Fig 1.
microRNA-dependent regulation of gene expression in developing thymocytes.
A) Schematic of T cell differentiation. Abbreviations are DN: double negative for CD4 and CD8; DP: double positive for CD4 and CD8; Periphery: peripheral lymphoid organs. The expression of mature microRNAs, including miR-181a, is reduced by ∼ 90% at the DP stage of development as demonstrated by northern blotting with U6 snRNA as a loading control, data from [26]. B) 3'UTR motifs and microRNAs associated with positive fold-change of transcript levels in Dicer-deficient DP thymocytes as determined by miReduce [27]. 'Average expr.' denotes the fold-change in the expression of mRNAs with the indicated 3'UTR motifs between Dicer-deficient and control DP thymocytes. C) Control and Dicer-deficient DP thymocytes as defined by staining for CD4 versus CD8a (left) and CD4 versus CD8b (middle). Both the mean expression and the CV of CD8a and CD8b are comparable between control and Dicer-deficient DP thymocytes. The ratio of CD8a/CD8b expression was calculated for individual control (left) and Dicer-deficient (middle) DP thymocytes. The CV of these ratios represents experimental noise (right) [25]. D) CD44, Ly6a and H2-Kb are examples of proteins encoded by transcripts that are deregulated in Dicer-deficient DP thymocytes. The panels show representative flow cytometry histograms of CD44, Ly6a and H2-K1 expression by Dicerlox/lox (black/grey) and DicerΔ/Δ (red/orange) DP thymocytes gated on low levels of T cell receptor (TCR) expression. The MHC class II antigen H-2Ab1 is not expressed by mouse T cells and served as a negative control. Numbers indicate the mean expression level and the coefficient of variation (CV). Representative of 3–5 biological replicates. E) Analysis of 10–30 biological replicates showed an increase in the CV of CD44 protein expression of 30% and an increase in the CV of Ly6a protein expression of 15% in Dicer-deficient versus control DP thymocytes. F) DP thymocytes were sorted by flow cytometry according to the level of CD44 and Ly6a expression by individual cells. Analysis of the sorted populations by quantitative RT-PCR indicates that protein expression as assessed by flow cytometry predicts mRNA expression.
Fig 2.
Increased CV of inducible CD69 expression in Dicer-deficient thymocytes.
A) Activation of CD4+ CD8+ DP thymocytes (oval, top) results in CD69 expression (horizontal line defines CD69+ cells, middle) and definition of CD69hi cells (oval, bottom). B) Histogram overlays of CD69 expression by CD69+ DP thymocytes activated for 18 hours with 125ng H57/ml. Mean and CV of CD69 expression are indicated. C) The CV of CD69 expression is higher in Dicer-deficient than in control CD69+ DP thymocytes over a range of activation conditions (n = 4–7 per data point, ** P<0.001). See S1 Table for additional data. D) The frequency of CD69hi CD25+ DP thymocytes is higher in Dicer-deficient than in control DP thymocytes. E) Histogram overlays of CD69 expression by CD69hi CD25+ thymocytes activated for 18 hours with 125ng H57/ml. Mean and CV of CD69 expression are indicated. See S1 Table for additional data. F) The CV of CD69 expression is higher in Dicer-deficient than in control CD69hi CD25+ DP thymocytes (n = 4 per data point, ** P<0.001). See S1 Table for additional data. G) Summary of CD69 CV data for Dicer-deficient DP thymocytes, CD69+ DP thymocytes and CD69hi CD25+ DP thymocytes.
Fig 3.
A dual fluorescence reporter system identifies endogenous microRNAs that target the Cd69 3'UTR in DP thymocytes.
