A Flow Cytometric Clonogenic Assay Reveals the Single-Cell Potency of Doxorubicin

Article abstract of DOI:10.1002/jps.24631

Standard cell proliferation assays use bulk media drug concentration to ascertain the potency of chemotherapeutic drugs; however, the relevant quantity is clearly the amount of drug actually taken up by the cell. To address this discrepancy, we have devel

Full text of DOI:10.1002/jps.24631

RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
A Flow Cytometric Clonogenic Assay Reveals the Single-Cell
Potency of Doxorubicin
KATIE F. MAASS,1,2 CHETHANA KULKARNI, MOHIUDDIN A. QUADIR, PAULA T. HAMMOND, ALISON M. BETTS,4
3
2
1,2
KARL DANE WITTRUP1
,2,5
1
Department of Chemical Engineering, Massachusetts Institute of Technology
David H. Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology
Oncology Medicinal Chemistry, Worldwide Medicinal Chemistry, Pfizer Worldwide Research and Development
Translational Research Group, Department of Pharmacokinetics Dynamics and Metabolism, Pfizer Worldwide Research and
2
3
4
Development
5
Department of Biological Engineering, Massachusetts Institute of Technology
Received 19 May 2015; revised 29 July 2015; accepted 4 August 2015
Published online 7 September 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.24631
ABSTRACT: Standard cell proliferation assays use bulk media drug concentration to ascertain the potency of chemotherapeutic drugs;
however, the relevant quantity is clearly the amount of drug actually taken up by the cell. To address this discrepancy, we have developed
a flow cytometric clonogenic assay to correlate the amount of drug in a single cell with the cell’s ability to proliferate using a cell
tracing dye and doxorubicin, a naturally fluorescent chemotherapeutic drug. By varying doxorubicin concentration in the media, length
5
10
of treatment time, and treatment with verapamil, an efflux pump inhibitor, we introduced 10 –10 doxorubicin molecules per cell; then
used a dye-dilution assay to simultaneously assess the number of cell divisions. We find that a cell’s ability to proliferate is a surprisingly
conserved function of the number of intracellular doxorubicin molecules, resulting in single-cell IC50 values of 4–12 million intracellular
doxorubicin molecules. The developed assay is a straightforward method for understanding a drug’s single-cell potency and can be used
for any fluorescent or fluorescently labeled drug, including nanoparticles or antibody–drug conjugates. ꢀC 2015 Wiley Periodicals, Inc.
and the American Pharmacists Association J Pharm Sci 104:4409–4416, 2015
Keywords: cancer chemotherapy; pharmacodynamics; drug effects; efflux pumps; cell lines
INTRODUCTION
it take to kill a cell?). This information is not directly avail-
able from potencies determined on a media-volume basis. The
assay described herein uses the amount of drug in an individ-
ual cell as the basis for cellular response rather than the drug
concentration in the cell growth media.
Routinely used in vitro proliferation assays provide a high-
throughput method for evaluating the potency of chemother-
1
,2
apeutic drugs. Typical proliferation assays use the bulk me-
dia concentration to determine the drug potency (i.e., IC50 or
IC90). However, the drug concentration on a media volumetric
basis would need to be freely in equilibrium with the drug’s
intracellular target for this to truly represent the drug’s intrin-
sic potency. This is not true in almost any case as drugs en-
counter membranous diffusion barriers and may be substrates
Standard chemotherapy potency assays include non-
clonogenic assays that are based on changes in cell mem-
brane permeability (Lactate Dehydrogenase or Trypan Blue),
mitochondrial function (MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide) or WST-1 Assay), or markers for
early (Annexin V) or late apoptosis (Terminal deoxynucleotidyl
transferase dUTP nick end labeling (TUNEL) or cytochrome
3
for active uptake or efflux transporters. The amount of drug
internalized into the cell is a more physiologically relevant ba-
sis for comparison than the bulk media concentration es-
pecially when considering drug delivery systems that involve
endosomal transport and processing steps, such as antibody–
7
c). In contrast, clonogenic assays measure a cell’s ability to
4
,5
proliferate after treatment. Traditionally, proliferation is mea-
sured by counting clones that have grown out after cells have
8
been plated at low density. Clonogenic assays capture all types
6
drug conjugates (ADCs) or liposome drug delivery systems. It
of cell death and include cell growth after reversible damage,
whereas non-clonogenic assays measure acute cellular toxicity,
often specific to one type of cell death. As clonogenic assays
capture the integrated effect of many different types of cellular
response to drug treatment, we focused on this assay type.
Here, we develop a flow cytometric dye-dilution clonogenic
assay to determine the relationship between the amount of drug
in a single cell and the cell’s ability to proliferate. Flow cytom-
etry enables high-throughput screening of thousands of indi-
vidual cells, resulting in analysis on a single-cell level rather
than a bulk population level. The assay uses a cell tracing dye
and a fluorescent drug. A cell tracing dye is used to track cell
proliferation via dye dilution. All cells are initially stained with
is now within the purview and capability of the drug designer
to attempt to alter a drug’s interaction with these transport
and processing machineries, in order to attain more efficient
delivery on target. However, a key piece of information in such
cases is the number of drug molecules on target necessary for
the desired effect (e.g., how many doxorubicin molecules does
Correspondence to: Karl Dane Wittrup (Telephone: +617-253-4578;
Fax: +617-253-1954; E-mail: wittrup@mit.edu)
This article contains supplementary material available from the authors upon
request or via the Internet at http://wileylibrary.com.
