epithelial-mesenchymal heterogeneity can emerge in a population
of cancer cells. Due to the cell-to-cell variation of kinetic parameters
driving EMT dynamics, cells can exhibit distinct gene expression
profiles that can broadly be grouped into epithelial, mesenchymal,
and hybrid E/M classes. The cell-to-cell variation in kinetic parameters is indicative of the differing exogenous signaling states in
different cells. While cells in the population exhibit different gene
expression profiles, the population does not consist of clones and
subclones with cells in each clonal population exhibiting a specific
EMT kinetic, i.e., the gene expression profile of a cell is not always
hereditary and can change in response to changes in the exogenous
signaling environment. Note that while our analysis reveals 2 groups
of steady states with co-expression of epithelial and mesenchymal
factors suggesting that 2 such hybrid E/M states exist, a different
analysis technique may reveal a greater number distinct types of
hybrid E/M phenotypes. Cells can likely be classified into an even
greater number of phenotypic groups by incorporating other
EMT-associated factors into the circuit topology [10, 11] which
would provide greater resolution as has been reported recently
[16]. Finally, the RACIPE framework can easily be used to probe
the contribution of each protein and micro-RNA and of each
regulatory relationship in driving epithelial-mesenchymal heterogeneity. One can edit the circuit topology file (extension .topo) to
add and/or delete EMT-associated factors and regulatory relationships and analyze the expression levels in the collection of steady
states obtained for the altered circuit.
5 Heterogeneity from Random Partitioning of Molecules During Cell Division
Another scenario in which phenotypic heterogeneity can emerge in
a population occurs if cells undergo stochastic changes in their
phenotypes. In general, for such stochastic changes to happen,
there must exist a mechanism to generate noise and a mechanism
to stabilize the decision reached in response to the noise [63]. One
mechanism which can generate noise is the random partitioning of
molecules (RNAs, proteins, etc.) in the parent cell among the
daughter cells at the time of cell division [53, 54, 64]. This mechanism is likely to be a prominent source of noise in tumors wherein
cells divide fast and uncontrollably. While phenotypic fluctuations
in cells in response to noise are usually small and transient, the
fluctuations can be amplified if the underlying response mechanism
exhibits multi-stability, i.e., co-existence of multiple steady states.
As described previously [9, 13], circuits which drive EMT and MET
exhibit multi-stable behavior. Thus, random partitioning of
EMT-associated factors during cancer cell division is likely to be a
key contributor toward the emergence of epithelial-mesenchymal
heterogeneity.
398
Shubham Tripathi et al.
of cancer cells. Due to the cell-to-cell variation of kinetic parameters
driving EMT dynamics, cells can exhibit distinct gene expression
profiles that can broadly be grouped into epithelial, mesenchymal,
and hybrid E/M classes. The cell-to-cell variation in kinetic parameters is indicative of the differing exogenous signaling states in
different cells. While cells in the population exhibit different gene
expression profiles, the population does not consist of clones and
subclones with cells in each clonal population exhibiting a specific
EMT kinetic, i.e., the gene expression profile of a cell is not always
hereditary and can change in response to changes in the exogenous
signaling environment. Note that while our analysis reveals 2 groups
of steady states with co-expression of epithelial and mesenchymal
factors suggesting that 2 such hybrid E/M states exist, a different
analysis technique may reveal a greater number distinct types of
hybrid E/M phenotypes. Cells can likely be classified into an even
greater number of phenotypic groups by incorporating other
EMT-associated factors into the circuit topology [10, 11] which
would provide greater resolution as has been reported recently
[16]. Finally, the RACIPE framework can easily be used to probe
the contribution of each protein and micro-RNA and of each
regulatory relationship in driving epithelial-mesenchymal heterogeneity. One can edit the circuit topology file (extension .topo) to
add and/or delete EMT-associated factors and regulatory relationships and analyze the expression levels in the collection of steady
states obtained for the altered circuit.
5 Heterogeneity from Random Partitioning of Molecules During Cell Division
Another scenario in which phenotypic heterogeneity can emerge in
a population occurs if cells undergo stochastic changes in their
phenotypes. In general, for such stochastic changes to happen,
there must exist a mechanism to generate noise and a mechanism
to stabilize the decision reached in response to the noise [63]. One
mechanism which can generate noise is the random partitioning of
molecules (RNAs, proteins, etc.) in the parent cell among the
daughter cells at the time of cell division [53, 54, 64]. This mechanism is likely to be a prominent source of noise in tumors wherein
cells divide fast and uncontrollably. While phenotypic fluctuations
in cells in response to noise are usually small and transient, the
fluctuations can be amplified if the underlying response mechanism
exhibits multi-stability, i.e., co-existence of multiple steady states.
As described previously [9, 13], circuits which drive EMT and MET
exhibit multi-stable behavior. Thus, random partitioning of
EMT-associated factors during cancer cell division is likely to be a
key contributor toward the emergence of epithelial-mesenchymal
heterogeneity.
398
Shubham Tripathi et al.