Research ArticleVASCULAR BIOLOGY

Endothelial cells decode VEGF-mediated Ca2+ signaling patterns to produce distinct functional responses

See allHide authors and affiliations

Sci. Signal.  23 Feb 2016:
Vol. 9, Issue 416, pp. ra20
DOI: 10.1126/scisignal.aad3188
  • Fig. 1 VEGF promotes multiple Ca2+ waveforms.

    (A) Time series showing YC3.6 FRET detection of cytosolic Ca2+ concentration in individual PAECs after VEGF stimulation. The FRET signal at each time point was normalized to that measured at the starting time. Scale bar, 50 μm. AU, arbitrary unit. (B) Representative traces illustrating different Ca2+ waveforms triggered by VEGF stimulation. Inset shows the three basic waveforms observed (LP, RS, and NR). (C) Distribution of Ca2+ waveforms as a function of VEGF concentration in PAECs and HUVECs. Data are means ± SEM for n = 3 independent experiments with greater than 100 cells for each concentration of VEGF. (D) Distribution of migration distances as a function of waveform after stimulation with VEGF for 1 hour (n = 92 cells from n = 3 independent experiments). (E) The distribution of time-averaged NFAT activation as a function of Ca2+ waveform in PAECs after stimulation with VEGF for 3 hours (n = 76 cells from n = 4 independent experiments). NFAT activation was determined as the ratio of nuclear to cytoplasm NFAT intensity. (F) HUVECs were starved overnight and exposed to VEGF and the NFAT inhibitor VIVIT for 24 hours. Nuclear BrdU (5-bromo-2′-deoxyuridine)incorporation was used to track cell cycle progression (Student’s t test, **P < 0.05, *P < 0.085, n = 3 independent experiments). (G) Schematic summarizing the role of Ca2+ signaling in phenotype selection. LP and RS signals are interpreted differently by cells resulting in either migration or NFAT signaling, which can promote cell proliferation.

  • Fig. 2 NFAT signaling is selectively enhanced by the RS Ca2+ waveform.

    (A) Example traces illustrating “staircase-like” NFAT response to Ca2+ spikes after stimulation with VEGF. (B) Representative images (from movie S3) showing simultaneous Ca2+ (FRET) amplitude and NFAT translocation for cells with RS (I) and LP (II) waveforms. VEGF was added after 1.5 min. Scale bars, 50 μm. (C and D) Ca2+ and NFAT regulation for the cells shown in (B) having RS (C) and LP waveforms (D). Arrows show the times corresponding to the sample images. (E) Comparison of average NFAT activation as a function of average increase in Ca2+ amount for individual cells exhibiting NR, LP, or RS waveforms. (F) Average NFAT activation as a function of the average spike frequency. The horizontal gray line indicates the boundary between the groups with basal and enhanced activation as determined by k-means analysis. The vertical lines mark the boundaries between cells exhibiting only basal NFAT activation, both basal and enhanced activation, and only enhanced NFAT activation. The red line illustrates a sigmoidal fit to the RS cell data [f(x) = 2.07 + 1/(0.1 + exp(−(x − 16.15))), R2 = 0.865]. Data in (E) and (F) were collected for 3 hours after application of VEGF (n = 76 cells from n = 4 independent experiments).

  • Fig. 3 Signaling that regulates enhanced cell motility requires Ca2+ concentrations to be continuously sustained above a threshold.

