Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006 ARTICLES 1777
Background: If cancer arises and is maintained by a small
population of cancer-initiating cells within every tumor,
understanding how these cells react to cancer treatment will
facilitate improvement of cancer treatment in the future.
Cancer-initiating cells can now be prospectively isolated from
breast cancer cell lines and tumor samples and propagated as
mammospheres in vitro under serum-free conditions. Methods:
CD24 − /low /CD44 + cancer-initiating cells were isolated from
MCF-7 and MDA-MB-231 breast cancer monolayer cultures
and propagated as mammospheres. Their response to radia-
tion was investigated by assaying clonogenic survival and by
measuring reactive oxygen species (ROS) levels, phosphory-
lation of the replacement histone H2AX, CD44 levels, CD24
levels, and Notch-1 activation using fl ow cytometry. All statis-
tical tests were two-sided. Results: Cancer-initiating cells
were more resistant to radiation than cells grown as mono-
layer cultures (MCF-7: monolayer cultures, mean surviving
fraction at 2 Gy [SF 2Gy ] = 0.2, versus mammospheres, mean
SF 2Gy = 0.46, difference = 0.26, 95% confi dence interval [CI] =
0.05 to 0.47; P = .026; MDA-MB-231: monolayer cultures,
mean SF 2Gy = 0.5, versus mammospheres, mean SF 2Gy = 0.69,
difference = 0.19, 95% CI = − 0.07 to 0.45; P = .09). Levels of
ROS increased in both mammospheres and monolayer cul-
tures after irradiation with a single dose of 10 Gy but were
lower in mammospheres than in monolayer cultures (MCF-7
monolayer cultures: 0 Gy, mean = 1.0, versus 10 Gy, mean = 3.32,
difference = 2.32, 95% CI = 0.67 to 3.98; P = .026; mammo-
spheres: 0 Gy, mean = 0.58, versus 10 Gy, mean = 1.46, difference =
0.88, 95% CI = 0.20 to 1.56; P = .031); phosphorylation of
H2AX increased in irradiated monolayer cultures, but no
change was observed in mammospheres. Fractionated doses
of irradiation increased activation of Notch-1 (untreated,
mean = 10.7, versus treated, mean = 15.1, difference = 4.4,
95% CI = 2.7 to 6.1, P = .002) and the percentage of the can-
cer stem/initiating cells in the nonadherent cell population of
MCF-7 monolayer cultures (untreated, mean = 3.52%, versus
treated, mean = 7.5%, difference = 3.98%, 95% CI = 1.67%
to 6.25%, P = .009). Conclusions: Breast cancer – initiating
cells are a relatively radioresistant subpopulation of breast
cancer cells and increase in numbers after short courses of
fractionated irradiation. These fi ndings offer a possible
mechanism for the accelerated repopulation of tumor cells
observed during gaps in radiotherapy. [J Natl Cancer Inst
2006;98: 1777 – 85 ]
One view of cancer is that it may arise from a single cell that
has the ability to self-renew and thus to maintain the growth of a
tumor, whereas the majority of its cellular progeny does not.
There is increasing evidence that such a cell population exists
and that these cells can be prospectively identifi ed in brain tu-
mors ( 1 ) , breast cancer ( 2 ) , prostate cancer ( 3 ) , and melanoma
Affi liation of authors: Department of Radiation Oncology, David Geffen School
of Medicine at the University of California at Los Angeles, Los Angeles, CA.
Correspondence to: Frank Pajonk, MD, PhD, Department of Radiation
Oncology, David Geffen School of Medicine at UCLA, 10833 Le Conte Ave.,
Los Angeles, CA 90095-1714 (e-mail: email@example.com ).
See “ Notes ” following “ References. ”
© The Author 2006. Published by Oxford University Press. All rights reserved.
For Permissions, please e-mail: firstname.lastname@example.org.
The Response of CD24 − /low /CD44 + Breast Cancer – Initiating
Cells to Radiation
Tiffany M. Phillips , William H. McBride , Frank Pajonk
( 4 ) . A considerable effort is going into determining unique prop-
erties of these cells with the assumption that this cell population,
more than any other, will determine the outcome of cancer treat-
ment. Because such cells might be expected to share properties
with adult stem cells in normal tissues, they are often termed can-
cer stem cells ( 5 ) . However, in spite of old ( 6 ) and more recent
( 1 , 2 , 7 – 9 ) evidence that cancer stem cells exist, there is still a
dearth of good phenotypic markers for such cells. In addition,
many of the self-renewing cancer cell populations that are stud-
ied may also contain early progenitor cells that are derived from
cancer stem cells but are also able to initiate and maintain tumor
growth. Therefore, we join others ( 2 ) in preferring to use the term
In breast cancer, a population of CD24 − /low /CD44 + cells has
been isolated that is highly enriched for cancer-initiating cells
( 2 ) . This population is 1000 times more tumorigenic than cell
populations that are depleted of CD24 − /low /CD44 + cells, and in-
jection of as few as 200 cells leads to tumor formation in SCID
mice ( 2 ) . Breast cancer – initiating cells can be established from
patients’ surgical specimens or breast cancer cell lines and can be
propagated in vitro as nonadherent mammospheres ( 7 ) .
Stem cell properties in normal tissues are tightly regulated by
the Wnt, Shh, and Notch signaling pathways ( 10 , 11 ) . In addition,
overexpression of Notch-1 was observed in breast cancer speci-
mens, and the level of expression was associated with prognosis
( 12 ) . Activation of the Notch-1 pathway is initiated by the bind-
ing of Notch-1 ligands, e.g., Jagged-1, to the extracellular do-
main of Notch-1. This binding causes a conformational change in
Notch-1 that allows the protease tumor necrosis factor alpha con-
verting enzyme to cleave the extracellular domain of the mole-
cule. Notch-1 is thereafter processed by γ -secretase – regulated
intramembrane proteolysis, which allows the intracellular do-
main of Notch-1 (Notch-1 ICD) to translocate into the nucleus
where it binds to and activates the transcriptional repressor CBF1.
