Diverse neuronal subtypes are generated by the gene regulatory activities of subtype-specific combinations of transcription factors (Allan and Thor, 2015; da Silva and Wang, 2011; Hobert et al., 2010). In postmitotic neurons, retrograde signaling continues to play prominent roles during neuronal maturation and through life in regulating terminal differentiation and synaptic function (Allan et al., 2003; Angley et al., 2003; Veverytsa and Allan, 2011; Berke et al., 2013; Chou et al., 2013; Kelly et al., 2013; Majdazari et al., 2013; Xiao et al., 2013; DuVal et al., 2014; McCabe et al., 2003; Guha et al., 2004; Lopez-Coviella et al., 2005; Hodge et al., 2007; Pavelock et al., 2007; Miguel-Aliaga et al., 2008; Eade and Allan, 2009; Henríquez et al., 2011). In Drosophila, retrograde BMP signaling occurs within most efferent neurons, regulating a diverse suite of target genes in a subtype-specific manner; for example, the neuropeptide genes unique to each neurosecretory subtype, and the partially overlapping battery of genes that support the growth and strengthening of motor neuron neuromuscular junction synapses (Veverytsa and Allan, 2011; Miguel-Aliaga et al., 2008; Allan et al., 2003; Kim and Marqués, 2010; Vuilleumier et al., 2019; Ball et al., 2010; Marqués et al., 2003). An unresolved question regards how BMP target genes are differentially specified within each of these BMP-activated neuron subtypes. Prevailing models suggest that subtype-specific transcription factors play the primary role in target gene selection, and BMP-responsive DNA motifs within those target gene enhancers encode generic responsiveness to BMP signaling in any neuron (Veverytsa and Allan, 2011; Miguel-Aliaga et al., 2008; Allan et al., 2005). Here we describe our identification of a low affinity BMP-responsive DNA motif that instructively contributes to the differential specification of BMP target genes within efferent neuron subtypes.

In Drosophila efferent neurons, retrograde BMP-signaling is triggered by Glass bottom boat (Gbb) ligand engagement at a presynaptic BMP-Receptor (BMP-R) complex of Wishful thinking (Wit), Thickveins (Tkv), and Saxophone (Sax). This leads to phosphorylation of Mad to pMad (vertebrate SMAD1/5/9) (Allan et al., 2003; McCabe et al., 2003; Aberle et al., 2002; Rawson et al., 2003), which then complexes with Medea (vertebrate SMAD4) (Gao and Laughon, 2007; Gao et al., 2005). This pMad/Medea complex translocates to the nucleus and regulates transcription by binding DNA in a sequence-specific manner at BMP-response elements (BMP-REs) (Kim et al., 1997; Xu et al., 1998; Shi and Massagué, 2003). In the nucleus, Schnurri (Shn) can act as a co-repressor with the pMad/Medea complex at silencer BMP-REs (BMP-SE) to repress target genes (Gao et al., 2005; Pyrowolakis et al., 2004; Charbonnier et al., 2015). Additionally, Brinker (Brk) is a DNA binding default repressor of genes in the absence of BMP-signaling (Barolo and Posakony, 2002; Affolter and Basler, 2007; Hamaratoglu et al., 2014).

Studies of BMP-dependent morphogenesis in the Drosophila embryo and wing imaginal disc have established core principles and sequence preferences for BMP-REs and transcriptional regulators of the BMP pathway. BMP-dependent gene activation is mediated directly by pMad/Medea acting at a bipartite 15 bp BMP-AE; GGCGCCA(N₄)GNCV (Weiss et al., 2010; Chayengia et al., 2019). BMP-dependent gene silencing is mediated directly by pMad, Medea, and Shn acting in a complex at a bipartite 15 bp BMP-SE; GRCGNC(N₅)GNCT (Gao et al., 2005; Pyrowolakis et al., 2004; Charbonnier et al., 2015). In these complexes, two pMad proteins bind to the 6 bp GC-rich sequence, and a single Medea protein binds to the 4 bp GNCV/GNCT sequence (Gao et al., 2005). An additional widely-utilized mechanism for BMP target gene activation in Drosophila involves brk-dependent de-repression (Affolter and Basler, 2007). Here BMP signaling results in pMad, Medea and Shn binding to multiple BMP-SE motifs in the brk locus (Pyrowolakis et al., 2004; Marty et al., 2000; Torres-Vazquez et al., 2000; Müller et al., 2003). Subsequent reduction in Brk levels relieves its default repression of many genes, which thereby become de-repressed in a BMP-dependent manner (Pyrowolakis et al., 2004; Jaźwińska et al., 1999; Campbell and Tomlinson, 1999; Winter and Campbell, 2004; Minami et al., 1999; Moser and Campbell, 2005).

