uorouracil required prolonged
exposure to produce cytotoxicity, dividing cells
were more a¡ected than con£uent cells, and its
less toxic analogue 5-chlorouracil appeared
innocuous in the long-term assay.
In summary, prolonged resazurin exposure
provides a convenient cytotoxicity assay for
hepatocytes and HepG2 cells. The long-term
resazurin-based cytotoxicity assay largely
re£ects the balance between resoru¢n forma-
tion by dead or dying cells, and metabolism of
resoru¢n by hepatic-derived cells. Although
the biochemical mechanisms involved remain
unproven, the long-term assay is clearly
mechanistically distinct from the traditional
short-term resazurin reduction assay, and two
valuable sets of cytotoxicity data can be
obtained with one addition of resazurin in
liver-derived cells.

References
Ahmed SA, Gogal RM Jr, Walsh JE. A new rapid and simple
non-radioactive assay to monitor and determine the prolif-
eration of lymphocytes: an alternative to [3H]thymidine
incorporation assay. J Immunol Methods. 1994;170:211^24.
Candeias LP, MacFarlane DPS, McWhinnie SLW, et al. The
catalyzed NADH reduction of resazurin to resoru¢n. J
Chem Soc Perkin Trans. 1998;2:2333^4.
Copin JC, Li Y, Reola L, Chan PH. Trolox and 6,7-dinitroqui-
noxaline-2,3-dione prevent necrosis but not apoptosis in
cultured neurons subjected to oxygen deprivation. Brain
Res. 1998;784:25^36.
Donato MT, Gomez-Lechon MJ, Castell JV. A microassay for
measuring cytochrome P450IA1 and P450IIB1 activities in
intact human and rat hepatocytes cultured on 96-well plates.
Anal Biochem. 1993;213:29^33.
Evans PJ. Decreased intracellular proteolysis correlates with the
maintenance of a speci¢c isozyme of cytochrome P-450.
Cell Biol Int. 1999;23:117^24.
Finstad HS, Myhrstad MC, Heimli H, et al. Multiplication and
death-type of leukemia cell lines exposed to very long-chain
polyunsaturated fatty acids. Leukemia. 1998;12:921^9.
Gazzano-Santoro H, Ralph P, Ryskamp TC, Chen AB, Mukku
VR. A non-radioactive complement-dependent cytotoxicity
assay for anti-CD20 monoclonal antibody. J Immunol
Methods. 1997;202:163^71.
Gumpricht E, Devereaux MW, Dahl RH, Sokol RJ. Glu-
tathione status of isolated rat hepatocytes a¡ects bile acid-
induced cellular necrosis but not apoptosis. Toxicol Appl
Pharmacol. 2000;164:102^11.
Hammond DK, Strobel HW. Ethoxyresoru¢n O-deethylase
activity in intact human cells. Toxicol in Vitro. 1992;6:41^6.
Harbell JW, Koontz SW, Lewis RW, Lovell D, Acosta D. IRAG
Working Group 4: cell cytotoxicity assays. Food Chem
Toxicol. 1997;35:79^126.
Kitson TM. A colourful journey through the chemistry of
resoru¢n^enzyme interactions. Chem NZ. 1999;63:7^12.
Lancaster MV, Fields RD. Antibiotic and cytotoxic drug
susceptibility assays using resazurin and poising agents. US
patent 1996; 5501959.
Lubet RA, Nims RW, Mayer RT, Cameron JW, Schechtman
LM. Measurement of cytochrome P450-dependent deal-
kylation of alkoxyphenoxazones in hepatic S9s and hepato-
cyte homogenates: e¡ects of dicumarol. Mutat Res. 1985;
142:127^31.
Manyike PT, Kharasch ED, Kalhorn TF, Slattery JT. Contribu-
tion of CYP2E1 and CYP3A to acetaminophen reactive
metabolite formation. Clin Pharmacol Ther. 2000;67:275^
82.
Masabuchi Y, Saito H, Horie T. Structural requirements for the
hepatotoxicity of nonsteroidal anti-in£ammatory drugs in
173
isolated hepatocytes. J Pharmacol Exp Ther. 1998;287:208^
13.
Mathew AJ, Hollister WR, Addona T, Baust JG, Van Buskirk
RG. Vitamin E and EDTA improve the e¤cacy of hypother-
mosol ^ implication of apoptosis. In Vitro Mol Toxicol.
1999;12:163^72.
Mershon MM, Rhoads LS, Van Buskirk RG. Alamar blue
discloses latent toxicity of vesicant in human epidermal
models and cells. J Toxicol Cutan Ocul Toxicol. 1994;13:
289^310.
Mizuoka H, Shikata N, Yang J, Takasu M, Inoue K, Tsubura A.
Biphasic e¡ect of cochicine on acute liver injury induced by
carbon tetrachloride or by dimethylnitrosamine in mice. J
Hepatol. 1999;31:825^33.
Nociari MM, Shalev A, Benias P, Russo C. A novel one-step,
highly sensitive £uorometric assay to evaluate cell-mediated
cytotoxicity. J Immunol Methods. 1998;213:157^67.
Page B, Page M, Noel C. A new £uorometric assay for
cytotoxicity measurements in vitro. Int J Oncol. 1993;3:
473^6.
Pitt BR, Schwarz M, Woo ES, et al. Overexpression of
metallothionein decreases sensitivity of pulmonary endothe-
lial cells to oxidant injury. Am J Physiol. 1997;273:L856^65.
Ramu K, Perry CS, Ahmed T, Pakenham G, Kehrer JP. Studies
on the basis for the toxicity of acrolein mercapturates.
Toxicol Appl Pharmacol. 1996;140:487^98.
Rasmussen ES. Use of £uorescent redox indicators to evaluate
cell proliferation and viability. In Vitro Mol Toxicol. 1999;
12:47^58.
Rasmussen ES, Nicolaisen GM. Stability of rezazurin in bu¡ers
and mammalian cell culture media. In Vitro Mol Toxicol.
1999;12:195^202.
Segura-Aguilar JE, Barreiro V, Lind C. Dicoumarol-sensitive
glucuronidation of benzo(a)pyrene metabolites in rat liver
microsomes. Arch Biochem Biophys. 1986;251:266^75.
Skehan P. Cell growth and cytotoxicity assays. In: GP Studzins-
ki, ed. Cell growth, di¡erentiation and senescence. Oxford:
Oxford University Press; 1999:33^60.
Wilson AP. Cytotoxicity and viability assays. In: JRW Masters,
ed. Animal cell culture, 3rd edn. Oxford: Oxford University
Press; 2000:175^219.
Zhi-Jun Y, Sriranganathan N, Vaught T, Arastu SK, Ansar
Ahmed S. A dye-based lymphocyte proliferation assay that
permits multiple immunological analyses: mRNA, cytoge-
netic, apoptosis, and immunophenotyping studies. J Immu-
nol Methods. 1997;210:25^39.
Address for correspondence: Michael McMillian, R.W. Johnson
Pharmaceutical Research Institute, OMP 2295, 1000 Route
202, Raritan, NJ 08869, USA.
E-mail: mmcmilli@prius.jnj.com