Supplementary Figure Legends
Supplementary Figure S1. Irradiation does not induce CD133 cell
surface expression in CD133- cells. CD133- cells derived from D456MG
xenografts and primary glioblastoma sample T3607 were left untreated as a
control or irradiated with 3 Gy of IR then assayed at 48 hours for CD133 cell
surface expression by FACS analysis with an APC-conjugated anti-CD133
monoclonal antibody. CD133- populations derived from xenografts and
primary human glioblastoma samples remained CD133- after irradiation,
suggesting that irradiation did not alter CD133 surface expression in CD133cells.
Supplementary Figure S2. Radioresistant tumour cell lines are enriched
for CD133+ cells.
a. Isogenic radioresistant glioma cultures (D54R3,
D456R3) were derived from parental short term cultures isolated from D54MG
and D456MG human glioblastoma xenografts grown in immunocompromised
rodents through sequential administration of ionizing radiation (6 Gy) followed
by recovery periods during which cell populations were allowed to regrow. b.
The percentage of CD133+ cells was increased in radioresistant glioma cell
populations. The CD133+ and CD133- cell subpopulations of isogenic
radioresistant cell populations (D54R3 and D456R3) and the parental cells
(D54MG and D456MG) were quantified and compared using FACS analysis
after labeling with an APC-conjugated anti-CD133 specific antibody. The
CD133+ fractions were significantly elevated in the radioresistant populations.
Results are means ± s.d. (n=3); *, P<0.001.
Supplementary Figure S3. Viable unfractionated tumour cells derived
from irradiated xenografts are enriched in CD133+ tumour cells and form
secondary intracranial tumours with reduced latencies.
a. The
percentage of CD133+ cells is increased in irradiated xenografts relative to
untreated xenografts. D456MG and T3379X (a patient specimen briefly grown
as a xenograft) xenografts in nude mice were untreated as a control or
irradiated (3Gy). Viable tumour cells were isolated from each xenograft 48
hours after irradiation. The percentages of CD133+ populations in untreated
and irradiated samples were analyzed by FACS with APC-conjugated antiCD133+ antibody.
b. Viable unfractionated tumour cells derived from
irradiated xenografts form secondary tumours with reduced latencies relative
to untreated controls.
Viable unfractionated tumour cells derived from
untreated or irradiated xenografts were transplanted into nude mice brains
through intracranial injection (500,000 cells/mouse, 5 mice per group). The
survival time after tumour cell implantation until the development of
neurological signs was recorded.
Upon development of signs, mice were
sacrificed and confirmation of tumour formation by gross inspection and
systematic histology review was performed (representative gross images are
presented).
Viable unfractionated tumour cells derived from irradiated
xenografts significantly reduced the latency of secondary intracranial tumour
relative to matched tumour cells derived from the untreated xenografts (mean
± s.d., n=5). *, P <0.002.
Supplementary Figure S4. CD133- cells isolated from glioma xenografts
contain human tumour cells without significant murine cellular
contamination.
CD133- cells isolated from D456MG and D54MG glioma
xenografts were disaggregated and immunostained with the FITC-conjugated
3B4 antibody that specifically recognizes human cells, and then analyzed by
FACS. Greater than 99% of CD133- cells isolated from the glioma xenografts
are human in origin, suggesting minimal contamination by normal murine
cells.
Supplementary Figure S5. Tumour cells isolated from primary human
glioblastoma patient specimens are tumour cells as confirmed by FISH
analysis.
