A conserved developmental patterning network produces quantitatively different output in multiple species of Drosophila

Abstract

Differences in the level, timing, or location of gene expression can contribute to alternative phenotypes at the molecular and organismal level. Understanding the origins of expression differences is complicated by the fact that organismal morphology and gene regulatory networks could potentially vary even between closely related species. To assess the scope of such changes, we used high-resolution imaging methods to measure mRNA expression in blastoderm embryos of Drosophila yakuba and Drosophila pseudoobscura and assembled these data into cellular resolution atlases, where expression levels for 13 genes in the segmentation network are averaged into species-specific, cellular resolution morphological frameworks. We demonstrate that the blastoderm embryos of these species differ in their morphology in terms of size, shape, and number of nuclei. We present an approach to compare cellular gene expression patterns between species, while accounting for varying embryo morphology, and apply it to our data and an equivalent dataset for Drosophila melanogaster. Our analysis reveals that all individual genes differ quantitatively in their spatio-temporal expression patterns between these species, primarily in terms of their relative position and dynamics. Despite many small quantitative differences, cellular gene expression profiles for the whole set of genes examined are largely similar. This suggests that cell types at this stage of development are conserved, though they can differ in their relative position by up to 3-4 cell widths and in their relative proportion between species by as much as 5-fold. Quantitative differences in the dynamics and relative level of a subset of genes between corresponding cell types may reflect altered regulatory functions between species. Our results emphasize that transcriptional networks can diverge over short evolutionary timescales and that even small changes can lead to distinct output in terms of the placement and number of equivalent cells.

Figure 1. Schematic of the regulatory relationships between 13 AP patterning network genes.

In this paper, we examine the expression patterns of a subset of anterior/posterior (AP) patterning genes; general information on the regulatory relationships between the genes in our dataset is shown. Not all regulatory relationships have been precisely defined; therefore, the network is shown as a hierarchy between classes of genes (maternal, gap, terminal and pair-rule) with known interactions within classes. For example, the gap genes are known to cross-repress one another , and the primary pair-rule genes are thought to regulate the secondary pair-rule genes , .

Figure 2. Average gene expression patterns for 13 AP patterning genes in D. melanogaster, D. yakuba, and D. pseudoobscura are qualitatively similar.

Fluorescent in situ hybridization, 2-photon microscopy and image analysis were used to measure the expression of 13 AP patterning genes at cellular resolution in D. melanogaster (see [7]), D. yakuba and D. pseudoobscura over 6 time points during the hour prior to gastrulation. Because the embryo is bilaterally symmetric, one half of a cylindrical projection (an “unrolled” embryo, dorsal side up and anterior to the left) is shown for each time point. High expression is red; low expression is black. Bcd is not expressed during the last three time points in D. yakuba and D. pseudoobscura and therefore not shown. Staining for cad was consistently low level and uniform in D. pseudoobscura and therefore not included in the dataset.

Figure 3. Blastoderm embryos of the three Drosophila species vary in size, shape, and number of nuclei.

(A) Silhouettes of species-specific embryo models are shown in both lateral view (anterior left, dorsal up,) and cross section (dorsal up), D. melanogaster (blue), D. yakuba (orange), D. pseudoobscura (green). (B) For each embryo in our datasets, the surface area was calculated and compared to the number of nuclei. Though both of these values vary within and between species, the relationship between them is linear. (C) Density patterns for early, middle and late stage blastoderm embryos are displayed as heat maps (red is high density, blue is low density), with corresponding contours on 2D cylindrical projections of the embryos. Anterior is to the left, posterior to the right. D = dorsal, L = left, V = ventral, R = right. In pairwise comparisons (Figure S2) the densities are statistically distinct across all three species with D. melanogaster and D. pseudoobscura containing the largest areas of similar nuclear density.

Figure 4. The expression distance metric can be used to search for corresponding cells.

A schematic of the algorithm to identify corresponding cells is shown. For a given query cell, the 30 nearest cells in 3D space in the target embryo are identified. The expression distance between the query cell and each of these target cells is calculated. The best corresponding cell is the target with the lowest expression distance score. This is often not the target cell nearest to the query cell in 3D space.

Figure 5. Even-skipped expression varies in relative position and intensity.