We recently demonstrated that the BMP-AE motif serves as a widely-deployed BMP-dependent activator of gene expression in larval Drosophila efferent neurons (Vuilleumier et al., 2019). In that study, BMP-AE motifs were shown to confer BMP-dependence on enhancers that were active in diverse patterns across the many subtypes of efferent neurons. This diversity did not appear to be encoded by sequence differences within BMP-AEs. For example, enhancers that included similar 15 bp BMP-AE sequences differed greatly in expression pattern (such as Van 50 and Van 27 reporters), and swapping BMP-AE sequences between enhancers failed to change expression pattern (Vuilleumier et al., 2019). This notion is supported by recent work in Drosophila follicular epithelium and wing imaginal disc showing that the cis-regulatory environment around BMP-AE motifs, bound by other transcription factors, determines the enhancer’s spatiotemporal activity, while the BMP-AE itself acts as a ‘monotonic’, or generic, interpreter of BMP activity (Chayengia et al., 2019).

The integration of BMP-activated pMad/Medea with subtype-specific transcription factors in Drosophila neurons has only been examined in detail in Tv4-neurons (Chayengia et al., 2019). Here a Tv4-neuron-specific transcription factor code and pMad/Medea were found to bind within a Tv4neuron-specific enhancer of the FMRFa gene (Allan et al., 2003; Allan et al., 2005; Miguel-Aliaga et al., 2004; Berndt et al., 2015; Benveniste et al., 1998), at a minimal 25 bp Homeodomain-Responsive Element, and a minimal 39 bp BMP-Responsive Element, respectively (Berndt et al., 2015). Misexpression of combinations of these transcription factors in other efferent neurons (with activated retrograde BMP signaling) led to widespread ectopic FMRFa gene activation (Allan et al., 2003; Eade and Allan, 2009; Allan et al., 2005; Miguel-Aliaga et al., 2004; Berndt et al., 2015; Benveniste et al., 1998; Baumgardt et al., 2009). By contrast, neither ectopic activation nor reduction in BMP signaling was found to alter Tv4-specific FMRFa expression. A parsimonious model emerging from these studies would suggest that activator BMP-RE motifs add a generic BMP-responsive input for an enhancer whose spatiotemporal activity is otherwise determined by subtype-specific transcription factors.

Here we provide evidence for a second model, in which low affinity BMP-RE motifs instructively contribute to subtype-specific gene expression within subsets of BMP-activated efferent neurons. By functional and biochemical dissection of the 39 bp BMP-responsive cis-element of the FMRFa enhancer, we identified a non-canonical minimal pMad/Medea binding motif [GGCGCC(N₅)GTAT], which has low affinity (LA) compared to BMP-AE and BMP-SE motifs, that we herein term BMP-LA. Our genetic analysis showed that pMad and Medea act through this motif to directly activate FMRFa expression, without any contribution from brinker or schnurri. Notably, by converting this motif into a high affinity BMP-AE motif, we observed ectopic expression of BMP-dependent reporter activity into other neuronal populations, and a loss of expression in the Tv4-neuron itself. Finally, in silico searches and subsequent enhancer activity analysis revealed other functional instances of this motif in the genome. Overall, our results identify a novel low affinity BMP-RE motif type (BMP-LA), and lead to our proposal that the relative strength of BMP-responsive motifs is utilized to confer subtype-specific expression of BMP target genes in neurons.