Journal of Pharmaceutical Sciences, Vol. 104, 4409–4416 (2015)
ꢀ
C 2015 Wiley Periodicals, Inc. and the American Pharmacists Association
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015
4409

4
410
RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
dye and the dye is diluted in half with each cell division. A flu- verapamil in the growth media. The replacement media after
orescent drug is used in order to measure the amount of drug 24 h also contained 20 :M verapamil.
taken up by each cell.
In this work, we used doxorubicin, a standard chemothera-
Data Analysis
9
10
peutic drug, which is also naturally fluorescent, as a model
drug to demonstrate application of the assay. Doxorobucin is
The raw flow cytometry data were processed in the following
manner in order to draw together the results from numerous
single-cell measurements from different treatment conditions.
FlowJo software (FlowJo, LLC, Ashland, Oregon) and MAT-
LAB (Mathworks, Natick, Massachusetts) were used for data
processing. First, the doxorubicin signal was calibrated as de-
scribed in section Calibration of Doxorubicin Signal. Next, we
normalized the CellTrace signal with respect to the median
CellTrace Signal for untreated cells as described in (1).
9
known to bind DNA and inhibit topoisomerase II and is widely
used as a front-line therapy for a number of different types of
1
1
cancer.
METHODS
Cell Lines and Materials
Eight different cell lines were used in this work: BT-474, HCT-
1
2
1
5, HT-29, IGROV-1, MDA-MB-231, NCI-N87, SK-BR-3, and
Proliferation factor
T-47D. Cell lines were purchased from American Type Culture
Collection (ATCC) (Manassas, Virginia). All cell lines except
HT-29 and SK-BR-3 were grown in RPMI (Corning Mediatech,
Manassas, Virginia) supplemented with 10% fetal bovine serum
and 5% penicillin–streptomycin. HT-29 and SK-BR-3 cells
were grown in Dulbecco’s Modified Eagle Medium (Corning
Mediatech) and McCoy’s 5A Medium Modified (Lonza, Basel,
Switzerland), respectively, supplemented in the same way. Dox-
orubicin hydrochloride and verapamil were purchased from
Sigma (St. Louis, Missouri).
Median CellTrace signal of untreated cells
=
(1)
CellTrace signal of sample
Note that the fluorescence signal from untreated cells is in the
numerator of the expression in (1). Cells that did not proliferate
at all have a high CellTrace signal because the CellTrace has
not been diluted by growth. Thus, the proliferation factor is low
for cells that had fewer cell divisions and is equal to one if cells
were unaffected by treatment. The theoretical minimum for the
−
n
proliferation factor with complete inhibition of growth is 2
where n is the number of doublings for untreated cells.
,
Assay Set-Up
Cells were stained using CellTraceTM Violet Cell Proliferation
With both fluorescence signals converted, the cells were
Kit (ThermoFisher Scientific (Invitrogen), Grand Island, New binned based on amount of intracellular doxorubicin. One hun-
4
York) following the “Standard Method for Labeling Cells in Sus- dred bins were used with even logarithmic spacing from 10
5
10
pension” as described in the product manual. Then, 10 cells to 10 intracellular doxorubicin molecules. Any bin with fewer
were plated per well in six-well tissue culture plates (BD Bio- than 100 cells was omitted. For each bin of cells, median prolif-
sciences, San Jose, California). The cells were treated with dox- eration factor was plotted versus the median number of intra-
orubicin hydrochloride at concentrations ranging from 10 nM cellular doxorubicin molecules resulting in a cellular response
to 5 :M in standard growth media. Control cells that were curve to doxorubicin treatment. The included plots show cellu-
either stained with CellTrace Violet only or unstained were lar response curves for either individual treatment conditions
plated at the same time. After 24 h, the cells were washed with or for data from all treatment conditions concatenated into one
phosphate-buffered saline (PBS) and the media was replaced response curve. When processing the fluorescence signal from
with fresh growth media. After an additional 3 days, the cells control cells, half of the untreated cells appear as if they have
were trypsinized and prepared for flow cytometry. Flow cytom- doxorubicin signal despite never being treated with doxorubicin
etry was performed using a BD FACSCanto II. The doxorubicin as the median doxorubicin signal for untreated cells was used
signal was measured using excitation with a 488 nm laser and to subtract out background fluorescence signal. In addition, the
detection with a 585 ± 42 nm filter. The CellTrace Violet signal lower half of the control cells appear to have a negative number
was measured using excitation with a 405 nm laser and detec- of doxorubicin cells based on the calibration of the doxorubicin
tion with a 450 ± 50 nm filter. We collected data for 10,000 cells signal and thus do not appear in the analysis plots as the plots
(
gated based on forward and side scatter) per condition, unless are log based.
there were an insufficient number of cells remaining.
Calculation of Standard and Single-Cell IC50
Treatment Length Study
The standard IC50 is the media doxorubicin concentration re-
quired for 50% of maximum reduction in proliferation factor.
The single-cell IC50 is the number of intracellular doxorubicin
molecules required for a 50% of maximum reduction in prolif-
eration factor. The IC50 values were calculated from a nonlin-
ear regression with the “log(inhibitor) versus response (three
parameter)” equation in GraphPad Prism software (GraphPad
Software, Inc., La Jolla, California). Median proliferation factor
for each treatment condition was used for standard IC50 and the
median values from bins for the concatenated data for each cell
For the dosing time study, MDA-MB-231 cells were plated as
described above in section Assay Set-Up. Initially, the media
either had a medium (0.3 :M) or high (5 :M) dose of doxoru-
bicin. The cells were washed at various time points (12, 24, 48,
72, and 96 h) and the media was replaced with fresh growth
media. All cells were read on the flow cytometer at the same
time after a total of 4 days after plating.