The expression distance score for each cell is plotted on a 2D representation of the embryo and the underlying gene expression profiles are illustrated in insets where gene expression is represented as a line trace over time. (1st row) For each D. melanogaster query cell, the expression distance score of the nearest target cell in D. yakuba (left) and D. pseudoobscura (right) is shown. (2nd row) The expression distance score for the best matched cell within the nearest 30 cells for both D. yakuba and D. pseudoobscura is shown. High expression distance scores, indicating poor matches, are darker. All cells scoring above 0.7 are colored the darkest blue; when the maximum value exceeds 0.7, the maximum value amongst all cells is reported at the top of the color map. Representative D. melanogaster cells are labeled, and their expression profiles are shown in the insets compared to their matches in D. yakuba or D. pseudoobscura (D. melanogaster cell in red, D. yakuba or D. pseudoobscura cell in blues - dark blue for nearest cell, light blue for best cell after local search). For each representative D. melanogaster cell, we list the label in the figure (a or b), the cell ID number, the target embryo to which it was matched (D. yakuba or D. pseudoobscura) and the expression distance score to the nearest target cell and the best matched target cell: a. 4314, D. yakuba, 0.719, 0.141; b. 5232, D. yakuba, 0.633, 0.557; a. 4314, D. pseudoobscura, 0.611, 0.066; b. 5232, D. pseudoobscura, 0.966, 0.524. (3rd row) For each D. melanogaster query cell, the distance and direction to the average position of the top 10 best corresponding target cells is shown. The correspondence is shown with a line that starts at the position of the query cell, and ends at the average position of the target cells. The end of the line is indicated with a black dot. Because the 2D projection distorts actual distance in 3D, the lines are color-coded to indicate actual distance traversed in 3D. Dark blue is a large distance, yellow is a small distance. (4th row) The distribution of expression distance scores using only the nearest cell (grey) and best-matched cell within the nearest 30 (blue) are shown; we plot the root of the expression distance score to separate values near zero. The distribution of expression distance scores narrows and the mean and median decrease after a local search (Table S3). To establish the significance of the calculated differences, we assembled two atlases from the D. melanogaster dataset, and compared these two atlases to each other (dotted lines).

Figure 6. The majority of cellular gene expression profiles are conserved.

The expression distance score for each cell is plotted on a 2D representation of the embryo and the underlying gene expression profiles are illustrated in graphs below, where gene expression is represented as a line trace over time for each gene in the dataset. (1st row) For each D. melanogaster query cell, the expression distance score of the nearest target cell in D. yakuba (left) and D. pseudoobscura (right) is shown. (2nd row) The expression distance score for the best matched cell within the nearest 30 for both D. yakuba and D. pseudoobscura is shown. The same representative cells from the top panel are indicated. High expression distance scores, indicating poor matches, are darker. All cells scoring above 2.5 are colored the darkest blue; when the maximum value exceeds 2.5, the maximum value amongst all cells is reported at the top of the color map. Representative D. melanogaster cells are labeled, and their expression profiles are shown in detail at the bottom. For each representative cell, we list the label in the figure (a, b, or c), the target embryo to which it was matched (D. yakuba or D. pseudoobscura) and the expression distance score to the nearest target cell and the best matched target cell: a, D. yakuba, 3935, 3.374, 1.018; b, D. yakuba, 4583, 0.881, 0.683; c, D. yakuba, 5644, 0.416, 0.355 (the nearest cell appears in the top 10 matches for this cell); a, D. pseudoobscura, 3630, 0.884, 0.712; b, D. pseudoobscura, 4583, 5.264, 0.811; c, D. pseudoobscura, 5644, 0.595, 0.529 (the nearest cell is the best match for this cell). (3rd row) For each D. melanogaster query cell, the distance and direction to the average position of the top 10 best corresponding target cells is shown. The correspondence is shown with a line that starts at the position of the query cell, and ends at the average position of the target cells. The end of the line is indicated with a black dot. Because the 2D projection distorts actual distance in 3D, the lines are color-coded to indicate actual distance traversed in 3D. Dark blue is a large distance, yellow is a small distance. (4th row) The distribution of expression distance scores using only the nearest cell (grey) and best-matched cell within the nearest 30 (blue) are shown; we plot the root of the expression distance score to separate values near zero. The distribution of expression distance scores narrows and the mode decreases after a local search (Table S3). To establish the significance of the calculated differences, we assembled two atlases from the D. melanogaster dataset, and compared these two atlases to each other (dotted lines). (5th row) The expression profiles of the representative cells labeled in the top and middle panels are represented as a series of chart maps , where each gene is a single box with a line trace indicating expression over time. All gene expression data is normalized to a maximum of 1.0 over the time course. The expression profile of the D. melanogaster query cell is shown in red, the cell in the target embryo (D. yakuba or D. pseudoobscura) is shown in blue (dark blue for the nearest cell, and light blue for the best cell after a local search).

Figure 7. The proportion of cell types varies between D. melanogaster, D. yakuba, and D. pseudoobscura.

(Left) To compare the allocation of cell-types across species, we calculated the expression distance score to every other cell in each species' atlas (left). The number of adjacent cells below a given expression distance score is a measure of the size of a neighborhood of similar cells; we call these cells “expression neighbors.” Cells with many expression neighbors are blue. Cells with fewer are yellow. These sets of expression neighbors per cell provide a means to compare the relative allocation of cell types in the different embryos. For each D. melanogaster cell, the number of D. melanogaster expression neighbors is compared to the number of expression neighbors for its best corresponding cell in D. yakuba and D. pseudoobscura (right). Values displayed are the log ratio of the neighborhood sizes for corresponding cells in the two species being compared. D. melanogaster cells with relatively more expression neighbors are blue. D. melanogaster cells with relatively fewer are red.