In locus analysis of patterning evolution of the BMP type II receptor Wishful thinking

Boundary formation is a fundamental requirement during animal development (Dahmann et al., 2011). In tissues, a handful of cell-to-cell signaling pathways inform cells of their position. Consequently, cells acquire different fates in order to form organs (Ashe and Briscoe, 2006; Hamaratoglu et al., 2014; Moussian and Roth, 2005; Wolpert, 1989). In the Drosophila ovaries, eggs are produced in an assembly line manner (Bastock and St Johnston, 2008; Hinton, 1981; Horne-Badovinac and Bilder, 2005; Spradling, 1993). During egg development, numerous cell signaling pathways, including the epidermal growth factor receptor (EGFR) and bone morphogenetic protein (BMP) signaling, control the dorsal-ventral and anterior-posterior axes of the fly (Berg, 2005; Deng and Bownes, 1997; Moussian and Roth, 2005; Neuman-Silberberg and Schupbach, 1994; Peri and Roth, 2000; Yakoby et al., 2008b). The BMP2/4 ligand homolog in Drosophila, Decapentaplegic (Dpp), activates the BMP pathway through a heteromeric complex of type I and type II receptors (Parker et al., 2004; Raftery and Sutherland, 1999; Wu and Hill, 2009). Previous work focused largely on the necessity and patterning of the BMP type I receptor Thickveins (Mantrova et al., 1999; Niepielko et al., 2011, 2012). Wishful thinking (Wit) was identified as the BMP type II receptor in the follicle cells of D. melanogaster (Marmion et al., 2013; Yakoby et al., 2008a). It was shown to be necessary for signaling and proper patterning of the eggshell. Here, we focus on gene patterning during oogenesis.

One way to identify common mechanisms of gene regulation is to search for overlapping spatiotemporal expression domains (Konikoff et al., 2012; Kumar et al., 2017). In D. melanogaster, the anterior domain of the follicular epithelium is patterned by BMP signaling (Twombly et al., 1996). Numerous genes are expressed in the anterior domain of the follicle cells (Jordan et al., 2005; Peri and Roth, 2000; Twombly et al., 1996; Yakoby et al., 2008b). The BMP inhibitor gene Daughters against dpp (Dad) is a known target of BMP signaling in the imaginal wing discs and the follicle cells (Marmion et al., 2013; Weiss et al., 2010). This gene is regulated directly by the activated intracellular BMP signaling transducer complex P-Mad/MED. At the same time, dpp is also expressed in an anterior pattern that is strikingly similar to the expression pattern of Dad (Twombly et al., 1996). However, dpp is not regulated by BMP signaling; rather, it is regulated by a combination of the EGFR and Janus kinase/signal transducer and activator of transcription (JAK/STAT) pathways (Peri and Roth, 2000; Xi et al., 2003). This was recently confirmed by cis-regulatory module (CRM) analysis screening using the FlyLight collection (Revaitis et al., 2017). The Dad CRM responded to ectopic BMP signaling, whereas the dpp CRM did not. This raises the following question: Are all genes expressed in the anterior domain regulated by the same mechanism?

The expression of wit is maintained in a positive-feedback loop at the anterior domain of the follicular epithelium (Marmion et al., 2013). However, the precise nature of the regulatory mechanism is not completely understood. Motivated by the subtle differences in the anterior pattern of wit across Drosophila species, we aimed to dissect the nucleotide-level changes in the regulatory domain of wit patterning between two fly species. We used CRISPR/Cas9 gene editing to test changes within the genomic context of wit in D. melanogaster. We found that wit is regulated indirectly by BMP signaling through the alleviation of Brinker (BRK) repression. This mechanism is maintained in D. virilis, a species that is separated from D. melanogaster by 45 million years. We recapitulated the local arrangement of other potential binding sites from D. virilis within D. melanogaster, but, surprisingly, this did not affect WIT expression. We conclude that a yet undiscovered positive element is changed between these two species.

