Feedback control of the EGFR signaling gradient: superposition of domain-splitting events in Drosophila oogenesis

Abstract

The morphogenesis of structures with repeated functional units, such as body segments and appendages, depends on multi-domain patterns of cell signaling and gene expression. We demonstrate that during Drosophila oogenesis, the two-domain expression pattern of Broad, a transcription factor essential for the formation of the two respiratory eggshell appendages, is established by a single gradient of EGFR activation that induces both Broad and Pointed, which mediates repression of Broad. Two negative-feedback loops provided by the intracellular inhibitors of EGFR signaling, Kekkon-1 and Sprouty, control the number and position of Broad-expressing cells and in this way influence eggshell morphology. Later in oogenesis, the gradient of EGFR activation is split into two smaller domains in a process that depends on Argos, a secreted antagonist of EGFR signaling. In contrast to the previously proposed model of eggshell patterning, we show that the two-domain pattern of EGFR signaling is not essential for specifying the number of appendages. Thus, the processes that define the two-domain patterns of Broad and EGFR activation are distinct; their actions are separated in time and have different effects on eggshell morphology.

Fig. 1.

Wild-type dynamics of dpERK and BR expression. Stage 10-11 Drosophila egg chambers stained for (A-F) phosphorylated ERK/MAPK (dpERK) or (A′-F′) BR; (A″-F″) merged channels. (A‴-F‴) Diagrams summarizing the levels of dpERK and BR at each stage at three levels of expression (low, basal or high). Cross-sections are also shown (along the arrow) of the MAPK and BR expression profiles. The BR expression profile becomes stable by mid stage 10B, and the MAPK profile is refined to the floor cells by stage 11. (A-A‴) High levels of dpERK are found in the dorsal midline and in an anterior band in stage 10A egg chambers. The cusp-like pattern is defined as the midline pattern. BR levels are initially uniform in the main body follicle cells that contact the oocyte. (B-B‴) High levels of dpERK in the midline of stage 10B egg chambers correlate with the repression of BR, which is expressed at a basal level in the posterior and ventral cells. (C-C‴) dpERK levels decrease in a narrow region in the dorsal anterior (arrow), but have expanded to include the roof domain as marked by BR, and in the midline space between the roof cells. (D-D‴) dpERK is found in the floor cells, roof cells (marked by BR) and in the cells that are found in the dorsal anterior corner. (E-E‴) dpERK expression increases in the floor cells and decreases in the roof cells. A posterior ring (`spectacle' pattern) surrounding the posterior boundary of BR expression also shows dpERK expression. (F-F‴) By stage 11, dpERK is found in the floor cells, which begin to slip under the roof cells as tube formation proceeds at stage 12. BR remains at high levels in the roof primordia.

Fig. 2.

The effects of argos on MAPK signaling and eggshell patterning. (A-A″) The argos^Δ7 FRT80B line, which was used for mosaic analysis, does not complement other mutant alleles of argos as demonstrated by the eye blister phenotype for argos^Δ7 FRT80B/argos²⁵⁷ (A), argos^Δ7 FRT80B/argos^W11 (A′) and argos²⁵⁷/argos^P1 (A″) flies. The arrow denotes the midline. (B-B″) An argos^Δ7 clone that spans all of the main body follicle cells, marked by loss of GFP (B), shows a single peak of dpERK staining (B′) at a time when comparable wild-type egg chambers show a loss of dpERK in a narrow midline region (see Fig. 1C-E). The midline is marked by an arrow. (C-C″) In another large argos^Δ7 clone spanning the dorsal domain, dpERK staining remains at high levels in the dorsal anterior (arrow, C′), in contrast to the pattern found in wild-type egg chambers (see Fig. 1E-E″), when the spectacle pattern is present. (D-D″) In a stage 12 egg argos^Δ7 clone, a single row of elevated dpERK is still seen (arrow, D′), which is where the argos transcript is normally expressed (data not shown). (E-E″) Examples of argos mutant eggs showing both wild-type and aberrant dorsal appendage (DA) morphologies. The majority of eggs examined showed wild-type DA morphology (E). Other phenotypes with low penetrance included reduced inter-appendage distance (E′) and appendages that were shorter (E″). Shown are eggshells laid by argos^Δ7/argos^W11 females.