All surgical glioblastoma specimens at Duke University Medical
Center undergo genetic analysis of chromosomes 7 and 10, Pten, and
epidermal growth factor receptor (EGFR) as a standard protocol. To confirm
that the cells derived from these tumour specimens are cancerous in origin,
we studied the genetic changes that were altered in the majority of tumour
cells in the clinical specimens (thus, each specimen is analyzed differentially
based on the initial genetic analysis). In a representative study using the cells
derived from the T3359 tumour specimen, CD133- cells were analyzed by
Fluorescent
In
Situ
Hybridization
(FISH)
with
specific
chromosome
centromere probes. CD133- tumour cells isolated from the T3359 specimen
displayed polysomy of chromosome 10 centromere in more than 85% of cells
when a specific chromosome 10 CEP probe (green) was used for FISH. The
nuclear DNA was counterstained with DAPI (blue). Several representative
cancer cells showing polysomy (more than two positive green spots in each
cell) of chromosome 10 centromere are shown (b-e). The normal control cells
contained only two positive spots for chromosome 10 centromere (a). This
genetic alteration matched with that detected by the pathologist on the same
primary tumour that displayed chromosome 10 centromere polysomy in more
than 80% of cancer cells but not in normal cells. These data confirm that the
majority of CD133- cancer cells isolated from the primary gliomas are cancer
cells.
Supplementary Figure S6.
CD133+ cells from D456MG glioblastoma
xenografts
primary
and
T3750
glioma
exhibit
multi-lineage
differentiation potential. CD133+ tumour cells were purified from D456MG
xenografts or T3750 primary glioblastoma and maintained in the neurobasal
medium with EGF/bFGF to form neurospheres. CD133+ cells from
neurospheres were subjected to differentiation conditions. CD133+ cells were
grown on poly-ornithine and laminin-coated coverslips in -MEM with 10%
FBS for 7 days to induce differentiation. Differentiated cells derived from a
single neurosphere of CD133+ tumour cells isolated from D456MG xenografts
(a)
or
T3750
primary
glioma
specimen
(b)
were
examined
by
immunofluorescent staining with specific antibodies against differentiation
markers for astrocytes (GFAP and S100), oligodendrocytes (O4 and GalC),
and neuronal progenitors (Map-2 and TUJ-1). FITC-conjugated anti-mouse
IgG or anti-rabbit IgG secondary antibodies were used for detection. Nuclei of
the differentiated cells were counterstained with DAPI. Differentiated cells
from a single neurosphere display positive markers for astrocytes,
oligodendrocytes, or neuronal progenitors suggesting capacity for multilineage differentiation.
Supplementary Figure S7.
CD133+ cells isolated from a primary
glioblastoma tumour specimen exhibit multi-lineage differentiation
potential that is maintained after irradiation with a clinical dose of IR. a.
CD133+ tumour cells derived from a human patient glioblastoma surgical
specimen (T3781) were subjected to differentiation conditions. CD133+ cells
were grown on poly-ornithine and laminin-coated coverslips in -MEM with
10% FBS for 7 days to induce differentiation. Differentiated cells derived from
a single CD133+ neurosphere were examined by immunofluorescent staining
with specific monoclonal antibodies against differentiation markers for
astrocytes (GFAP and S100), oligodendrocytes (O4 and GalC), and
neuronal progenitors (Map-2 and TUJ1). A FITC-conjugated anti-mouse IgG
or anti-rabbit IgG secondary antibody was used for detection. Nuclei of the
differentiated cells were counterstained with DAPI. Differentiated cells from a
single neurosphere display positive markers for astrocytes, oligodendrocytes,
or neuronal progenitors, suggesting capacity for multi-lineage differentiation.
b. CD133+ tumour cells irradiated with a clinical dose of IR maintain the multilineage differentiation potential in vitro. CD133+ cells isolated from the human
primary glioblastoma surgical specimen were irradiated (2 Gy) and permitted
to recover for 48 hours before induction of differentiation. Differentiated cells
derived from the irradiated CD133+ cancer cells were examined by
immunofluorescent staining with specific monoclonal antibodies against
astrocytic markers (GFAP and S100), oligodendrocytic markers (O4 and
GalC), and neuronal progenitor markers (Map-2 and TUJ1). A FITCconjugated anti-mouse IgG or anti-rabbit IgG secondary antibody was used
for detection. Nuclei of the differentiated cells were counterstained with DAPI.