In D. melanogaster, wit is expressed in a band of cells that is on average 2.4±0.1 (mean±s.e.m.) cells wide (Fig. 1A) (Yakoby et al., 2008a). The corresponding P-Mad pattern at stage 10A is on average 1.7±0.1 (mean±s.e.m.) cells wide (Fig. 1B). Previously, we demonstrated the conservation of this simple expression pattern in closely related species, including D. sechellia and D. simulans (Marmion et al., 2013). We hypothesized that in more evolutionarily divergent species of Drosophilid, we might find different patterns of wit, as has been shown for the type I BMP receptor thickveins (tkv) (Niepielko et al., 2011, 2012). We selected D. virilis to test patterning changes, which accounts for ∼45 million years of evolution. We found that not only was wit patterning wider (Fig. 1C) but it is correlated with changes in the width of P-Mad in this species (Fig. 1D). Specifically, wit is expressed in a band that is 4.8±0.2 (mean±s.e.m.) cells wide on average. The P-Mad pattern band had an average width of 3.0±0.1 (mean±s.e.m.) cells. These are both significantly different from the corresponding expression patterns in D. melanogaster (P=3.3×10⁻¹⁴ and 1.4×10⁻¹⁰, respectively).

Intrigued by the wider expression pattern of wit in D. virilis, we aimed to understand whether the patterning change is regulated in cis or is unique to D. virilis and thus regulated in trans. To examine the mechanism of gene patterning within the endogenous context, we engineered a knockout allele of wit (Fig. 2A, Fig. S1). We replaced the entire wit locus in D. melanogaster with a φC31 site, including most of the previously identified follicle cell-associated wit enhancer (only 217 bp remaining out of 1261 bp of witZ) (Marmion et al., 2013). Importantly, as we show later in Fig. 3, a functional short fragment of witZ, which we name wit6 (Fig. 3E,F), was completely removed (Fig. 2A, Figs S1 and S2). This φC31 site served as a hub for genetic modification within the wit locus (Fig. 2B). The recovered knockout allele of wit is homozygous lethal; pharate adults fail to eclose from the pupal casings (Fig. S1C), consistent with other null alleles of wit (Aberle et al., 2002; Marqués et al., 2002).

First, we generated a construct that contained the wild-type wit locus from D. melanogaster, which was inserted into the φC31 site. This allele is homozygous viable and reproduced the wild-type anterior pattern of WIT (Fig. 2D). Also, a CFP-tagged wit was able to rescue the null fly (Fig. S3). The corresponding pattern of P-Mad was detected in a band of an average of 1.9±0.1 (mean±s.e.m.) cells wide (Fig. 2E), which was not significantly different from OreR (P=0.34) (Fig. 1B). Next, we created a construct that includes the entire wit locus from D. virilis (Fig. 2C), and tested whether it can rescue a D. melanogaster wit null fly. This construct was targeted to the φC31 site in the wit locus. Consistent with the endogenous pattern of wit in D. virilis (Fig. 1C), the vir-wit was expressed in an anterior pattern of 4.4±0.2 (mean±s.e.m.) cells wide in addition to a low level of uniform expression (Fig. 2F). This suggests that the regulatory elements of vir-wit are included in this locus and the differences between the species are changes in cis. Surprisingly, unlike the wide pattern of P-Mad in D. virilis (Fig. 1D), the P-Mad domain is similar to the pattern found in D. melanogaster (Fig. 1D and Fig. 2G) (Yakoby et al., 2008b). We reason that the anteriorly secreted DPP ligands bind to a uniformly expressed TKV (Mantrova et al., 1999; Yakoby et al., 2008b) and the anteriorly expressed vir-WIT; consequently, BMP signaling is activated proximal to the DPP source.