A) Dual fluorescence reporter based on retroviral vectors encoding eGFP followed by a multiple cloning site and mCherry for normalisation. Introduction of a 3'UTR containing relevant microRNA sites is predicted to downregulate eGFP expression relative to the mCherry control. The 842 nt 3’ UTR of Cd69 contains predicted binding sites for miR-181, miR-130 and miR-17-20 starting at positions 255, 354 and 391, respectively, which were mutated alone and in combination. B) Representative log-log dot plot of mCherry and eGFP-Cd69 3'UTR expression by control (black) and Dicer-deficient mature CD4+ T cells isolated from lymph nodes (red). Fitted lines were used to calculate de-repression of eGFP. See S3A Fig. for additional data. C) Expression of eGFP obtained with empty vector and the indicated 3'UTR constructs in control (black) and Dicer-deficient mature CD4+ T cells isolated from lymph nodes (red, n = 4–14 per data point, mean ± SE). D) Expression of eGFP obtained with the indicated 3'UTR constructs in control DP thymocytes (n = 4–14 per data point, mean ± SE). See S3B Fig. for additional data.
Fig 4.
miR-181a controls cell-to-cell variation in CD69 expression.
A) Histogram overlays of CD69 expression by control (black) and miR-181a-deficient (red) DP thymocytes activated for 18 hours with 125ng H57/ml. Histograms are gated on CD69+ cells. B) The CV of CD69 expression is higher in miR-181a-deficient than in control DP thymocytes (n = 7–8, ** p<0.005, *** p<0.0005). C) The frequency of CD69 hi CD25+ DP cells is higher in miR-181a-deficient than in control DP thymocytes (n = 12, P<10–5). D) The CV of CD69 expression in miR-181a-deficient and control CD69 hi CD25+ DP thymocytes is slightly higher than in control thymocytes. E) Model for the action of miR-181 upstream of TCR signaling and on Cd69 mRNA.
Fig 5.
miR-17 and miR-20 form an incoherent positive feedback loop with the target mRNA Cd69.
A) The strength of activation signals (0.1, 1, 10 μg/ml H57) determines the expression of Cd69, miR-17 and miR-20a (normalised to snRNA-135 and -202, n = 2–3, mean ± SE). B) Perceived signal strength varies among individual DP thymocytes and determines the expression of Cd69, miR-17 and miR-20a (microRNA expression is normalised to snoRNA-135 and snoRNA-202). At a fixed extracellular signal of 1u/ml H57, the fold-change in miR-17 and miR-20 relative to CD69 negative DP and normalised to snoRNA-135 was proportional to the expression of CD69 (n = 3, mean ± SD). C) The microRNA target mRNA Cd69 and microRNAs of the miR-17-92 cluster are co-regulated in response to activation signals and form an incoherent feed-forward loop downstream of the TCR. D) Modeling CD69 protein expression with and without microRNA feed-forward regulation. The state of the system is described by five major variables: the number of mRNAs transcribed from the TF gene, the number of TF molecules, the number of miRNAs, the number of mRNAs and the number of target proteins. Of these variables we can estimate the number of mRNA copies and the number of microRNA copies. As detailed in the legend to S4 Fig., the number of Cd69 mRNA copies was estimated as 0 in resting and 6 in activated cells, miR-17 and miR-20 were estimated as 6–12 copies per cell the resting state and 30–60 copies per cell after activation. Simulations of transcriptional networks were carried out using the Gillespie exact stochastic simulation algorithm, programmed and analysed using R based on a microRNA feed-forward model [8] to simulate CD69 protein expression in resting T cells (unfilled histogram), in a scenario where activation increases Cd69 mRNA but the expression of miR-17 and miR-20a remained the same as in resting T cells (activated without microRNA FFL, filled red histogram), and in a scenario where activation increases both Cd69 mRNA and miR-17 and miR-20a expression (activated with microRNA FFL, filled grey histogram). The plot represents 10,000 simulations. The model predicts that thymocyte activation with co-regulation of Cd69 mRNA and miR-17/miR-20a reduces the mean (887 versus 1300) and the CV (10.2 versus 14.6) of CD69 expression compared to the regulation of Cd69 mRNA alone (P<10–4). See S4 Fig. for details of the underlying circuitry, the parameters used, and a model based on microRNA effects on mRNA degradation [49].