In the Markov model illustrated using roulettes in figure 2(d) of Liao et al. A, stochastic fluctuations are depicted as though they occur independently in different cells. One cell’s spin of a wheel of fortune is not affected by the spins of other wheels at the same time, and vice versa. In other words, stochasticity is cell intrinsic (). However, this perspective is likely to be often an oversimplification. Just because a cell is depicted in a cartoon as a well-defined container does not mean that it is “statistically” isolated.
Figure 5 Stochastic fluctuations can be integrated at various scales. (a) Some fluctuations in the abundances of some molecules may be localized to individual cells. These fluctuations are cell-intrinsic. (b) Local signaling may propagate the effects of fluctuations (more ...)
Molecular fluctuations may propagate through clusters of cells connected by paracrine signaling (). Kim et al.
have described a paracrine signaling loop in Wnt1-induced mouse mammary tumors where luminal cells provide Wnt1 signaling for basal cells presenting the Lrp5 receptor [37
]. In principle, transient fluctuations in Wnt1 signaling secreted by the luminal subpopulation could manifest cell-extrinsic effects, including transient losses in tumor-initiating capability. In a study of human embryonic stem cells (hESCs), Bendall et al.
proposed a model in which hESCs differentiated into a fibroblast-like population (hdFs). The hdFs in turn secrete signaling factors, such as IGF-II, that sustain the hESCs in a self-renewing, pluripotent state [38
]. In these two examples, a transient loss of signaling from one cell type could result in a loss of stem-like phenotypes in another cell type, causing the system to “differentiate out.”
Fordyce et al.
have shown that DNA damage stress in primary human mammary epithelial cells increases the secretion of Activin A, which can increase the levels of Activin A in surrounding cells [39
]. Human mammary epithelial cells (HMEC) respond by secreting molecules (prostaglandins) that increase the motility of surrounding epithelial cells. Thus fluctuations in Activin A may ripple outward in a bed of stationary cells, as well as be carried along by newly mobilized vehicles (). In the presence of cell-cell signaling and cell motility, the fundamental “stochastically fluctuating units” most relevant to consider for therapy may be cell communities in a tissue, rather than individual cells in a population.
For the particular case of metronomic therapy, this perspective offers a direction for increasing our understanding of the role of the microenvironment. As discussed in Liao et al. A, rationales that have historically been associated with high-frequency therapeutic dosing schedules have included targeting “non-epithelial” populations and processes such as angiogenesis, carcinoma-associated fibroblasts, and immune modulation. Our current discussion suggests going beyond simply regarding stromal cells as secondary targets for metronomic therapy. We propose that combinations of the constituents of the microenvironment and the frank carcinoma may need to be regarded together as the basic, cohesive units, in which stochastic fluctuations appear, propagate, and integrate.