Verapamil Treatment
Using HCT-15 cells, the study with verapamil treatment was line were used for single-cell IC50. Confidence intervals were
set up as described above in section Assay Set-Up with 20 :M also calculated in the GraphPad Prism software.
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015
DOI 10.1002/jps.24631

RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
4411
Calibration of Doxorubicin Signal
washed with brine three times (3 × 5 mL). The organic phases
were pooled together and dried over anhydrous magnesium sul-
fate. After filtration and washing of the solid filter cake with ad-
ditional dichloromethane (2 × 5 mL), the filtrate and washings
were combined and concentrated in vacuo to viscous oil. Silica
gel column chromatography of the residue (eluted with 0%–
In order to convert the fluorescence signal from doxorubicin to
numbers of intracellular doxorubicin molecules, we used cali-
TM
ꢀR
bration beads (Quantum Simply Cellular anti-Human IgG;
Bangs Laboratories, Inc., Fishers, Indiana) and a trastuzumab–
doxorubicin conjugate (Tras–Dox). In future work, we will char-
acterize the pharmacodynamics of Tras–Dox; however, for the
purposes of the present discussion, it serves only as a useful cal-
ibration standard. BT-474, NCI-N87, and SK-BR-3 cells were
used as they have a high expression of HER2, the antigen target
5
8
% methanol/dichloromethane) yielded a red solid product at
9% yield: Matrix-Assisted Laser Desorption/Ionization-Time-
of-Flight Mass Spectrometry (MALDI-TOF MS) m/z 785.559
+
+
[
(
with Na , the K adduct also observed at 801.528]; R
95/5 CH Cl /MeOH); UV 8max 204, 251, 479, 490 nm. Major
H NMR signals were found to be identical with previously
f
0.31
TM
2
2
for trastuzumab. These cells were fixed using BD Cytofix Fix-
1
ation Buffer following the recommended protocol. For each cell
1
3
1
reported compound (3): H NMR (400 MHz, DMSO-d
6
) * =
5
line, 1.25 × 10 cells were stained with 60 nM Alexa Fluor-647-
0
2
5
1
.91 (m), 1.19 (m), 1.62 (m, 4H), 1.84 (m, 1H), 2.20 (m), 3.21 (d,
labeled trastuzumab (Tras–647) in staining buffer (PBS with
H), 3.81 (m, 1H), 4.21 (m, 4H), 4.62 (s, 2H), 4.80 (d, 1H), 4.94,
.24 (s, 1H), 5.80 (s, 1H), 7.40 (d, 1H), 7.67 (d, 1H), 7.96 (m, 2H),
2.43 (s, 1H), and 13.21 (s) ppm.
0
1
.2% bovine serum albumin and 0.09% sodium azide). Another
.25 × 10 cells were stained with 50 nM Tras–Dox in stain-
5
ing buffer. Cells were stained overnight at 37°C. After staining,
cells were washed twice with cold stain buffer before being read
on the flow cytometer.
Preparation of Tras–Dox
A trastuzumab antibody mutant with an engineered cysteine
site for conjugation, Trastuzumab A114C, was prepared accord-
Calibration beads were stained with 100 nM Tras–647 fol-
lowing the manufacturer’s recommended protocol. The calibra-
tion beads and Tras–647-labeled cells were read on a BD Accuri
C6 Flow Cytometer. Using the linear fit from calibration, the
HER2 receptor expression level on each of the cell lines was
determined from the fixed cells stained with Tras–647. Cells
stained with Tras–Dox were read on a BD FACS Canto II.
From the HER2 expression level, Tras–Dox drug-to-antibody
ratio (DAR), and doxorubicin fluorescence intensity of cells sat-
urated with Tras–Dox, the number of doxorubicin molecules
per fluorescence signal unit was determined. This was used as
a scaling factor to convert fluorescence signal from cells to num-
ber of intracellular doxorubicin molecules. The fluorescence sig-
nal above background (median fluorescence signal of untreated
cells) was considered signal from doxorubicin.
1
4
ing to literature procedure. Trastuzumab A114C antibody
7 mg/mL final concentration) was reacted with six equivalents
(
of doxorubicin–SMCC in PBS with 15% dimethylacetamide.
The reaction tube was incubated in a Thermomixer (Eppen-
dorf, Hamburg, Germany) at 25°C and 700 rpm for 2 h. Ini-
tial clean-up and concentration of the reaction mixture were
completed using a PD-10 desalting column (GE Healthcare,
Little Chalfont, United Kingdom) and an Amicon centrifu-
gal filter unit (EMD Millipore, Billerica, Massachusetts), re-
spectively. An AKTA size-exclusion chromatography system
(GE Healthcare) was employed for purification. The final con-
jugate had a DAR of 2, based on LC–MS characterization
results.
Synthesis of Doxorubicin–SMCC
Preparation of Tras–647
Doxorubicin–succinimidyl 4-[N-maleimidomethyl] cyclohex- A trastuzumab antibody mutant with an engineered cysteine
ane-1-carboxylate (SMCC) was prepared using a modification of site for conjugation, Trastuzumab A114C, was prepared accord-
1
3
14
a previously published procedure as illustrated in Scheme 1. ing to literature procedure. Trastuzumab A114C antibody
Doxorubicin HCl salt (1, 0.050 g, 0.086 mmol, 1.0 equiv.), SMCC (1 mg/mL final concentration) was reacted with five equiva-
(
(
2, 0.035 g, 0.104 mmol, 1.2 equiv.), and diisoprpylethylamine lents of NHS-Alexa Fluor 647 (ThermoFisher Scientific (Invit-
H u¨ nig’s base, 0.065 mmol, 1.5 equiv.) were added to anhydrous rogen)) in PBS with 10% 1 M borate buffer (pH 9). The re-
dimethylformamide (3.0 mL) in a 100 mL round-bottomed flask. action tube was affixed to a rotator at room temperature for
The reaction was allowed to run in the dark at room tempera- 2.5 h. Initial clean-up and concentration of the reaction mixture
ture under inert conditions overnight.