A 445 bp enhancer region mediates Tv4-neuron-specific expression of the FMRFa gene (Benveniste et al., 1998; Benveniste and Taghert, 1999). We recently sub-mapped this enhancer and defined two short cis-elements that are together necessary for enhancer activity, but are not individually sufficient (Berndt et al., 2015). Toward defining the information encoded by each cis-element, we found that concatemers for both cis-elements independently generate highly specific Tv4-neuron activity in late embryos. Thus, both contain sufficient information for Tv4-specific activity. This allowed us to explore their transcription factor inputs, leading to their definition as a homeodomain-responsive cis-element (HD-RE) and a BMP-responsive cis-element (BMP-RCE). The HD-RE recruits and is activated by the LIM-homeodomain transcription factor, Apterous, but does not require BMP-signaling for activation. The BMP-RCE requires BMP-signaling for its activity and recruits pMad sequence-specifically at a GGCGCC site (Berndt et al., 2015; Figure 1A).

Figure 1 with 1 supplement see all

Download asset Open asset

pMad and Medea are direct activators of FMRFa

The lack of a canonical Medea site, necessary for BMP-RE motif activity, led us to test if Medea is even required for FMRFa expression. In two different null Medea backgrounds, immunoreactivity to the FMRFa prepropeptide was entirely lost (Figure 1D–F). Anti-Eya staining revealed that all four Tv neurons were generated. We also examined EYFP reporter expression driven from the full length 445 bp Tv^WT-EYFP enhancer, and from the concatemerized BMP-RCE and HD-RE reporters (Berndt et al., 2015). As expected, Tv^WT-EYFP and BMP-RCE-EYFP expression were eliminated, while the BMP-insensitive HD-RE-EYFP exhibited expression comparable to controls (Figure 1—figure supplement 1D–I). Thus, Medea is selectively required for the BMP input that activates FMRFa expression.

The apparent dichotomy of Mad and Medea-dependence in the absence of a canonical BMP-RE motif led us to test whether brinker (brk) de-repression accounts for BMP-dependent FMRFa expression (Figure 2A). In numerous Drosophila tissues outside the nervous system, brk-mediated de-repression results in BMP-dependent gene expression (Hamaratoglu et al., 2014). In the absence of BMP-signaling, Brk binds GGCGYY motifs to default repress many genes that are activated upon the onset of BMP signaling (Zhang et al., 2001). In the presence on BMP signaling, pMad, Medea, and Shn bind BMP-SE motifs to repress brk (Pyrowolakis et al., 2004; Marty et al., 2000; Torres-Vazquez et al., 2000; Müller et al., 2003), lowering its expression and resulting in the de-repression of BMP gene targets (Pyrowolakis et al., 2004; Jaźwińska et al., 1999; Campbell and Tomlinson, 1999; Winter and Campbell, 2004; Minami et al., 1999; Moser and Campbell, 2005).

Figure 2

Download asset Open asset

The BMP-responsive GGCGCC sequence in the FMRFa BMP-RCE matches this Brk consensus motif (Zhang et al., 2001). If a Brk de-repression model were correct for FMRFa expression (Figure 2H), BMP-induced pMad, Medea, and Schnurri (Shn) would be required to silence brk, resulting in FMRFa de-repression (Marty et al., 2000; Müller et al., 2003). Thus, we would expect that Brk expression would be increased in wit nulls, resulting in FMRFa repression. To test this model genetically, we first examined brk^XA/Y hemizygotes, but observed no change in either anti-FMRFa immunoreactivity or Tv4^WT-EYFP reporter expression (Figure 2B,E). Second, we tested brk;;wit double mutants to test if the absence of FMRFa in wit mutants is due to upregulation of the brk repressor. Contrary to this hypothesis, FMRFa expression was absent in brk/Y;;wit^A12/wit^B11 double mutants (Figure 2D,E), phenocopying wit mutants (Figure 2C,E).

However, this analysis did not rule out the possibility that FMRFa activation requires de-repression by brk as well as direct activation by pMad/Medea, as occurs at BMP-AE motifs regulating dad (Weiss et al., 2010). In such a model, the absence of FMRFa in brk; ;wit nulls may be due to a lack of activation by pMad/Medea. Therefore, we tested FMRFa expression in shn nulls. In this genotype, Brinker and pMad/Medea would all be expressed. In shn (Allan and Thor, 2015) null mutants, we found that FMRFa immunoreactivity was wildtype in late stage 17 embryos, the latest age testable in these mutants due to shn (Allan and Thor, 2015) late embryonic lethality (Figure 2G). Thus, brk de-repression cannot function as the primary mechanism for BMP-dependent FMRFa expression. We conclude that FMRFa is directly activated by pMad/Medea (Figure 2I), with no apparent involvement for shn and brk-mediated de-repression.