Activation of Notch-1 signaling leads to increased transcription
of ErbB2 ( 13 ) , cyclin D1 ( 14 ) , CDK2 ( 14 ) , and Notch-4 ( 15 ) .
ErbB2 is related to radiation resistance ( 16 ) , whereas cyclin D1
and CDK2 promote the transition from G1 to S phase of the cell
cycle and thus promote proliferation. Notch-1 signaling promotes
the self-renewal of mammary stem cells ( 17 ) , and there is strong
evidence that Notch-1 is involved in the carcinogenesis of breast
cancer ( 17 ) . In addition, Notch-1 maintains the malignant pheno-
type of Ras-transformed cells ( 15 ) , and overexpression of Notch
induces mammary tumors in mice ( 18 ) .
at :: on November 13, 2012
1778 ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006
Radiation therapy (RT) is an integral part of the multimodal
treatment concept for breast cancer. Its success depends on the
complete elimination of all cancer stem cells. Radiation oncolo-
gists have been advocating the existence of stem cells in normal
tissues and cancers for decades ( 6 ) . Accelerated repopulation —
the increase in the rate of growth as a result of time between
treatments ( 19 , 20 ) — is a cancer stem cell – related phenomenon
that occurs during fractionated RT. During accelerated repopula-
tion, each day of a treatment gap decreases the effi cacy of RT by
approximately 0.6 Gy, making it one of the major reasons for lo-
cal failure of RT. Accelerated repopulation was fi rst described for
head and neck epithelial tumors ( 21 ) , but it also occurs in breast
cancer even though it may be diffi cult to detect ( 22 – 24 ) .
In this study, we investigated the radiation response of CD24 − /low /
CD44 + breast cancer – initiating cells, the population of cancer cells
that are likely to be critical for success or failure of cancer therapy.
We characterized the radiation sensitivity of these cells and the size
of this cell population after clinical fractions of radiation and ex-
plored possible mechanisms for the failure of radiotherapy.
MCF-7 and MDA-MB-231 breast cancer cells (American
Type Culture Collection; Manassas, VA) were cultured in log-
growth phase in modifi ed Eagle medium (MEM) (supplemented
with 0.1 mM nonessential amino acids and 1 mM sodium pyru-
vate; Cellgro, Kansas City, MO) and Dulbecco’s modifi ed Eagle
medium (DMEM) (Cellgro), respectively, supplemented with
10% heat-inactivated fetal calf serum (FCS) and 0.01 mg/mL
bovine insulin (Sigma, St Louis, MO) at 37 °C in a humidifi ed
atmosphere (5% CO 2 ). To obtain cancer-initiating cells and to
propagate them as mammospheres, cells fl oating in the superna-
tant of 2-day-old cultures were collected by centrifugation for
5 minutes at 500 g , washed in Hanks’ buffered salt solution, and
resuspended in phenol red – free DMEM – F12 (Cellgro) supple-
mented with 0.4% bovine serum albumin (BSA, Sigma),
5 μ g/mL bovine insulin (Sigma), 20 ng/mL basic fi broblast
growth factor 2 (bFGF, Sigma), and 10 ng/mL epidermal growth
factor (EGF, Sigma) at a density of 1000 cells/mL. Growth fac-
tors were added to the mammosphere cultures every 3 days. To
mimic mammosphere culture conditions in cells grown as mono-
layer cultures, cells were plated in MEM or DMEM media con-
taining 10% FCS supplemented with 5 μ g/mL bovine insulin, 20
ng/mL bFGF, and 10 ng/mL EGF.
For clonogenic assays, cells derived from monolayer cultures
or 5-day-old mammospheres were enzymatically dissociated
with trypsin – EDTA (monolayer cultures) or mechanically disso-
ciated with a Pasteur pipette (mammospheres), both passed
through a 40- μ m sieve, and immediately irradiated (10 6 cells/
mL) at room temperature with a 137 Cs laboratory irradiator (Mark I,
JL Shephard, San Fernando, CA) at a dose rate of 4.95 Gy/minute
for the time required to generate a dose curve of 0, 2, 4, 6, and
8 Gy. Corresponding controls were sham irradiated. Colony-
forming assays were performed immediately after irradiation by
plating cells into triplicate 100-mm culture dishes. After 28 days,
cells were fi xed with 75% ethanol and stained with 1% crystal
violet, and colonies containing more than 50 cells were counted.
To generate a radiation survival curve, the surviving fraction at
each radiation dose was normalized to that of the sham-irradiated
control, and curves were fi tted using a linear – quadratic model
( surviving fraction = e ( − α dose − β dose 2 ) , in which α is the number of
logs of cells killed per gray from the linear portion of the survival
curve and β is the number of logs of cells killed per [gray] 2 from
the quadratic component) ( 25 ) . Three independent experiments
To evaluate H2AX phosphorylation, single-cell suspensions
were irradiated as above with 0, 2, or 10 Gy. Cells were harvested
by centrifugation (500 g for 5 minutes at 4 °C) at 5 and 60 min-
utes after irradiation.
To measure reactive oxygen species (ROS) accumulation,
100 000 cells were treated with 0, 2, or 10 Gy. Cells were imme-
diately analyzed as described below.
To measure Notch-1 activation and Jagged-1 expression, cells
were treated with single and fractionated doses of radiation. Cells
(400 000 per dish) were plated onto 100-mm tissue culture dishes
and allowed to grow for 24 hours. Cultures were then irradiated
as monolayers at room temperature with 3 Gy daily for 5
consecutive days (days 2 – 6) or with a single dose of 10 Gy on
day 6. Control cells were sham irradiated. Nonadherent and ad-
herent cells were harvested 48 hours after the last irradiation
(on day 8).