were completed using a PD-10 desalting column (GE Health-
After evaporation of the solvent under vacuum at 35°C– care) and an Amicon centrifugal filter unit (EMD Millipore),
0°C, the residue was dissolved in 15 mL dichloromethane and respectively. An AKTA size-exclusion chromatography system
4
O
O
OH
OH
O
O
O
OH
O
O
O
OH
N
N
OH
OH
,
DMF
O
O
OH
O
O
O
r.t., 14 h
O
OH
O
O
O
O
HN
HO
N
NH2.HCl
O
HO
1)
N
O
(
(2)
(3)
Scheme 1. Synthesis of Doxorubicin–SMCC.
DOI 10.1002/jps.24631
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015

4
412
RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
Figure 1. Data processing steps. (a) The raw flow cytometry results for the control population (stained with CellTrace, no doxorubicin treatment)
are shown in red and the highest doxorubicin treatment population (5 :M doxorubicin in media) is shown in blue. (b) Conversion of fluorescence
signals to number of intracellular doxorubicin molecules and the proliferation factor are shown for cell populations shown in a. The horizontal
lines illustrate growth equivalent to untreated cells and maximum inhibition with a cell doubling rate of 24 h for untreated cells. Panels (c) and
(
d) demonstrate the same transformation as in (a) and (b) with the full range (10 nM to 5 :M) of media doxorubicin treatment concentrations.
In (d), each color represents an individual doxorubicin media concentration. The solid black line is the resulting response curve from the
concatenation of all cell treatment data.
(
GE Healthcare) was employed for purification. The final conju- as great an extent. The control cells are included as a reference
gate had a fluorophore-to-antibody ratio of 1.4, based on LC–MS for background signal and variability.
characterization results.
Figures 1c and 1d illustrate results for a typical experiment
when cells were treated with a range (10 nM to 5 :M) of media
doxorubicin concentrations. The data from all treatment con-
ditions were combined and processed together, resulting in one
cellular response curve, as illustrated in Figure 1d (black line).
For reference, the response curves for individual doxorubicin
treatment conditions are also shown in Figure 1d (each color
represents one media doxorubicin concentration). The horizon-
tal dashed lines shown in Figures 1b and 1d indicate where
the proliferation factor is equal to 1, when cells are proliferat-
ing as if they were untreated, as well as the theoretical lower
limit (maximum inhibition) for the proliferation factor with an
assumed cell doubling time of 24 h.
RESULTS
Data Processing for Flow Cytometric Clonogenic Assay
For meaningful comparisons, the raw flow cytometry data were
processed as illustrated in Figure 1. The doxorubicin signal was
converted to number of intracellular doxorubicin molecules us-
ing the calibration method as described in the methods section.
The doxorubicin signal from calibration beads (105 antibody
binding sites) saturated with Tras–Dox was too dim to use for
calibration (data not shown). Instead, we measured the signal
for saturated, fixed cells (10 antibody binding sites) stained
6
with Tras–Dox and correlated this signal with the known num-
ber of doxorubicin molecules per antibody and number of HER2
receptors per cell, which was measured using calibration beads
Cell Proliferation Response with a Range of Doxorubicin
Treatment
and Tras–647. This calibration yielded a scaling factor for the The growth response of six cell lines (BT-474, HCT-15, HT-29,
doxorubicin signal observed and number of intracellular dox- IGROV-1, MDA-MB-231, and T-47D) to varying concentrations
orubicin molecules. With our flow cytometry settings, 100,000 of doxorubicin was determined on the most commonly used cul-
±
19,000 (standard error of the mean) doxorubicin molecules ture volumetric basis for the drug (Fig. 2a). There appears to
result in one flow cytometric fluorescence unit. See Supple- be substantial variation in potency of doxorubicin across these
mental Table 1 for the flow cytometry data used to reach this cell lines, ranging from 0.8 nM to 1.1 :M IC50 as listed in
conversion factor.
Table 1. This variation may be because of the differences in
Figures 1a and 1b illustrate the data processing for un- transport efficiency of the drug to the nucleus amongst cell
treated control cells and cells treated with the highest dox- lines, or instead because there are differential responses to a
orubicin treatment (5 :M doxorubicin in media). As expected, given quantity of doxorubicin in the nucleus. From the infor-
the cells treated with doxorubicin did not proliferate at the rate mation available in this plot, it is not possible to discriminate
of control cells, and consequently did not dilute the tracer dye to between these two fundamentally different possibilities.
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015
DOI 10.1002/jps.24631

RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
Table 2. Doxorubicin Single-Cell IC50 Values
4413
a
Single-Cell IC50
# of Intracellular
95% Confidence
Interval on
(
Cell Line
Doxorubicin Molecules)
Single-Cell IC50
6
8.6 × 106 to 14.8 × 10
6
BT-474
HCT-15
HT-29
IGrov-1
MDA-MB-231
T-47D
11.3 × 10
6
6
6
6
6
6
6
6.9 × 10
5.6 × 10 to 8.4 × 10
6
6
8.2 × 10
7.3 × 10 to 9.1 × 10
6
6
6.5 × 10
5.9 × 10 to 7.3 × 10
6
6
5.2 × 10
4.5 × 10 to 6.0 × 10
6
3.5 × 106 to 5.1 × 10
4.2 × 10
d
c
in 50% of maximum reduction in proliferation. It is notewor-
thy how similar the IC50 is on a per-cell basis, as opposed to
a culture-volumetric basis, indicating that differential subcel-
lular trafficking rather than differential drug responsiveness
accounts for the apparent variation in sensitivity in Figure 2a.