The 15 bp FMRFa BMP-RE has reduced pMad/Medea binding relative to canonical BMP-REs

These data led us to test if the pMad/Medea complex is recruited by a non-canonical motif within the FMRFa BMP-RCE. We used an Electrophoretic Mobility Shift Assay (EMSA) to examine nucleotides essential for pMad/Medea binding. We activated BMP signaling in S2 cells, by transfecting S2 cells with FLAG::Mad, Myc::Medea, and activated BMP type I receptor Thickveins (Tkv^QD) (Gao et al., 2005). From these cells, we obtained total cell lysates to perform EMSA tests on IRDye700-tagged DNA oligonucleotides containing a 27 bp BMP-RE sequence. In Figure 3, we show that co-transfection of all three plasmids was required for a band shift of the FMRFa BMP-RCE probe by BMP-activated S2 cell lysates. Addition of either an anti-Myc or anti-FLAG IgG to the lysate super-shifted the band, whereas a control IgG had no effect. Thus, a BMP-activated pMad/Medea complex isolated from S2 cells bind and band shift the FMRFa BMP-RCE.

Figure 3

Download asset Open asset

Next, we tested which sequences within the BMP-RCE are required for pMad/Medea recruitment. We generated an IRDye-700 tagged wildtype FMRFa BMP-RCE probe comprising a reduced 27 bp FMRFa BMP-RCE (AGAGGCGCCACAATGTATCCCGTTACA), necessary for appropriate in vivo expression (Figure 1—figure supplement 1A). We first tested candidate pMad and Medea binding sites by generating untagged probes with mutant sequence to test their ability to outcompete the wildtype tagged BMP-RE probe when pre-incubated at 10× or 100× excess. We confirmed that mutation of the pMad-binding sequence GGCGCC eliminated binding of the activated pMad/Medea complex (Figure 4A), as previously demonstrated (Berndt et al., 2015). Next, we tested candidate Medea-binding sequences by mutating the GTAT sequence (to TTAT) and the sequence CCCG (to AATT). We found that only mutation of the GTAT sequence reduced binding of the activated pMad/Medea complex (Figure 4B). This indicated that a 15 bp GGCGCC(N5)GTAT motif likely represents the pMad/Medea-recruitment motif for the FMRFa BMP-RCE. We further corroborated these data by comparing the ability of the pMad/Medea complex to bind IRDye-700-tagged BMP-RE oligonucleotides that were either wildtype or contained GTAT mutations predicted to abrogate Medea binding (ACTA or TTAT). Our results demonstrated that pMad/Medea had reduced binding to the mutant sequences (Figure 4—figure supplement 1).

Figure 4 with 1 supplement see all

Download asset Open asset

We wished to define the sequence requirements for this novel 15 bp BMP-Response Element (BMP-RE). Therefore, we examined band shifts of BMP-activated S2 cell lysates pre-incubated with tagged wildtype BMP-RE DNA oligonucleotides and a 100× stoichiometric excess of cold mutants in which a single nucleotide through the 15 bp motif was substitution mutagenized (Figure 4C–E). Mutagenesis of the entire GGCGCC sequence, or any single nucleotide therein, severely reduced binding of pMad/Medea, as shown by an inability to reduce the band shift of the tagged BMP-RE probe (Figure 4C). In addition, mutation of the entire GTAT sequence, or of any nucleotide except the A, greatly reduced the binding of pMad/Medea, as shown by retention of a strong band shift of the tagged BMP-RE probe (Figure 4E). By contrast, mutation of any nucleotide within the linker sequence ACAAT only minimally reduced pMad/Medea binding and the band shift was only minimally retained (Figure 4D). We conclude that a minimal 15 bp element of GGCGCCacaatGTaT is essential for pMad/Medea recruitment in vitro (with capitalized letters being essential).