Following up on the pattern of vir-wit in D. melanogaster, we sought to determine the mechanism underlying wit regulation. We screened two potential enhancers from D. virilis using a lacZ reporter in D. melanogaster (Fig. 3A). The Brinker (BRK, GGCGYY) and phosphorylated Mothers against DPP (P-Mad, GRCGNC) binding sites have much in common (Hamaratoglu et al., 2014; Zhang et al., 2001). We characterized two fragments that contain a BRK/P-Mad binding site (GGCGCC). The first fragment tested (vir1) failed to produce any expression in the follicle cells (not shown). However, the second enhancer tested (vir2) produced expression that had a distinct wide anterior expression pattern on top of a uniform background expression (Fig. 3B). This pattern is similar to the endogenous pattern of vir-wit in D. virilis and D. melanogaster (Fig. 1C and Fig. 2F), but different from the anteriorly restricted pattern of the previously characterized 1.26 kb fragment of the wit enhancer (witZ) from D. melanogaster (Fig. 3C) (Marmion et al., 2013).

Target genes of BMP signaling are regulated by direct binding of the P-Mad/MEDEA (P-Mad/MED) complex or by repression of the transcriptional inhibitor BRK (Kirkpatrick et al., 2001; Müller et al., 2003; Pyrowolakis et al., 2004; Weiss et al., 2010). The vir2 enhancer contains only one MED site adjacent to the BRK/P-Mad site (there are no nucleotides between the predicted P-Mad and MED binding sites) (Fig. 3A). The previously described activation/silencer element (AE/SE) requires a nucleotide linker of 5 nt. Although this linker length is variable, a linker as short as 3 nt reduces binding considerably (Pyrowolakis et al., 2004). Based on the position of the MED binding site in the vir2 enhancer in relation to the position of the BRK/P-Mad binding site, we did not expect that the P-Mad/MED complex would be able to bind and activate expression through this site (Weiss et al., 2010) and would instead be regulated by BRK. Measured laterally, the intensity of the vir2 anterior band was 70±4.3 a.u. (mean±s.e.m.) and the uniform level was 36±1.1 a.u. (mean±s.e.m.) (Fig. 3B). At the same time, in the dpp>BRK background, the anterior band was decreased to 44±3.0 a.u. (mean±s.e.m.) and the uniform level was 32.5±1.7 a.u. (mean±s.e.m.) (Fig. 3D). We note that the intensity of anterior expression of vir2 in the dpp>BRK background (Fig. 3D) was reduced compared with the wild type (Fig. 3B) (P=0.0002). As expected, the intensity of uniform expression remained unchanged (P=0.08). On the dorsal anterior side, the approximately four-cell-wide band of vir2 expression was reduced to an approximately one-cell-wide fragmented band (Fig. 3B,D). These results further support the role of BRK repression in the regulation of the D. virilis wit gene. Our attempts to separate this enhancer to three smaller fragments did not produce expression (not shown).

We previously reported that brk mutant clones did not lead to detectable WIT expression in D. melanogaster (Marmion et al., 2013). This suggests that in D. melanogaster, wit is regulated by P-Mad/MED binding. To compare between the enhancers from the two species, we aimed to dissect further the regulation of wit in D. melanogaster. We divided the witZ into nine smaller fragments (Fig. 3E). The consensus between BRK and P-Mad is largely overlapping, so the regulation could be through the contribution of MED binding sites and/or the competition between BRK and P-Mad binding. The average size of each of the nine fragments covering witZ is ∼240 bp (Fig. 3E). Of all tested fragments, only fragment 6 (wit6) generates a pattern that is similar to the witZ pattern (Fig. 3F). The wit6 fragment contains an exact match to the BRK/P-Mad binding site (GGCGCC).