    (A) Migration as a function of the minimum Ca2+ concentrations reached during the later portion of Ca2+ signaling (20 to 60 min). Cells that returned to baseline were separated from those with sustained Ca2+ concentrations (vertical gray line). The horizontal gray line indicates the boundary between basal and enhanced motility groups as determined by k-means analysis (n = 92 cells from n = 3 independent experiments). (B) Ca2+ traces for cells displaying LP waveforms that either fully adapt back to baseline (red) or maintained low amounts of signaling (blue) are compared to NR cells, used as a control (black). Corresponding migration is shown in the inset. Data were collected for 1 hour after simulation with VEGF. (C) Migration as a function of integrated Ca2+ concentrations. (D) Representative images showing that cells with RS waveforms (movie S5) did not change shape and displayed limited actin recruitment to small lamellipodia and filopodia (white arrows). Cells exhibiting LP waveforms (movie S4) markedly changed shape and formed large lamellipodia with greater actin recruitment (white arrows). (E) Distribution of migration values for cells exhibiting LP waveforms after VEGF stimulation alone (black, n = 58 cells from n = 3 independent experiments) and with 30-min preincubation with the MLCK-specific inhibitor peptide 18 (red, n = 57 cells from n = 3 independent experiments). (F) MLCK activation corresponding to Ca2+ waveforms with LP (top) and RS (bottom) was synchronized in time. Traces are representative of n = 64 cells from three independent experiments. (G) Average MLCK activation, taken after 1 hour of VEGF stimulation, shown as a function of the average amount of Ca2+ for both LP and RS (n = 64 cells from n = 3 independent experiments).

  • Fig. 4 Waveform discrimination through signal “ratcheting.”

    (A) Schematic description of Ca2+ waveform discrimination that translates VEGF input into specific phenotypes. The inset depicts the theoretically possible activities of Ca2+ effectors responding to Ca2+ spikes (red) in analogy to a ratchet. An effective ratchet (orange) resists deactivation, and an ineffective ratchet (blue) readily deactivates. (B) Conceptual model of information processing by a ratchet-like pathway (left, orange) and a non-ratcheting pathway (right, blue). The red lines and arrows denote how the sustained Ca2+ signaling after an RS waveform is translated into a response, and the black lines show the same for LP. The threshold for triggering Ca2+ effector activation differs for NFAT and cell motility phenotypes (dashed lines). (C) Illustration of the probabilistic model structure when considering the distributions of f and td either dependent on VEGF (“expanded model,” black) or independent of VEGF (“simplified model,” red). Also shown is the model structure without stochasticity downstream of waveform (w) selection (“idealized,” blue). (D) Predicted percentage of each enhanced phenotype, either NFAT activation (N, left) or migration (m, right). Lines are colored corresponding to the model illustrations in (C). Experimental measurements are denoted by the green triangle.

  • Fig. 5 A combination of stochastic and deterministic decoding governs phenotype selection in a gradient of VEGF.

    (A) Migration trajectories from cells simulated in a 0.0625 ng ml−1 μm−1 gradient of VEGF. Cells were assumed to reevaluate decisions every 15 min for these simulation results and those described below. (B) Predicted migration velocity vmig shown as a function of td for cells simulated in a gradient of VEGF (correlation coefficient of 0.54, P < 0.01). Each point shows data from an individual cell and is colored by starting VEGF concentration. Histograms showing the distribution of VEGF concentrations at discrete values of td are shown at the top of the plot and are also colored by VEGF concentration. (C) Experimentally measured vmig as a function of td for cells placed in a microfluidic gradient of VEGF (0.0625 ng ml−1 μm−1) (correlation coefficient of 0.59, P < 0.01) (n = 91 cells from n = 3 independent experiments). Range of simulation results is superimposed (gray envelope) for comparison. (D) Predicted NFAT activation for cells as a function of the duration of RS waveforms in a gradient of VEGF.

  • Fig. 6 Repeated Ca2+ spikes reduce tip cell migration during angiogenesis in developing zebrafish.