Differentiated cells from a single radiated neurosphere display positive
markers for astrocytes, oligodendrocytes or neuronal progenitors suggesting
capacity for multi-lineage differentiation in vitro after the treatment with clinical
dose IR.
Supplementary Figure S8. CD133+ tumour cells display radioresistance
and lower sensitivity to radiation-induced apoptosis than CD133- tumour
cells dependent on checkpoint kinase activity.
a. CD133+ glioma cells
exhibit greater survival potential after irradiation than matched CD133- tumour
cells derived from D54MG glioma xenografts. Cell survival in response to
irradiation was assessed by colony formation. Identical numbers of purified
matched CD133+ and CD133– cells were left untreated as a control or
subjected to a dose of IR (5 Gy) alone, a specific Chk1/Chk2 low molecular
weight inhibitor alone (debromohymenialdisine, DBH, 3 M), or a combination
of the two treatments in NBM without EGF/bFGF for 24 hours, and then
cultured in the zinc optional media with 10% FBS in 6-well plates until visible
colony formation. Each treatment for each cell type from each source was
performed in triplicate. Representative images of colony formation derived
from all treatments of CD133+ and CD133- cells are displayed. CD133+ cells
formed a greater number of colonies after IR treatment than the CD133- cells
supporting increased cellular resistance. CD133- cells were minimally
sensitized to ionizing radiation by checkpoint inhibition with DBH, but CD133+
cells displayed a much greater sensitization to radiation after checkpoint
inhibition with DBH.
b. CD133+ cells purified from in vivo irradiated
xenografts also display less caspase-3 activation than the CD133- cells
derived from the same irradiated xenografts. Mice bearing subcutaneous
D456MG glioma xenografts were untreated or treated with a single dose of 2
Gy or 5 Gy IR. Matched CD133+ and CD133- tumour cell populations were
isolated from the xenograft 24 hours after in vivo IR treatment. Whole cell
lysates were collected, resolved by SDS-PAGE, and immunoblotted with an
antibody specific for cleaved caspase-3 that is an apoptotic indicator. Equal
loading was confirmed by tubulin immunoblot.
Both doses of radiation
potently induced caspase-3 cleavage in the CD133- cells, an effect that was
greatly reduced in the CD133+ cells, suggesting CD133+ cells are also more
resistant to the IR-induced apoptosis in vivo.
CD133+ cells also
demonstrated lower rates of IR-induced Annexin V staining than matched
CD133- cells (Supplementary Fig. S11).
Supplementary Figure S9.
Survival of CD133+ and CD133- cells in
neurobasal media without growth factors did not differ within a five-day
period under basal conditions but differed after IR treatment (3 Gy).
CD133+ and CD133- cells derived from D54MG glioma xenografts (a) and
primary glioblastoma T3781 (b) were untreated or irradiated with 3 Gy IR and
incubated in the neurobasal media without EGF/FGF. The fraction of surviving
cells from each sample on each day (1-5 days) was analyzed in triplicate. The
survival of CD133+ and CD133- cells in neurobasal media without growth
factors in the first 5 days showed no significant difference in basal conditions.
In contrast, after 3 Gy IR treatment, the survival of CD133+ and CD133- cells
displayed a significant difference in the neurobasal medium without growth
factors. The maximal difference was detected on day 3 (48 hours after IR).
Results are means ± s.d. (n=3) *, P<0.002; **, P<0.001.
These data
demonstrate that CD133+ cells are more resistant than CD133- cells to
radiation-induced cell death.
Supplementary Figure S10. CD133+ cells show lower rates of apoptotic
cell death after irradiation than CD133- cells regardless of media
conditions.
