Research ArticleSYNTHETIC BIOLOGY

Numerical operations in living cells by programmable RNA devices

See allHide authors and affiliations

Science Advances  21 Aug 2019:
Vol. 5, no. 8, eaax0835
DOI: 10.1126/sciadv.aax0835
  • Fig. 1 Scheme for a multivariate linear combination in living cells.

    (A) Two steps in the calculation of an intracellular miRNA profile. The left panel shows a canonical approach, such as high-throughput analysis, in which the detection of a large dataset for miRNA molecules in a cell is followed by multivariate statistics to recapitulate the profile with few parameters. The right panel shows our approach, in which the profile is recapitulated with an mRNA device before signal detection. (B) In our device, mRNA responds to multiple miRNAs independently and with high sensitivity to output the coded fluorescent protein (FL1 or FL2). The values of ρ denote miRNA activity that represses the reporter protein expression and are tuned by specific tuning factors, k. A linear combination of multiple miRNA activities in a cell is obtained as the ratio of two fluorescent reporters.

  • Fig. 2 Quantitatively additive detection of miRNA activity by a synthetic five-slot mRNA.

    (A) The design of a synthetic mRNA that contains five slots for miRNA target sequences complementary to the miRNAs in the 5′UTR. The bottom part shows five-slot mRNAs responding to two or three miRNAs. Colored boxes indicate occupation of the slots by a target sequence as follows: gray, miR-34-a-5p; blue, miR-17-5p; red, miR-92a-3p; and green, miR-21-5p. Blank boxes depict empty slots, which are sequences of the same length as the target sequence and free from an miRNA target sequence. (B) An example result of a five-slot mRNA that responds to miR-17-5p and miR-92a-3p in HeLa cells. The design of the slots is shown above. Relative expressions are defined as the reporter expression normalized by the expression in the presence of miRNA inhibitors to both miRNAs (+/+). Values are presented above the bars. Error bars indicate the mean ± SD (n = 3). Calculation of the estimated expression is depicted above the chart. (C) Comparison of the relative expression with the estimated expression. A dot in the plot indicates the result of a five-slot mRNA responding to two or three miRNAs. Three independent experiments of 12 mRNAs are shown. Correlation coefficient (r) = 0.99 (for the means).

  • Fig. 3 Independently tunable detection of miRNA activity by a synthetic five-slot mRNA.

    (A) A series of five-slot mRNAs that contain one to three copies of the identical miRNA target sequences. A part of the series responding to miR-92a-3p (red boxes) is illustrated. The 5′UTR sequences are shown in table S1. The equation below represents the estimated expression, which is the product of the relative expressions repressed by the miRNA at each slot (ρ). (B) Comparison of the relative expression with the estimated expression. A dot in the plot indicates the result of a five-slot mRNA responding to the indicated miRNA with one slot or more. The values are the mean of three independent analyses. r = 0.98 (for all dots). (C) Relationship between the distance of a slot from the start codon (d) and the repression by the indicated miRNAs at the slot (local repression). Error bars indicate the mean ± SD (n = 3). Dotted lines are curves of the exponential model with the global constant = −0.56 (eqs. S5 and S6 in Supplementary Text). r = 0.98 (for all dots). nt, nucleotides.

  • Fig. 4 Classification of living cells by a set of five-slot mRNAs based on miRNA activity profiles.

    (A) Schematic illustration of the miRNA activity screening. A set of three single-slot mRNAs that respond to distinct miRNAs (miR-a, miR-b, and miR-c) and encode hmAG1, tagBFP, and hdKRed, respectively, and control hmKO2 mRNA were transfected into target cultured cells on a 24-well plate. The cells were analyzed by flow cytometry 24 hours later. The right table depicts miRNA activity profiles in analyzed cell types, which are determined on the basis of the expressions of the single-slot reporter mRNAs. Four plates (96 transfections) covered the screening of 270 miRNAs, as well as control transfections. (B) Obtained reporter expressions in hiPSCs from two independent screenings. Green, blue, and purple dots indicate single-slot mRNAs encoding hmAG1, tagBFP, and hdKRed, respectively. (C) Comparison of normalized miRNA activity profiles in NHDFs and hiPSCs. The normalized expressions are the mean of two screenings. See Supplementary Text and fig. S3 for details. (D) Classification of eight cell types by PCA of the 270 miRNAs. The cell types are plotted on the first two components (components 1 and 2). (E) The design of a set of four five-slot mRNAs that maximizes the variance of the eight cell types. Slots are occupied by the target sequence to the indicated miRNAs. Blank boxes denote empty slots. (F) Simulated classification of the eight cell types with the transfection of the mRNA set shown in (E). Axes are the fluorescence ratios of hmAG1/hmKO2 and tagBFP/hdKRed. (G) Live cell classification with the set of the four synthetic mRNAs and flow cytometry. Merged density plots of the transfected cells on the two fluorescence ratios are shown. The indicated cells are presented in red density, other cells in black density.