For primary mammosphere formation assays, cells were
irradiated with 3 Gy daily for 5 consecutive days (days 2 – 6) or
with a single dose of 10 Gy on day 6. Control cells were sham
3-( 4 , 5 -Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium
Bromide Assays to Measure Cell Proliferation
For 3-( 4 , 5 -dimethylthiazol-2-yl)-2,5-diphenyltetrazolium
bromide (MTT) assays, MCF-7 cells in monolayer culture were
irradiated; incubated for indicated times in MEM media supple-
mented with 10% FCS, 20 ng/mL bFGF, and 10 ng/mL EGF;
washed twice with PBS; incubated with trypsin – EDTA; resus-
pended in MEM (containing 10% FCS); counted; and plated in
100 μ L MEM (10% FCS) at 2000, 10 000, 15 000, and 20 000
cells per well into 96-well plates. After 7 days, 20 μ L of MTT
solution (5 mg/mL in PBS) was added to each well, and cells
were incubated for 4 hours at 37 °C. Then 50 μ L sodium dodecyl
sulfate solution (20% sodium dodecyl sulfate, 0.01% HCl) was
added to each well, and plates were incubated at 37 °C overnight.
Absorbance was measured at 560 nm in a fl uorescence plate
reader (Spectrafl uor, Tecan, San Jose, CA).
Flow Cytometry to Measure CD24, CD44, and
Jagged-1 Expression; Notch-1 Activation; and H2AX
CD24 and CD44 expression was analyzed in cells derived
from monolayer cultures or in 5-day-old primary mammo-
spheres following incubation in trypsin – EDTA or dissociation
with a Pasteur pipette and passage through a 40- μ m sieve. At
least 10 5 cells were pelleted by centrifugation at 500 g for 5 min-
utes at 4 °C, resuspended in 10 μ L of monoclonal mouse anti-
human CD24 – fl uorescein isothiocyanate (FITC) antibody (BD
Pharmingen, San Jose, CA) and a monoclonal mouse anti-
human CD44 – phytoerythrin (PE) antibody (BD Pharmingen),
at :: on November 13, 2012
Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006 ARTICLES 1779
and incubated for 20 minutes at 4 °C. Ten independent experi-
ments were performed.
To measure Jagged-1 expression and Notch-1 activation, cells
were permeabilized with 4% formaldehyde and pelleted by
centrifugation as above. Cells were then incubated with 0.25 μ g
of PE/Cy5-conjugated monoclonal mouse anti-human CD44
antibody, 10 μ L of monoclonal mouse anti-human CD24 – FITC
antibody, and 200 μ L of either polyclonal rabbit anti-human
Jagged-1 antibody (Santa Cruz Biotechnology, Santa Cruz, CA)
or polyclonal rabbit anti-human Notch-1-ICD antibody (Cell Sig-
naling, Danvers, MA) that had been diluted 1 : 200 in PBS con-
taining 2% BSA for 20 minutes at 4 °C. Cells were then washed
with PBS/4% BSA and incubated with a secondary, PE-con-
jugated polyclonal goat anti-rabbit antibody (BD Pharmingen).
For analysis of H2AX phosphorylation, cells were centrifuged
for 5 minutes at 500 g and resuspended in 0.3 mL of PBS. To fi x
the cells, 0.7 mL of ethanol (99%) was added to the tube while
vortexing, and samples were stored for 30 minutes at − 20 °C.
Cold Tris-buffered saline (TBS, pH 7.4, 1 mL) was added, and
cells were pelleted by centrifugation at 500 g and resuspended
in 1 mL cold TST (TBS containing 4% FBS and 0.1% Triton
X-100) for 10 minutes to permeabilize and rehydrate the cells.
Cells were pelleted again and resuspended in 200 μ L of monoclo-
nal mouse anti- γ H2AX – FITC antibody (Upstate, Charlottesville,
VA) diluted 1 : 500 in TST, incubated on a shaker platform for
2 hours at room temperature, and washed twice in TST. Three
independent experiments were performed.
Flow cytometry and cell sorting were performed on a
FACScalibur fl ow cytometer (Becton Dickinson, Franklin Lakes,
NJ). The CellQuest (Becton Dickinson) software package was
Reactive Oxygen Species Formation Assay
Cells derived from monolayer cultures or 5-day-old mammo-
spheres were incubated with trypsin – EDTA or dissociated me-
chanically using a Pasteur pipette, respectively, resuspended in
modifi ed HBSS (10 mM HEPES, 1 mM MgCl 2 , 2 mM CaCl 2 ,
2.7 mM glucose), passed through a 40- μ m sieve, counted, and
diluted to a fi nal concentration of 10 6 cells/mL in 15-mL Falcon
tubes (Becton Dickinson). Aminophenyl fl uorescein (Cell Tech-
nology, Mountain View, CA) was added to a fi nal concentration
of 10 μ M, and cells were incubated for 30 minutes in the dark
and irradiated as indicated above. A total of 100 000 cells per
well were plated into black 96-well plates, and fl uorescence was
measured in a fl uorescence plate reader (Spectrafl uor, Tecan; ex-
citation: 480 nm, emission: 520 nm). Fluorescence was normal-
ized to the fl uorescence readings of untreated monolayer culture
cells. Three independent experiments were performed, each in
Primary Mammosphere Formation Assay
The ability of cells in the nonadherent population of mono-
layer cultures to initiate mammosphere formation after irradia-
tion was assessed by harvesting, washing, and resuspending
nonadherent cells in phenol red – free DMEM – F12 medium
(supplemented with 0.4% BSA, 20 ng/mL bFGF, and 10 ng/mL
EGF). Cells were then passed through a 40- μ m sieve, counted,
diluted, and plated into 96-well plates at clonal densities. Mam-
mospheres were counted on day 5.
All data are represented as means and differences of the means
with 95% confi dence intervals (CIs). P values of .05 or less, cal-
culated using a paired two-sided Student’s t test, were considered
to indicate statistically signifi cant differences.
Response of CD24 − /low /CD44 + Breast Cancer – Initiating
Cells to a Single Dose of Radiation
We established nonadherent mammosphere cultures from both
MCF-7 and MDA-MB-231 breast cancer cells and analyzed the
percentage of CD24 − /low /CD44 + cells on day 5 by fl ow cytome-
try. In general, by day 5, MCF-7 ( Fig. 1, A ) and MDA-MB-231
(data not shown) mammospheres showed dramatically elevated
percentages of CD24 − /low /CD44 + cells.
