Non-redundant selector and growth-promoting functions of two sister genes, buttonhead and Sp1, in Drosophila leg development

Abstract

The radically distinct morphologies of arthropod and tetrapod legs argue that these appendages do not share a common evolutionary origin. Yet, despite dramatic differences in morphology, it has been known for some time that transcription factors encoded by the Distalless (Dll)/Dlx gene family play a critical role in the development of both structures. Here we show that a second transcription factor family encoded by the Sp8 gene family, previously implicated in vertebrate limb development, also plays an early and fundamental role in arthropod leg development. By simultaneously removing the function of two Sp8 orthologs, buttonhead (btd) and Sp1, during Drosophila embryogenesis, we find that adult leg development is completely abolished. Remarkably, in the absence of these factors, transformations from ventral to dorsal appendage identities are observed, suggesting that adult dorsal fates become derepressed when ventral fates are eliminated. Further, we show that Sp1 plays a much more important role in ventral appendage specification than btd and that Sp1 lies genetically upstream of Dll. In addition to these selector-like gene functions, Sp1 and btd are also required during larval stages for the growth of the leg. Vertebrate Sp8 can rescue many of the functions of the Drosophila genes, arguing that these activities have been conserved, despite more than 500 million years of independent evolution. These observations suggest that an ancient Sp8/Dlx gene cassette was used in an early metazoan for primitive limb-like outgrowths and that this cassette was co-opted multiple times for appendage formation in multiple animal phyla.

Figure 1. Generating a deficiency for btd and Sp1.

(A) Genomic organization of the btd and Sp1 genomic region. While btd only encodes for one isoform, Sp1 encodes for two, a small one (Sp1-PB (Flybase) or Sp1^S,) and a larger one (Sp1-PD (Flybase) or Sp1^L). Two FRT-containing P elements, d01932 and e03908, are situated 5′ and 3′ of btd and Sp1, respectively. The PCR primers used to molecularly confirm the deficiency are indicated (1 and 2). (B) PCR confirmation of Df(btd,Sp1). Using the primers shown in (A), no product is observed in the original P element stocks, while a 2.7 kb product is observed in two independently generated Df(btd,Sp1) stocks. (C,D) btd and Sp1 expression patterns visualized by btd-Gal4>UAS-lacZ and Sp1 in situ hybridization, respectively, in third instar leg (C) and antennal (D) imaginal discs. Note the absence of btd or Sp1 expression in the presumptive body wall of the leg (arrows). (E) Df(btd,Sp1) mutant clones (absence of signal) are difficult to recover in the btd/Sp1 expression domain when they generated before the third instar (<72 hrs AEL). Twin spots (white arrows) and clones in proximal regions (asterisks) can be observed, as can clones in the wing or eye discs (red arrows). Only twin spots are recovered in the medial antenna (white arrows). The images of the antenna and eye discs represent two different confocal planes of the same disc. (F) Df(btd,Sp1) clones generated 48–72 hrs AEL positively marked by ß -Gal staining (green) in the leg survive poorly and tend to segregate from the surrounding tissue. The disc is co-stained for Discs large (Dlg) which labels all cell membranes (red). (G) MARCM Df(btd,Sp1); Sp1^S+ mutant clones generated in parallel to those in (F) are recovered more frequently than Df(btd,Sp1) mutant clones, indicating rescue. (H) Quantification of rescue. Sp1 rescued the number of clones in the leg disc (only telopodite clones were scored). Note that Sp1^S rescued better then Sp1^L. Clones were induced 48–72 hrs AEL. The rescue experiments with Sp1^S or Sp1^L, but not with btd+ (p>0.05), show a statistically significant difference from the control experiment (Df(btd,Sp1); * p<0.05 and ** p<0.001 with Student's t-test).

Figure 2. btd and Sp1 control leg growth.