b
Proliferation Response as a Function of Length of Treatment
Next, we considered doxorubicin dose rate effects on cell pro-
liferation. All experiments described up to this point involved
a one-day treatment period followed by a three-day growth pe-
riod. Figure 3 illustrates how cells respond to varying treatment
period length. At the high dose (5 :M doxorubicin), cells take
up sufficient drug to stop proliferation within the first 12 h.
i
d
m
Figure 2. Cell proliferation response to doxorubicin treatment. Flow
cytometric clonogenic assay results are shown with (a) media doxoru-
bicin concentration as the basis and (b) intracellular doxorubicin con-
centration as the basis. Data were processed in same manner described
in Figure 1. Lines represent the curves for IC50 value fits. In (a), the
data points represent the median proliferation factor for a given media
doxorubicin treatment, whereas in (b), the data points represent bins
from the concatenation of all treatment conditions and replicates.
We used our flow cytometric clonogenic assay to determine
cell proliferation response to a particular number of intracel-
lular doxorubicin molecules for the same six cell lines, as illus-
trated in Figure 2b. Note that the response data points shown
in Figure 2b are the product of data concatenated from all treat-
ment conditions from replicate experiments. Response curves
for each individual treatment condition are shown in Supple-
mental Figure 1. More variability was seen in the BT-474 cell
response to doxorubicin treatment than in the responses of
other cell lines. The BT-474 cell line is known to have a vari-
1
5
able response to chemotherapy, which may be due to protein
1
6
fluctuations.
Table 2 lists the single-cell IC50 for each cell line. Thus, cel-
lular uptake of 4–12 million molecules of doxorubicin results
Table 1. Doxorubicin IC50 Values (on a Media Concentration Basis)
IC50 (Media Doxorobucin
Concentration, M)
95% Confidence Interval
on IC50 (M)
Figure 3. Effect of length of drug exposure on cell proliferation.
MDA-MB-231 cells were treated with a moderate drug concentration
(0.3 :M, solid lines) or high drug concentration (5 :M, dashed lines).
(a) Flow cytometric clonogenic assay with cell proliferation as a func-
tion of amount of doxorubicin in cells. Length of doxorubicin exposure
is given by different colors as described in legend. (b) Proliferation as a
function of the length of doxorubicin exposure with either 0.3 :M me-
dia doxorubicin concentration (solid line) or 5 :M media doxorubicin
concentration (dashed line).
Cell Line
1.8 × 10−8
−10
−7
−6
−6
−7
−8
−7
−8
BT-474
HCT-15
HT-29
IGrov-1
MDA-MB-231
T-47D
1.3 × 10
to 2.7 × 10
−
−
−
−
6
8
8
8
8
1.3 × 10
7.1 × 10
2.9 × 10
4.5 × 10
7.4 × 10 to 2.3 × 10
−8
2.2 × 10 to 2.3 × 10
−8
1.3 × 10 to 6.7 × 10
−8
1.78 × 10 to 1.2 × 10
1.1 × 10−
−9
1.8 × 10 to 7.2 × 10
DOI 10.1002/jps.24631
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015

4
414
RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
Additional treatment time resulted in increased uptake of dox- interpretation of the exact quantitative values should consider
orubicin, but no change in proliferation. However, at the mod- these factors.
erate dose (0.3 :M doxorubicin), cell proliferation decreased in
In this work, we assume that the intracellular doxorubicin
response to increased doxorubicin treatment time in a man- is all available to bind to DNA; however, there may be a frac-
ner dependent on the amount of doxorubicin in the cell. The tion of drug bound to proteins and lipids in the cell. The free
cellular response to doxorubicin is one continuous function of drug (i.e., not bound to proteins or lipids) is the relevant quan-
3
0
the amount of doxorubicin in the cell, independent of the rate tity in relation to efficacy. As doxorubicin is a relatively hy-
3
1
at which that level of doxorubicin was obtained. This effect is drophilic drug (log D7.4 = −1.98), this cellular protein bound
not observed when the length of treatment was used as the drug accounts for less of the intracellular drug than for more
metric to evaluate cell proliferation instead of the amount of hydrophobic drugs, where drug bound to cellular proteins could
3
2
doxorubicin in the cells, as in Figure 3b.
account for more than 90% of the intracellular drug. Plasma
protein binding of drugs could be used as a correlate for cellu-
Doxorubicin Uptake and Cell Proliferation in Pgp Expressing Cells lar protein binding. Methods are established for plasma protein
binding as well as published values for doxorubicin and a num-
Doxorubicin treatment is known to induce multidrug resistance
via expression of the drug efflux pump P-glycoprotein (Pgp).
3
2–34
ber of other drugs.
Hydrophobicity can also affect a drug’s
1
7–19
ability to diffuse into cells, which drives the difference between
intracellular and extracellular drug concentrations.
In Figure 4, we observed how treatment with a Pgp inhibitor,
verapamil, affected doxorubicin uptake and cell proliferation
in Pgp-expressing cells. Of the cell lines that we tested, only
HCT-15 cells express Pgp without any pre-treatment with
Another key consideration when interpreting results from
this assay is the fact that cells that have died by the time of
analysis are not seen in the flow cytometry results. (The absence
of dead cells is intrinsic to all clonogenic assays.) Thus, the
reported intracellular doxorubicin values are for cells that are
still intact. Likely, for many treatment conditions, there were
cells that took up more doxorubicin and died before analysis
was completed. Lower cell number and altered cell morphology
as indicated by forward and side scatter patterns were observed
for cells treated at the highest doxorubicin concentrations.
