Recurrent modification of a conserved cis-regulatory element underlies fruit fly pigmentation diversity

Abstract

The development of morphological traits occurs through the collective action of networks of genes connected at the level of gene expression. As any node in a network may be a target of evolutionary change, the recurrent targeting of the same node would indicate that the path of evolution is biased for the relevant trait and network. Although examples of parallel evolution have implicated recurrent modification of the same gene and cis-regulatory element (CRE), little is known about the mutational and molecular paths of parallel CRE evolution. In Drosophila melanogaster fruit flies, the Bric-à-brac (Bab) transcription factors control the development of a suite of sexually dimorphic traits on the posterior abdomen. Female-specific Bab expression is regulated by the dimorphic element, a CRE that possesses direct inputs from body plan (ABD-B) and sex-determination (DSX) transcription factors. Here, we find that the recurrent evolutionary modification of this CRE underlies both intraspecific and interspecific variation in female pigmentation in the melanogaster species group. By reconstructing the sequence and regulatory activity of the ancestral Drosophila melanogaster dimorphic element, we demonstrate that a handful of mutations were sufficient to create independent CRE alleles with differing activities. Moreover, intraspecific and interspecific dimorphic element evolution proceeded with little to no alterations to the known body plan and sex-determination regulatory linkages. Collectively, our findings represent an example where the paths of evolution appear biased to a specific CRE, and drastic changes in function were accompanied by deep conservation of key regulatory linkages.

Figure 1. Abdomen pigmentation correlates with the regulatory activity of dimorphic element alleles.

(A) The A5 and A6 segment dorsal tergites of D. melanogaster males are fully pigmented, (B–H) whereas the female A5 and A6 tergite pigmentation varies from “Light” to a male-like “Dark” phenotype. (A′–H′) GFP-reporter transgene activity was measured in transgenic pupae at 85 hours after puparium formation (hAPF) and activity measurements were represented as the % of the D. melanogaster Canton^S allele female A6 mean ± SEM. (A′) The regulatory activity of a male Canton^S pupae. The regulatory activity of alleles from the following locations were measured: (B′) Oaxaca, Mexico (called Light 2), (C′) Crete, Greece, (D′) Kuala Lumpur, Malaysia (called Light 1), (E′) Mumbai, India, (F′) Kisangani, Africa, (G′) Uganda, Africa (called Dark 1), and (H′) Bogota, Columbia (called Dark 2).

Figure 2. bab locus allelic variation underlies phenotypic variation.

(A) The A5 and A6 tergite phenotype for F1 females were intermediate to those from the parental Light 1 and Dark 1 stocks. F2 females had pigmentation phenotypes that were (B) “Light”, (C) “Intermediate”, or (D) “Dark”. (E–P) Complementation tests for population stock bab loci with a bab locus null allele. (E) The Light 1 stock complemented the bab locus null allele with regards to abdomen tergite pigmentation, whereas the (F) Dark 1, and (G) Dark 2 stocks failed to complement the null allele in segments A5 and A6 but complemented the null allele for the A3 and A4 segments. Light 1, Dark 1, and Dark 2 stocks complemented the bab locus null allele for (I–K) posterior abdomen phenotypes and (M–O) for the development of the leg tarsal segments. Females with a homozygous bab locus null genotype displayed (F) ectopic pigmentation on segments A3 through A6, and (L) lacked bristles on the A6 and A7 ventral sternites and the genitalia (g) had altered bristles and morphology. (P) Individuals with a homozygous bab locus null genotype had tarsal segments 5, 4, and 3 fused, and altered bristle morphology on tarsal segments 2 and 3. Red arrowheads and black arrows respectively indicate the location abnormal posterior abdomen and tarsus features.

Figure 3. Population level differences in Bab paralog expression.

(A–C) The expression of Bab1 in the dorsal abdomens of female pupae at 85 hAPF. (A) Light 1 females display uniform Bab1 expression throughout segments A2-A6, whereas expression is reduced in the A5 and A6 segments of (B) Dark 1 and (C) Dark 2 females. (D and E) Expression of Bab1 in the female genitalia (g) and analia (a) at 29 hAPF. (F–H) Bab2 expression in the dorsal abdomen of female pupae is at 85 hAPF. Bab2 expression is (F) uniform throughout the A2–A6 segments of Light 1 females, (G) reduced in the A5 and A6 segments of Dark 1 females, and (H) uniform throughout the A2–A6 of Dark 2 females. (I and J) Expression of Bab2 in the female genitalia (g) and analia (a) is at 29 hAPF. Red arrowheads indicate segments where expression is reduced compared to more anterior segments, whereas yellow arrowheads indicate the segments where Bab2 is expressed at a higher level than that observed for Bab1 for Dark 2 females.

