Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Dec 27;5(6):549-563.e5.
doi: 10.1016/j.cels.2017.10.019. Epub 2017 Nov 29.

Linear Integration of ERK Activity Predominates over Persistence Detection in Fra-1 Regulation

Taryn E Gillies  1 , Michael Pargett  1 , Marta Minguet  1 , Alex E Davies  2 , John G Albeck  3
Affiliations

Linear Integration of ERK Activity Predominates over Persistence Detection in Fra-1 Regulation

Taryn E Gillies et al. Cell Syst. .

Abstract

ERK signaling regulates the expression of target genes, but it is unclear how ERK activity dynamics are interpreted. Here, we investigate this question using simultaneous, live, single-cell imaging of two ERK activity reporters and expression of Fra-1, a target gene controlling epithelial cell identity. We find that Fra-1 is expressed in proportion to the amplitude and duration of ERK activity. In contrast to previous "persistence detector" and "selective filter" models in which Fra-1 expression only occurs when ERK activity persists beyond a threshold duration, our observations demonstrate that the network regulating Fra-1 expression integrates total ERK activity and responds to it linearly. However, exploration of a generalized mathematical model of the Fra-1 coherent feedforward loop demonstrates that it can perform either linear integration or persistence detection, depending on the basal mRNA production rate and protein production delays. Our data indicate that significant basal expression and short delays cause Fra-1 to respond linearly to integrated ERK activity.

Keywords: EGF; EGFR; FOSL1; FRET; KTR; MAPK; Ras; c-Fos degradation; signal transduction dynamics; transcription.