CD133+ and CD133- cells derived from T3781 primary
glioblastoma were untreated or treated with 3 Gy IR and then incubated in
neurobasal media with or without EGF/bFGF. 20 hours after IR, apoptosis
was assessed with an Annexin-V-FITC kit according to the manufacturer’s
instructions. Annexin-V-FITC labeled samples were analyzed by flow
cytometry, and the results from three experiments were combined for
statistical analysis. IR-induced apoptotic cell death in the CD133- cells is
consistently greater (4-5 fold) than that in the CD133+ cells in the neurobasal
media with or without EGF/bFGF, but EGF/FGF provided a modest protective
benefit for both CD133- and CD133+ cells from IR-induced apoptosis. Data
are means ± s.d. (n=3; *, P<0.001; **, P<0.05).
Supplementary Figure S11. CD133+ cells displayed less apoptotic cell
death in response to radiation than matched CD133- cells by Annexin-VFITC staining. CD133+ and CD133- cells derived from primary glioblastoma
T3781 were untreated or irradiated with 3 Gy IR and incubated in neurobasal
media with EGF and FGF. 20 hours after IR, cells were collected and labeled
with Annexin-V-FITC according to the manufacturer’s instructions. A fraction
of the Annexin-V-FITC labeled cells from each sample was fixed and
counterstained with DAPI and examined using fluorescent microscopy (a).
Triplicate Annexin-V-FITC labeled samples were analyzed by flow cytometry
(b).
These data indicate that the IR-induced apoptotic cell death in the
CD133- cells is significantly greater (4-5 fold) than that in CD133+ cells,
confirming that CD133+ cells are more resistant to the clinical dose radiation
than CD133- population.
Results are means ± s.d. (n=3; *, P<0.001; **,
P>0.5).
Supplementary Figure S12.
CD133+ tumour cells display preferential
repopulation potential after ionizing radiation. CD133+ and CD133- cells
derived from D456MG xenografts were labeled with CFSE (green) and
CMTRX (red) fluorescent dyes separately and mixed in defined ratios (2%
CD133+). As the dyes are maintained after cell division, cells derived from
CD133+ cells were green while those derived from CD133- cells were red.
Triplicate parallel cultures were left untreated as a control or treated with IR (5
Gy) in NBM without EGF/bFGF for 24 hours then cultured in zinc option media
with 10% FBS after recovery.
The resulting growth patterns of each tumour
cell population at each indicated time point was visualized using fluorescent
microscopy (a) and quantified by FACS (b).
c. The percentage of cells
derived from CD133- and CD133+ cells at days 1, 4, and 8 under untreated or
irradiated conditions was quantified (mean ± s.d., n=100 cells X 3 trials).
CD133+ cells (green) demonstrated a modest repopulation advantage relative
to CD133- cells (red) under the untreated control conditions, but the
repopulation advantage of CD133+ cells was much greater than that of
CD133- cells after IR treatment. *, P<0.002; **, P<0.001.
Supplementary Figure S13.
Irradiated CD133+ cells derived from
xenografts and primary human glioblastoma surgical specimens are
capable of initiation of brain tumours in vivo. a. Irradiated CD133+ tumour
cells with a clinical IR dose (2 Gy) retain tumourigenecity in vivo. CD133+
cells were purified from D456MG xenografts and primary glioblastomas
(T3359, T3379) and then left untreated or irradiated with 2 Gy IR. Viable
irradiated or untreated CD133+ tumour cells were transplanted into the right
frontal lobes of BalbC athymic nude mice (10,000 cells/mouse) at 48 hr after
IR. The neurological status of the mice after implantation was closely
monitored. Upon development of neurological signs, mice were sacrificed,
and the brains were harvested. Tumour formation was confirmed by gross
inspection and systematic histology of the entire brain. Descriptive statistics
are displayed (mean ± s.d., n=5). No statistically significant differences in
tumour latency between the group injected with irradiated CD133+ cells and
the control group injected with untreated CD133+ tumour cells were noted.
b. Tumour latency is not significantly different between mice injected with
CD133+ cells derived from untreated or irradiated xenografts. D456MG or
T3379X (derived from an uncultured patient specimen directly implanted into
the flanks of athymic nude mice) xenografts were either left untreated as a
control or radiated (3 Gy) in vivo. Viable CD133+ cells were purified from both
control and irradiated xenografts and implanted into the brains of athymic
nude mice (10,000 cells/mouse, 5 mice per group). The time until the
development of neurological signs was recorded.