Based on the assumption that wit is positively regulated by BMP signaling (Marmion et al., 2013), we investigated the role of proximal MED sites (GNCN) in regulation of the proper gene patterning of wit. Three predicted MED binding sites exist proximal to the BRK/P-Mad site; N-nucleotide spacers of −3, +1 and +6 are found within wit6 (Fig. 3E). Based on our previous analysis of witZ (Marmion et al., 2013), we were surprised that mutation of any of the three MED sites did not perturb the expression pattern (Fig. 3G-I). We next mutated the BRK/P-Mad site in order to reduce binding of both factors, expecting to lose or considerably reduce expression. However, we were surprised to find a uniform expansion of the tested enhancer (Fig. 3J). Next, we tested this further by mutating only the BRK binding (P-Mad binding is still present). The uniform expansion remained (Fig. 3K). These results support an alternative regulation mechanism by the BMP negative regulator BRK (Hamaratoglu et al., 2014), similar to what was found for vir2.

The enhancer analysis is contrary to our previous analysis of brk null clones that failed to detect an increase in wit expression (Marmion et al., 2013). Thus, we aimed to resolve this discrepancy. The pattern of P-Mad and BRK expression is found in anti-correlated graded domains (Fig. 4A-C). We repeated experiments for brk clones; however, this time, we halved the time of dissection and permeabilization during the immunohistochemistry procedure (see Materials and Methods for more details). Under these modified conditions, we detected WIT in cells null for brk (Fig. 4D,F). We reason that the stability of WIT in various regions of the follicle cells is different. By reducing the time from dissection to fixation, the levels of WIT were above the detection level. We did not detect changes in the P-Mad pattern in this perturbation (Fig. 4E).

To further explore the regulation of wit by BRK, we utilized the φC31 site in the wit locus (Fig. 2) to test specific modifications in the binding sites within the wit locus (Fig. 4G, Fig. S4). Consistent with our prediction of patterning by BRK, complete removal of the BRK binding site resulted in a uniform high expression of WIT (Fig. 4H). P-Mad was detected in a band of 2.2±0.1 (mean±s.e.m.) cells wide on average, similar to the wild-type pattern (Fig. 1B). In about 12% of the egg chambers, the removal of this site also resulted in ectopic detection of P-Mad (Fig. 4I,I′). Interestingly, this allele is homozygous viable and generated no changes in eggshell morphologies. Unlike ectopic expression of the type I receptor TKV, which mainly restricts DPP diffusion and consequently confines signaling near the ligand source (Crickmore and Mann, 2006; Niepielko et al., 2012), ectopic expression of the type II receptors (WIT and/or Punt) led to activation of BMP signaling (not shown). We reason that the levels of ectopic WIT expression were sufficient to activate the pathway above the detection limit. In this case, a complete removal of the BRK binding site was sufficient to increase the levels of wit expression, and consequently increase BMP signaling levels above the detection limit within a portion of the egg chambers (Fig. 4I′).

The noteworthy differences in the CRMs for D. melanogaster and D. virilis are the number of MED binding sites and their proximity to the BRK/P-Mad site (Fig. 3A,E). As shown above, wit is regulated by BRK in both species. However, we aimed to determine whether the patterning divergence between the two species depends on the proximal environment of the BRK/P-Mad binding site. We replaced the BRK/P-Mad/MED site in D. melanogaster, which spans 21 nucleotides, with 21 nucleotides that are proximal to the BRK/P-Mad site in D. virilis (Fig. 3A and Fig. 4G). This virCRM allele is homozygous viable and WIT expression was a similar width to that induced by the wild-type rescue construct (P=0.79) (Fig. 4J). The P-Mad width was consistent with WIT expression with a pattern that was 1.8±0.1 (mean±s.e.m.) cells wide (Fig. 4K). Thus, the sequence around the BRK/P-Mad site in D. virilis cannot account for the difference in WIT patterning between species.