(A) Time line showing when removing the function of btd and Sp1 affects leg growth (orange shadow). (B) A large Df(btd,Sp1) M+ clone in a T1 leg induced 48–72 hrs AEL and marked by yellow (y) bristles (the clone boundary is indicated by the white dotted line in the left leg). For comparison, the wild type right T1 leg is included in this image. The mutant tissue still maintains leg identity scored by the presence of bracted bristles (arrows, inset). Asterisks mark the segments affected by the clone. The same leg segment nomenclature has been used for all the figures: coxa (cox), trochanter (tro), femur (fem), tibia (tib) and tarsus (tar). (C) Wild type antenna with 1^st antennal segment (a1), 2^nd antennal segment (a2), 3^rd antennal segment (a3) and arista (ar). (D) Df(btd,Sp1) M+ clone induced 48–72 hrs AEL results a strong reduction in size of the a1 and a2 antennal segments, while a3 and the ar are normal. The clone is marked by y. (E) A large btd^XG81 M+ clone in a T2 leg induced 48–72 hrs AEL and marked by y (clone is outlined by white dots) results in a small growth defect in the femur (fe) and tibia (tib), which are also partially fused (arrow). The tarsus (tar), trochanter (tro) and coxa (cox) are unaffected. (F, G, H) The downregulation of Sp1 beginning at the second instar using RNAi affects the growth of the entire leg. Two different Gal4 drivers were used to examine different regions of the leg. (F) The medial part of the leg is strongly reduced in size in dac-Gal4; UAS-Sp1i flies. (G) The distal part of the leg is strongly reduced in size in Dll-Gal4; UAS-Sp1i flies. In this experiment we blocked Gal4 activity prior to the second instar using tub-Gal80^ts. (H) Shows a schematic representation of the expression patterns of the two Gal4 drivers used to downregulate Sp1 function (dac in blue and Dll in red). (I, J) Df(btd,Sp1) M+ clones induced 48–72 hrs AEL and examined in 3^rd instar leg discs. (I) A subset of Df(btd,Sp1) M+ clones (marked by the absence of GFP) show de-repression of Dll in the Dac domain and de-repression of dac in the Dll domain (white arrows). Note that these clones do not affect the expression of Dll and dac in their normal expression domains (green arrows). (J) A subset of Df(btd,Sp1) M+ clones (marked by the absence of GFP) show derepression of hth (arrow). Clones that do not derepress hth are indicated with asterisks.

Figure 3. Sp1 is required for leg development.

(A) Top: time line showing when the functions of btd and Sp1 (orange shadow) were removed. Bottom: In these experiments we initiated clone induction during embryogenesis using Dll-Gal4; UAS-flp to generate Df(btd,Sp1) M+ or btd M+ clones. This method results in excellent survival of animals that have all six legs completely, or nearly completely, mutant. For the rescue experiments, an additional UAS transgene (e.g. UAS-btd) was included. (B) Wild type third instar leg imaginal disc showing the expression patterns of Dll and dac. (C) Third instar Df(btd,Sp1) mutant leg generated using the genotype schematized in (A); mutant tissue is marked by the absence of GFP. These discs are much smaller than wild type and show a nearly complete loss of Dll and dac expression. White bar is 75 µm. (D) Wild type T1 adult leg with the segments indicated. (E) T1 adult leg entirely mutant for btd (marked by y) generated as shown in (A). Only the size of the femur and the tibia are affected and are partially fused together. (F) btd-Gal4; UAS-Sp1i reduces the size of the entire leg, from the coxa to the distal tip. Shown here is a T1 leg. (G,G') T1 adult legs entirely (G') or nearly entirely (G) mutant for btd and Sp1, generated as described in (A). Mutant tissue is marked by y. In (G), only a small patched of mutant tissue is visible and is associated with some non-mutant (y+) coxa tissue. In (G'), no leg tissue is observed. (H–K) Rescue of Df(btd,Sp1) T1 mutant legs (marked by y) generated as described in (A) where (H) Sp1^S, (I) Sp1^L, (J) btd and (K) Dll were expressed under the control of Dll-Gal4. (H'-K') show examples of leg imaginal discs of the same genotypes. White bars represents 75 µm. (H) Sp1^S is able to rescue the Df(btd,Sp1) mutant phenoype and restore the Dll and dac expression domains. (I) Sp1^L is able to partially rescue the adult Df(btd,Sp1) mutant phenoype and completely restore the Dll and dac expression domains. (J) btd is unable to rescue the adult Df(btd,Sp1) mutant phenoype and partially rescues the Dll and dac expression domains. (K) Dll is unable to rescue the adult Df(btd,Sp1) mutant phenoype but partially rescues Dll and dac expression domains.

Figure 4. Dorsal to ventral transformations resulting from the loss of btd and Sp1.