The responses of cells from monolayers and CD24 − /low /
CD44 + -enriched mammospheres (day 5) to radiation were com-
pared by clonogenic assay. The plating effi ciencies of MCF-7
cells derived from monolayer cultures and mammospheres with-
out irradiation were similar (mean = 7.2%, 95% CI = 0.73 to
13.7, and mean = 11%, 95% CI = 8.8 to 13.2, respectively). How-
ever, cells derived from MCF-7 mammospheres were more ra-
dioresistant than cells derived from monolayer cultures
(monolayer-derived cells: α = 0.79, β = 0.011, mean surviving
fraction at 2 Gy [SF 2Gy ] = 0.2, versus mammospheres: α = 0.30,
β = 0.044, mean SF 2Gy = 0.46, difference = 0.26, 95% CI = 0.05
to 0.47; P = .026, n = 9; Fig. 1, B ). Comparable results were
found for mammospheres that were derived from MDA-MB-231
cells (monolayer-derived cells: α = 0.65, β = 0.0014, mean
SF 2Gy = 0.5, versus mammospheres: α = 0.31, β = 0.035, mean
SF 2Gy = 0.69, difference = 0.19, 95% CI = − 0.07 to 0.45; P = .09,
n = 6).
One hallmark of the recognition and repair of double-strand
DNA breaks is phosphorylation of the replacement histone
H2AX ( 26 ) . Single-cell suspensions from MCF-7 mammo-
sphere and monolayer cultures were irradiated with 0, 2, or 10
Gy, and H2AX phosphorylation ( γ H2AX) was measured by
fl ow cytometry at 5 and 60 minutes after irradiation (n = 2).
MCF-7 cells derived from monolayer cultures showed a time-
dependent increase of γ H2AX after irradiation, whereas cells
derived from primary mammospheres showed little change in
γ H2AX ( Fig. 1, C ). At 60 minutes, the increase in γ H2AX for
cells derived from monolayer cultures was dose dependent, with
10 Gy being more effective than 2 Gy (monolayer cultures: rela-
tive to control, 2 Gy, mean = 1.63-fold, not statistically signifi -
cant, 10 Gy, mean = 3.2-fold, difference = 2.2, 95% CI = 1.46 to
2.92, P = .006, n = 3; mammospheres: relative to 0 Gy, 2 Gy,
mean = 1.06-fold, not statistically signifi cant, 10 Gy, mean =
1.13-fold, not statistically signifi cant, n = 3); however, even the
10-Gy dose did not affect the phosphorylation of H2AX in mam-
mospheres ( Fig. 1, D ).
The lack of γ H2AX staining, and hence repair of DNA double-
strand breaks in mammospheres, after irradiation could be due to
a very rapid repair, failure of detection, or initially low induction
of DNA double-strand breaks. Therefore, we next investigated
whether irradiation induced the formation of ROS in MCF-7
monolayer cultures and mammospheres. Single-cell suspensions
were irradiated with 0, 2, or 10 Gy. Cells derived from MCF-7
at :: on November 13, 2012
1780 ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006
monolayer cultures consistently showed dose-dependent forma-
tion of ROS (0 Gy, mean = 1.0, 2 Gy, mean = 1.45, difference =
0.45, not statistically signifi cant; 10 Gy, mean = 3.32, difference =
2.32, 95% CI = 0.67 to 3.98; P = .026, n = 3; Fig. 1, E ). Cells de-
rived from primary mammospheres did so as well, but at levels
that were approximately 50% of those formed by cells from mono-
layer cultures (0 Gy, mean = 0.58, 2 Gy, mean = 0.647, difference =
0.067, not statistically signifi cant; 10 Gy, mean = 1.46, difference =
0.88, 95% CI = 0.20 to 1.56; P = .031, n = 3; Fig. 1, E ).
Because EGF and bFGF, which were used to generate mam-
mospheres, both decrease the radiation sensitivity of cells
( 27 , 28 ) , independent of the presence of CD24 − /low /CD44 + can-
cer- initiating cells, MCF-7 and MDA-MB-231 cells were cul-
tured as monolayers for 5 days in MEM or DMEM (10% FCS),
supplemented with EGF and bFGF. As expected, cells derived
from EGF- and bFGF-treated monolayer cultures exhibited in-
creased radiation resistance that was similar to that of cells de-
rived from primary mammospheres ( Fig. 2, A and B ), which
could be ascribed to the increased percentage of CD24 − /low /
CD44 + cancer-initiating cells in the nonadherent fraction (i.e.,
supernatant) of monolayer cultures after cytokine treatment ( Fig.
2, C ). To further exclude an acute direct radioprotective effect of
EGF and bFGF, we exposed MCF-7 monolayer cultures to
growth medium supplemented with or without EGF and bFGF
Fig. 1. Radiation response of cells in monolayer
culture and mammospheres. A ) Fluorescence-
activated cell-sorting (FACS) analysis to measure
CD44 and CD24 expression of cells derived from
MCF-7 monolayer cultures ( left ) or primary
mammospheres ( right ). Cells were incubated
with trypsin – EDTA (monolayer cultures) or
dissociated mechanically (mammospheres) and
incubated with a fl uorescein isothiocyanate
(FITC) – conjugated monoclonal mouse anti-
human CD24 and a phytoerythrin-conjugated
monoclonal mouse anti-human CD44 antibody.
After 20 minutes, cells were washed with
phosphate-buffered saline and analyzed on a fl ow
cytometer. Data from one of 10 experiments are
shown. B ) Clonogenic survival assay of cells
derived from MCF-7 monolayer cultures (MCF-7)
or 5-day-old mammospheres (MCF-7S). Cells
were irradiated as single-cell suspensions and
plated to allow colony formation. After 28
days, cells were fi xed and stained with crystal
violet, and colonies consisting of more than
50 cells were counted for each dose point. To
determine surviving fractions, counts were
normalized using the plating effi ciency of the
unirradiated corresponding control. Means and
95% confi dence intervals are shown for three
experiments, each performed in triplicate (n =
9). P = .026 difference at 2-Gy dose using a
paired two-sided Student’s t test. C ) H2AX
phosphorylation ( γ H2AX) was measured in
MCF-7 cells derived from monolayer cultures
or mammospheres using fl ow cytometry and a
specifi c FITC-conjugated anti- γ H2AX antibody.