The anterior patterning of the follicle cells seems simple (Yakoby et al., 2008a); however, various mechanisms control this process. The BMP inhibitor Dad is regulated directly by BMP signaling in the imaginal wing discs and the follicle cells (Marmion et al., 2013; Weiss et al., 2010). At the same time, dpp expression is regulated by a combination of the EGFR and JAK/STAT pathways (Peri and Roth, 2000; Xi et al., 2003). This was recently confirmed by a CRM analysis screen using the FlyLight collection (Revaitis et al., 2017). The same screen found additional CRMs with different extents of anterior coverage, including an anterior portion of the stretched cells, all stretched cells, and stretched cells and centripetally migrating follicle cells. This suggests additional complexity in the regulation of genes in the anterior domain. Identifying these CRMs opens an opportunity to answer whether additional mechanisms control the anterior patterning of the follicle cells.

The vir2 enhancer controls a wide anterior expression pattern of the reporter, in addition to a lower uniform pattern throughout the follicle cells (Fig. 3B). However, replacing the 21 nucleotides proximal to the BRK binding site of D. virilis in the context of D. melanogaster generated an anterior pattern of WIT that is indistinguishable from the endogenous pattern of WIT in D. melanogaster (Fig. 4J). In this case, the vir2 fragment contains additional information for a uniform expression that is independent of the predicted BRK binding site. This type of uniform expression during oogenesis is common to other genes, including the early patterns of wit, thickveins (tkv) and br (Deng and Bownes, 1997; Fuchs et al., 2012; Mantrova et al., 1999; Marmion et al., 2013; Yakoby et al., 2008a,b). The early expression of br was shown to be regulated by Notch signaling (Jia et al., 2014), but the regulator of uniform expression is still unknown. Interestingly, these patterns are dynamically regulated at later stages of development. Although the three genes remain in an overlapping pattern in the ventral anterior domain of the follicle cells, br and tkv also appear on the dorsal side as two patches on either side of the dorsal midline (Deng and Bownes, 1997; Fuchs et al., 2012), and wit becomes absent from the dorsal anterior domain. The dorsal repression of wit is likely to be due to high levels of BRK induction as a result of activation of EGFR signaling in the dorsal anterior domain (Chen and Schüpbach, 2006). The wit^brk− reporter drives patchy dorsal anterior expression (Fig. 3K), which may indicate that the modified BRK binding site can still bind BRK, especially at the dorsal anterior domain where BRK expression is highest. Further support for this assumption is the uniform expression of WIT in the follicle cells when the BRK binding site is completely removed (Fig. 4H). We hypothesize that an as-yet-uncharacterized transcriptional activator has stronger or more abundant binding sites within the D. virilis CRM, accounting for the cis changes in the anterior pattern.

The eggshells of D. melanogaster and D. virilis differ in several aspects of their morphology, including the number of respiratory dorsal appendages (two and four, respectively), and a larger operculum in D. virilis (Niepielko et al., 2011). The wider anterior pattern of P-Mad in D. virilis was reasoned to be due to the late uniform expression of TKV, in contrast to the anterior clearing of TKV in D. melanogaster. Here, although WIT is uniformly expressed when the BRK binding site is removed, it still does not change the pattern of P-Mad or eggshell morphologies. In the future, it will be interesting to determine whether the uniform expression of both WIT and TKV in D. virilis is required for the expansion of the P-Mad domain as well as the operculum size.

Oregon R (OreR) Bloomington #25211 was utilized as wild type for D. melanogaster. D. virilis was provided by the Drosophila Species Stock Center at UC San Diego (stock #15010-1051.88). Clonal analysis of Brinker was conducted using the Flp/FRT system (Theodosiou and Xu, 1998) utilizing: FRT^19A BRK^M68/FRT^19A ubiGFP; e22c-GAL4 UAS-flp. A UAS-BRK (a gift from J. Duffy, Worcester Polytechnic Institute, MA, USA) was expressed with the oogenesis dpp:GAL4 driver (Revaitis et al., 2017). The trap lacZ of Brinker was used to mark BRK expression (X47[PlacZ(ry+)]) (Campbell and Tomlinson, 1999). All other fly stocks were generated for this study. Four- to seven-day-old flies were used in this study. Only female flies were dissected.

