A quantitative method for staging mouse embryos based on limb morphometry

Abstract

To determine the developmental stage of embryonic mice, we apply a geometric morphometric approach to the changing shape of the mouse limb bud as it grows from embryonic day 10 to embryonic day 15 post-conception. As the ontogenetic sequence results in the de novo emergence of shape features not present in the early stages, we have created a standard ontogenetic trajectory for limb bud development - a quantitative characterization of shape change during limb morphogenesis. This trajectory of form as a function of time also gives us the reverse function: the ability to infer developmental stage from form, with a typical uncertainty of 2 h. We introduce eMOSS (embryonic mouse ontogenetic staging system) as a fast, reliable, convenient and freely available online tool for staging embryos from two-dimensional images of their limb buds, and illustrate its use in phenotyping early limb abnormalities.

Keywords: Limb growth; Morphometrics; Mouse development; Osterix; Sonic hedgehog; Staging system; eMOSS.

Conflict of interest statement

Competing interestsThe authors declare no competing or financial interests.

© 2018. Published by The Company of Biologists Ltd.

Figures

Fig. 1.

The limb developmental trajectory. (A) Superposition of data points for the whole dataset separated into equally spaced intervals of 12 h, from 249 to 381 h. The red lines represent the average shapes, for each of the harvesting age groups. (B) Interpolation of one reference shape to the next in the time sequence. The red lines indicate the shapes that correspond to the experimental averages for each age group.

Fig. 2.

The staging system. The result of the global fit of a limb as shown in the web application. The scaled curve (see Eqn 1) and the corresponding probability density function are shown in the panels on the right.

Fig. 3.

Distributions of different fit variables. (A) Distribution of the fit score for the whole dataset of hindlimbs. (B) Distribution of the estimated error as a function of the staging time. (C) Distribution of differences of the stage estimates for the left and the right limbs (dLR=aL−aR) as a function of the average staging time (aL+aR)/2. (D,E) Distributions of dLR/σd in the ranges below and above 290 h (mE12:02). Arrows indicate the position of the Gaussian peak (at −0.5 and −0.1, respectively).

Fig. 4.

Distributions of the measured stages in the dataset. The distribution of all 864 samples of limbs. The y-axis gives the estimated stage and the x-axis corresponds to the 12 harvesting age groups. The inset plot shows the estimated stage of limbs in the age group harvested at 273 h (11 days and 9 h post-conception) for different litters along the horizontal axis. Each vertical set of color-coded points represents a litter and each point represents the limb of one individual.

Fig. 5.

Comparison of Osx-GFP mice, arranged according to harvesting time and staging time. (A) Image of Osx-GFP mouse with fluorescent expression of Osx in ossification centers (left lateral view). (B,C) Each image shows the lateral view of an embryo's head with the eye (dashed line) in the bottom left quadrant. Ossification centers of the frontal and parietal bones are outlined to make clear the region of Osx expression. This sample includes images from three different litters collected at 6 h intervals (E12.75, E13.0 and E13.25). Yellow numbers (in the bottom right quadrant of each panel) show order of the embryos when their sequence is determined by the harvesting age (i.e. top row, B). These numbers are reproduced in the bottom row (C) once the sequence of embryos has been re-ordered based on the morphometric staging system, to highlight how the sequence has changed. The embryos exhibit substantial variation in the rate of development but ordering specimens according to estimated developmental stage (using eMOSS) provides a coherent view of the expected expansion of Osx expression over developmental time.

Fig. 6.

Correspondence of the staging outcome and other morphological quantities. (A) The number of somites (red in schematic) of the embryos as a function of the age of the hindlimb. Blue points correspond to data from Richardson et al. (2014); error bars represent the range of variation in the somite counting. (B) Calibration curve between the response of the staging system algorithm on hind- and forelimbs. Forelimb is marked in red in schematic. In A and B, shaded green areas represent the ±1 sigma error in the linear fit. (C) Distribution of the fit score as a function of age for limbs presenting the Ssq mutation. Red circles correspond to limbs that carry the mutated gene, superimposed to the wild-type distribution in blue. Red circles at low values correspond to limbs that carry the genetic mutation but have a normal phenotype (as previously reported by Sharpe et al., 1999). The dashed line corresponds to a linear fit to points above , where the phenotype is recognizable by visual inspection. Insets show example images of two limbs.