Bicoid gradient formation mechanism and dynamics revealed by protein lifetime analysis

Abstract

Embryogenesis relies on instructions provided by spatially organized signaling molecules known as morphogens. Understanding the principles behind morphogen distribution and how cells interpret locally this information remains a major challenge in developmental biology. Here, we introduce morphogen-age measurements as a novel approach to test models of morphogen gradient formation. Using a tandem fluorescent timer as a protein age sensor, we find a gradient of increasing age of Bicoid along the anterior-posterior axis in the early Drosophila embryo. Quantitative analysis of the protein age distribution across the embryo reveals that the synthesis-diffusion-degradation model is the most likely model underlying Bicoid gradient formation, and rules out other hypotheses for gradient formation. Moreover, we show that the timer can detect transitions in the dynamics associated with syncytial cellularization. Our results provide new insight into Bicoid gradient formation and demonstrate how morphogen-age information can complement knowledge about movement, abundance, and distribution, which should be widely applicable to other systems.

Keywords: Drosophila melanogaster; SPIM; embryogenesis; fluorescent timers; morphogen gradient.

Figure 1. Protein age can distinguish alternative models of morphogen gradient formation with similar concentration profiles

1. Cartoon of the morphogen hypothesis: a spatially varying concentration of signaling molecule can result in precise readout of positional information.

2. Outline of models considered. Morphogen RNA (grays) and protein (green dots) distribution are shown in early (top) and cycle 14 (middle) rows for each model considered. The magnitude of protein movement is represented by length of green arrows. Degradation of protein/RNA is represented by green/gray arrows. Bottom row shows a schematic of morphogen RNA and protein concentration profiles in cycle 14.

3. Normalized Bcd concentration profiles for the models considered in (B) at time = 2.5 h. Inset: same on log scale. See Appendix and Appendix Fig S1 for extended discussion of models and parameters.

4. Data points correspond to mean output of stochastic Monte Carlo simulation results for the average protein age as function of position, 2.5 h after initiation (see Appendix and Materials and Methods for details). Colored lines correspond to theoretical predictions for protein age in each model (Appendix Section B).

5. As (D) but showing solutions over larger parameter space. Solid line represents mean solution, and dashed lines represent 1 SD. Parameter range described in text. SDD: synthesis, diffusion, degradation model, NucSh: nuclear shuttling model, RNA‐grad: RNA gradient model, and RNA‐diff: RNA diffusion model.

Figure 2. tFT‐Bcd reporter reflects average protein age

1. A

   Schematic of tFT‐Bcd reporter with mCherry and sfGFP fluorophores.

2. B

   Cartoon of the fluorescent protein maturation states in the tFT‐Bcd reporter. m_g and m_r represent the sfGFP and mCherry maturation rates, respectively (note, m_r is an effective rate as mCherry has a two‐step maturation process).

3. C–E

   Examples of embryos expressing the tFT‐Bcd reporter with (C) mCherry‐sfGFP‐Bcd, (D) fmCherry‐sfGFP‐Bcd, and (E) mCherry‐sfGFP. For all, (i) images of “shells” of embryos in early n.c. 14. 3D images of the embryos were generated, and then, the interior of the embryo was erased, leaving only the embryo cortex to improve clarity (see Materials and Methods for more details). (ii) Mean AP intensity profile of each color for the embryo in (i). Shade region represents ± 1 SD. Inset shows same profiles after multiplication of red intensity by constant factor. Data binned into 10‐μm bins (n = 4 embryos). (iii) Ratio of green over red signal, reflecting protein age. The thin lines represent individual embryos, while the thick solid line is the mean. The solid dashed line depicts the mean green/red ratio for a line with the tFT‐Bcd reporter but lacking endogenous Bcd (Bcd^E1 mutant, n = 4. The scale is the same for all embryos. The scale bar is 50 μm long).

Source data are available online for this figure.

Figure 3. The SDD model is most consistent with the tFT‐Bcd reporter

1. Model fitting to the experimental tFT‐Bcd ratio in early n.c. 14 as function of AP position for the different models outlined in Fig 1B. Black line represents mean measured ratio of tFT‐Bcd reporter with mCherry‐sfGFP‐Bcd in y/w flies. The shadowed area represents ± 1 SD. Inset is same as main panel, except for the tFT‐sfGFP‐fmCherry‐Bcd reporter.

2. Increasing protein age results in increasing tandem reporter ratio. The fluorophore states of the tFT reporter are shown in the inset: (i) immature; (ii) only sfGFP matured; (iii) only intermediate mCherry matured; (iv) sfGFP and intermediate mCherry matured; (v) only mCherry matured; and (vi) both mCherry and sfGFP matured. k_g, k_r1, k_r2 denote the maturation rates of sfGFP, intermediate mCherry, and fully matured mCherry, respectively. The different curves correspond to sets of values for k_g, k_r1, k_r2 selected from Gaussian distributions with half‐time means 27, 40, and 9 min and standard deviations 2.7, 4, and 0.9 min, respectively.