witZ1-9 fragments were amplified in pairs of forward and reverse primers and cloned into a gateway entry vector, PCR8. For the D. virilis enhancers, split1 was amplified with vir_big_F and vir_split1_R, and split2 was amplified with vir_split2_F and vir_small_R. The full sequences of the corresponding primers can be found in Table S1. As reported previously, prospective enhancers were subcloned into a lacZ reporter vector by Gateway reaction and inserted by φC31 injection into the genomic site attP2 at 68A4 of chromosome III (Marmion et al., 2013).

Two CRISPR targets were chosen that cut the wit locus within the 3′ UTR and in the upstream intergenic region using online prediction software (Housden et al., 2015; Ren et al., 2013). CRISPR chiRNAs were cloned into U6-chiRNA (Gratz et al., 2014). This was achieved by annealing sense and antisense unphosphorylated oligos at 10 μM in ligation buffer with the following thermocycler settings: 95°C for 5 min, then ramping to 25°C at a rate of −0.1°C/s. Annealed oligos were diluted (1:100) and 30 fmol of each chiRNA were ligated to 10 fmol of vector that had been digested with BbsI [New England Biolabs (NEB), R3539S], excluding the dephosphorylation step in a 10 μl reaction (cttcGTCTTGGACAAGAGCGAAAC and aaacGTTTCGCTCTTGTCCAAGAC, cttcGCGCTCAGCTATGCTCCCAT and aaacATGGGAGCATAGCTGAGCGC). Similarly, homology arms were amplified such that they were directly adjacent to the CRISPR cut sites. Homology arms were cloned into pdsRed_attP using cloning sites AarI (Thermo Fisher Scientific, ER1581) and SapI (NEB, R0569S) in subsequent cloning/transformation reactions. Flies containing a nos:Cas9 in the attP40 site of chromosome II were injected with a mixture of all three plasmids by Rainbow Transgenic Flies. Flies positive for repair by homology directed repair had red glowing eyes (dsRed). PCR confirmation of a heterozygous knockout was confirmed by mixture of primers wit_crispr_conf_left, wit_crispr_conf_right, and wit_crispr_WT_right. (The full sequence of the corresponding primers can be found in Table S2.) Separate PCR reactions were used to verify each side of the mutation: left arm by wit_crispr_conf_right and wit_outside_1_for and sequenced with wit_outside_1_for, and right arm by wit_crispr_conf2_rev and wit_outside_2_rev and sequenced with wit_outside_2_rev.

The entire wit locus was amplified using the primers wit_locus_left and wit_locus_right from gDNA and purified by phenyl-chloroform extraction and then cloned into pCR8. An FRT variant of the locus was amplified using the primers wit_locus_L_FRT and wit_locus_R_FRT. These flies were subsequently balanced to a fly that contained a nos:φC31 on the X chromosome (Bloomington Drosophila Stock Center, 34771). The full sequence of the corresponding primers can be found in Table S3.

Gibson assemblies were conducted using NEB HiFi Assembly Master Mix (2621S) containing 10 fmol of each fragment in a 10 μl reaction. Assemblies created a circular product with the vector backbone pSC. pSC is a linear PCR product of the primers pSC_F and pSC_R, using a pSC cloning kit template (Stratagene, 240205). Assemblies were cloned and screened by restriction fragment mapping in DH10B Escherichia coli with EcoRI (NEB, R3101S) or EcoRV (NEB, R3195S). Assembly of a D. melanogaster wit locus with a substituted D. virilis SMAD binding site utilized the primers pSC2vir_wit_L_F, vir_wit_L2R_R, vir_wit_R2L_F and vir_wit_R2pSC_R. Assembly of a CFP fusion of WIT utilized the primers wit_locus_left_R, wit_locus_right_F, pSC2vir_wit_L_F, vir_wit_R2pSC_R, wit_CFP_F and wit_CFP_R. The insertion site of CFP followed the signal polypeptide as predicted computationally (Tamura et al., 2011). A deleted SMAD locus was created utilizing the primers pSC2vir_wit_L_F, vir_wit_R2pSC_R, assembly_delSmad_for and assem_dSmad_rev_sho. See Fig. S4 for diagrams of all wit rescue constructs. Primer sequences can be found in Table S3.