Single-cell suspensions were exposed to 0 Gy
and harvested immediately ( fi lled histogram ) or
to 2 Gy and harvested 5 minutes ( solid line ) or 60
minutes ( dashed line ) after exposure. Data from
one of two independent experiments are shown.
D ) γ H2AX levels in MCF-7 monolayer cultures
( black ) or mammospheres ( gray ) 60 minutes
after exposure to 0, 2, or 10 Gy. Means and
95% confi dence intervals of three independent
experiments (n = 3) are shown. Relative
fl uorescence was calculated by normalizing
all data to the fl uorescence of corresponding
unirradiated control cells. P values comparing
data from experimental and corresponding
control samples were determined using a paired
two-sided Student’s t test. n.s. = not statistically
signifi cant. E ) Free radical formation in MCF-7
cells derived from monolayer cultures ( black )
and mammospheres ( gray ) after exposure to 0,
2, or 10 Gy as measured by fl ow cytometry. Cells
were incubated with aminophenyl fl uorescein at
10 μ M concentration for 30 minutes to allow the
fl uorogenic dye entering into the cells. Cells were then irradiated, and fl uorescence was immediately assessed in a fl uorescence plate reader (excitation: 480 nm,
emission: 520 nm). Relative fl uorescence was calculated by normalizing all data to the fl uorescence of unirradiated control cells. Cells were irradiated as single-
cell suspensions. Means and 95% confi dence intervals of three independent experiments performed in triplicate (n = 9) are shown. P values comparing data from
experimental and corresponding control samples were determined using a paired two-sided Student’s t test. n.s. = not statistically signifi cant.
at :: on November 13, 2012
Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006 ARTICLES 1781
for 0.5, 1, 2, 3, 4, or 6 hours after irradiation with 4 or 6 Gy (n =
2). Cells were then washed twice with PBS, incubated with tryp-
sin – EDTA and plated into 96-well plates at 2000, 10 000, 15 000,
or 20 000 cells per well. Although cytokine treatment increased
the viability of the cells as assessed by MTT assays on day 7, we
could not detect a radioprotective effect of the EGF/bFGF treat-
ment (data not shown).
Response of CD24 − /low /CD44 + Breast Cancer Cells to
To determine whether the CD24 − /low /CD44 + -enriched cancer-
initiating cells were truly more radioresistant than their non –
CD24 − /low /CD44 + -enriched monolayer cell counterparts, or if
they might even increase in numbers after clinical fractions of
radiation (accelerated repopulation), monolayer cultures of MCF-7
cells were irradiated with either a single dose of 10 Gy on day
6 or fi ve daily doses of 3 Gy on days 2 – 6. When unirradiated
cells were analyzed for CD24 and CD44 expression on day 8, the
size of the CD24 − /low /CD44 + population of cells in the nonadher-
ent fraction was, as expected, higher than that of the adherent
cells (mean = 3.52% versus mean = 0.86%, difference = 2.66%,
95% CI = 0.63 to 4.7; n = 5, P = .02; Fig. 3, A and B ). For mono-
layer cultures, the percentages of CD24 − /low /CD44 + cells on day
8 in the adherent and nonadherent cell populations were not al-
tered by a single dose of 10 Gy given on day 6. In addition, after
fi ve fractions of 3 Gy, the percentage of CD24 − /low /CD44 + cells
in the adherent cell population did not change ( Fig. 3, A and B );
however, the percentage of CD24 − /low /CD44 + cells in the super-
natant (nonadherent cells) increased (untreated, mean = 3.52%,
versus treated, mean = 7.5%, difference = 3.98%, 95% CI =
1.67% to 6.25%; n = 5, P = .009; Fig. 3, A and B ).
To further explore the biologic relevance of the increase of
the proportion of CD24 − /low /CD44 + cells after fractionated
Fig. 2. Effects of epidermal growth factor (EGF)/
basic fi broblast growth factor (bFGF) treatment
on breast cancer cells. Clonogenic survival
assays using A ) MCF-7 and B ) MDA-MB-231
monolayer cultures treated with EGF (10 ng/mL)
and bFGF (20 ng/mL) for 5 days. Cells were
irradiated as single-cell suspensions and plated
to allow colony formation. After 28 days, cells
were fi xed and stained, and colonies consisting
of more than 50 cells were counted for each dose
point. To determine surviving fractions, counts
were normalized using the plating effi ciency
of the unirradiated corresponding control.
Means and 95% confi dence intervals from three
independent experiments are shown (n = 3).
C ) Fluorescence-activated cell-sorting (FACS)
analysis of MCF-7 monolayer cultures treated
with EGF and bFGF for 5 days. Cells were
incubated with trypsin – EDTA and incubated
with a fl uorescein isothiocyanate – conjugated
monoclonal mouse anti-human CD24 and a
phytoerythrin-conjugated monoclonal mouse
anti-human CD44 antibody. After 20 minutes,
cells were washed with phosphate-buffered
saline and analyzed on a fl ow cytometer. Data
from one of six experiments are shown.
at :: on November 13, 2012
1782 ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006
Fig. 3. Effect of fractionated irradiation on
breast cancer cells. A ) Fluorescence-activated
cell-sorting (FACS) analysis was used to
measure percentages of cells with CD44 + /
CD24 − /low in adherent ( black ) and nonadherent
(fl oating, gray ) monolayer MCF-7 cell
cultures. Cells were treated with a single dose
of 10 Gy or with fi ve daily doses of 3 Gy. Cells
were incubated with trypsin – EDTA (adherent
cells) or harvested (fl oating cells) 48 hours
after the last irradiation followed by incubation
with a fl uorescein isothiocyanate – conjugated
monoclonal mouse anti-human CD24 and
a phytoerythrin-conjugated monoclonal mouse
anti-human CD44 antibody. After 20 minutes,
cells were washed with phosphate-buffered
saline and analyzed on a fl ow cytometer. Means
and 95% confi dence intervals are shown from
four (10 Gy) and fi ve (5 × 3 Gy) independent
experiments. P values were determined using
the two-sided Student’s t test. B ) Data from
one experiment of each group of cells in ( A ).