Figure 4. Dimorphic element alleles diverged from an ancestral state.

(A–E) To scale representations of various dimorphic elements, including the (A, Concestor) inferred allele for the most recent common ancestor of extant D. melanogaster populations, and alleles from populations with Light (B, Light 1; C, Light 2) and Dark (D, Dark1; E, Dark 2) female pigmentation phenotypes. Dark blue and yellow rectangles respectively represent the fourteen ABD-B and two DSX binding sites. Thin and thick red lines respectively represent derived point and indel mutations. (A′–E′ and A″–E″) Comparison of GFP-reporter gene activities in female transgenic pupae was at 85 hAPF. Activity measurements are represented as the % of the D. melanogaster Concestor element female (A′–E′) A6 mean ± SEM and (A″–E″) A7 mean ± SEM. Red upward and downward arrows respectively indicate segments with increased and decreased regulatory activity. Yellow arrowhead indicates a region of expanded regulatory activity. Lowercase letter “g” indicates expression in the genitalia.

Figure 5. Functionally-relevant mutations in dimorphic element alleles.

(A) Dimorphic element allele phylogeny, including the outgroup species D. simulans (D. sim.). Alignment of sequences encompassing the (B) “D” mutation, (C) “E” mutation, (D) “F” mutation, (E) and the “L” mutation. Black background color for the E mutation indicates the 1 base pair overlap for the derived deletion and the adjacent DSX binding site. (F–J) Comparison of GFP-reporter activity in female transgenic pupae at 85 hAPF, represented as the % of the D. melanogaster Concestor element female A6 mean ± SEM. Red upward and downward arrow respectively indicate segments with increased and decreased regulatory activity. Yellow arrowhead indicates expanded regulatory activity. Regulatory activities differing from the Concestor element due to the following derived mutations: (G) D mutation; (H) E mutation; (I) F mutation; and (J) L mutation. (K) Summary for the female A6 regulatory activities for modifications to the E mutation region. The Concestor element sequence is provided and the introduced modifications indicated by red bases. (L) Gel shift assays for annealed oligonucleotide probes containing the wild type (Concestor element, lanes 1–7), E mutation (lanes 8–14), and mutant (Dsx1 KO, lanes 15–19) Dsx1 binding site. The binding site sequences are included with mutant bases in red. For the Concestor element and E mutation probes, binding reactions used increasing amounts of the DSX protein (from left to right: 0 ng, 8 ng, 16 ng, 31 ng, 63 ng, 125 ng, 250 ng, and 500 ng). For the Dsx1 KO probe, binding reactions used the following amounts of protein (from left to right: 0 ng, 8 ng, 31 ng, 125 ng, 500 ng). Blue and red arrowheads point to the respective locations of single or pair of DSX monomers bound to the probe.

Figure 6. Interspecific evolution of pigmentation and dimorphic element activity.

(A) Phylogeny for species that differ in the extent of sexually dimorphic pigmentation. (B–I) Dorsal view of adult abdomens, pigmentation of the (E) D. yakuba female A5 and A6 segments is more (D) male-like, whereas pigmentation is altogether absent on the A5 and A6 segments of (G) D. fuyamai females. (J–Q) Comparison of GFP-reporter gene activity in female transgenic pupae at 85 hr APF. Activity measurements are represented as the % of the (K) Concestor element female A6 mean ± SEM for (M) D. yakuba, (O) D. fuyamai, and (Q) D. auraria.

Figure 7. Pigmentation gene network model and the evolution of an ancestral CRE regulatory logic.

(A–C) Schematic of the hierarchical structure of the D. melanogaster pigmentation gene network. Direct regulation is represented as solid connections and dashed connections represent connections where regulation has not been shown to be direct. Activation and repression are respectively indicated by the arrowhead and nail-head shapes. This network includes an (A) upper level of patterning genes, including Abd-B and dsx respectively of the body plan and sex-determination pathways, (B) a mid-level tier that integrates patterning inputs, (C) and a lower level that includes pigmentation genes whose encoded products function in pigment metabolism. Although Abd-B directly regulates the pigmentation gene yellow, sexually dimorphic expression of the yellow and tan genes results from the sexually dimorphic output of the bab locus that acts to repress tan and yellow expression in females. (D) A model for the evolution of diverse dimorphic element regulatory activities. The common ancestor of D. melanogaster populations and related species possessed a dimorphic element with both DSX and ABD-B regulatory linkages and that drove expression in the female A6–A8 segments. This ancestral regulatory logic was recurrently modified to increase the levels and expand the segmental domain of activity, or to decrease and contract activity. These changes occurred amidst the preservation of the core ABD-B and DSX regulatory linkages, perhaps though the loss (TF 3) and/or gain (TF 4) of other transcription factor linkages.