It is intriguing that the position 14 A nucleotide in the putative GTAT Medea binding motif is the only nucleotide not required for pMad/Medea recruitment to this motif in vitro, as a C in this position was shown to be important for pMad/Medea recruitment to BMP-AE and BMP-SE motifs (Pyrowolakis et al., 2004; Weiss et al., 2010). Due to this loss of a key pMad/Medea recruitment nucleotide, we postulate that this position 14 C > A switch reduces affinity by introducing a nucleotide that fails to contribute to binding, resulting in the observed reduction of pMad/Medea recruitment.

Hereafter, we term the FMRFa BMP-RE motif as a BMP-Low Affinity Activation motif (BMP-LA). We wished to determine if the C > A switch that distinguishes the BMP-LA from the BMP-AE/SE motifs has an impact on BMP-activated pMad/Medea recruitment. To examine this, we tested whether modifying the BMP-LA Medea site to a BMP-AE-like or BMP-SE-like sequence indeed increases its ability to bind pMad/Medea. We generated a series of tagged and untagged DNA oligonucleotides comprising 27 bp of the BMP-LA sequence that was either wildtype at the Medea site, GTAT, or mutated at this site to GACG (BMP-AE-like) or GTCT (BMP-SE-like). By competition EMSA assays, we tested the relative ability of these untagged DNA oligonucleotides to compete for pMad/Medea binding when at 1×, 5×, and 10× stoichiometric ratios relative to the tagged wildtype probe (Figure 4F–G). We found that a wildtype untagged BMP-LA sequence reduced but did not eliminate the band shift generated by its tagged counterpart by 10×. In contrast, both the AE-like and the SE-like sequence mutants proved to be much stronger competitors, totally outcompeting the BMP-LA tagged probe by 5×. Moreover, at equimolar ratios, the untagged wildtype BMP-LA competitor did not substantially alter the tagged probe band shift; however, both AE-like and SE-like mutants reduced the band shift. These data indicate that the GTAT sequence displays low affinity pMad/Medea binding activity relative to the characterized BMP-AE and BMP-SE motifs.

Conversion of the low affinity FMRFa BMP-RE to a high affinity BMP-AE sequence results in ectopic neuronal BMP-dependent activity in vivo

Collectively, our genetic and biochemical data indicated that the BMP-LA motif exhibits low affinity pMad/Medea recruitment. This raised the question as to why the BMP-LA motif is attenuated in this manner, yet conserved across all Drosophila species. To address this, we tested for functional relevance for this motif’s low affinity.

First, we mutated the 445 bp Tv4-enhancer (within the Tv4-nEGFP nuclear-localized reporter) at the GTAT sequence into ACTA (Tv4^mGTAT>ACTA-nEGFP) (sequence shown in Supplementary file 1a), in order to eliminate Medea recruitment. We found that the Tv4^mGTAT>ACTA-nEGFP reporter exhibited a total loss of expression in Tv4-neurons, and no ectopic expression was detected in other neurons (Figure 5A,B). We conclude that the GTAT sequence is essential. Second, we tested the effect of converting the GTAT motif of BMP-LA to an optimal BMP-AE-like sequence (GTAT > GACG), within the Tv4-nEGFP reporter (Tv4^mGTAT>GACG-nEGFP) (sequence shown in Supplementary file 1a). This conversion eliminated Tv4-enhancer activity in Tv4-neurons (Figure 5A,C) but led to strong ectopic reporter activity in other neurons of the VNC (Figure 5D). We noted that all EGFP positive cells were also anti-pMad immunoreactive (Figure 5D’–D”’), suggesting that we had indeed created a high affinity BMP-RE that recruits pMad/Medea to generate BMP-responsive Tv4-enhancer activity in an expanded population of neurons. To confirm this, we tested the wit-dependence of this ectopic reporter activity, by examining the expression of the mutant BMP-AE-like Tv4-enhancer reporter (Tv4^mGTAT>GACG-nEGFP) in a wit mutant background (Figure 5D,E). As expected, this eliminated all reporter activity, demonstrating that the low-to-high affinity conversion created a functional BMP-AE that generated BMP-dependent reporter activity in other BMP-activated neurons of the VNC.