C ) Primary mammosphere formation assay
of nonadherent MCF-7 cells that were plated
at increasing dilutions and treated with 0 Gy,
a single dose of 10 Gy, or fi ve daily doses of
3 Gy. Cells were plated at a starting density of
256 cells per well into 96-well plates in serum-
free Dulbecco’s modifi ed Eagle medium – F12
media supplemented with bovine serum
albumin, epidermal growth factor, and basic
fi broblast growth factor and subsequently
diluted (1 : 1). Mammosphere counts refl ect the
total number of spheres in eight wells for each
dilution. Mammospheres were counted on day
5. Means and 95% confi dence intervals of three
independent experiments are shown.
irradiation, we performed a primary mammosphere formation as-
say, which allows estimation of the number of breast cancer cells
that exhibit self-renewal capacity ( 7 ) . Primary mammosphere
formation by nonadherent cells from cultures irradiated with a
single dose of 10 Gy was similar to that of unirradiated control
cultures. ( Fig. 3, C ). However, primary mammosphere formation
capacity was increased in nonadherent populations from cultures
that received fi ve fractions of 3 Gy, although this increase did not
reach statistical signifi cance ( Fig. 3, C ).
Fractionated Irradiation and the Notch-1 Pathway
Self-renewal and lineage differentiation in normal mam-
mary stem cells is regulated by the developmental Notch signal
at :: on November 13, 2012
Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006 ARTICLES 1783
transduction pathway ( 29 ) , which may also be involved in can-
cer stem cell self-renewal. As measured by fl ow cytometric
analysis, adherent MCF-7 cells expressed detectable levels of
Jagged-1 that increased after fractionated irradiation (5 × 3 Gy)
on day 8 (untreated, mean = 14.2, versus treated, mean = 35.3,
difference = 21.1, 95% CI = 1.23 to 41.0; n = 7, P = .04). Jag-
ged-1 expression in nonadherent cells was slightly greater than
that of adherent cells but did not statistically signifi cantly
change ( Fig. 4, A ). No increase in Jagged-1 expression was
observed in either cell population after a single 10-Gy dose.
Activated Notch-1, as measured by levels of intracellular
Notch-1-ICD, increased after radiation in both adherent and
nonadherent populations after fi ve fractions of 3 Gy (adherent
cells, mean = 10.0 versus mean = 18.5, difference = 8.48, 95%
CI = 3.0 to 13.9; n = 7, P = .009, and nonadherent cells, mean =
10.7 versus mean = 15.1, difference = 4.4, 95% CI = 2.7 to 6.1;
n = 5, P = .002) but not after a single dose of 10 Gy ( Fig. 4, B ).
The increase in Jagged-1 expression and Notch activation
(Notch-ICD) was not observed in the population of CD24 − /low /
CD44 + cells. However, the percentage of CD24 − /low /CD44 +
cells was low in the supernatant and even lower in the adherent
population. Radiation increased the number of cells in this
population by twofold in the supernatant, but the fraction was
still less than 10% in both populations. Therefore, the histo-
grams for Jagged-1 and Notch-ICD of the gated populations
were statistically unreliable.
Using previously published techniques ( 2 , 7 ) and MCF-7 and
MDA-MB-231 breast cancer cell lines, we isolated and propa-
gated populations of cells that contain potential breast cancer
stem cells and are tumor initiating ( 2 , 7 ) . We investigated the re-
sponse of this cell population to both a single dose and a 5-day
course of radiation and found that CD24 − /low /CD44 + -enriched
cancer-initiating cells were more resistant to radiation than cells
in monolayer culture. During a fractionated course of radiation,
the number of breast cancer – initiating cells increased. This in-
crease was accompanied by radiation-induced Jagged-1 expres-
sion and subsequent activation of Notch-1, suggesting that
radiation activates this developmental pathway.
We observed that the plating effi ciency of cells that were de-
rived from mammospheres was similar to that of cells derived
from monolayer cultures and was much higher than the frequency
of primary mammosphere formation. One possible interpretation
is that mammospheres are not exclusively formed by breast can-
cer stem cells but also contain early progenitor cells. Such a situ-
ation would support the use of the term breast cancer – initiating
cells for this population rather than breast cancer stem cells. Al-
ternatively, breast cancer stem cells could be the only population
capable of forming mammospheres but could also give rise to
non – stem cells within the mammosphere. Although non – stem
cells would be incapable of mammosphere formation and thus
self-renewal, they would still be considered as clonogenic in clo-
nogenic survival assays.
We found that breast cancer – initiating cells were more radio-
resistant than non – breast cancer initiating cells. Interestingly, the
radiation survival curve of cells derived from mammospheres
had a shoulder that is characterized by a comparably higher resis-
tance at lower and thus clinically more relevant doses of radia-
tion. This shoulder indicates an enhanced capacity to repair
potentially lethal damage. Resistance to apoptotic stimuli, in-
cluding radiotherapy, was recently reported for nonproliferating
CD34 + chronic myeloid leukemia progenitor cells when com-
pared with normal CD34 + cells ( 30 ) .