Previous studies that have quantified intracellular doxoru-
1
9
doxorubicin. Again, the cell proliferation response to intra-
cellular doxorubicin is one continuous response curve, rather
than different response curves in the presence or absence of ve-
rapamil treatment. In contrast, the cell proliferation response
curves with respect to the media doxorubicin concentration are
two distinct curves in the presence or absence of verapamil
treatment, as shown in Figure 4b. Thus, the key factor for
HCT-15 cell proliferation is the amount of doxorubicin in the
cell. It is irrelevant in terms of cell proliferation whether the
amount of doxorubicin in the cell resulted from a higher media
doxorubicin concentration or from a lower media doxorubicin
concentration with verapamil treatment.
7
9
bicin report 10 –5 × 10 intracellular doxorubicin molecules
6
,35–39
after 2–12 h treatment periods. Many of these reported
values are higher than the single-cell IC50 values shown here,
because the respective studies use higher doxorubicin media
concentrations for treatment and the intracellular doxorubicin
was quantified immediately after treatment rather than after
a 3-day growth period as in this assay. Most of the previous
DISCUSSION
In this work, a flow cytometric clonogenic assay was devel- intracellular doxorubicin quantification studies were not cor-
2
2
oped and the application of this assay was demonstrated with related with cell proliferation. However, Kerr et al. reported
the model drug doxorubicin. Intracellular doxorubicin, rather a LD90 (lethal dose for 90% of colonies) of 39 million intracel-
than doxorubicin concentration in the cell growth media, was lular doxorubicin molecules. This value corresponds well with
demonstrated as the key determinant of cell proliferation in- single-cell IC50 values reported here.
hibition. Whether the amount of intracellular doxorubicin was
Delivery of tens of millions of doxorubicin molecules to a cell,
achieved by varying doxorubicin media concentration, treating although easily accomplished via diffusion of the free drug,
cells with verapamil, or increasing treatment length was ir- could be harder to achieve with an ADC. ADCs are designed
relevant. These results are consistent with previous studies of to carry a drug payload specifically to cancer cells and re-
doxorubicin cellular pharmacodynamics.20–24
lease the drug once internalized into the cell. Consider a cell
6
The doxorubicin single-cell IC50 values were 4–12 million with high antigen expression at 10 surface antigens per cell
intracellular doxorubicin molecules. To put this number in con- and an internalization half time of 12 h. With a DAR of 4, a
text, this represents ꢀ3% loading of the total number of binding cell would internalize ꢀ4 million drug molecules per day. This
sites on doxorubicin’s primary intracellular target, DNA, based estimated delivery is below many of the reported single-cell
2
5
on one doxorubicin molecule binding per ten DNA base pairs
IC50 values for doxorubicin and does not take into account the
and approximately three billion base pairs per cell in humans.26 additional processing steps between ADC internalization and
It is important to note that the quantification of intracellular doxorubicin binding to DNA. The single-cell IC50 values mea-
doxorubicin molecules we report here is at the time of analy- sured in this work for doxorubicin support the generally ac-
4
0–43
sis, which is 3 days after ending doxorubicin treatment. During cepted rule that ADCs need a potent drug to be successful.
this incubation time, doxorubicin may efflux from the cell, be This assay provides a method for determining the required
diluted by growth, or be degraded. Doxorubicin fluorescence is amount of drug that must be delivered intracellularly via
known to be quenched once bound to DNA, resulting in a 30– an ADC.
0,27,28
4
0-fold decrease in fluorescence signal.1
It has also been
This flow cytometric clonogenic assay can be generalized to
shown that some doxorubicin degradation products fluoresce study any fluorescent drug or fluorescently tagged drug. Fluo-
2
7,29
in the same emission wavelength window as doxorubicin.
Single-cell IC50 values provide a snapshot of the order of cellular uptake of a drug in assays such as this flow cytomet-
rescent analogs of drugs can be very useful tools for tracking
4
4,45
magnitude of intracellular doxorubicin molecules, but the ric assay and others.
However, fluorescent analogs may be
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015
DOI 10.1002/jps.24631

RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
4415
Figure 4. Effect of verapamil treatment on cell proliferation. HCT-15 cells, which express the drug efflux pump P-glycoprotein (Pgp), were
treated with doxorubicin in the presence (dashed lines) and absence (solid lines) of 20 :M verapamil, a Pgp inhibitor. (a) Flow cytometric
clonogenic assay with cell proliferation as a function of amount of intracellular doxorubicin in cells. Doxorubicin media concentration is given by
different colors as described in legend. (b) Proliferation as a function of the media doxorubicin concentration with (dashed line) or without (solid
line) 20 :M verapamil.
difficult to synthesize, have reduced potency, and have different Fellowship. C.K. was supported by the Pfizer Worldwide Re-
trafficking behavior when compared to the parent drug. Drug search & Development Post-Doctoral Program. This work was
delivery systems that are taken up by cells, such as liposomes, also supported by a research grant from Pfizer and in part by
nanoparticles, and ADCs, can also be used in this clonogenic the Koch Institute Support (core) Grant P30-CA14051 from the
assay with a fluorescent tag on the delivery system as proxy for National Cancer Institute. We thank the Koch Institute Swan-
how much drug was taken up, rather than using a fluorescent son Biotechnology Center for technical support, specifically the
drug or fluorescent analog of a drug.
Flow Cytometry Core.