Upon development of
signs, mice were sacrificed, and confirmation of tumour formation by gross
inspection and systematic histology review was performed (representative
gross images are presented). Data are means ± s.d. (n=5).
c. Irradiation
completely inhibits the tumourigenecity of CD133- cells derived from D456MG
xenografts. Untreated CD133- cells derived from D456MG xenografts formed
occasional small tumours in two of five mice brains. In contrast, the matched
irradiated CD133- cells (2 Gy) demonstrated no capacity to form any tumours.
Representative gross images of brains from each group are displayed.
Supplementary Figure S14. Preferential activation of the DNA damage
checkpoint in CD133+ glioma tumour cells.
a, b. CD133+ tumour cells
displayed increased checkpoint activation in response to DNA damage
relative to CD133- cells derived from glioma D456MG xenograft and human
primary glioblastoma T3359. CD133+ and CD133- cells were isolated from
malignant D456MG glioma xenograft and human primary glioblastomas after
tumour dissociation, labeled with an APC-conjugated anti-CD133 antibody,
and sorted. The activation state of the checkpoint response was assessed in
each cell line without (-) and 1 hour after NCS (neocarzinostatin, a
radiomimetic agent, 100 ng/ml) treatment (+).
Whole cell lysates were
prepared, resolved by SDS-PAGE, and immunoblotted with the indicated
antibodies.
The phosphorylation levels and total expression levels of
checkpoint proteins (ATM, Rad17, Chk2 and Chk1) were examined by
Western blot analysis. c. CD133+ tumour cells demonstrate increased ATM
kinase activity in response to irradiation relative to CD133- cells. CD133+ and
CD133- cells isolated from a human primary glioblastoma surgical specimen
(T3618) were untreated or irradiated (3 Gy). Whole cell lysates were prepared
one hour after IR.
ATM complexes were immunoprecipitated (IP) from
untreated or irradiated CD133+ and CD133- cells and incubated with a GSTRad17 substrate in an in vitro ATM kinase assay. The phosphorylation of the
GST-Rad17 substrate was analyzed by immunoblotting with a phosphospecific antibody for pS645-Rad17. Equal amounts of GST-Rad17 substrate
input was determined by immunoblotting with an antibody against total Rad17.
The equal amount of lysate protein for ATM IP was determined by
immunoblotting with anti-tubulin antibody. The increase of phosphorylated
Rad17 mediated by ATM kinase from irradiated CD133+ cells is significantly
higher than that from irradiated CD133- cells, indicating increased ATM
kinase activity in CD133+ cells.
Supplementary Figure S15.
CD133+ glioma tumour cells repair
radiation-induced DNA damage more efficiently than CD133- cells.
a.
The IR-induced DNA damage was examined using the alkaline Comet assay.
In this assay, the presence of DNA damage at sequential time points after
damage was assessed by single cell gel electrophoresis assay under alkaline
conditions. The repair capacity of each type cell was measured by the
disappearance of the comet tail after damage. Comet tails reflect damaged
DNA fragments that move faster than the intact DNA under single cell
electrophoresis. CD133+ and CD133- tumour cells were derived from
D456MG malignant glioma xenografts. Separate cultures were exposed to a
single dose of ionizing radiation (3 Gy) and permitted to recover in NBM with
EGF/bFGF. a. Representative images of the single cell gel electrophoresis of
CD133+ and CD133- cells at the indicated time points after IR are displayed.
b. The percentage of cells with comet tails at different time points after IR in
CD133+ and CD133- populations derived from a D456MG xenograft were
quantified (means ± s.d., n=100 cells X 3 trials). CD133+ cells displayed a
significantly lower percentage of cells with comet tails at 18 hours and 30
hours after IR than the matched CD133- populations. Thus, CD133+ cells
recovered from DNA damage at a faster rate than matched CD133- cells. *,
P<0.001; **, P<0.002.
