The vertebrate segmentation clock drives segmentation by stabilizing Dusp phosphatases in zebrafish

Abstract

Pulsatile activity of the extracellular signal-regulated kinase (ERK) controls several cellular, developmental, and regenerative programs. Sequential segmentation of somites along the vertebrate body axis, a key developmental program, is also controlled by ERK activity oscillation. The oscillatory expression of Her/Hes family transcription factors constitutes the segmentation clock, setting the period of segmentation. Although oscillation of ERK activity depends on Her/Hes proteins, the underlying molecular mechanism remained mysterious. Here, we show that Her/Hes proteins physically interact with and stabilize dual-specificity phosphatases (Dusp) of ERK, resulting in oscillations of Dusp4 and Dusp6 proteins. Pharmaceutical and genetic inhibition of Dusp activity disrupt ERK activity oscillation and somite segmentation in zebrafish. Our results demonstrate that post-translational interactions of Her/Hes transcription factors with Dusp phosphatases establish the fundamental vertebrate body plan. We anticipate that future studies will identify currently unnoticed post-translational control of ERK pulses in other systems.

Keywords: Dusp; ERK; clock; oscillation; pattern formation; post-translational; segmentation; somitogenesis; wave.

Figure 1.. Segmentation clock increases Dusp proteins and triggers their oscillations.

(A) The Clock-dependent Oscillatory Gradient (COG) Model: The segmentation clock reciprocates its oscillations on FGF signaling (ppERK) by inhibiting ERK activity. Spatial fold-change detection (SFC) converts oscillatory ERK activity gradient to discrete positional information. Discrete positional information dynamics are necessary and sufficient for somite segmentation. (B-E) Protein expression dynamics following heatshock induction of her1 clock expression (10 min, blue, 20 min, orange, 30 min, green, and 60 min, magenta). Controls included wild-type embryos without heatshock (light gray), wild-type embryos with 10 min or 60 min heatshock (dark gray), or transgenic embryos without heatshock (black). All embryos were collected out of heatshock simultaneously at the 16-somite stage for immediate fixation. (B) Representative immunofluorescence images of Dusp4. (C) Average Dusp4 levels throughout the heatshock. N=9, 17, 22, 41, 37, 29, and 29 resp. p>0.9999 for non-significant comparisons and p<0.0001 for others. (D) Representative immunofluorescence images of Dusp6. (E) Average Dusp6 levels throughout the heatshock. n=40, 30, 28, 43, 36, 36, 43 and 49 resp. Transgenic no heatshock levels are significantly lower than 20 min heatshock (p=0.0009), 30 min heatshock (p<0.0001), and 60 min heatshock (p=0.0016) conditions. p>0.6053 for non-significant comparisons. (F) Summary of the results shown in (B-E). Orange line: indirect transcriptional activation of dusp6 by ppERK; dashed black arrows: post-translational increase of Dusp4 and Dusp6 by Her1; cyan lines: post-translational regulation of ERK. (G) Schematic diagram showing 12 somite stage embryos sorted into three phases (Phase I, orange, Phase II, purple, Phase III, black) according to their PSM lengths as a proxy for developmental time. (H-I) Dusp4 levels quantified throughout the PSM in clock-intact sibling embryos (n=9, 10, and 11 for 3 phases resp., H) or homozygous clock-deficient mutants (n=11, 9, and 10 for 3 phases resp., I). (J-K) Dusp6 levels quantified throughout the PSM in clock-intact sibling embryos (n=17, 15, and 17, J) or homozygous clock-deficient mutants (n=11, 11, and 12, K). All microscopy images are lateral views with dorsal towards the bottom of the figure. Scale bars are 100 μm. Posterior is left. Lines and shaded error bars indicate mean±s.e.m. Average heatshock levels data are presented as 10^th – 90^th percentile box-whisker plots with individual outlier data points (plus signs represent mean). See also Figure S1, Figure S2 and Figure S4.

Figure 2.. Segmentation clock post-transcriptionally increases Dusp4.