    (A) Time-lapse images showing angiogenic sprout formation originating from the dorsal aorta (DA). GCaMP6m was used to monitor cytoplasmic Ca2+ regulation and is shown in green. mCherry was used to determine the location of vascular cells and is shown in red. Sprouts eventually split and fuse to form the dorsal longitudinal anastomotic vessel (DLAV) and remain as intersegmental vessels (Se). Examples of migrating tip cells are labeled TC1 and TC2. Scale bars, 50 μm. (B) Detailed Ca2+ signaling and migration profiles for tip cells TC1 and TC2 are also indicated in (A). Black circles indicate Ca2+ peaks that were distinguished from noise. The tip cell migration distance (middle panel) is measured from the tip of the sprouting vessel to the branch point along the DA. The migration velocity (top panel) is calculated from the tip cell migration distance over time (determined as a rolling average, see Materials and Methods). (C) Maximum tip cell migration velocity at each observed Ca2+ spike frequency (n = 8 angiogenic sprouts from n = 2 zebrafish embryos). (D and E) Ca2+ signaling during cell division. Individual frames from the image tile in (E) are indicated at the top of the plot. (E) Image tile showing stalk cell division (white arrow) during sprout formation. Scale bars, 50 μm.

Supplementary Materials

  • www.sciencesignaling.org/cgi/content/full/9/416/ra20/DC1

    Text S1. Detailed description of the probabilistic model.

    Fig. S1. Distribution of averaged Ca2+ concentrations.

    Fig. S2. Additional analysis of cell motility.

    Fig. S3. Probabilistic model simulation of responses to spatially uniform VEGF.

    Fig. S4. Analysis of cell migration in a microfluidic gradient.

    Fig. S5. Probabilistic model simulation of cells in a gradient of VEGF.

    Movie S1. Time-lapse video showing NR, RS, and LP waveforms in PAECs.

    Movie S2. Time-lapse video showing NR, RS, and LP waveforms in HUVECs.

    Movie S3. Time-lapse video showing simultaneous Ca2+ and NFAT signaling.

    Movie S4. Cells exhibiting LP waveforms formed large lamellipodia with pronounced actin translocation.

    Movie S5. Cells exhibiting RS waveforms displayed only limited actin recruitment and small filopodia.

    Movie S6. Angiogenic sprout formation in zebrafish.

  • Supplementary Materials for:

    Endothelial cells decode VEGF-mediated Ca2+ signaling patterns to produce distinct functional responses

    David P. Noren, Wesley H. Chou, Sung Hoon Lee, Amina A. Qutub, Aryeh Warmflash, Daniel S. Wagner, Aleksander S. Popel,* Andre Levchenko*

    *Corresponding author. E-mail: andre.levchenko{at}yale.edu (A.L.); apopel{at}jhu.edu (A.S.P.)

    This PDF file includes:

    • Text S1. Detailed description of the probabilistic model.
    • Fig. S1. Distribution of averaged Ca2+ concentrations.
    • Fig. S2. Additional analysis of cell motility.
    • Fig. S3. Probabilistic model simulation of responses to spatially uniform VEGF.
    • Fig. S4. Analysis of cell migration in a microfluidic gradient.
    • Fig. S5. Probabilistic model simulation of cells in a gradient of VEGF.
    • Legends for movies S1 to S6

    [Download PDF]

    Technical Details

    Format: Adobe Acrobat PDF

    Size: 1.17 MB

    Other Supplementary Material for this manuscript includes the following:

    • Movie S1 (.mov format). Time-lapse video showing NR, RS, and LP waveforms in PAECs.
    • Movie S2 (.mov format). Time-lapse video showing NR, RS, and LP waveforms in HUVECs.
    • Movie S3 (.avi format). Time-lapse video showing simultaneous Ca2+ and NFAT signaling.
    • Movie S4 (.avi format). Cells exhibiting LP waveforms formed large lamellipodia with pronounced actin translocation.
    • Movie S5 (.avi format). Cells exhibiting RS waveforms displayed only limited actin recruitment and small filopodia.
    • Movie S6 (.avi format). Angiogenic sprout formation in zebrafish.

    Citation: D. P. Noren, W. H. Chou, S. H. Lee, A. A. Qutub, A. Warmflash, D. S. Wagner, A. S. Popel, A. Levchenko, Endothelial cells decode VEGF-mediated Ca2+ signaling patterns to produce distinct functional responses. Sci. Signal. 9, ra20 (2016).

    © 2016 American Association for the Advancement of Science