Endocytosis is required for Toll signaling and shaping of the Dorsal/NF-kappaB morphogen gradient during Drosophila embryogenesis

Abstract

Dorsoventral cell fate in the Drosophila embryo is specified by activation of the Toll receptor, leading to a ventral-to-dorsal gradient across nuclei of the NF-κB transcription factor Dorsal. Toll receptor has been investigated genetically, molecularly, and immunohistologically, but much less is known about its dynamics in living embryos. Using live imaging of fluorescent protein chimeras, we find that Toll is recruited from the plasma membrane to Rab5(+) early endosomes. The distribution of a constitutively active form of Toll, Toll(10b), is shifted from the plasma membrane to early endosomes. Inhibition of endocytosis on the ventral side of the embryo attenuates Toll signaling ventrally and causes Dorsal to accumulate on the dorsal side of the embryo, essentially inverting the dorsal/ventral axis. Conversely, enhancing endocytosis laterally greatly potentiates Toll signaling locally, altering the shape of the Dorsal gradient. Photoactivation and fluorescence recovery after photobleaching studies reveal that Toll exhibits extremely limited lateral diffusion within the plasma membrane, whereas Toll is highly compartmentalized in endosomes. When endocytosis is blocked ventrally, creating an ectopic dorsal signaling center, Toll is preferentially endocytosed at the ectopic signaling center. We propose that Toll signals from an endocytic compartment rather than the plasma membrane. Our studies reveal that endocytosis plays a pivotal role in the spatial regulation of Toll receptor activation and signaling and in the correct shaping of the nuclear Dorsal concentration gradient.

Fig. 1.

Toll-GFP localizes to an early endosomal compartment in living embryos. (A–C) Surface views of Toll-GFP at the syncytial and cellular blastoderm. Toll localization shifts from predominantly vesicular and cytoplasmic during nuclear cycles 10 (A), 11, and 12 to predominantly plasma membrane in nuclear cycles 13 (B) and 14 (C). (D) Toll-GFP in a side view at nuclear cycle 14. (E–G) Coimaging at nuclear cycle 12 of Toll (green) and mCherry-Rab5 (red) shows significant overlap (yellow in F). (H) mCherry-Rab5 is distributed in particles concentrated in the cortical cytoplasm (arrow) at nuclear cycle 12. (I) Antibody staining of heat-fixed whole-mount embryos at nuclear cycle 14 using anti-Toll antibody showing both plasma membrane and particulate localization. (J) Toll moves from the plasma membrane to early endosomes and selectively partitions to discrete endosomal domains. Toll-paGFP was photoactivated (green) and appears first within the plasma membrane and subsequently in Rab5⁺ early endosomes (yellow). The last panel shows a higher magnification view of the region around the arrow in the preceding 196.4-s panel. PRE, preinjection; POST, postinjection.

Fig. 2.

Constitutively active Toll^10b-GFP is shifted from the plasma membrane toward an internal cytoplasmic distribution. (A) Toll-GFP in an anterior sagittal section, nuclear cycle 14. (B) Toll^10b-GFP in an anterior sagittal section, nuclear cycle 14. (C) Toll-GFP surface view, nuclear cycle 14. (D) Toll^10b-GFP surface view, nuclear cycle 14. (E) Dorsal-mCherry distribution in an early nuclear cycle 14 embryo expressing Toll^10b-GFP showing Dorsal protein in all nuclei along the d/v axis. (F–H) Overlap of Toll^10b-GFP (green) with mCherry-Rab5 (red) showing that Toll^10b is colocalized (yellow) with Rab5.

Fig. 3.

Inhibition of endocytosis ventrally reduces nuclear accumulation of Dorsal proximally and can alter the polarity of the Dorsal gradient. Live imaging of transgenic embryos expressing Dorsal-GFP at nuclear cycle 14. (A) Mock-injected control embryo expressing Dorsal-GFP showing the WT (wt) nuclear Dorsal gradient. (B) Embryo microinjected with Dynasore, a dynamin inhibitor, as a small bolus on the ventral midline in which nuclear accumulation of Dorsal is attenuated near the site of injection and slightly expanded dorsally (arrow). (C) Embryo injected with synthetic mRNA encoding Rab5S43N, a dominant negative form of Rab5 on the ventral midline. (D) Embryo injected as in C and allowed to develop for 40 min, illustrating a progressive shifting of nuclear translocation to the dorsal side of the embryo. (E) Rab5S43N ventrally injected embryo showing complete inversion of the Dorsal gradient. In all cases, ventral is down, dorsal is up, and red circles mark the site of injection of drug or synthetic mRNA.

Fig. 4.

(A) Mock-injected control embryo expressing Dorsal-GFP showing the WT (wt) Dorsal gradient. The Dorsal gradient is retracted toward the ventral midline (arrow). (B) Dorsal-GFP–expressing embryo injected near the ventral midline with synthetic mRNA encoding WT Rab5 (red circle). The Dorsal gradient is retracted toward the ventral midline (arrow). (C) Dorsal-GFP–expressing embryo injected ventrolaterally (red circle) with mRNA encoding constitutively active Rab5Q88L. The Dorsal gradient is expanded dorsolaterally but exhibits normal polarity. (D) Mock-injected Dorsal-GFP control embryo viewed at the same relative ventrolateral position as the embryo in E. (E) Dorsal-GFP–expressing embryo injected with synthetic mRNA encoding WT Rab5 axially at 25% egg length (red circle). The Dorsal gradient is shifted away from the ventral midline toward the site of injection only at the anterior of the embryo. (F) Dorsal-GFP–expressing embryo injected centrally (red circle) with synthetic mRNA encoding Rab5Q88L. Dorsal accumulates in all nuclei. In all cases, ventral is down, dorsal is up, red circles mark the site of injection, and embryos are in nuclear cycle 14.

Fig. 5.

Toll-GFP exhibits limited diffusion within the plasma membrane and is compartmentalized within the endocytic pathway. (A) Photoactivation of Toll-paGFP on the plasma membrane surface. Toll within the plasma membrane at nuclear cycle 14 diffuses one to two cell diameters within 20 min at 22 °C. PRE, preinjection; POST, postinjection. (B) FRAP of both the membrane and particulate fractions of Toll (bleach box in red). (C) Fluorescence recovery curves for the plasma membrane fraction (blue) and the particulate fraction (red) illustrate slow recovery for membrane-localized Toll over a distance of one future blastoderm cell. The recovery rate for particulate Toll (red) is 20% of the recovery rate for the plasma membrane fraction.

Fig. 6.

Inversion of the d/v axis by ventral microinjection of dominant negative Rab5 S43N leads to selective accumulation of Toll in endosomes on the dorsal signaling center. (A) Toll-GFP–expressing embryo injected ventrally with Rab5S43N under conditions that invert the d/v axis. The arrows show a shift in the distribution and accumulation of Toll at the opposite (dorsal) side of the embryo, where a newly created signaling center is located. pi, postinjection. (B) Dorsal view of another embryo injected as in A. (C) Ventral view of the embryo injected as in A. (D) Dorsal surface view of an embryo injected as in A, showing accumulation of Toll in a compartment with morphology characteristic of Rab5⁺ early endosomes.