In conclusion, we have presented here a straightforward flow
cytometric clonogenic assay using a cell tracing dye and a flu-
orescent drug. The first application of the assay revealed the REFERENCES
single-cell potency of doxorubicin via single-cell IC50 values of
1
. Fotakis G, Timbrell Ja. 2006. In vitro cytotoxicity assays: Compari-
4–12 million intracellular doxorubicin molecules for the cancer
son of LDH, neutral red, MTT and protein assay in hepatoma cell lines
following exposure to cadmium chloride. Toxicol Lett 160:171–177.
cell lines evaluated. Single-cell potency is a key parameter for
molecular design of targeting agents where subcellular traffick-
ing and processing can be affected by altered design attributes.
2
. Weyermann J, Lochmann D, Zimmer A. 2005. A practical note on
the use of cytotoxicity assays. Int J Pharm 288:369–376.
. Gottesman MM. 2002. Mechanisms of cancer drug resistance. Annu
3
Rev Med 53:615–627.
ACKNOWLEDGMENTS
4. Pirie CM, Hackel BJ, Rosenblum MG, Wittrup KD. 2011. Convergent
potency of internalized gelonin immunotoxins across varied cell lines,
antigens, and targeting moieties. J Biol Chem 286:4165–4172.
We thank Lindsay King, Nahor Haddish-Berhane, and mem-
bers of the Wittrup Lab for their technical suggestions and
5. Sundman-Engberg B, Tidefelt U, Liliemark J, Paul C. 1990. Intra-
review. K.F.M. was supported by a Hertz Foundation Fellow- cellular concentrations of anti cancer drugs in leukemic cells in vitro
ship and a National Science Foundation Graduate Research vs in vivo. Cancer Chemother 252–256.
DOI 10.1002/jps.24631
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015

4
416
RESEARCH ARTICLE – Pharmacokinetics, Pharmacodynamics and Drug Transport and Metabolism
6
. Eliaz RE. 2004. Determination and modeling of kinetics of cancer 26. A brief guide to genomics. Accessed April 14, 2014, at
cell killing by doxorubicin and doxorubicin encapsulated in targeted <http://www.genome.gov/18016863>.
liposomes. Cancer Res 64:711–718. 27. Fiallo M, Laigle A. 1993. Accumulation of degradation products
. Sumantran VN. 2011. Cellular chemosensitivity assays: An of doxorubicin and pirarubicin formed in cell culture medium within
overview. Methods Mol Biol 731:219–236. sensitive and resistant cells. Biochem Pharmacol 45:659–665.
. Franken N, Rodermond H, Stap J. 2006. Clonogenic assay of cells 28. Hajian R, Shams N, Mohagheghian M. 2009. Study on the inter-
7
8
in vitro. Nat Protoc 1:2315–2319.
action between doxorubicin and Deoxyribonucleic acid with the use of
9
. Minotti G, Menna P, Salvatorelli E, Cairo G, Gianni L. 2004. An- methylene blue as a probe. J Braz Chem Soc 20:1399–1405.
thracyclines: Molecular advances and pharmacologic developments in 29. Hovorka O, Subr V, Vetvicka D, Kovar L, Strohalm J, Strohalm
antitumor activity and cardiotoxicity. Pharmacol Rev 56:185–229. M, Benda A, Hof M, Ulbrich K, Rihova B. 2010. Spectral analysis of
0. Karukstis K, Thompson E. 1998. Deciphering the fluorescence sig- doxorubicin accumulation and the indirect quantification of its DNA
1
nature of daunomycin and doxorubicin. Biophys Chem 73:249–263.
intercalation. Eur J Pharm Biopharm 76:514–524.
11. Saltiel E, McGuire W. 1983. Doxorubicin (adriamycin) 30. Smith Da, Di L, Kerns EH. 2010. The effect of plasma protein bind-
cardiomyopathy—A critical review. West J Med 139:332–341.
ing on in vivo efficacy: Misconceptions in drug discovery. Nat Rev Drug
1
2. B e´ nard J, Da Silva J, De Blois MC, Boyer P, Duvillard P, Chiric Discov 9:929–939.
E, Riou G. 1985. Characterization of a human ovarian adenocarcinoma 31. Law V, Knox C, Djoumbou Y, Jewison T, Guo AC, Liu Y, Maciejew-
line, IGROV1, in tissue culture and in nude mice characterization of a ski A, Arndt D, Wilson M, Neveu V, Tang A, Gabriel G, Ly C, Adamjee
human ovarian adenocarcinoma line, IGROV1, in tissue culture and in S, Dame ZT, Han B, Zhou Y, Wishart D. 2014. DrugBank 4.0: Shedding
nude mice. Cancer Res 45:4970–4979.
3. Chen Q, Gabathuler R. 2004. Efficient synthesis of doxorubicin 32. Zhang F, Xue J, Shao J, Jia L. 2012. Compilation of 222 drugs’
melanotransferrin p97 conjugates through SMCC linker. Synth Com- plasma protein binding data and guidance for study designs. Drug
mun 34:2407–2414. Discov Today 17:475–485.
4. Junutula JR, Raab H, Clark S, Bhakta S, Leipold DD, Weir S, Chen 33. Mateus A, Matsson P, Artursson P. 2013. Rapid measurement of
new light on drug metabolism. Nucleic Acids Res 42:1091–1097.
1
1
Y, Simpson M, Tsai SP, Dennis MS, Lu Y, Meng YG, Ng C, Yang J, Lee intracellular unbound drug concentrations. Mol Pharm 10:2467–2478.
CC, Duenas E, Gorrell J, Katta V, Kim A, McDorman K, Flagella K, 34. Greene RF, Collins JM, Jenkins JF, Speyer JL, Myers CE. 1983.