Consistent with the increased radioresistance, treatment
with ionizing radiation caused lower levels of ROS in cells
derived from mammospheres compared with cells derived
from monolayer cultures. This decrease in ROS levels indi-
cated high intracellular levels of radical scavengers. Although
cells in primary mammospheres may exhibit certain levels of
hypoxia, as reported for spheroid cultures of tumor cells ( 31 ) ,
the absence of oxygen did not account for the observed effect
in our experiments because cells derived both from monolayer
cultures and mammospheres were irradiated as single-cell
To our knowledge, this is the fi rst study to directly investigate
the radiation resistance of breast cancer – initiating cells. Three-
dimensional in vitro culturing techniques for tumor cells have
been used previously using nonselective serum-containing con-
ditions that caused cells to aggregate. These tumor spheroids also
exhibit resistance to radiation ( 32 ) and chemotherapeutic drugs
Fig. 4. Flow cytometry analysis of Jagged-1 expression and Notch-1 activation
in adherent and nonadherent MCF-7 cells after irradiation. Monolayer cultures
were exposed to fi ve fractions of 5 Gy on days 2 – 6 or a single fraction of 10 Gy
on day 6. Nonadherent (fl oating) cells were harvested 48 hours later, and adherent
cells were incubated with trypsin – EDTA. Cells were fi xed and stained with a
fl uorescein isothiocyanate – conjugated monoclonal mouse anti-human CD24, a
phytoerythrin/Cy5-conjugated monoclonal mouse anti-human CD44 antibody,
and either a polyclonal rabbit anti-human Jagged-1 antibody ( A ) or a polyclonal
rabbit anti-human Notch-1-intracellular domain antibody ( B ). Means and 95%
confi dence intervals are shown from seven (adherent cells 0 and 5 × 3 Gy), fi ve
(fl oating cells 0 and 5 × 3 Gy), or three (1 × 10 Gy) independent experiments.
P values were determined using the two-sided Student’s t test.
at :: on November 13, 2012
1784 ARTICLES Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006
and have enhanced colony-forming effi ciency ( 33 ) . The mecha-
nisms leading to increased resistance and enhanced colony-
forming effi ciency were incompletely understood. However,
given the size of these aggregates and the gradient of cytokine
concentrations and nutrients from the periphery to the center of
these spheroids, the previous techniques may have also selected
for therapy-resistant cancer stem cells.
Our observation of increased radiation resistance in the
cancer-initiating cell population most likely underestimates the
resistance of breast cancer stem cells because the cycling popula-
tion of progenitor cells present in the mammospheres are not
necessarily as resistant as breast cancer stem cells. However,
these results indicate that therapies that specifi cally target path-
ways that are deregulated in breast cancer stem cells may en-
hance the effi ciency of radiotherapy in the future. In addition,
these fi ndings may have an impact on future design of predictive
assays for drug or radiation sensitivity because the therapeutic
response of cancer stem cells may not be refl ected by the re-
sponse of an unselected tumor cell population.
Using an in vitro system, we mimicked a week of clinical
fractionated radiotherapy followed by a typical weekend gap of 2
days. This treatment increased the proportion of breast cancer –
initiating cells. The observation that the increase in cell number
was observed in the supernatant (i.e., nonadherent cells) of con-
fl uent monolayer cultures but not in adherent cells and the fact
that these cells formed primary mammospheres at a higher rate
than untreated cells indicate that the increase in the proportion of
cancer-initiating cells was not caused by simple selection of a
radioresistant subpopulation but by an absolute increase in the
number of viable breast cancer – initiating cells with increased ca-
pacity for self-renewal.
Activation of the developmental Notch-1 signal transduction
pathway promotes self-renewal of early progenitor cells derived
from normal mammary stem cells ( 29 ) . In the present study, we
investigated whether ionizing radiation interfered with the Notch-
1 signaling pathway directly. Our observation that radiation in-
duced Jagged-1 expression on the surface of cells from monolayer
cultures and activated Notch-1 in cells in the supernatant is the
fi rst demonstration, to our knowledge, of an acute radiation effect
on this developmental signaling pathway. Future studies will be
necessary to defi ne the population of cells in which this pathway
is targeted by radiation.
The study has several limitations. Breast cancer stem cells
have not yet been identifi ed directly, although they can be enriched
for and propagated in vitro with the techniques we used in this
study. Still, thes e enriched populations are heterogeneous, and
the results of our study may therefore actually underestimate the
differences between breast cancer stem cells and non – stem cells.
The number of different breast cancer cell lines used limits the
conclusions for clinical radiotherapy that can be drawn from our
study. Selection for a specifi c phenotype may have occurred dur-
ing the establishment and maintenance of these lines, and thus,
these cells may not accurately refl ect the behavior of breast can-
cer cells in human tumors. Thus, our data need validation on
mammospheres derived directly from patients’ tumor specimens.
Taken together, our data indicate that breast cancer – initiating
cells exhibit increased radiation resistance resulting from de-
creased ROS induction, followed by decreased double-strand
break formation. Additionally, fractionated irradiation appeared
to activate the Notch-1 developmental pathway, which may have
caused the numbers of breast cancer – initiating cells to increase,
offering a mechanism for accelerated repopulation during radia-
tion therapy treatment gaps.
(1) Singh SK, Hawkins C, Clarke ID, Squire JA, Bayani J, Hide T, et al.
Identifi cation of human brain tumour initiating cells. Nature 2004 ; 432 :
396 – 401.
(2) Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF.
Prospective identifi cation of tumorigenic breast cancer cells. Proc Natl
Acad Sci U S A 2003 ; 100 : 3983 – 8.
(3) Collins AT, Berry PA, Hyde C, Stower MJ, Maitland NJ. Prospective
identifi cation of tumorigenic prostate cancer stem cells. Cancer Res
2005 ; 65 : 10946 – 51.
(4) Fang D, Nguyen TK, Leishear K, Finko R, Kulp AN, Hotz S, et al. A tumori-
genic subpopulation with stem cell properties in melanomas. Cancer Res
2005 ; 65 : 9328 – 37.
(5) Reya T, Morrison SJ, Clarke MF, Weissman IL. Stem cells, cancer, and can-
cer stem cells. Nature 2001 ; 414 : 105 – 11.
(6) Lange CS, Gilbert CW. Studies on the cellular basis of radiation lethality.
3. The measurement of stem-cell repopulation probability. Int J Radiat Biol
Relat Stud Phys Chem Med 1968 ; 14 : 373 – 88.
(7) Ponti D, Costa A, Zaffaroni N, Pratesi G, Petrangolini G, Coradini D, et al.
Isolation and in vitro propagation of tumorigenic breast cancer cells with
stem/progenitor cell properties. Cancer Res 2005 ; 65 : 5506 – 11.
(8) Liu S, Dontu G, Mantle ID, Patel S, Ahn NS, Jackson KW, et al. Hedgehog
signaling and Bmi-1 regulate self-renewal of normal and malignant human
mammary stem cells. Cancer Res 2006 ; 66 : 6063 – 71.
