MicroRNA-222 Regulates Melanoma Plasticity

Abstract

Melanoma is one of the most aggressive and highly resistant tumors. Cell plasticity in melanoma is one of the main culprits behind its metastatic capabilities. The detailed molecular mechanisms controlling melanoma plasticity are still not completely understood. Here we combine mathematical models of phenotypic switching with experiments on IgR39 human melanoma cells to identify possible key targets to impair phenotypic switching. Our mathematical model shows that a cancer stem cell subpopulation within the tumor prevents phenotypic switching of the other cancer cells. Experiments reveal that hsa-mir-222 is a key factor enabling this process. Our results shed new light on melanoma plasticity, providing a potential target and guidance for therapeutic studies.

Keywords: cancer stem cells; hsa-mir-222; kinetic equations; melanoma; phenotypic switching; reaction diffusion; tumor plasticity.

Figure A1

The size $c_{staz}$ of the steady state CSC concentration as a function (a) of the switching rate $\sigma$, (b) of the rate difference $ϵ$, (c) of the rate $k_d$ and (d) of the number g of generations of CC.

Figure A1

The size $c_{staz}$ of the steady state CSC concentration as a function (a) of the switching rate $\sigma$, (b) of the rate difference $ϵ$, (c) of the rate $k_d$ and (d) of the number g of generations of CC.

Figure A2

(a) The quantity $R_{T,m_0}$ for 2 × 1$0^5$ sorted cancer cells as a function of T for different choices of $m_0$. (b) The same quantity plotted in (a),but assuming different possibilities for $c_0$ and making the experiments at $T=20$ days.

Figure A3

Expression of miRNA in conditioned medium (CM) obtained from WT IgR39 after 16 h in comparison to the level of expression of miRNA in the cells. The graph shows the average expression over two replica as counts per million [CPM]. Colors correspond to fold change expression in conditioned medium compared to the level found inside the cells. Gray dots identify miRNA detected only in one of the two conditions. The gray dashed line is a visual guide.

Figure A4

Efficiency of hsa-miRNA-222-5p knockdown in IgR39 cells. MicroRNAs silencing efficiency was evaluated by measuring hsa-miRNA-222-5p in IgR39 sponged cells (sh-mi222 5-p) according to the Materials and Methods section by relative expression and standard errors over three replica with qRT-PCR (a) (** p < 0.005) and microRNAs abundance (b). Significance has been assessed using EdgeR (* p-value < 0.05). Cells sponged with a scramble miRNA sequence (Scramble, see the Materials and Methods section) was used as a control.

Figure A5

Level of expression of CXCR6 in IgR39 cells transfected with hsa-miR-222-5p inhibitor (sh-mir222) or scramble sequence (scramble) as control. Sh-mir222 and scramble cells were obtained as described in the Materials and Methods section. Cells were analyzed for phycoerythrin (PE) anti-human CXCR6 (cod. MAB699 R&D System, Minneapolis, MN). For each flow cytometry evaluation, a minimum of ${10}^6$ cells were stained. Flow cytometry and analysis was performed using a Gallios flow cytometer (Beckman Coulter Indianapolis, IN, USA). (a) Average fraction of CXCR6 positive cells and standard errors over three replica. (b–e) Distribution of fluorescence signal for control cells stained with isotype antibody (b,d) and CXCR6-stained cells (c,e) in a typical experiment. Positive, selected cells are reported in green in both dotplots and histograms. Median and average fluorescence intensity is reported below each plot.

Figure A6

Expression of miRNAs 10b-5p, 15b-5p, 221-3p and 378a-3p for WT (a) and Sponge hsa-mir222-5p (b) IgR39 cells, compared to the Scramble condition, see Materials and Methods section for details. All comparisons lead to non-significant p-values when using a two-sample t-test.

Figure 1

Mean-field model (a) Reaction network for the mean-field model, involving cancer stem cells (CSCs) (S), cancer cells (CCs) ($N_j$) and inhibitor molecule m. (b) The experimental fraction $c\left(t\right)$ of CSCs after 2 days (blue bars) and after 6 days (red bars) from their sort out, as a function of their remaining fraction $c\left(0\right)$. The curves indicate the simulated results. (c) The dynamics of the concentration $c\left(t\right)$ of CSCs with different choices of the initial value $c\left(0\right)$ ($N_0={10}^6$, $m_0=0$).

Figure 2

Two-dimensional model (a) A sketch of the lattice model used for the calculations is shown on the left. Top panels show snapshots displaying the spatial arrangements of CCs of different ages and of CSCs for different values of the diffusion constant D. Below each panel, we display CSCs clusters, subtracted their density halo and colored it differently for each cluster for the same values of D. All the panels are obtained after 30 days of growth. (b) The radial density profile of the fraction of CSCs after 30 days of growth, with $\tau =4$, for different values of D and for the model without switching. (c) The histogram of densities of the different clusters.

Figure 3

Impact of mir222-5p knock-down on cell migration. (a) PIV analysis of wound healing assays for IgR39 cells transfected with hsa-miR-222-5p inhibitor (sponge miR222 [sh-mir222]) or scramble sequence (scramble) as control. Arrows indicate the local velocity and the color represents the local vorticity calculated with PIV as discussed in the Materials and Methods section. (b) The distribution of local velocities shows a reduction of the velocities upon mir222-5p knock down. (c) Growth of 3D spheroids in a collagen network sh-mir222 and scramble IgR39 cells. After digitization, we quantify the area occupied by cells as described in the Materials and Methods section. (d) The spread of the spheroid is quantified by measuring ${\left(\langle {(R-R_c)}^2\rangle \right)}^{1/2}$, where $R_c$ is the center of mass of the spheroid. Spreading of sh-mir222 cells is strongly impaired.

Figure 4

Impact of mir222-5p on the Wnt pathway. (a) The level of expression of LEF1 in different conditions in IgR39 cells, Results from [1]. The reported values are normalized with respect to the value of LEF1 in unsorted WT condition, whose value is therefore set to one. (b) A cartoon highlighting the role of mir222 on the Wnt pathway. Binding of Wnt to Frizzled leads to the stabilization of $\beta$-catenin and its translocation to the nucleus. Here, $\beta$-catenin can interact with the transcription factor LEF1, activating it. LEF1 allows the expression transcription factor MITF and, as a consequence, of MITF targets, including mir222. The latter interact with Frizzled, repressing Fzd-7 mRNA translation. Mir-222 is also released outside the cell (dashed black line) acting on paracrine cells.