PubMed Disclaimer

Figures

Figure 1
Figure 1. Different ERK reporters measure activity in different ranges
A. Schematic of the pathway and reporters used in this study. EKAR3 and ERKTR indicate ERK activity with minute-scale resolution, providing a readout of the PTM stage of EGF response. Genomic knock-in of a YFP fusion tag at the FOSL1 locus results in production of a fluorescent signal that responds to ERK activity at the level of both transcription and protein stabilization (orange arrows). NCd2 expressed from a viral promoter provides a control for general changes in gene expression capacity. B. Expected behavior of a persistence detector. As the duration of ERK activity increases, Fra-1 expression responds selectively to longer periods of stimulation. In the absence of mechanisms imposing this selectivity, a saturating linear response would be expected (blue dotted line). C. Sample EKAR3 and ERKTR measurements in single cells starved of growth factors and then stimulated at time 0 with 20 ng/mL EGF. Paired measurements from three individual cells are shown, and the mean for 489 cells below. D. Scatter plot of initial peak amplitude on EGF stimulation, for 5935 single cells, stimulated with different EGF concentrations (indicated by color). Each point represents the peak EKAR3 and ERKTR signals for an individual cell within a 1.5-hour window after stimulation. E. Spontaneous ERK activity pulses occurring in a single cell, stimulated at time 0 with 0.02 ng/mL EGF, reported by both ERKTR and EKAR3. Dotted orange and blue lines indicate the peak signal immediately following stimulation for ERKTR and EKAR3, respectively. Green and purple bars indicate the differences of subsequent peaks relative to the initial response. F. Histograms of the peak ERKTR (left) and EKAR3 (right) initial pulse amplitude reached in single cells, when stimulated with various concentrations of EGF. For comparison, shaded gray regions show the fully stimulated distribution (20 ng/mL EGF) in all panels. Red lines indicate the mean values of the responding population in each condition; responders were are defined as cells whose peak ERK activity is below the 95th percentile of the EGFRi (gefitinib) treated population (dashed lines). See also Fig. S1. G. Quantification of responding cells as detected by EKAR3 and ERKTR for multiple doses of EGF (as defined in F). H. Individual cell traces of EKAR3 following EGF stimulation, imaged at higher time resolution (2 min), demonstrating the increase in initial slope of EKAR3 with EGF concentration. I. Distributions of initial slope (maximum derivative within 30 min following stimulation) in EKAR3 following stimulation by EGF at different concentrations. Red bars indicate medians.
Figure 2
Figure 2. Measuring Fra-1 production as a function of ERK activity
A. Mean and representative single-cell time course measurements of ERKTR, Fra-1::Venus and NCd2 in the same cells following stimulation with 20, 0.2, or 0 ng/mL EGF. Prior to imaging, cells were cultured in the absence of growth factors for 48 hours, then stimulated with EGF at different concentrations. Four single cells and the mean (with 25th–75th percentile range as gray lines) are shown for each concentration of EGF, on both long and short time scales. B. Correlation between stimulated ERK activity and Fra-1 production rate. The maximum amplitude of ERKTR is plotted against the maximum slope of Fra-1::mVenus fold change within the first 60 minutes following EGF stimulation for each cell. Cells from multiple conditions are combined on a single plot. Red line represents a linear regression fit. P: Pearson’s correlation; slope is indicated with 95% confidence intervals. C. Correlation between stimulated ERK activity and NCd2 production rate. Plot was constructed as in (B), using NCd2 instead of Fra-1::mVenus. D. Contributions of ERK parameters to a PLSR model of Fra-1 expression. A PLSR model was constructed using the mean, maximum amplitude, mean derivative, and squared terms of ERKTR over the first 60 minutes following EGF treatment as inputs, and maximum slope of Fra-1 intensity as an output. Parameter weight distributions were generated by bootstrapping 8629 cells with replacement 10,000 times. ERKTR was used because of its broader amplitude response E. Prediction of Fra-1 rate by linear modeling of ERK signal terms, plotted against measured Fra-1 value. Values falling on the red line are predicted exactly by the model.
Figure 3
Figure 3. Fra-1 induction as a function of ERK activity duration
A–C. Response of Fra-1 to ERK pulses of varying duration. MCF10A cells expressing Fra-1::mVenus, ERKTR, and NCd2 were cultured in the absence of growth factors for 48 hours and then stimulated by 20 ng/ml EGF. At various times following EGF treatment, ranging from 30 minutes to 10 hours, 100 nM PD0325901 was added to acutely inhibit ERK activity. Stimulation duration 0 received PD 15 hours prior to EGF stimulation. Plots indicate the population mean values for ERKTR cytoplasmic/nuclear ratio (A), Fra-1::mVenus intensity (B), and NCd2 intensity (C); lines are color coded according to the interval between stimulation and inhibition. D, E. Plot of population-averaged AUC of NCd2 (D) or Fra-1::mVenus normalized to NCd2 (E) as a function of the interval between EGF addition and PD325901 addition. Error bars show 25th – 75th percentile range. Red line indicates linear fit. Data were also fit to a library of nonlinear models, and the linear model was chosen as the most parsimonious by Akaike’s Information Criterion. AUC was used to reflect the net Fra-1 output for the single defined ERK pulse. Regressed slopes are indicated with 95% confidence intervals. See also Fig. S3. F. Scatter plot of single cell measurements of Fra-1::mVenus (normalized as in E) plotted against ERKTR area under the curve. Red line indicates linear fit.
Figure 4
Figure 4. Fra-1 expression in response to pharmacological modulation of ERK activity patterns
A. Single-cell responses of ERK activity to inhibitors. MCF10A cells expressing EKAR3 were cultured in imaging medium containing 20 or 0.2 ng/ml EGF and treated with inhibitors of EGFR (gefitinib), MEK (trametinib), or ERK (SCH772984) at time 0 (arrows). Each plot shows the mean value in black at the bottom, with lighter gray lines representing the 25th and 75th percentile. Colored lines above the mean depict four randomly selected single cells for each condition, plotted with the same scaling as the mean. The higher-sensitivity EKAR3 is used here to show low ERK activity under inhibition; see Fig. S4 for corresponding ERKTR data. B. Population responses of ERK activity and Fra-1 expression to the indicated inhibitors. Cells were cultured in the presence of EGF and treated at time 0, at concentrations indicated by line color. C,D. Correlation of integrated ERK activity with Fra-1 expression across pharmacological conditions. In (C), each point represents the population average, for one drug condition, of Fra-1 intensity plotted against integrated ERKTR cytoplasmic/nuclear ratio; replicate experiments are shown individually. Measurements were made for a 20-hour window beginning 17 hours following drug treatment. Fra-1 intensity is quantified as AUC during the final hour of the 20-hour window to reflect the steady state response to the preceding ERK activity, which is represented as AUC over the 20-hour window. In (D) each point represents a single cell measurement of ERKTR and Fra-1, calculated over the same interval as in (C). Cells from all conditions are pooled, a total of 24,620 cells. ERKTR is used here to capture the wide dynamic range across all conditions; corresponding EKAR3 data shown in Fig. S4.
