A Model-System to Address the Impact of Phenotypic Heterogeneity and Plasticity on the Development of Cancer Therapies

Abstract

The main challenges in developing effective anti-cancer therapies stem from the highly complex and heterogeneous nature of cancer, including the presence of multiple genetically-encoded and environmentally-induced cancer cell phenotypes within an individual. This diversity can make the development of successful treatments difficult as different phenotypes can have different responses to the same treatment. The lack of model-systems that can be used to simultaneously test the effect of therapies on multiple distinct phenotypic states further contributes to this problem. To mitigate these challenges, we suggest that in vitro model-systems that consist of several genetically-related but phenotypically distinct populations can be used as proxies for the several phenotypes (including adherent and circulating tumor cells) present in a patient with advanced disease. As proof of concept, we have developed such a model and showed that different phenotypes had different responses to the same challenge (i.e., a change in extracellular pH) both in terms of sensitivity and phenotypic plasticity. We suggest that similar model-systems could be developed and used when designing novel therapeutic strategies, to address the potential impact of phenotypic heterogeneity and plasticity of cancer on the development of successful therapies. Specifically, the effect of a therapy should be considered on more than one cancer cell phenotype (to increase its effectiveness), and both cell viability as well as changes in phenotypic state (to address potential plastic responses) should be evaluated. Although we are aware of the limitations of in vitro systems, we believe that the use of established cell lines that express multiple phenotypes can provide invaluable insights into the complex interplay between therapies and cancer's heterogeneous and plastic nature.

Keywords: H2122; experimental evolution; extracellular pH; metastasis; microenvironment; phenotypic plasticity; selection; therapy.

Figure 1

An in vitro model-system to address the effect of the same microenvironmental change on genetically-related but phenotypically distinct cancer cell populations. (A) A non-small cell lung cancer line (H2122_ANC) was used to select for an adherent line (H2122_AS) and a suspension line (H2122_SS) through 15 and 40 serial passages of only the adherent or suspension populations, respectively. (B) Experimental design to address the effect of extracellular acidic and alkaline pH on the ANC_Adherent and ANC_Suspension populations of the (i) ANC line, and the (ii) Adherent Selected (AS) and (iii) Suspension Selected (SS) lines.

Figure 2

The effect of extracellular pH on the size (as number of live cells) and phenotypic composition (as percentage of live cells in the adherent and suspension cell fractions relative to the entire population) of four genetically-related but phenotypically distinct cancer cell populations subjected for 24 h to acidic (pH 6.4), control (pH 7.4), and alkaline (pH 8.4) media. (A) H2122_ANC_Adherent Population. (B) H2122_ANC_Suspension Population. (C) H2122_AS line. (D) H2122_SS line. (E,F) Changes in the proportion of the suspension and adherent fractions in the Adherent and Suspension populations of the ANC line, as a function of pH [data are from (A,B)]. (H) Attached H2122_SS cells exhibit round morphology following treatment with alkaline media; (G) Attached H2122_ANC_Suspension cells exhibit a mixture of round and epithelial-like morphology following treatment with alkaline media. All treatments were performed in 1.5 ml media, with 3 biological replicates. Viability was assessed using SYTO 9 and Propidium Iodide (PI) (Molecular Probes) and the Countess II FL Automated Cell Counter (ThermoFisher Scientific). SYTO 9 stains DNA in both dead and live cells, while PI is only permeant to necrotic and late apoptotic cells. Because this method could miss early apoptotic cells with intact membranes and apoptotic cells that disintegrated, numbers of dead cells are not reported here. All data were expressed as means ± SD of 3 biological replicates (each with 3 technical replicates). ^* and ^† indicate significant differences in the numbers of live cells of the suspension and adherent fractions, respectively, between treatments and control (^*p < 0.0332, ^**/^††p < 0.0021, ^***/^†††p < 0.0002, ^****p < 0.0001). Total population sizes did not show statistically significant differences, except in one case (see text).