A multiscale investigation of bicoid-dependent transcriptional events in Drosophila embryos

Abstract

Background: Morphogen molecules form concentration gradients to provide spatial information to cells in a developing embryo. Precisely how cells decode such information to form patterns with sharp boundaries remains an open question. For example, it remains controversial whether the Drosophila morphogenetic protein Bicoid (Bcd) plays a transient or sustained role in activating its target genes to establish sharp expression boundaries during development.

Methodology/principal findings: In this study, we describe a method to simultaneously detect Bcd and the nascent transcripts of its target genes in developing embryos. This method allows us to investigate the relationship between Bcd and the transcriptional status of individual copies of its target genes on distinct scales. We show that, on three scales analyzed concurrently--embryonic, nuclear and local, the actively-transcribing gene copies are associated with high Bcd concentrations. These results underscore the importance of Bcd as a sustained input for transcriptional decisions of individual copies of its target genes during development. We also show that the Bcd-dependent transcriptional decisions have a significantly higher noise than Bcd-dependent gene products, suggesting that, consistent with theoretical studies, time and/or space averaging reduces the noise of Bcd-activated transcriptional output. Finally, our analysis of an X-linked Bcd target gene reveals that Bcd-dependent transcription bursts at twice the frequency in males as in females, providing a mechanism for dosage compensation in early Drosophila embryos.

Conclusion/significance: Our study represents a first experimental uncovering of the actions of Bcd in controlling the actual transcriptional events while its positional information is decoded during development. It establishes a sustained role of Bcd in transcriptional decisions of individual copies of its target genes to generate sharp expression boundaries. It also provides an experimental evaluation of the effect of time and/or space averaging on Bcd-dependent transcriptional output, and establishes a dosage compensation mechanism in early Drosophila embryos.

Figure 1. Simultaneous detection of Bcd and nascent hb transcripts in embryos.

(A) Shown is a merged Confocal image of a wt embryo at early nuclear cycle 14 detecting the nuclear envelope (red), Bcd protein (blue) and the nascent hb transcripts as intron dots (green). Shown on the right is a magnified view of a section of the expression region. (B) Shown is the radial distribution of detected hb intron dots. Data shown here were extracted from one single wt embryo with 2,556 identified nuclei (with a mean diameter l = 6.06±0.80 µm) and 1,537 detected intron dots. Intron dots that appear to be outside the boundaries of this illustrative “average” nucleus are due to the fact that real nuclei are not all perfectly round in shape. As explained in the text, our intron dot dataset used in all the analyses described in this work were extracted from within the nuclei. (C) Shown are the measured distances between two detected hb intron dots inside individual nuclei. The measured mean distance between two intron dots inside individual nuclei is 2.70±0.18 µm for all 14 tested embryos (represented by different colors). Each error bar is one standard deviation among the nuclei for a single embryo.

Figure 2. Expression profiles for hb intron dots and Hb proteins.

(A) Shown is a 3-D scatter plot of normalized A-P positions (x/L), mean Bcd intensities of the nucleus (B _nuc) and the normalized intensities of hb intronic mRNA within identified dots for a single embryo. The projection onto the x/L-B _nuc plane (red) gives the profile of the nuclear Bcd intensity as a function of the A-P position. (B) Shown is a 3-D scatter plot of normalized A-P positions (x/L), mean Bcd intensities of the nucleus (B _nuc) and the normalized intensities of nuclear Hb protein for a single embryo. The projection onto the x/L-B _nuc plane (red) gives the profile of the nuclear Bcd intensity as a function of the A-P position. (C) Shown are the noise (η) profiles of hb intron dot density (ρ) for a single embryo (blue) and the noise for Hb protein intensity for a single embryo (red) as a function of x/L. Error bars were obtained from bootstrapping.

Figure 3. Investigating the role of Bcd in hb transcriptional events on embryonic and nuclear scales.

(A) Shown is the number density of hb intron dots (ρ) as a function of the A-P position (x/L). An error bar is one standard deviation. (B) Shown is the profile of ρ as a function of mean B _nuc (on a log scale) of different bins along the A-P axis. An error bar is one standard deviation. (C) Shown is the measured noise η as a function of either mean B _nuc (on a log scale) of different bins or the bin position x/L (inset). Error bars were obtained from bootstrapping. (D) Shown are the mean Bcd intensities within the inactive nuclei (blue) and the active nuclei (red) at different A-P positions. For the anterior hb expression region (x/L = 0.16∼0.46), where hb expression is dependent on Bcd concentration, p-values<0.0001 (Student's t-tests); at the posterior of the embryos (x/L = 0.82∼0.96), where a Bcd-independent hb expression domain is detected, p-values>0.1; for the non-expressing regions of the embryos (x/L = 0.46∼0.82 and 0.96∼1), where the burst probability is too low (<0.05) to allow reliable computation of the mean Bcd intensities within the active nuclei, only B _nuc within inactive nuclei are shown.

Figure 4. Investigating the role of Bcd in hb transcriptional events on the local scale.

(A) A zoom-in merged view of a nucleus showing a single detected hb intron dot (green) and Bcd intensities (red). (B) Shown are the probability density functions of Bcd pixel intensity, P(B _pix), inside the inactive nuclei (blue), the active nuclei (red), or the areas of the detected intron dot sites (green). Data from three different A-P positions are shown in different colors. (C) Ratio of P(B _pix) inside the areas of the detected intron dot sites to P(B _pix) within the active nuclei. A base line at ratio of 1 is drawn (see text for details). Different colors represent data at different A-P positions: x/L = 0.2∼0.22 (blue), x/L = 0.3∼0.32 (red) and x/L = 0.4∼0.42 (green).

Figure 5. Multiscale investigation of the transcriptional properties of otd.

(A) Shown are the mean Bcd intensities measured from the inactive nuclei (blue), the active nuclei (red) and the areas of the detected otd intron dots (green) as a function of x/L. Data were extracted from all 12 tested embryos (females and males). Student's t-tests were conducted to determine p-values for each A-P bin of nuclei. Comparing red with blue, all p-values<0.0001; comparing red with green, p-values are (in A-P order at the locations shown): 0.2, 0.1, 0.1, 0.01, 0.02, 0.06, 0.001, 0.02, 0.07 and 0.08. (B) Shown is the number density of otd intron dots (ρ) as a function of x/L for 8 female embryos (blue) or 4 male embryos (red, inset). An error bar is one standard deviation. (C) Shown is the number density of otd intron dots (ρ) as a function of mean B _nuc of different bins along the A-P axis for 8 female embryos (blue) or 4 male embryos (red, inset). An error bar is one standard deviation. Hill coefficient for the posterior boundary in response to the Bcd input is 4.0±1.9, corresponding to fewer binding sites for Bcd molecules in the otd enhancer than those in the hb enhancer . (D) Shown is the raw intensity of otd intron dots (I) as a function of mean B _nuc of different bins along the A-P axis for 8 female embryos (blue) or 4 male embryos (red, inset). An error bar is one standard deviation.