Venook R, Ross S, Spencer SD,Wong WL, Lowman HB, Vandlen R, Plasma pharmacokinetics of adriamycin and adriamycinol: Implica-
Sliwkowski MX, Scheller RH, Polakis P, Mallet W. 2008. Site-specific tions for the design of in vitro experiments and treatment protocols
conjugation of a cytotoxic drug to an antibody improves the therapeutic plasma pharmacokinetics of adriamycin and adriamycinol: Implica-
index. Nat Biotechnol 26:925–932.
tions for the design of in vitro experiments and treatment. Cancer Res
15. Holliday DL, Speirs V. 2011. Choosing the right cell line for breast 43:3417–3421.
cancer research. Breast Cancer Res 13:215.
35. Anderson AB, Gergen J, Arriaga EA. 2002. Detection of doxorubicin
1
6. Spencer SL, Gaudet S, Albeck JG, Burke JM, Sorger PK. 2009. Non- and metabolites in cell extracts and in single cells by capillary elec-
genetic origins of cell-to-cell variability in TRAIL-induced apoptosis. trophoresis with laser-induced fluorescence detection. J Chromatogr B
Nature 459:428–432. Analyt Technol Biomed Life Sci 769:97–106.
7. Gottesman M, Pastan I. 1993. Biochemistry of multidrug resistance 36. Deng B, Wang Z, Song J, Xiao Y, Chen D, Huang J. 2011. Analysis
mediated by the multidrug transporter. Annu Rev Biochem 62:385– of doxorubicin uptake in single human leukemia K562 cells using capil-
1
4
27.
lary electrophoresis coupled with laser-induced fluorescence detection.
18. Shen F, Chu S, Bence A, Bailey B, Xue X, Erickson PA, Montrose Anal Bioanal Chem 401:2143–2152.
MH, Beck WT, Erickson LC. 2008. Quantitation of doxorubicin uptake, 37. Eichholtz-Wirth H. 1980. Dependence of the cytostatic effect of adri-
efflux, and modulation of multidrug resistance (MDR) in MDR human amycin on drug concenration and exposure time in vitro. Br J Cancer
cancer cells. Pharmacology 324:95–102.
886–891.
19. Lai G-M, Chen Y-N, Mickley La, Fojo AT, Bates SE. 1991. 38. Sakai-Kato K, Saito E, Ishikura K, Kawanishi T. 2010. Anal-
P-glycoprotein expression and schedule dependence of adriamycin cy- ysis of intracellular doxorubicin and its metabolites by ultra-high-
totoxicity in human colon carcinoma cell lines. Int J Cancer 49:696– performance liquid chromatography. J Chromatogr B Analyt Technol
7
2
03.
0. Durand R, Olive P. 1981. Flow cytometry studies of intracellular 39. Xiong G, Chen Y, Arriaga EA. 2005. Measuring the doxorubicin
content of single nuclei by micellar electrokinetic capillary chromatog-
1. El-Kareh AW, Secomb TW. 2005. Two-mechanism peak concentra- raphy with laser-induced fluorescence detection. Anal Chem 77:3488–
tion model for cellular pharmacodynamics of doxorubicin. Neoplasia 3493.
Biomed Life Sci 878:1466–1470.
adriamycin in single cells in vitro. Cancer Res 41:3489–3494.
2
7
:705–713.
40. Adair JR, Howard PW, Hartley Ja, Williams DG, Chester Ka. 2012.
22. Kerr DJ, Kerr aM, Freshney RI, Kaye SB. 1986. Comparative in- Antibody-drug conjugates – a perfect synergy. Expert Opin Biol Ther
tracellular uptake of adriamycin and 4’-deoxydoxorubicin by non-small 12:1191–1206.
cell lung tumor cells in culture and its relationship to cell survival. 41. Alley SC, Okeley NM, Senter PD. 2010. Antibody-drug conjugates:
Biochem Pharmacol 35:2817–2823.
Targeted drug delivery for cancer. Curr Opin Chem Biol 14:529–537.
23. Praet M, Stryckmans P, Ruysschaert J. 1996. Cellular uptake, cy- 42. Carter PJ, Senter PD. 2008. Antibody-drug conjugates for cancer
totoxicity, and transport kinetics of anthracyclines in human sensitive therapy. Cancer J 14:154–169.
and multidrug-resistant K562 cells. Biochem Pharmacol 51:1341–1348. 43. Teicher Ba. 2009. Antibody-drug conjugate targets. Curr Cancer
24. Yu T, Xiong Z, Chen S, Tu G. 2005. The use of models in ‘target’ Drug Targets 9:982–1004.
theory to evaluate the survival curves of human ovarian carcinoma 44. Thurber GM, Yang KS, Reiner T, Kohler RH, Sorger P, Mitchi-
cell line exposure to adriamycin combined with ultrasound. Ultrason son T, Weissleder R. 2013. Single-cell and subcellular pharmacokinetic
Sonochem 12:345–348.
5. Tarasiuk J, Fr e´ zard, F, Garnier-Suillerot A, Gattegno L. 1989. An- 45. Yang J, Chen H, Vlahov IR, Cheng J-X, Low PS. 2006. Evaluation
thracycline incorporation in human lymphocytes. Kinetics of uptake of disulfide reduction during receptor-mediated endocytosis by using
imaging allows insight into drug action in vivo. Nat Commun 4:1504.
2
and nuclear concentration. Biochim Biophys Acta 1013:109–117.
FRET imaging. Proc Natl Acad Sci USA 103:13872–13877.
Maass et al., JOURNAL OF PHARMACEUTICAL SCIENCES 104:4409–4416, 2015
DOI 10.1002/jps.24631