A versatile and robust cell purification system with an RNA-only circuit composed of microRNA-responsive ON and OFF switches

MicroRNA-responsive ON and OFF switches provide a safe and time-saving method for purifying a wide range of cell types.


Fig. S3. Schematic illustration of the gene expression distribution in each cell type.
The expression level of the lethal gene from the switch in target cells (survive) is lower than that in non-target cells (death), because the endogenous miRNA expression shows a shifted distribution (green and orange distributions). The amount of introduced mRNA into individual cells is variable and difficult to be controlled precisely. In addition, the translation activity depends on the cell status, such as size, cell cycle, and health, resulting in a wide distribution of the expression level in each cell. Thus, the expression level in the target cells overlaps that in non-target cells (overlap). When a small amount of the switch is transfected to a heterogeneous cell population, the lethal gene is expressed in non-target cells, but not to the lethal threshold in some, resulting in contamination by non-target cells. To remove the contamination, we need to increase the amount of the switch transfected. Higher amounts of the transfected switch shift the expression distribution in both target-and non-target cells towards a higher lethal gene expression level (rightward). In this situation, the purity of the target cells increases but also results in a lower yield, because the expression level of the lethal gene in some target cells exceeds the lethal threshold. Dose-dependency of the selection efficiency of miR-21-5p-Bim-OFF switch. Co-cultured HeLa and 293FT were treated with miR-21-5p-OFF switch encoding the apoptotic protein Bim. The number of cells for each cell type was counted in a similar manner to Fig. 7B and 7F, and the percentage of each cell line was calculated. Closed circles and triangles are HeLa and 293FT, respectively. Error bars represent the mean ± SD (n=3), and the data of each biological replicate is shown as a point. (C) Merged fluorescence microscopic images of co-cultured cells. HeLa cells with a stable and high expression of hmAG1-M9 and 293FT cells with a stable and high expression of iRFP670-M9 were used in order to distinguish the two cell lines. The co-cultured cells were transfected with miR-21-5p-OFF switch encoding Bim. HeLa cells were expected to survive, because high miR-21-5p activity should decrease the Bim expression. However, both cells were removed by the switch. Scale bar indicates 200 μm.

Fig. S5. Phase diagram showing the relationship between cell death and the expression levels of Barnase (Bn) and Barstar (Bs) using single-or dual-switch selection.
Bn and Bs are a lethal RNase and its inhibitor, respectively. The pink region indicates the region where cells die. Light blue indicates the area where the cells survive. On the left, by using a single switch, the overlap between non-target (red) and target cells (green) will be observed due to the leaky expression of Bn in the target cell. In contrast, by using dual switches (right), Bn and Bs expression levels from the co-transfected mRNAs show a proportional relationship on a twodimensional plot. When the miRNA-ON switch encoding Bn and the miRNA-OFF switch encoding Bs are introduced, cells with high miRNA activity show the expression pattern indicated by the red dots. On the other hand, cells with weak or no miRNA activity show the expression pattern indicated by the green dots. When the total amount of transfected RNA changes, the expression pattern moves in the direction indicated by the double-edged arrows.

Fig. S6. Co-transfection ratio and efficiency of mRNAs (A)
The dose-dependency of the EGFP expression, relative expression (EGFP/iRFP670), and transfection efficiency in HeLa, 293FT and iPS cells. The cells were transfected with 50, 100, and 200 ng of EGFP mRNA and iRFP670 mRNA (total 100, 200, and 400 ng, respectively). The relative expressions were normalized by the value of the cells transfected with 50 ng of each mRNA. (B) Scatter plots of cells co-transfected with EGFP mRNA and iRFP670 mRNA. The cells inside the rectangles were defined as transfected cells and calculated for the expression level of EGFP. The iPS-EGFP cells originally showed a fluorescence intensity of 10 4 . Changes in the fluorescence cannot be detected when the amount of transfected mRNAs is low (region where the co-transfected iRFP670 signal is less than 10 4 ), because the expression level of EGFP is too low compared with the originally expressed EGFP.

Fig. S7. Schematic illustration for detecting intracellular Bn activity via EGFP translation and selecting specific cell types.
The expression from miR-21-5p-OFF switch encoding Bs (miR-21-5p-Bs-OFF) and miR-21-5p-ON switch encoding Bn (miR-21-5p-Bn-ON) selectively kill HeLa cells because of the high miR-21-5p activity and high Bn activity. The active Bn in HeLa cells suppresses the EGFP translation from co-transfected EGFP mRNA. In contrast, the switches should not kill 293FT cells due to the high Bs activity. HeLa and 293FT cells were transfected with miR-21-5p-Bn-OFF and -Bs-ON switches (or miR-21-5p-Bn-ON and -Bs-OFF switches) in addition to EGFP mRNA. The Bn/Bs ratio indicates the ratio of miR-21-5p-Bn and -Bs switches. The cells inside the red rectangles showed different EGFP expressions and cytotoxicity between HeLa and 293FT cells.  The number of cells for each cell type was counted inside the squares in (A), and the percentage of each cell line was calculated. Error bars represent the mean ± SD (n=3), and data of each biological replicate is shown as a point. ****P < 0.001. (C) Merged fluorescence images of the cells treated with the switches without passage. iPS-EGFP and HeLa-iRFP670-M9 are colored in green and red, respectively. Scale bar, 200 μm.

Fig. S11. Scatter plots of cells after purification by miR-302a-5p-ON and -OFF switches.
Representative two-dimensional flow cytometry plots. Co-cultured iPSCs expressing GFP and HeLa cells expressing iRFP670-M9 were treated with miR-302a-5p-Bn-OFF and -Bs-ON switches. After the passage of the co-cultured cells, all cells were analyzed by flow cytometry.
The squares indicate the gates of iPSCs and HeLa cells.

Fig. S12. Gene expression analysis of iPSCs transfected with RNA switches. (A)
Scatter plots of iPSCs at 72, 96 and 120 hours after selection. The miR-302a-5p-target genes predicted by TargetScan and pluripotency critical genes (PCGs) are highlighted in blue and red, respectively. The genes whose normalized expression level is less than 100 were excluded. (B) Boxplots of PCGs (31), human ESC essentialome genes (hESCE) (32), and target genes predicted by TargetScan (29) or miRTarBase (30). (C) Relative read counts of Bsd mRNA normalized by the value at 72 hours after the transfection. The data of each biological replicate (n=2) is shown as a point. (D) Mean percentage of TRA-1-60 positive cells at 10 days after the transfection with the three mRNAs (miR-302a-5p-OFF switch, miR-302a-5p-ON switch, and Bsd mRNA) (Switch mRNA). The cells transfected with mRNA encoding Bsd (Bsd mRNA) were used as transfection control. The data of each biological replicate (n=2) is shown as a point.

Fig. S14. cTNT assay of purified iPSC-derived cardiomyocytes (A)
Percentage of iPSC-cardiomyocytes positive for cTNT after selection. Treatment with RNA switches (miR-208a-3p-Bn-OFF switch, and miR-1-3p-Bs-ON switch, iRFP670 mRNA, and aph mRNA) increased the ratio of cTNT positive cells (94.8%) compared with untransfected cells (25.1%). Error bars represent the mean ± SD (n=3), and the data of each biological replicate (n=3) is shown as a point. ****P < 0.0005. (B) Representative scatter plots of cTNT stained cells after purification by miRNA-switches. The population in the area surrounded by the line is defined as cTNT positive cells. (C) Summary of the cells before and after the purification process. Table S1. UTR sequences of the miRNA-ON switch.