The donor vector, pGE-attB-GMR_GW, was created by modifying pGE-attB-GMR (Huang et al., 2009) in order to add a Gateway cloning cassette. This was accomplished by digestion with EcoRI (NEB, R3101S) and KpnI (NEB, R3142S), followed by Gibson assembly of a PCR-amplified Gateway cassette with gggcgcgtactccacGTTTATCACCACTTTGTACAAG and catacattatacgaagttatGTTTATCACAAGTTTGTACAAAAAAG. The opposite Gateway cassette was used to make pGE-attB-GRM_GWflip. Mutant loci were subcloned with wit_locus primers including or excluding FRT sites into pCR8 and LR recombined into pGE-attB-GMR. Loci were gateway assembled such that the rescue fragment was in the same orientation as wit. Destination vectors were injected into null flies containing nos:φC31 and positive flies were selected by expression of red pigmentation in the eye (Bloomington Drosophila Stock Center, 34771).

Species-specific cDNA libraries were constructed and wit was amplified using the degenerate primers reported previously (Marmion et al., 2013). In situ hybridization was performed as reported previously (Niepielko et al., 2014). Dissections and fixation were performed as reported by others (Pacquelet and Rørth, 2005). We reduced the times for dissection and permeabilization by half in order to detect proteins with high turnover rate or low abundance at the cell membrane. Specifically, four flies were dissected at a time and moved into a fixation solution. After four cycles, the fixation solution was replaced with a fresh solution, and an additional 10 min of fixation was completed before washes. Primary antibodies were: mouse anti-Wit [23C7; 1:500, Developmental Studies Hybridoma Bank (DSHB)], rabbit anti-phosphorylated-Smad1/5/8 (1:3000, a generous gift from D. Vasiliauskas, S. Morton, T. Jessell and E. Laufer, Columbia University, NY, USA), mouse anti-β-galactosidase (1:1000, Promega, Z3781), rabbit anti-β-galactosidase (1:1000, Invitrogen, A-11132) and sheep anti-GFP (1:2000, Biogenesis, AHP2984). Secondary antibodies were Alexa Fluor 488-, 568- and 647- conjugated (1:1000, Molecular Probes, A-21206, A-21202, A-11015, A10042, A10037 and A-31571). DAPI (1:10,000, Invitrogen, D1306) was used to stain nuclei. Images were captured with a Leica SP8 Confocal microscope. Images were processed with ImageJ (Rasband, 1997–2009) and Gimp (GNU Image Manipulation Program, 1995–2008). The intensity of the vir2 reporter line was measured by a box on the lateral edge of approximately ten cells wide and four cells tall (marked in Fig. 3B,D). A plot profile of intensity was measured from the anterior boundary of the oocyte-associated follicle cells into the main body follicle cells (ImageJ), and expressed in arbitrary units (a.u.). All statistical analysis was conducted with unpaired two-tailed t-tests. Sample sizes are indicated in figure panels.

We thank V. Veikkolainen and G. Pyrowolakis for continued discussions, and greatly appreciate G. Pyrowolakis for his careful reading of the manuscript. We also thank the members of the Yakoby Lab for insightful discussions. We are grateful to D. Vasiliauskas, S. Morton, T. Jessell and E. Laufer for providing the P-SMAD antibody. We are thankful for the Drosophila species fly stock of D. virilis and the Developmental Studies Hybridoma Bank for antibodies.