Frequency-modulated pulses of ERK activity transmit quantitative proliferation signals

Abstract

The EGF-stimulated ERK/MAPK pathway is a key conduit for cellular proliferation signals and a therapeutic target in many cancers. Here, we characterize two central quantitative aspects of this pathway: the mechanism by which signal strength is encoded and the response curve relating signal output to proliferation. Under steady-state conditions, we find that ERK is activated in discrete, asynchronous pulses with frequency and duration determined by extracellular concentrations of EGF spanning the physiological range. In genetically identical sister cells, cell-to-cell variability in pulse dynamics influences the decision to enter S phase. While targeted inhibition of EGFR reduces the frequency of ERK activity pulses, inhibition of MEK reduces their amplitude. Continuous response curves measured in multiple cell lines reveal that proliferation is effectively silenced only when ERK pathway output falls below a threshold of ~10%, indicating that high-dose targeting of the pathway is necessary to achieve therapeutic efficacy.

Figure 1. Steady-state signaling and proliferation in mammary epithelial cells

A. Determining signal-response relationships. Left, a signal-response curve (yellow) relates the strength of pharmacological signal inhibition (e.g., of ERK) to the resulting reduction in proliferation (arrows). Right, this curve can be evaluated using steady-state measurements of signal intensity and proliferation rate at various EGF concentrations. B. EGFR-ERK signaling pathway measurements used in this study. C. Design of FIRE. Amino acids 163–271 of Fra-1 were fused in-frame to the C-terminus of mVenus, with a nuclear localization sequence (NLS) at the N-terminus. D. Induction of FIRE fluorescence. Curves depict the median fluorescence intensity for populations of 100–1000 cells treated with EGF at time 0 (green) or untreated (black) cells. E. Decay of FIRE fluorescence upon ERK pathway inhibition. MCF-10A cells expressing FIRE were grown in the presence of 20,000 pg/ml EGF and treated with 1 µM gefitinib (EGFR inhibitor), 1 µM PD325901 (MEKi3), or 1 µM PD184161 (MEKi4) at the indicated time. Curves depict the median fluorescence intensity for a population of 100–1000 cells for EGF-treated (green) or untreated (black) cells. Associated time-lapse images are shown in Supplemental Movie S1. F. Live-cell measurements of ERK output (FIRE reporter) and cell cycle progression (RFP-geminin) in a single cell deprived of growth factors for 48 hours and re-stimulated at time 0 with 20,000 pg/ml EGF. Medium was replaced at ~24-hour intervals (dashed lines). G. Population-level measurements of signaling and proliferation in mammary epithelial cells. Curves represent the median intensity of FIRE signal (green; shaded region indicates the 25^th – 75^th percentile) and the fraction positive for RFP-geminin (f-GMNN+, red) as determined by automated live-cell microscopy for cells treated as in (C). Yellow shaded region indicates a time period over which both FIRE and f-GMNN+ signals remain stable. Measurements were made under high-volume culture conditions (see Materials and Methods) H. Fluctuations in signaling and proliferation under standard culture conditions. Measurements were performed as in (G), using standard culture conditions (see Materials and Methods). I. Establishment of multiple proliferative steady states. Cells were treated and analyzed as in (G), with varying concentrations of EGF. J. Percentage of pRb-positive cells at steady state. Error bars indicate standard deviation of six replicates. See Supplemental Fig. S1 for characterization of pRb dynamics in proliferation. Data shown are from one experiment representative of three or more independent replicates.

Figure 2. Signaling states under chronic EGF stimulation

A. Mammary epithelial cells were treated as in Fig. 1G, fixed at 58 hours after treatment, and analyzed for the indicated proteins by HCIF. Histograms indicate the distribution for a population of >5000 cells; the distribution in the absence of EGF is shown in gray for comparison. See Supplemental Fig. S2 for HCIF analysis of additional proteins. B. Representative HCIF images for cells analyzed as in (A). C. Quantitation of pRb⁺,pERK⁻ cells cultured at steady state with 200 pg/ml EGF and analyzed as in (A). Data shown are from one experiment representative of three or more independent replicates.

Figure 3. EGF-stimulated, frequency-modulated ERK activity pulses

A, B. ERK activity in single cells under steady-state conditions. Individual MCF-10A cells expressing EKAR-EV were treated as in Fig. 1G and imaged from 48 to 70 hours after stimulation. A. EKAR-EV signal (CFP/YFP ratio) for single cells; note that the Y-axis is inverted as ERK activity induces a decrease in CFP/YFP ratio. B. EKAR-EV signal for 29 cells in each EGF concentration; each row in the heat maps represents one cell. Data shown are from one experiment representative of five independent replicates. Associated time-lapse imaging is shown in Supplemental Movie S2. C. Comparison of nuclear and cytosolic ERK activity for an individual cell. Cells stably expressing EKAR-EV were transferred from full growth medium to 20 pg/ml EGF immediately prior to imaging. The ratio of CFP/YFP fluorescence was measured in discrete regions within the nucleus or cytosol; results are representative of all cells examined. D. Immunofluorescence of phospho-ERK under steady-state conditions. MCF-10A cells were grown in the presence of 20 pg/ml EGF for 60 hours prior to fixation and staining. Note that in pERK-positive cells, pERK is present throughout the cell.