Figure 5
Figure 5. PLSR modeling of the ERK-Fra-1 relationship across all conditions
A. ERK activity metrics for single cells. Triangles indicate peak position; IPI, inter-pulse interval. For subsequent PLSR analysis, per-peak metrics were summarized on an individual cell basis, using the average or maximum as indicated in (C), over all measured peaks for that cell. B. Predictive capabilities of PLSR models for Fra-1 expression (measured by percent of total variance explained) using different combinations of ERK activity measurements (by EKAR3). For the “scrambled” model, pairings between input and output measurements for each cell were randomly reassigned. Models generated using ERKTR data (Fig. S5) showed similar trends but lower predictive capacity, possibly due to sampling bias within the dataset that included large numbers of low-ERK conditions. C. Contributions of ERK parameters to the PLSR model. Distributions of parameter coefficients generated by bootstrapping with replacement 10,000 times. D. Prediction of Fra-1 maximum value by linear modeling of ERK signal terms, plotted against measured maximum Fra-1 value. The PLSR model was constructed using the EKAR3 features in (A) over the 20 hour window, with the endpoint value of Fra-1 intensity as the output. Non-linear dependence was assessed by including mean squared in the PLSR. Red line represents perfect prediction of the data by the model.
Figure 6
Figure 6. Model analysis predicts linear response with baseline Fra-1
A. Network diagram of the delay differential equation analyzed and simulated. Delay (clock symbol) is included for the stimulated production of Fra-1 by activated transcription factor (TFP). See STAR Methods for a full description of the model. B. Minimal model simulations with various relative levels of baseline Fra-1 production (percentage given by 100 * kb/(kb + kF)). Total Fra-1 (shaded regions) is plotted against left axes; lower (green) and upper (blue) regions correspond to naïve and stabilized Fra-1 respectively. Fra-1 accumulation rate (dashed line) is plotted against right axes. Models were simulated with ERK activity beginning at time 0 and a 15 time-point delay for Fra-1 production; the effect of baseline production is most apparent in this window at the start of each simulation. See also Fig. S6. C. Simulated experiments with varied ERK pulse width, using the same parameter values as plots in B. The total Fra-1 AUC was calculated for square waves of ERK activity, performed with different relative baseline production levels (indicated in legend). Fra-1 AUC is plotted against the duration of ERK activity (as in Figure 3E). D. Sample image and pixel intensity histograms from cells expressing the Fra-1::mVenus fusion, after growth factor withdrawal for 48 hours and treatment with MEK inhibitor for another 15 hours. Histograms shown for regions indicated in the image: background only (red), cells (blue), whole frame (dashed line). E. Time course measurement of Fra-1 mRNA in MCF-10A cells, by RT-qPCR. Cells were cultured without growth factors for 48 hours, then stimulated with 20 ng/ml EGF and lysed for RNA extraction at ten-minute intervals up to 2 hours, followed by 1-hour intervals to 6 hours. Data were compiled from 3 replicate RNA extractions, with at least 2 replicate qPCR measurements per sample. NTC: No Target Control, NRT: No Reverse Transcription; IQR: Inter-Quartile Range, from 25th to 75th percentile. F. Comparison of RT-qPCR measurements of Fra-1 mRNA in unstimulated cells (48 hour growth factor withdrawal) to NTC (stars) and NRT (triangles), shown on a log scale. These data are not normalized to the control, HINT1. Red line indicates median. G. Early Fra-1 response following EGF stimulation (red line). Data for Fra-1::mVenus in cells cultured in the absence of growth factors for 48 hours and then stimulated with 20 ng/ml EGF are re-plotted from Fig. 3 to clarify early dynamics. Bold line indicates the population mean, and lighter lines represent 6 individual cells. Asterisk (*) indicates first time point significantly greater than baseline (p < 0.05). Blue line indicates the mean intensity of background (non-cell) pixels. Inset shows further magnification of the mean signal on a shorter time scale.
Figure 7
Figure 7. Comparative analysis of ETG responses between configurations and genes
A. Schematic of FIRE/Fra-1 comparison experiment. FIRE consists of the Fra-1 degradation domain fused to mCerulean and expressed from a stably integrated constitutive viral promoter; relative to endogenous Fra-1 it lacks stimulated mRNA production. B. Simultaneous measurements of Fra-1 and FIRE signals in five individual cells treated with 100 ng/ml EGF at time 0. C. Distributions of correlations between Fra-1 and FIRE within individual cells. Pairwise Pearson’s correlation coefficients were computed between Fra-1::mVenus and FIRE signals for 4383 cells over a period of 60 hours following treatment with different EGF concentrations; mean values are shown as red lines and numbers. To control for the correlation of FIRE with general expression increases, the same calculations were made between FIRE and total (nuclear+cytosolic) ERKTR-RFP signal, which is not directly controlled by ERK activity (right panels). D. Median intensities of Fra-1 and FIRE in cells treated with 1 ng/ml EGF. Shaded region indicates interquartile range across all cells. Inset plot shows median (across all time points) coefficient of variation between cells. E. Measurements of c-Fos and Egr-1 mRNA by RT-qPCR. Cells were stimulated, analyzed, and annotated as in Fig. 6(E). F. Measurements of c-Fos and Egr-1 protein abundance by immunofluorescence following growth factor withdrawal for 48 hours and stimulation with 20 ng/ml EGF at ten-minute intervals up to 2 hours, followed by 1-hour intervals to 6 hours. c-Fos data are compiled from 3 replicates. Egr-1 data are compiled from 2 replicates. G. Comparison of RT-qPCR measurements of Egr-1 and c-Fos mRNA in unstimulated cells (48 hour growth factor withdrawal) to NTC (stars) and NRT (triangles), shown on a log scale. These data are not normalized to the control, HINT1. Red line indicates median.
See this image and copyright information in PMC

