Abstract:
Clonal heterogeneity in tumors is a critical problem, because it leads to increased
metastasis and treatment resistance. In order to better understand the role of tumor
heterogeneity in cancer malignancy, we developed a cell culture model to study
morphological changes that occur in two clones of the widely studied U87MG
glioblastoma cell line, C11 and E8, when grown separately versus when grown together.
When cultured alone, E8 cells show a moderate amount of clustering at full confluency
while C11 cells cluster very little. However, when grown together, C11 and E8 cells form
tight, spheroid-like clusters. Removing FBS, which is normally part of the growing
medium for these cells, for two hours at the time of cell seeding ("serum shock") resulted
in cell scattering and growth in an even monolayer. This phenotypical change was
permanent, with no cell clustering resumed for one week. To quantify this change in cell
clustering observation, we developed a novel method using Trainable Weka
Segmentation and ImageJ. This cluster analysis confirmed that both, C11 and E8 cells, cluster more in mixed culture than when alone (p < 0.0001) and serum shocking caused a
significant reduction in cell clustering in both clones when grown alone and when in
mixed culture (p < 0.0001). Migration assays revealed that clonal interaction during cell
migration, which normally leads to an inhibition of E8 cells by C11, was reduced due to
serum shock induced cell scattering. Reverse Phase Protein Array (RPPA) analysis
showed that PKCd T505, a downstream effector of the c-Met signaling pathway,
increased in mixed culture after serum shocking. Serum fractionation and mass
spectrometry lead to the identification of Apolipoprotein H (ApoH) as the component of
FBS that caused cell clustering. We verified that treating the clones with ApoH restored
clonal clustering and migration patterns to their normal levels. We discovered that
phosphorylation of five c-Met effectors, namely, STAT3 T505, FAK Y576/577, PKCd
T505, mTOR Ser2448 and Ezrin T567, were similar in mixed culture, when grown with
FBS and ApoH, but reduced in mixed culture after serum shock. We also examined the
effect of hypoxia on clonal interactions by simulating hypoxia using cobalt chloride
hexahydrate (CoCl2). We found that chemically induced hypoxia caused E8 cells that
were grown alone, to cluster more and migrate less than C11 cells that were grown alone.
In contrast to how they behave in normal conditions, E8 migration was less inhibited
when co-cultured with C11 cells. RPPA analysis also suggests that E8 cells increased Ecadherin
dependent cell adhesion after CoCl2 induced hypoxia, which may have activated
pro-survival signaling pathways via GAB1 Y627 phosphorylation. In conclusion, our
findings highlight the complex nature of interactions between tumor clones and
demonstrate the impact the microenvironment can have on these interactions.