(A) A T2 adult segment comprised mostly of Df(btd,Sp1) y tissue. These animals are generated via the genotype shown in Figure 3A. Dorsal is up. The normal notum, wing, and hinge are at the top; the bottom half of the tissue shows a transformation of ventral fates towards dorsal fates, including an ectopic wing, hinge, and notum (indicated by asterisks). Note that the normal notum, wing, and hinge are mutant (marked by y) but appear wild type. (B) Third instar leg imaginal disc of the same genotype as in (A) stained for GFP (absence marks the mutant tissue), Vg (red) and Eygone (Eyg; blue), which are markers for wing and notum fates, respectively. (C,D) Ectopic expression of Sp1^L (C) or mouse Sp8 (D), under the control of dpp-Gal4 result in the transformation of wing towards leg in the adult. Arrows indicate leg tissue. Remaining notum (N) and wing (W) tissue are indicated. Insets show a high magnification of the leg tissue, with bracts (small arrows). Note the appearance of leg structures also in the pronotum in (E) (arrowhead). (E,F) Ectopic expression of Sp1^L (E) or mouse Sp8 (F), under the control of dpp-Gal4 results in the induction of leg fates in the wing imaginal disc. These discs were stained for dpp-Gal4 expression (green), Dll (red), Dac (blue), or Vg (E, right-most panel). dpp>Sp8 also results in dramatic overgrowths that are not observed in the dpp>Sp1^L wing discs.

Figure 5. Sp1, not btd, is required for Dll-LT activity.

(A) Df(btd,Sp1) M+ clone (outlined in green) generated 72–96 hrs AEL shows the absence of LT-lacZ expression, but no affect on Dll. (B) Clones expressing Sp1i strongly reduced LT-lacZ expression. (C) btd^XG81 mutant clones generated 72–96 hrs AEL do not affect LT-lacZ expression. (D–F) Ectopic expression of btd (D), Sp1^L (E), or mouse Sp8 (F) in the wing disc activates Dll and LT-lacZ (green arrows). These flip-out clones were generated 48–72 hrs AEL.

Figure 6. Different dependencies on Sp1 for early and late Dll enhancer activities.

(A,B) Stage 11 wild type (A) and Df(btd,Sp1) (B) embryos stained for Dll (green) and Dll304-lacZ (red). Dll304-lacZ remains active in the absence of both factors. Expression of Dll in the antennal primordia, however, is nearly absent in Df(btd,Sp1) embryos (arrows). T1, T2, and T3 mark the three thoracic segments. (C) Wild type stage 14 embryo stained for Dll (green), Hth (blue), and Dll-LT-lacZ (red). The white dots mark the prd-Gal4 expression domain. (D, E) Df(btd,Sp1); prd-Gal4; UAS-Sp1^L (D) or UAS-Sp8 (E) stage 14 embryos. In T1 and T3, where prd-Gal4 is not expressed, Dll expression is greatly reduced (asterisks) and LT activity is completely absent. The remaining Dll expression is likely derived from the Dll304 early enhancer. In T2, where prd>Sp1^L (D) or prd>Sp8 (E) both Dll and LT-lacZ expression are rescued. btd and Sp1^S can also rescue embryonic Dll and LT-lacZ expression (not shown). (F) Df(btd,Sp1) M+ mutant leg disc generated using the scheme shown in Figure 3A, rescued with UAS-Sp8. Significant rescue of the Dll (red) and dac (blue) expression domains is observed. Mutant tissue is marked by the absence of GFP (green).

Figure 7. btd and Sp1 require Dll to induce distal and medial leg development.

(A,B) btd+ MARCM clones in the wing disc activated the expression of Lim1, dac (A) and hth (B) (arrows). Clones were generated 48–72 hrs AEL and are marked by GFP+ (green). (C,D) Dll-; btd+ MARCM clones in the wing disc were unable to induce Lim1 or dac (C; arrows), but can still able activate hth close to its own domain (arrow) but not in the center of the wing pouch (D; arrowhead). Clones were generated 48–72 hrs AEL and are marked by GFP (green). (E) btd+ MARCM clones in the adult wing blade induced the formation of leg-like tissue, including distal leg identities (arrows and inset). (F) Dll-; btd+ MARCM clones failed to induce distal leg-like tissue, although they generate tissue that might correspond to proximal leg tissue (arrows and inset). (G) Schematic representation of the differential requirements for btd and Sp1 during leg development. At embryonic 11 stage, Dll (via the 304 enhancer), btd, and Sp1 are all activated independently in the appendage primordia. A few hours later, Dll expression is restricted to the telopodite precursors cells of the leg (via the LT enhancer) and depends on Sp1 activity. At this stage, Sp1 is required to promote the formation of the ventral appendage primordia (legs) and inhibit the formation of the dorsal primordia (wing and haltere). Dll is required for the entire telopodite domain. During larval second instar stage (L2), Dll expression no longer requires btd and Sp1. Dll is required only for distal leg development while btd and Sp1 are required for the growth of the entire leg but have no function in the body wall.