Fig. 3.

The effects of rho on MAPK signaling and BR expression. (A-A″) In early stage 10B Drosophila egg chambers, rho^7M clones show no effect on dpERK levels in the midline (arrow, A′) or on repression of BR in the midline (A″). (B-B″) rho is locally required for MAPK activity in the floor cells (B′, arrowhead), as shown in a clone that covers the dorsal edge of the floor domain (B,B′, arrow). The BR domain is unaffected (B″).

Fig. 4.

kek1 and sty modulate the size of the roof domain and inter-appendage distances. (A-A″) kek1^-/- Drosophila egg chambers have DAs that are further from each other (A) than in wild type, consistent with the increased size of the midline domain (∼4-5 cells), which is marked by low levels of BR (A′), and with the reduced size of the roof domain (A″, stage 10B; see also Fig. 6C). (B-B″) A small percentage of eggshells exhibits multiple, ectopic appendages (B, arrowheads), consistent with the creation of ectopic boundaries of BR with the midline (B′,B″, arrow). BR remains repressed in small clones located in the midline (B′,B″, arrowhead). (C-C″) The most common eggshell phenotype in unmarked sty^Δ5 clones exhibits thinner DAs that are located more laterally (C). This phenotype is consistent with a large expansion in the midline (∼10-11 cells; see also Fig. 5E) and a significantly reduced BR patch size (C′,C″). (D,D′) kek1^-/- egg chambers with small sty^-/- clones show a superposition of patterning phenotypes: the dorsal midline has increased, similar to the kek1^-/- pattern, and sty^-/- clones located within the dorsal half of the BR patch lead to a loss of elevated BR expression (arrowheads). sty^-/- clones located in the ventral side of the roof domain are unaffected (arrow).

Fig. 5.

kekkon-1 and sprouty are not required for splitting the peak of MAPK activity. (A-A″) In stage 10B kek1^-/- Drosophila egg chambers, a single gradient of dpERK is detected during specification of the midline (loss of BR) and of the two separated BR domains, similar to as in Ore R. (B-B″) The split domains of activated MAPK (green) are detected in older kek1^-/- egg chambers. The only difference from the wild type is the increased separation between the two domains of DA primordia. (C-C″) In large sty^-/- clones, a single large domain of dpERK spans the midline. (D-D″) The increase in dpERK in small clones located in the midline is not fully cell-autonomous (D′), in contrast to the effect of sty clones on BR. Note that this effect is evident after the two BR domains have already been specified. In some cases, the level of BR is variable at stage 10B in sty^-/- clones (D″). (E-E″) Although elevated dpERK staining is detected in large sty^-/- clones in late stage 10B/11 egg chambers, the highest levels of dpERK (E′, arrows) are detected in two sets of floor cells, anterior to the roof domain, as marked by BR (E″).

Fig. 6.

Negative feedbacks modulate the output of an incoherent feedforward loop. (A-A″) Drosophila pnt^Δ88 mosaic eggs show a single DA that includes the dorsal midline, which is consistent with elevated BR expression in clones that span the midline (A,A′). Ectopic BR is found in the clone spanning the midline. The follicle cells that form the anterior-most two rows over the oocyte show repression of BR that is independent of PNT repression. The arrow indicates the approximate location of the midline. (B,B′) Model for specifying the two domains of roof cells. BR is activated by GRK-induced EGFR signaling during stage 10B in a wide domain of dorsal follicle cells. High levels of EGFR signaling lead to repression of BR, which is mediated by PNT. Cell-autonomous inhibitory feedback by KEK1 and STY modulates the location of the dorsal domain of BR. The strength of inhibitory feedback, α, is stronger for STY than for KEK1. (B′) As a result, the size of the midline increases at the expense of the roof domain in egg chambers mutant for either kek1 or sty. (C) Quantification of the effect of each inhibitor on the size of the BR patch. Argos has a negligible effect, whereas kek1^-/- and sty^-/- egg chambers have a reduced number of BR cells in each DA primordial (see text for P-values).