Supplementary Figure S16.
CD133+ tumour cells resolve foci of
phosphorylated H2AX after DNA damage more efficiently than CD133+
tumour cells. a. The presence of double stranded DNA breaks after IR was
assessed by immunoflourescent analysis of phosphorylated H2AX foci.
Matched CD133- and CD133+ tumour cell cultures were derived from
D456MG xenografts and grown under control conditions or irradiated with 3
Gy then assessed at 1 and 24 hours after irradiation.
Cells were
permeabilized, immunoblotted with an anti-phosphorylated H2AX antibody,
and visualized under a fluorescent microscope. Representative images of
phosphorylated H2AX under untreated control or IR treatment conditions are
shown. b. The resolution of the phosphorylated H2AX nuclear foci in CD133+
glioma cells is significantly faster than that in the matched CD133- cells after
radiation. The number of cells with nuclear foci of phosphorylated H2AX at
different time points after IR (3 Gy) in CD133- and CD133+ cells derived from
D456MG gliomas were quantified. Descriptive statistics are graphed (mean ±
s.d., n=100 cells X 4 trials). The presence of phosphorylated H2AX nuclear
foci in CD133- cells is significantly higher than that in CD133+ cells at 24 hr
after IR. Accelerated resolution of the phosphorylated H2AX nuclear foci in
radiated CD133+ tumour cells suggests preferential repair in this tumour
subpopulation *, P<0.001.
Supplementary
Figure
S17.
CD133+
tumour
cells
display
radioresistance in vivo relative to CD133- cells dependent on checkpoint
kinase activity.
CD133+ and CD133- glioma tumour cells were separately
labeled with CMTPX (red) and CFSE (green) fluorescent dyes, mixed in a
defined ratio (1:5 CD133+:CD133-), and implanted orthotopically into the right
frontal lobes of the brains of athymic nude mice (5 mice/group). Injected mice
were then left untreated as a control or treated with external beam ionizing
radiation (IR, 5 Gy), a checkpoint inhibitor (DBH), or a combination of the two
treatments. Mice were sacrificed after 8 days and brains were assessed for
the relative proportion of cells derived from the CD133+ (red) and CD133(green) populations by fluorescent microscopy of the brain section and by
FACS analysis. a. The paraformadehyde (PFA)-fixed brain tumour sections
from four groups of mice (untreated, IR treated, DBH treated, IR+DBH
treated) were examined under a fluorescent microscope. The regions of the
brains injected with the mixed cancer cells were analyzed and photographed.
Tumour cells in red were derived from CD133+ cells, and tumour cells in
green were derived from CD133- cells. b. Descriptive statistics of the total
number of cells in each group quantified by FACS analysis from three set of
animal experiments are presented (mean ± s.d., n=3). The region of the brain
injected with the labeled cells from each mouse was dissected and
disaggregated to isolate the total tumour population, and the cancer cells
derived from CD133+ (red) and from CD133- (green) were quantified by
FACS. *, P<0.001; **, P>0.01. c. The ratio of in vivo tumour cells derived from
CD133+:CD133- cells demonstrated that CD133+ tumour cells have a growth
advantage in vivo that is significantly augmented after IR treatment. Data are
means ± s.d. (n=3; *, P<0.001 IR treated relative to the untreated control
group). The relative resistance to irradiation of the CD133+ tumour cells was
reversed by the combination of IR and DBH (Chk1/Chk2 inhibitor) treatments
in vivo. IJ=the ratio of CD133+/CD133- cells for the initial intracranial injection.