(9) Singh SK, Clarke ID, Terasaki M, Bonn VE, Hawkins C, Squire J, et al.
Identifi cation of a cancer stem cell in human brain tumors. Cancer Res
2003 ; 63 : 5821 – 8.
(10) Beachy PA, Karhadkar SS, Berman DM. Tissue repair and stem cell renewal
in carcinogenesis. Nature 2004 ; 432 : 324 – 31.
(11) Taipale J, Beachy PA. The Hedgehog and Wnt signalling pathways in
cancer. Nature 2001 ; 411 : 349 – 54.
(12) Parr C, Watkins G, Jiang WG. The possible correlation of Notch-1 and
Notch-2 with clinical outcome and tumour clinicopathological parameters
in human breast cancer. Int J Mol Med 2004 ; 14 : 779 – 86.
(13) Chen Y, Fischer WH, Gill GN. Regulation of the ERBB-2 promoter by
RBPJkappa and NOTCH. J Biol Chem 1997 ; 272 : 14110 – 4.
(14) Ronchini C, Capobianco AJ. Induction of cyclin D1 transcription and CDK2
activity by Notch(ic): implication for cell cycle disruption in transformation
by Notch(ic). Mol Cell Biol 2001 ; 21 : 5925 – 34.
(15) Weijzen S, Rizzo P, Braid M, Vaishnav R, Jonkheer SM, Zlobin A, et al. Ac-
tivation of Notch-1 signaling maintains the neoplastic phenotype in human
Ras-transformed cells. Nat Med 2002 ; 8 : 979 – 86.
(16) Stackhouse MA, Buchsbaum DJ, Grizzle WE, Bright SJ, Olsen CC,
Kancharla S, et al. Radiosensitization mediated by a transfected anti-erbB-
2 single-chain antibody in vitro and in vivo. Int J Radiat Oncol Biol Phys
1998 ; 42 : 817 – 22.
(17) Politi K, Feirt N, Kitajewski J. Notch in mammary gland development and
breast cancer. Semin Cancer Biol 2004 ; 14 : 341 – 7.
(18) Dievart A, Beaulieu N, Jolicoeur P. Involvement of Notch1 in the develop-
ment of mouse mammary tumors. Oncogene 1999 ; 18 : 5973 – 81.
(19) Withers HR, Reid BO, Hussey DH. Response of mouse jejunum to multi-
fraction radiation. Int J Radiat Oncol Biol Phys 1975 ; 1 : 41 – 52.
(20) Steel GG, McMillan TJ, Peacock JH. The 5Rs of radiobiology. Int J Radiat
Biol 1989 ; 56 : 1045 – 8.
(21) Withers HR, Maciejewski B, Taylor JM, Hliniak A. Accelerated repopula-
tion in head and neck cancer. Front Radiat Ther Oncol 1988 ; 22 : 105 – 10.
(22) Kirkpatrick JP, Marks LB. Modeling killing and repopulation kinetics of
subclinical cancer: direct calculations from clinical data. Int J Radiat Oncol
Biol Phys 2004 ; 58 : 641 – 54.
(23) Johnson A. The timing of treatment in breast cancer: gaps and delays in
treatment can be harmful. Breast Cancer Res Treat 2000 ; 60 : 201 – 9.
(24) Bese NS, Sut PA, Ober A. The effect of treatment interruptions in the post-
operative irradiation of breast cancer. Oncology 2005 ; 69 : 214 – 23.
(25) Albright N. Computer programs for the analysis of cellular survival data.
Radiat Res 1987 ; 112 : 331 – 40.
at :: on November 13, 2012
Journal of the National Cancer Institute, Vol. 98, No. 24, December 20, 2006 ARTICLES 1785 Download full-text
(26) Olive PL. Detection of DNA damage in individual cells by analysis of his-
tone H2AX phosphorylation. Methods Cell Biol 2004 ; 75 : 355 – 73.
(27) Wollman R, Yahalom J, Maxy R, Pinto J, Fuks Z. Effect of epidermal growth
factor on the growth and radiation sensitivity of human breast cancer cells in
vitro. Int J Radiat Oncol Biol Phys 1994 ; 30 : 91 – 8.
(28) Paris F, Fuks Z, Kang A, Capodieci P, Juan G, Ehleiter D, et al. Endothe-
lial apoptosis as the primary lesion initiating intestinal radiation damage in
mice. Science 2001 ; 293 : 293 – 7.
(29) Dontu G, Jackson KW, McNicholas E, Kawamura MJ, Abdallah WM,
Wicha MS. Role of Notch signaling in cell-fate determination of human
mammary stem/progenitor cells. Breast Cancer Res 2004 ; 6 : R605 – 15.
(30) Holtz MS, Forman SJ, Bhatia R. Nonproliferating CML CD34+ progenitors
are resistant to apoptosis induced by a wide range of proapoptotic stimuli.
Leukemia 2005 ; 19 : 1034 – 41.
(31) West CM. Size-dependent resistance of human tumour spheroids to photo-
dynamic treatment. Br J Cancer 1989 ; 59 : 510 – 4.
(32) Sutherland RM, Durand RE. Radiation response of multicell spheroids — an
in vitro tumour model. Curr Top Radiat Res Q 1976 ; 11 : 87 – 139.
(33) Frankel A, Buckman R, Kerbel RS. Abrogation of taxol-induced G2-M
arrest and apoptosis in human ovarian cancer cells grown as multicellular
tumor spheroids. Cancer Res 1997 ; 57 : 2388 – 93.
Support for this research was provided by the Jonssen Comprehensive Cancer
Center Foundation at the University of California, Los Angeles to F. Pajonk and
by a National Institute of Biomedical Imaging and BioEngineering Training
Grant (# 5 T32 EB002101) to T. M. Phillips. The sponsors had no role in the study
design, data collection and analysis, interpretation of the results, or the preparation
of the manuscript.
Manuscript received June 27, 2006 ; revised October 2, 2006 ; accepted
October 24, 2006.
at :: on November 13, 2012