Figure 5

Download asset Open asset

These data show that the low affinity of BMP-LA is essential for the spatially restricted activity of this cis-element to Tv4-neurons.

Identification of additional functional BMP-responsive BMP-LA motifs through the genome

Following the identification of the novel BMP-LA motif regulating FMRFa expression in Tv4-neurons, we examined if this cis-regulatory motif was unique to FMRFa regulation, or functions to confer BMP-dependence to other cis-regulatory regions. To this end, we identified 178 BMP-LA motifs in the D. melanogaster genome using the motif discovery tool HOMER (v4.10) (Heinz et al., 2010). These were filtered for sequence conservation across 24 sequenced Drosophila species using PhastCons scores (Siepel et al., 2005), reducing the list to 128 BMP-LA with an average PhastCons score over 0.55. Of these 128 BMP-LA, 103 were highly conserved with a score of over 0.9 (Supplementary file 1b).

We selected 24 of the BMP-LA motifs for functional testing in vivo. Of the 24 prioritized motifs, 20 were highly conserved (average PhastCons score >0.9) and chosen in order to optimize the chance of characterizing functional BMP-RE's. Additionally, four motifs with lower scores 0.75–0.55 were tested for functionality (Table 1, Supplementary file 1c). We examined reporter activity driven from these genomic fragments in late third instar larvae (L3). Of the 24 reporters, five showed no expression in the ventral nerve cord (VNC) (Figure 6—figure supplement 1A), while the remaining 19 reporters exhibited low to high reporter activity in the VNC (Table 1). Of these active reporters, 10 exhibited expression in subsets of pMad-positive cells in the VNC, and also numerous pMad-negative glia and neurons (Table 1). The remaining nine active reporters were restricted to pMad-negative glia and neurons (Figure 6—figure supplement 1B). We tested the BMP-responsiveness of the 10 reporters expressed in pMad-positive cells, by placing them in a wit mutant background. Out of those 10 reporters, seven showed reduced reporter expression (Figure 6B), including two (CM5 and CM7) with motifs of lower PhastCons score. Three reporters showed no significant change (Figure 6A, Figure 6—figure supplement 1B). Quantification of reporter expression in the VNC of controls and their corresponding wit mutants revealed a loss of up to 88% of reporter-expressing cells in certain genotypes (Figure 6—figure supplement 2B).

Table 1

DNA fragment	VNC expression	Reporter/pMad stain overlap	Wit responsive	VNC expression details
CM5	+++	√	√	neurons and glia
CM1	+++	√	√	medial and lateral neurons
CM4	+++	√	√	medial and lateral neurons
CM3	+++	√	√	medial and lateral neurons
CM2	++	√	√	medial and lateral neurons
CM7	+	√	√	sparse
CM6	+	√	√	sparse
CM8	+++	√	-	medial neurons
CM9	++	√	-	lateral neurons
CM10	+	√	-	sparse
CM11	++	-	-	neurons and glia
CM12	+	-	-	sparse
CM13	+	-	-	sparse
CM14	+	-	-	sparse
CM15	+	-	-	sparse
CM16	+	-	-	sparse
CM17	+	-	-	low intensity
CM18	+	-	-	low intensity
CM19	+	-	-	low intensity
CM20	-	-	-	none
CM21	-	-	-	none
CM22	-	-	-	none
CM23	-	-	-	none
CM24	-	-	-	none

Figure 6 with 2 supplements see all

Download asset Open asset

Next, we tested whether the activity of the identified wit-responsive fragments was dependent on the BMP-LA motif within these genomic fragments. We selected four wit-responsive fragments (CM1, CM2, CM3, and CM7) and one non-wit responsive fragment (CM9; Figure 7A). We introduced nucleotide substitution mutations into the pMad-binding site of the LA (GGCGCC > TGATGA). In all four wit-responsive fragments tested, there was a significant loss of reporter expressing cells (Figure 7B). Finally, having established the necessity of Medea for BMP-dependent activity of the FMRFa BMP-RE, we tested the requirement for Medea in reporter expression these same BMP-dependent CM1, CM2, CM3, and CM7 reporters. In Medea mutant third instar (L3) larvae, reporter expression was reduced in the same pattern as observed in a wit mutant background (Figure 7—figure supplement 1).