3. Least‐squares fitting of each model in n.c. 14 to: (left) the sfGFP only; (middle) the tFT‐Bcd reporter ratio only; and (right) both the sfGFP profile and the tFT‐Bcd reporter ratio simultaneously. Each fit is performed with the fluorophore maturation rates drawn from Gaussian distributions as in (A). See Materials and Methods for details of fitting.

4. Quality of SDD model fitting to tFT‐Bcd ratio (shown in A) for different Bcd diffusion coefficients and lifetimes. For each pair of these parameters, the fluorescence correction ε is left as a free parameter for fitting. The colormap is the log of the fitting function (Materials and Methods). The white lines correspond to parameter combinations that produce a gradient with a length constant λ = 75 μm (solid) and λ = 95 μm (dashed).

Source data are available online for this figure.

Figure 4. Bcd is degraded by the proteasomes in vivo

1. A–C

   Injection of the proteasome inhibitor MG132 in embryos increases both the amount and the average age of Bcd (gray, open squares, N = 10). Control embryos were injected with DMSO (black, filled circles, N = 9). All embryos are y/w and were injected while in cycle 14 and imaged 60 min later. The shaded regions correspond to the mean ± SD.

2. D

   The age of Bcd and its gradient are unaltered in fsd mutant. Comparison of a representative y/w embryo (left) and loss‐of‐function mutation fsd^KG02393 (right) are shown. Both embryos have the tFT‐Bcd construct in the same locus, are in early n.c. 14, were imaged side‐by‐side under identical conditions, and are displayed with the same settings. Embryo “shells” as in Fig 2 are displayed. The scale bar corresponds to 50 μm.

3. E

   Quantification of embryos as in (D). Fluorescence intensity from sfGFP (green) and mCherry (magenta) for wt (N = 8, filled circles) and fsd (N = 6, open squares) at early cycle 14. Inset: tFT‐Bcd ratio for wt (black) and FSD (gray) embryos. The shaded regions depict the mean ± SD.

Source data are available online for this figure.

Figure 5. The average age of Bcd proteins increases continuously during the cleavage divisions and the gastrulation

Time‐course of a representative y/w embryo with the tFT‐Bcd reporter from the division cycle 10 to the early gastrulation. A movie of the same embryo with 6‐min time resolution is available as Movie EV1.

1. SPIM images of the embryo early development. Embryo “shells” as in Fig 2 are displayed. Left: Green fluorescence, center: Red fluorescence, right: Intensity‐weighted ratiometric image (see Materials and Methods) of the mCherry/sfGFP ratio, reporting average protein age. The scale bar corresponds to 50 μm.

2. Quantification of the same embryo as is (A). Each line represents the fluorescence profile along the anterior–posterior axis of the embryo for a time frame (separated by 6 min). In the lower panel, where the mCherry/sfGFP ratio is shown, the posterior half of the embryo is not shown because the signal drops to the background levels (due to softer imaging conditions for the long term time‐course).

3. The same data are shown as a function of time for different embryo segments. The shaded regions show the mean value ± SD.

Source data are available online for this figure.

Figure 6. At the onset of cellularization Bcd production and degradation rates decrease

1. Simulation of SDD model, where after reaching steady state (black line, insets), either production was stopped (left) or degradation increased (right). The gray arrow shows the direction of time, and the color of the lines reflects the mCherry/sfGFP ratio (y‐axis).

2. As (A), where the SDD model was run until reaching steady state (red circle), and then, production and/or degradation parameters were perturbed (see legend in the plot). The y‐axis shows the mean mCherry/sfGFP ratio (proxy for average protein age) in the anterior region of the embryo (0–25% of embryo length), and the x‐axis represents the mean green fluorescence intensity in the same region (age/levels diagram).

3. During the cleavage divisions, the embryos define reproducible trajectories in the age/levels diagram. Experimental data from quantification of movies of y/w embryos as in Fig 4. The trajectories of four embryos from n.c. 10 to gastrulation are shown (each embryo depicted with a different symbol and in a different panel). The time resolution of the time‐course was 6 min. The data points are colored according to the division cycle. A time window of approximately 30 min where the trajectories stay (compatible with a steady state) is marked with dotted gray circles. All imaged embryos that managed to start gastrulation were included in the analysis. Embryo 2 is the same from Fig 4.

4. The normalized trajectories from the embryos in panel (C) overlap. The intensity values were normalized for each embryo to the intensity values at the mid n.c. 14 (showed in gray circles in panel C). The gray line depicts the average trajectory.

Source data are available online for this figure.