  • Fig. 5 Tracking changes in hiPSCs by a set of five-slot mRNAs based on miRNA activity profiles.

    (A) Schematic illustration of the second screening. According to differences between hiPSCs and differentiated hiPSCs revealed by the first screening, 54 of 270 miRNAs were selected for the second screening. Together with control transfections, 54 single-slot mRNAs made 24 sets of mRNA transfection. (B) Time course for tracking changes in hiPSCs. hiPSCs were cultured without bFGF for up to 14 days. The cells were transfected with 24 sets of mRNAs at the indicated days (filled triangles) and, 24 hours later, analyzed by flow cytometry (open triangles). Before the transfection at day 14, the cells were reseeded on new culture plates. (C) Reporter expressions between days 1 and 3. Green, blue, and purple dots denote single-slot mRNAs encoding hmAG1, tagBFP, and hdKRed, respectively. (D) Classification of hiPSCs in the time course by PCA of the 54 miRNAs. For comparison, hiPSCs cultured with bFGF for the indicated days were also subjected to the second screening and the analysis (blue dots). The cells are plotted on the first two components. (E) The design of a set of four five-slot mRNAs that maximizes the variance of the hiPSCs. (F) Simulated classification of the hiPSCs using the set shown in (E). (G) Density plots of the transfected hiPSCs. Plots are shown as in Fig. 4G except that the x axis represents hmKO2/hmAG1.

  • Table 1 Summary of the model accuracy.

    ModelFigureDetermination
    coefficient (R2)
    Root mean
    square error
    AdditivityFig. 2C0.990.029
    Positional effect
    (overall)
    Fig. 3B0.960.066
      miR-34a-5p0.750.044
      miR-17-5p0.920.067
      miR-21-5p0.920.071
      miR-92a-3p0.910.071
    Distance
    dependency
    (overall)
    Fig. 3C0.950.053
      miR-34a-5p0.960.014
      miR-17-5p0.700.072
      miR-21-5p0.970.045
      miR-92a-3p0.990.062

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/8/eaax0835/DC1

    Supplementary Text

    Fig. S1. The effect of miRNA inhibitors on the measurement.

    Fig. S2. The positional effect of miRNA target sequences in synthetic mRNAs for miRNA-mediated repression.

    Fig. S3. Steps in the normalization of the screening data in Fig. 4.

    Fig. S4. Live cell classification with four synthetic mRNAs.

    Fig. S5. Tracking of hiPSCs with four five-slot mRNAs.

    Table S1. 5′UTR sequences of the five-slot mRNAs used in this study.

    Table S2. 5′UTR sequences of the single-slot mRNAs used in this study.

    Table S3. List of primers, single-stranded oligo DNAs, and plasmids for PCR amplification.

    Table S4. List of experimental conditions.

  • Supplementary Materials

    This PDF file includes:

    • Supplementary Text
    • Fig. S1. The effect of miRNA inhibitors on the measurement.
    • Fig. S2. The positional effect of miRNA target sequences in synthetic mRNAs for miRNA-mediated repression.
    • Fig. S3. Steps in the normalization of the screening data in Fig. 4.
    • Fig. S4. Live cell classification with four synthetic mRNAs.
    • Fig. S5. Tracking of hiPSCs with four five-slot mRNAs.
    • Table S1. 5′UTR sequences of the five-slot mRNAs used in this study.
    • Table S2. 5′UTR sequences of the single-slot mRNAs used in this study.
    • Table S3. List of primers, single-stranded oligo DNAs, and plasmids for PCR amplification.
    • Table S4. List of experimental conditions.

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article