Figure 4. Stimulation of cell cycle progression by sustained ERK activity pulses

A. Imaging of ERK activity pulses and cell cycle progression. MCF-10A cells expressing EKAR-EV and RFP-geminin were imaged in the presence of 50 pg/ml EGF. Pairs of sister cells were analyzed in which one sister cell entered a subsequent round of DNA replication (as indicated by RFP-geminin induction) >5 hours prior to the other sister. One representative cell pair is shown; see Supplemental Fig. S3 and Movie S3 for additional examples and time-lapse images. B. Automatic detection of ERK activity pulses. Top, regions of increased EKAR-EV activity (red circles) identified by a peak detection algorithm for a representative cell pair. Bottom, heat map of EKAR-EV signal in the same cell pair, with binary EKAR-EV values (solid blue and yellow) shown for each pair. C. Patterns of EKAR-EV pulse dynamics prior to S-phase entry. Binarized EKAR-EV measurements for 223 cells were ordered from top to bottom by the length of the interval between the previous mitosis (green lines) and the subsequent induction of RFP-geminin (red line) in the first cell of the pair (cell 1, left); the corresponding sister cells not entering S-phase are shown at right. At bottom, the median EKAR_on pulse length for a sliding time window is shown; dotted line indicates the overall median pulse length for all cells. Cyan line indicates 12 hours prior to RFP-geminin induction, where median pulse length increases sharply for the cell 1 population. D. Distribution of ERK activity parameters in cells committing to S-phase earlier (cell 1) or later (cell 2). Mean pulse length, longest pulse, and fraction of time in the ERK_on state were calculated for each cell within the 12 hours preceding RFP-geminin induction (for cell 1), and within the same 12 hour period for cell 2. Red lines indicate the median, boxes the 25^th – 75^th percentile, dashed lines the total range, and (+) the outliers for each population. See Table S1 for values and additional analysis. E. Pairwise comparison of ERK activity parameters. For each pair of sister cells, the ratio of the indicated ERK pulse parameters for the earlier- (cell 1) and later- (cell 2) committing cells was calculated. The logarithm of the cell 1/cell 2 ratio was plotted as a histogram to show the frequency of pairs in which the indicated parameter was larger in cell 1 (pink, >0) or in cell 2 (blue, ≤0). P-values were calculated using a paired t-test. Data shown are collated from three independent experiments (n=24, 59, and 140).

Figure 5. Modulation of ERK frequency and amplitude by signaling inhibitors

A. Distributions of pERK at steady state in the presence of 20,000 pg/ml EGF and the indicated concentrations of inhibitors. Histograms represent >5000 cells measured by HCIF after 48 hours of treatment with the indicted inhibitors. For gefitinib-treated cells, vertical red line represents the division between pERK+ and pERK- cells. For PD-treated cells, vertical red lines indicate the approximate median pERK fluorescence intensity (or “amplitude”). B. Live-cell measurements of EKAR-EV signal (CFP/YFP ratio) upon EGFR or MEK inhibition. Mammary epithelial cells growing in the presence of 20,000 pg/ml EGF were treated with the indicated concentrations of inhibitors immediately prior to imaging at time 0. C. Simultaneous modulation of pERK frequency and amplitude. Cells were grown at steady-state with the indicated concentrations of PD and EGF and analyzed by HCIF. Dashed lines indicate the approximate median of the pERK-positive population; a shift in this value indicates decreased amplitude of the ERK-on state. Data shown are from one experiment representative of two or more independent replicates.

Figure 6. Integration of ERK pulses by downstream effectors

A. Decay of endogenous pERK and Fra-1 levels upon MEK inhibition. Cells were grown in full growth medium and treated with MEK inhibitor for the indicated times prior to fixation and HCIF analysis. B. Simulation of pERK and Fra-1 decay upon MEK inhibition. See Supplemental Experimental Procedures and Tables S2 and S3 for details of model construction and simulation. C. Simulation of Fra-1 concentration in response to varying frequencies of ERK activity pulses. D. Simulated steady-state levels of Fra-1 as a function of the fraction of time pERK spends in the “on” state. E. Simulation of slow fluctuations in Fra-1 levels driven by variation in ERK pulse frequency over time. F. Representative individual cell trace of FIRE expression at 50 pg/ml. Cells were cultured as in Fig. 1F. G. Scatter plot of pRb and Fra-1 intensities detected for HCIF for MCF-10A cells growing at steady state (59 hr) with 50 pg/ml EGF. Pink line indicates the threshold for determining pRb+ or pRb- status; gray vertical lines indicate bins dividing Fra-1 into graded expression levels. H. Frequency of pRb+ cells as a function of Fra-1 expression. For each Fra-1 expression bin (gray lines), the fraction of pRb+ cells was determined (red circles); error bars indicate the standard deviation for triplicate samples. The overall Fra-1 distribution is shown in green. Data shown are from one experiment representative of three or more independent replicates.

Figure 7. Signal-response relationship between ERK output and proliferation rate

A. Correspondence between Fra-1 intensity and pERK at steady state. HCIF measurements for Fra-1 intensity were plotted against the fraction of pERK-positive cells under FM conditions (titration of EGF) or median pERK intensity under AM conditions (titration of PD). B. HCIF measurements of Fra-1 intensity and f-pRb+ for combinations of EGF and PD. Values represent the mean, and error bars the standard deviation, for triplicate wells with >5000 cells each. C. Relationship between ERK output (Fra-1 intensity) and f-pRb+ for the measurements shown in (B). Yellow region indicates the segment of the dynamic range over which proliferation changes rapidly as a function of ERK pathway output. D. Representative HCIF images of pERK, Fra-1, and pRb of mammary epithelial cells at steady state (60 hours post-stimulation) in the presence of 20,000 pg/ml EGF and varying concentrations of PD. E. Quantitation of fluorescence for images shown in (D). F,G. Relationship between ERK output (Fra-1 intensity) and f-pRb+ for 184A1 mammary epithelial cells and SUM159 breast cancer cells. Cells were grown in varying concentrations of EGF and MEK inhibitor for 48 hours prior to fixation and staining. pRb and Fra-1 were measured and plotted as in (C). Data shown are from one experiment representative of two or more independent replicates.