(A) A confocal z-section for smFISH signal from probe against dusp4 RNA (cyan) overlaid with immunostaining of cell membrane against CAAX-mCherry (magenta) and DAPI nuclear (gold) staining. Lateral view of 16 somites stage embryo PSM tissue. (B) Alpha blending 3D projection of confocal z-stacks. (C) dusp4 RNA density per cell is quantified as average number of RNA detected per cell nuclei falling within the ellipsoids sliding (purple) through the PSM and avoiding neural tube (green) and notochord (blue). (D), dusp4 expression gradient through the PSM quantified for 50 min heatshock treated transgenic embryos (red; n=11) and three controls: 1-) no heatshock transgenic embryos (black; n=13), 2-) 50 min heatshock wild-type sibling embryos (dark gray; n=15), and 3-) no heatshock wild-type sibling embryos (light gray; n=19). Data is mean±s.e.m. (E) dusp4 expression density along the PSM, as shown for three positions. All data from individual embryos are shown in box (median and quartiles) and whisker (min to max) plots. p=0.73, 0.49, and >0.9999 for positions from posterior towards anterior. (F-G) Alpha blending 3D projection of confocal z-stacks of smFISH signals from probes against her7 RNA (red) (F) and dusp4 (cyan) RNA (G) in 10 somites stage wild-type embryos. Exemplary images from three flat mounted PSM tissues at different phases of clock expression are shown. Dorsal view. (H-I) Average her7 (H) and dusp4 (I) expression per cells are quantified along the PSM for each clock phase (n=26 in total). Data is mean±s.e.m. (J) Immunofluorescence signal from cell nuclei (gray) and Her7-Venus (red) and Dusp4 (teal) proteins. Dorsal view. (K) Pseudo-kymograph representation of double immunostaining data, every horizontal line represents protein levels from a single embryo along the PSM. Vertical Pseudo-time axis is formed by sorting embryos according to their Her7-Venus clock expression profiles (n=75). Scale bars are 100 μm. Posterior is left. (L) Dusp4 and ppERK (n=15, 18, and 17 for three phases resp., N=3 co-immunostaining in wild-type embryos. ppERK (bottom) and Dusp4 (top) profiles across the PSM (posterior left) from same set of phase-sorted embryos. Protein levels for a posterior PSM position (x=110 μm) is also plotted at the right column as box-whisker plots (10^th – 90^th percentile) indicating oscillatory signal experienced by posterior PSM cells within a single somite stage. See also Figure S1 and Figure S4.

Figure 3.. Clock proteins physically interact with and stabilize short-lived Dusp4.

(A) Wild-type and her1 heatshock overexpression transgenic line embryos are heat shocked for 1hr and treated with 100 μg/mL cycloheximide to block protein translation right after the heatshock. CHX treated embryos are fixed 8, 10, 12, 14, 16, 20, and 24 min (B) Dusp4 levels are quantified by immunofluorescence in the posterior PSM (N=2, n= 25, 18, 16, 22, 17, 28, 26 for wild-type and 19, 11, 14, 13, 13, 17, 17 for transgenic embryos). (C) In bimolecular fluorescence complementation assay, two tested proteins (magenta and teal) are fused to complementary fragments of Venus protein (yellow). When proteins interact, Venus signal is observed. (D-K) Representative maximum intensity projection confocal images from embryos injected with Her7-Venus^1–154 plus Hes6-Venus^155–239 RNAs (n=38, D), Her7-Venus^1–154 plus Venus^155–239 RNAs (n=25, E), Her7-Venus^1–154 plus Dusp4-Venus^155–239 RNAs (n=63, F), Her7-Venus^1–154 plus Dusp4^mutant-Venus^155–239 RNAs (n=36, G), Venus^1–154 plus Dusp4-Venus^155–239 RNAs (n=28, H), Her7-Venus^1–154 plus Hes6-Venus^155–239 RNAs (n=38, D), Her7-Venus^1–154 plus Dusp6-Venus^155–239 RNAs (n=33, I), Venus^1–154 plus Dusp6-Venus^155–239 RNAs (n=28, J), Her7-Venus^1–154 plus Dusp1-Venus^155–239 RNAs (n=38, H). Scale bar is 200 μm. (L) Quantification of median Venus fluorescence signal from wild-type embryos with all injections. 10^th – 90^th percentile box and whisker plots with individual dots for outliers. All significant comparisons are p<0.0001 and all non-significant comparison are p>0.9999 using Kruskal-Wallis ANOVA test. (M) Regulatory network as envisioned from obtained data. Orange lines: transcriptional activation or repression; black arrows: post-translational activation; cyan blocking bars: post-translational inhibition. (N) Clock (red), Dusp4 (purple), ppERK (green), and Dusp6 (orange) profiles are shown for two phases (U-domain (light colors) and midPSM (dark colors) corresponding to 20 min difference) of the clock in simulations. See also movie S1. Posterior is left. Error bars indicate mean±s.e.m. See also Figures S3 and Figure S4, and Table S1.

