Overshoot during phenotypic switching of cancer cell populations

Abstract

The dynamics of tumor cell populations is hotly debated: do populations derive hierarchically from a subpopulation of cancer stem cells (CSCs), or are stochastic transitions that mutate differentiated cancer cells to CSCs important? Here we argue that regulation must also be important. We sort human melanoma cells using three distinct cancer stem cell (CSC) markers - CXCR6, CD271 and ABCG2 - and observe that the fraction of non-CSC-marked cells first overshoots to a higher level and then returns to the level of unsorted cells. This clearly indicates that the CSC population is homeostatically regulated. Combining experimental measurements with theoretical modeling and numerical simulations, we show that the population dynamics of cancer cells is associated with a complex miRNA network regulating the Wnt and PI3K pathways. Hence phenotypic switching is not stochastic, but is tightly regulated by the balance between positive and negative cells in the population. Reducing the fraction of CSCs below a threshold triggers massive phenotypic switching, suggesting that a therapeutic strategy based on CSC eradication is unlikely to succeed.

Figure 1. Overshoot of CXCR6 and CD271 in human melanoma sorted cells.

CXCR6- or CD271- negative cells are sorted from IgR39 cells as described in Methods and plated under standard growth conditions. Cells are collected 3, 10 and 20 days after sorting, and analyzed by flow cytometry. Non-specific mouse IgG is used as isotype control (Isotype). The distribution of the markers in unsorted cells is also reported (U). Flow cytometry is performed using a FACSAria flow cytometer (Becton, Dickinson and Company, BD, Mountain View, CA). Data are analyzed using FlowJo software (Tree Star, Inc., San Carlos, CA). For each flow cytometry evaluation, a minimum of 5 × 10⁵ cells are stained and at least 50000 events are collected and analyzed. (a) The level of expression of each markers in unsorted cells (U) and at different times after sorting is reported as flow cytometric analysis of a minimum of 5 independent experiments. (b) Unsorted cells (U), CXCR6-positive, and CXCR6-negative cells after 10 days sorting are fixed with 3.7% paraformaldeide and incubated with polyclonal anti-CXCR6 antibody (1:400, Abcam, Ab8023) overnight. Then the cells are incubated with the secondary antibody (anti rabbit Alexa488 1:250) for 1 h and the nuclei counterstained with DAPI. The slides are mounted with Pro-long anti fade reagent (Invitrogen) and the images acquired with a Leika TCS NT confocal microscope.

Figure 2. The rate of phenotypic switching depends on the initial concentration of positive cells.

(a) CXCR6-positive and -negative IgR39 cells are sorted by flow cytometry and mixed with the fraction of positive cells varying from 0.2 to 1%. Cells are collected 2 and 6 days after sorting and the level of expression of CXCR6 is quantified by flow cytometry. (b) A representative example of flow cytometry for different initial fractions of CXCR6 positive cells. (c) The percentage of positive cells obtained in the different experimental conditions.

Figure 3. Identification of differentially expressed miRNAs before and during the overshoot and the corresponding pathways involved.

(a) The color map shows miRNAs that are are differentially expressed with respect to unsorted cells for CXCR6- or CD271- negative cells at 3, 10 and 20 days after sorting. The cells are also analysed immediately after sorting (0). We filter by p-value (<10⁻⁵), fold change(>2.0), and absolute expression (>0.01%). Increased levels are reported in red, decreased levels in blue. (b) For each condition in panel a, we identify the significantly affected pathways using the Diana-MirPath database as described in Methods. The colormap shows the estimated p-value for each pathway at different times. The p-value represents the probability that a given pathway is significantly enriched with the presence of mRNA targets of each miRNA.

Figure 4. Interaction network between miRNAs and pathways.

(a) The network formed by differentially expressed miRNAs in CXCR6-negative cells before the overshoot (3 days after sorting) and pathways identified to be significantly targeted by those miRNAs. (b) The same network as in a) restricted to Wnt and PI3K signalling pathways. (c) The network formed by differentially expressed miRNAs in CXCR6-negative cells at the overshoot (10 days after sorting) and pathways identified to be significantly targeted by those miRNAs. (d) The same network as in (c) restricted to Wnt and PI3K signalling pathways. For all panels the miRNAs are shown in red when they increase and in blue when they decrease; the arrows are red when they induce an increase of the targeted pathway and in blue when they induce a decrease.

Figure 5. Expression of regulatory factors during phenotyping switching.

(a) Colormap of real time PCR has been carried out on unsorted cells and CXCR6-negative cells 3 and 10 days after sorting as described in Methods. The colormap shows the log of fold change of each factor at different times after sorting. (b) Western blot of unsorted cells and CXCR6-negative cells 10 and 20 days after sorting as described in Methods. Tubulin has been used as housekeeping gene. (c) Immunofluorescence of βcatenin (green) and Falloidin (red) of unsorted cells (US) and CXCR6-negative cells 10 days after sorting (10 days).

Figure 6. Mathematical model explains CSC overshoot in phenotypic switching.

(a) A schematic representation of the model considering CSC, cancer cells (CC) and senescent cells, with a possibility of phenotypic switching depending on the level of miRNA. (b) Simulations of the model for cell populations with different initial fraction of CSC display an overshoot in the fraction CSC for sufficiently low initial concentrations, in agreement with experiments.

Figure 7. mRNA and miRNA expression in breast tumors with ESC signature.

mRNA expression data from 1980 breast tumors and miRNA expression data from 1283 breast tumors are separated in two classes depending on the presence of an ESC signature. (a) We report a set, corresponding to the factors in Fig. 5a, of significantly changed mRNA between ESC and non-ESC tumors. The color represents the relative fold-change of each mRNA for the two sets of tumors. (b) A large set of miRNA are significantly over-expressed (OE) or under-expressed (UE) between ESC and non-ESC tumors. We report the pathways putatively targeted by these two sets of miRNAs. The color represent the p-value. A low p-value suggests that the pathway is likely to be targeted by the set of OE or UE miRNAs.