Comment in

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Albeck JG, Mills GB, Brugge JS. Frequency-modulated pulses of ERK activity transmit quantitative proliferation signals. Mol Cell. 2013;49:249–261. - PMC - PubMed
    1. Altan-Bonnet G, Germain RN. Modeling T cell antigen discrimination based on feedback control of digital ERK responses. PLoS Biol. 2005;3:e356. - PMC - PubMed
    1. Amit I, Citri A, Shay T, Lu Y, Katz M, Zhang F, Tarcic G, Siwak D, Lahad J, Jacob-Hirsch J, et al. A module of negative feedback regulators defines growth factor signaling. Nat Genet. 2007;39:503–512. - PubMed
    1. Aoki K, Kumagai Y, Sakurai A, Komatsu N, Fujita Y, Shionyu C, Matsuda M. Stochastic ERK activation induced by noise and cell-to-cell propagation regulates cell density-dependent proliferation. Mol Cell. 2013;52:529–540. - PubMed
    1. Bakiri L, Macho-Maschler S, Custic I, Niemiec J, Guio-Carrion A, Hasenfuss SC, Eger A, Muller M, Beug H, Wagner EF. Fra-1/AP-1 induces EMT in mammary epithelial cells by modulating Zeb1/2 and TGFbeta expression. Cell Death Differ. 2015;22:336–350. - PMC - PubMed

Publication types

MeSH terms

Substances