Figure 7 with 1 supplement see all

Download asset Open asset

We conclude that the BMP-LA is utilized by numerous cis-regulatory regions in order to generate BMP-dependent enhancer activity in neurons.

Retrograde BMP signaling, and BMP responsive DNA motifs in target gene enhancers, have been viewed as functionally equivalent across efferent neuron populations. Therefore, the differential expression of BMP target genes across efferent neuron populations has been attributed to the activity of subtype-specific transcription factors (Veverytsa and Allan, 2011; Miguel-Aliaga et al., 2008; Allan et al., 2005). Here we describe our identification of a low affinity BMP-responsive DNA motif that appears to instructively contribute to differential specification of BMP target genes across efferent neuron subtypes.

Our previous work had identified a 39 bp BMP responsive cis-element within the Tv4-enhancer of FMRFa that is absolutely required for Tv4-enhancer activity, and encodes sufficient information for selective activity in Tv4-neurons (Berndt et al., 2015). Here we reveal that pMad and Medea are recruited to this cis-element by an essential low affinity motif (the BMP-LA). Converting this motif to a high affinity BMP-AE motif led to Tv4-enhancer activity in a broader set of efferent neurons, as would be expected from BMP-AE’s generic responsiveness across many efferent neurons and Drosophila tissues (Vuilleumier et al., 2019; Chayengia et al., 2019). These results provide evidence that low affinity BMP responsive motifs instructively restrict BMP target genes within subsets of efferent neuron.

Low affinity motifs serve widespread roles in restricting gene expression in space and time (Crocker et al., 2015). They function to reduce enhancer activity in order to prevent ectopic expansion of enhancer activity (Wharton et al., 2004; Farley et al., 2015; Jiang and Levine, 1993). They are also required to generate enhancer activity with the appropriate expression domain; for example, Hedgehog regulation of numerous wing imaginal disc and embryonic enhancers requires Ci transcription factor binding to low affinity motifs; increasing their affinity abolished enhancer activity (Ramos and Barolo, 2013). Also, clustered low affinity motifs have been shown to improve discrimination between transcription factors with similar binding preferences, and to locally concentrate transcription factors to ensure robust target gene expression (Crocker et al., 2015).

How does the BMP-LA confer Tv4-neuron specificity? We propose that the BMP-LA does not singularly specify Tv4 neuron activity, but rather imposes a requirement for pMad/Medea cooperative interactions with subtype-specific transcription factors bound to adjacent motifs for combinatorial activation of FMRFa expression. Four primary lines of evidence inform our conclusion. First, lowering or ectopically hyperactivating BMP signaling in other neurons failed to trigger ectopic Tv4-enhancer or FMRFa expression, without also co-misexpressing Tv4-specific transcription factors (Allan et al., 2003; Allan et al., 2005; Miguel-Aliaga et al., 2004; Eade et al., 2012; Berndt et al., 2015). This rules out a model whereby BMP-LA’s lower affinity is tuned for a specific level of BMP signaling only observed in Tv4-neurons. Second, the minimal BMP-responsive, Tv4-specific cis-element within the Tv4-enhancer cannot dispense with a highly conserved sequence 3’ of the BMP-LA. This sequence likely contains binding sites for other transcription factors that synergize with pMad/Medea to specify FMRFa expression in Tv4-neurons. These transcription factor(s) remain unidentified at this time. Third, BMP-LA motifs in other genomic regions confer BMP-dependent enhancer activity across other efferent neurons, suggesting that the BMP-LA imposes a requirement for pMad/Medea synergistic interactions with other transcription factors that combinatorially specifies neuronal subtype expression. Fourth, converting the BMP-LA to a BMP-AE resulted in a loss of reporter expression within its appropriate domain (Tv4-neurons), but an inappropriate expansion of BMP-responsive activation into other efferent neurons. This strongly suggests that the BMP-LA imposes transcription factor interactions that the BMP-AE is unable to replicate in Tv4-neurons; yet are unnecessary for generic BMP-AE activity in other efferent neurons. These specific interactions may occur between transcriptional regulators that require a specific pMad/Medea conformation that is imposed by the BMP-LA sequence. Precedent for this comes from analysis of Schnurri binding to pMad and Medea bound at the BMP-SE, which requires a pMad/Medea conformation that is imposed by a precise 5nt linker between the Mad and Medea binding motifs and also a terminal T at position 15 of the BMP-SE motif (Gao et al., 2005; Pyrowolakis et al., 2004; Müller et al., 2003). No such sequence constraints appear to be required for pMad/Medea activity at the BMP-AE (Weiss et al., 2010; Chayengia et al., 2019). Alternatively, but not mutually exclusively, these specific interactions may occur between other transcription factors bound to adjacent DNA motifs and a pMad/Medea conformation imposed by the BMP-LA sequence. Precedent for such a model comes from evidence that conversion of a low affinity Pax6 site in the chicken DC5 enhancer to a high-affinity site disrupted the formation of an activation competent Sox2-Pax6 complex, and decreased enhancer activity (Kamachi et al., 2001; Uchikawa et al., 2003). Similarly, single nucleotide differences in NF-κB binding sequences determine cofactor specificity for NF-κB dimers (Leung et al., 2004; Pan et al., 2010). A similar instructive role for low affinity motifs in generating exquisitely subtype-specific neuronal gene expression has been proposed in Drosophila (Jafari and Alenius, 2015; González et al., 2019). The exclusive expression of odorant receptor genes in specific olfactory sensory neurons is encoded by arrays of low affinity binding sites for multiple transcription factors. The authors propose that these low affinity motifs provide the necessary environment for the weak and somewhat promiscuous binding for numerous transcription factors that ultimately establishes the cooperative interactions that stabilizes target gene activation in a single neuronal subset.