Figure 4.. Dusp activity is necessary for the clock-dependent ppERK inhibition and successful somite segmentation.

(A) 16 somites stage embryos were fixed after 90 min of 20 μM BCI, Dusp inhibitor, treatment during which various durations (0–50 min) of heatshock applied. (B-E) Her1 clock induction driven by heatshock as quantified from HA-tag immunofluorescence for vehicle (B, n=13, 28, 15, 23, 15 resp. For 50 min heatshock wild-type p>0.9999, 10 min heatshock p<0.0001, 30 min heatshock p=0.0002, and 50 min heatshock p=0.0001, all compared versus no heatshock transgenic fish) and BCI treated embryos (D, n=25, 17, 20, 15, 18 resp. For 50 min heatshock wild-type p=0.9988, 10 min heatshock p=0.0001, 30 min heatshock p=0.0094, and 50 min heatshock p<0.0001, all compared versus no heatshock transgenic fish) and Dusp4 levels throughout the heatshock for vehicle (C, n=15, 28, 20, 25, 17 resp. For 50 min heatshock wild-type p=0.3641, 10 min heatshock p=0.0022, 30 min heatshock p=0.0398, and 50 min heatshock p=0.0227, all compared versus no heatshock transgenic fish) and BCI treated embryos (E, n=25, 19, 25, 20, 20 resp. For 50 min heatshock wild-type p=0.5172, 10 min heatshock p<0.0001, 30 min heatshock p=0.0227, and 50 min heatshock p=0.0039, all compared versus no heatshock transgenic fish). Heatshock data are presented as Tukey box-whisker plots with individual outlier data points (plus signs represent mean). (F) Immunofluorescence signals from antibody against ppERK for BCI and vehicle treated embryos. Scale bar is 100 μm. Lateral view. Posterior left. Dorsal bottom. (G-H) Nuclear ppERK profile through posterior PSM for DMSO (G; n=33, 26, and 22 resp.) or BCI (H; n=27, 29, and 24 resp.) treated embryos. Posterior is left in PSM plots. Lines and shaded error bars indicate mean±s.e.m. (I) 20 μM BCI treatment for 45 min beginning at 14 somites stage. (J) xirp2a staining of somite boundaries at 45 hours post-fertilization (hpf) fish. Failed somite segmentation due to Dusp inhibition is highlighted with stars. (K) Quantification of broken boundaries for 45 min inhibition of Dusp activity (n=34). (L) Pulsatile inhibition of Dusp activity with 20 μM BCI for 15 min followed by 20 min washouts. (M) Quantification of broken boundaries for 6× pulses of BCI inhibition of Dusp activity at various concentrations (n=67, 96, 62, and 56 for DMSO, 10 μM, 15 μM, and 20 μM resp.). (N) Description of dusp4 ^ci305 mutant fish. (O) xirp2a staining of somite boundaries at 45 hpf for mutant fish. (P) xirp2a staining of somite boundaries for dusp4 mutant (top) and wild-type (bottom) fish treated with low dose (10 μM) BCI for 90 min from 14 somite stage. (Q) Quantification of broken boundaries for DMSO treated mutants or 90 min low dose inhibition of Dusp activity in dusp4 mutants and wild-type (p<0.0001, n=108, 87, and 119 resp.). Number of broken boundaries are reported as 10^th-90^th percentile box and whisker plots. Plus sign represents mean. See also Figure S2 and Figure S4.