Supplementary Methods
Side Population: For SP analyses, sub-confluent cultures were collected with cell
dissociation buffer (Gibco), resuspended in PBS with 2% of FBS at 1or 2x106 cells/mL,
and incubated with 5 µg/mL Hoechst 33342 (Invitrogen, Eugene, Oregon, USA) for 90
min at 370C in a water bath. The side population was analyzed based on the exclusion of
Hoecsht 33342 using an LSRII flow cytometer (BD Bioscience).
SP gating was
determined by treatment with 200 µM verapamil (Sigma-Aldrich), which is used as a
negative control because it inhibits the transporters responsible for effluxing Hoescht dye
in SP cells.
Propidium iodide (1 µg/mL) was used to discriminate live cells. Early
passage cells were analyzed three separate times.
Quantitative RT-PCR Array: Total mRNA was isolated from Dek+/+ and Dek-/- MEFs
using Trizol extraction, then reverse transcribed using the Quantitect kit (Qiagen,
Valencia, CA) and analyzed with an ABI-7500 quantitative PCR machine (Applied
Biosystems, Carlsbad, CA). The mouse Wnt pathway quantitative RT-PCR array (RT2
Profiler PCR Array, SABiosciences/Qiagen) was performed and analyzed according to
manufacturer’s instructions.
Supplementary Figure Legends
Supplementary Figure 1:
HGFL treatment of R7 cells activates Ron receptor-
mediated signaling. Serum-starved (-FBS) R7 cells were treated with 100 ng/ml HGFL
from two different manufacturing lots for 30 minutes then analyzed by western blotting.
Increased p-ERK1/2 indicates activated signaling while total ERK1/2 and actin are
loading controls.
Supplementary Figure 2: Dek proficient murine breast cancer cell lines have more
side population cells and can form xenograft tumors. (A) Dek expression correlates
with the size of the cancer stem cell pool. Cells were treated with Hoechst dye and
analyzed by flow cytometry to identify dye-excluding side population (SP) cells.
Representative plots for SP cells (highlighted box) are shown, including 200 M
verapamil treatment as a negative control to eliminate the SP fraction. (B) Cell lines
generated from RontgDek-/- and RontgDek+/+ tumors were injected into the inguinal
mammary fat pad of female nude or NSG immunodeficient mice and monitored for tumor
growth for 10 weeks. Most cell lines were injected into four separate glands.
Supplementary Figure 3: Immunohistochemical staining of lung metastases and
primary murine tumors. (A) Lung metastases from RontgDek-/- and RontgDek+/+ mice
stain for cytokeratin-5, a marker of epithelial tissues. (B,C) There is no consistent
difference in MMP2 and MMP9 expression in Dek proficient and deficient cell lines and
tumors. (B) Western blotting of whole cell lysates of Dek knockdown and complemented
cell lines. Blots were probed for MMP2 (OE=over-exposed), MM9, Dek, and Actin. The
location of the protein ladder is shown on the right. The predicted size of MMP2 is 72
kDa while MMP9 is approximately 105 kDa. (C) Immunohistochemical staining of MMP9
(left) and MMP2 (right) in RontgDek-/- and RontgDek+/+ primary tumors and R7
NTsh/Deksh xenograft tumors. Representative images are shown from at least 5 tumors
tested in each group.
Supplementary Figure 4: Quantitative RT-PCR screening for Wnt pathway
perturbations in Dek deficient cells. (A) A quantitative RT-PCR array was probed
using cDNA from Dek+/+ and Dek-/- mouse embryonic fibroblasts. Relative expression
compared to the average of five housekeeping genes is shown. (B) The transcripts with
>2-fold change in Dek-/- MEFs compared to Dek+/+ MEFs from the Q RT-PCR array are
shown.
Blue indicates down-regulation and red indicates up-regulation.
(C) Dek
complementation in an additional RontgDek-/- cell line results in Wnt10b and cyclin D1
over-expression as determined by western blotting.
Supplementary Figure 5:
The contribution of DEK and RON, individually and
combined, for predicting -catenin staining and patient outcomes.
(A) RON
expression, and to a lesser extent DEK expression, correlate with high -catenin staining
in a tissue microarray of infiltrating breast ductal carcinomas. When combined, tissues
that had high levels of RON and/or DEK were highly predictive of positive -catenin
staining. (C-E) DEK expression is highly predictive of patient outcomes, especially when
combined with RON expression.
DEK and RON expression were assessed
independently and together for their association with relapse-free survival in patients that
had not received systemic therapy (C), distant metastases in systemically-treated
patients (D), and post-progression survival in all patients (E). Data was acquired from a
publically available online database, Kaplan-Meier Plotter (www.kmplot.com).
Supplementary Figure 6: DEK expression in head and neck squamous cell
carcinomas (HNSCCs) correlates with cell growth, -catenin activity, and invasive
potential. (A) DEK knockdown in a K14E7 Dek+/+ murine tumor cell line reduces cellular
growth. Tumor line was derived from an established HNSCC mouse model, wherein 4nitroquinoline-1-oxide (4-NQO) was utilized as a tumor initiator and given to mice
expressing HPV E7 oncogene under the control of a K14 (cytokeratin 14) promoter. This
esophageal tumor was excised and cultured in vitro on irradiated mouse fibroblasts as a
feeder layer. These Dek+/+ cells were then depleted for DEK with lentiviral DEKshRNA as
confirmed by western blot. Cells were plated at equal densities post-selection in
puromycin and counted 48, 72, and 96 hours post plating. Data represents three
independent experiments with SEM displayed. (B) HNSCC DEK overexpressing cells
have elevated -catenin activity. CCHMC-HNSCC1 cell populations were derived from a
primary human HNSCC tumor. Cells were retrovirally transduced with control (R780) or
overexpressing (R780-DEK) and sorted for GFP and then assayed by luciferase reporter
for TOP/FOP activity. (C) HNSCC DEK knockdown cells have reduced -catenin activity.
CCHMC-HNSCC3 cell populations, derived from a different patient than CCHMCHNSCC1 described above, were lentivirally depleted for DEK and assayed for TOP/FOP
luciferase activity. Data is presented as fold change and SD represented. (D) DEK
overexpressing cells from (B) have increased invasion through Matrigel transwell
assays. Data represents the total number of cells invaded.