Our results add to growing evidence that BMP-dependent gene regulation can be profoundly altered by subtle nucleotide substitutions within the 4-nucleotide Medea-recruitment site of these response elements; a BMP-dependent activator (GNCV), a BMP-dependent repressor (GNCT), and a low affinity activator that generates restricted activation (GTAT). The sequence similarity yet diverse functional output of these three BMP-REs reveals the diversity of responses that can be generated from subtle sequence deviations from a core BMP-RE motif. This raises a challenge for computational approaches for BMP-RE motif discovery; a specific high affinity motif will not capture all functional motif instances, but a degenerate motif built from multiple BMP-RE motifs would increase background noise to unacceptably high rates. Moreover, our discovery here of a novel motif with predictive value only increases the promise that more BMP-REs with predictive value will be discovered. However, as demonstrated by numerous studies, accounting for low affinity sites has proven to be beneficial in identifying functional enhancers (Zandvakili et al., 2018; Gurdziel et al., 2015). Therefore, it remains important to identify novel BMP-responsive motifs with demonstrated predictive value in BMP-RE discovery. Regardless, considerable degeneracy has been found in the sequences of BMP-REs (Ross and Hill, 2008) and it is important to acknowledge that in silico BMP-RE discovery will always under-represent functional BMP-REs in any system.

We started this study by sub-mapping a minimal BMP-responsive, Tv4-specific cis-element to a short 27 bp region that contains the canonical GGCGCC pMad-binding site, but no canonical Medea-binding site. After dismissing the expected model that this motif would mediate Brk default repression and BMP-dependent de-repression, we defined a minimal 15 bp GGCGCC(N5)GTAT bipartite motif required for pMad/Medea recruitment. The low affinity of this motif arises from the C > A nucleotide switch at position 14; removing the C nucleotide required for pMad/Medea recruitment to BMP-AE and BMP-SE (Pyrowolakis et al., 2004; Weiss et al., 2010), for an A nucleotide that plays no role in recruitment (Figure 4E). Interestingly, this loss of recruitment activity at motif position 14 leads to increased nucleotide stringency at positions 2, 5, 12, and 13 for pMad/Medea recruitment (Figure 4E), which are either unnecessary or less necessary for recruitment to BMP-AE and BMP-SE (Gao et al., 2005; Pyrowolakis et al., 2004; Weiss et al., 2010). Thus, transversion of one core nucleotide requires compensatory pMad/Medea recruitment activity from other nucleotides within the overall 15 bp motif. This could be viewed as exerting an evolutionary pressure on these remaining nucleotides to not exhibit degeneracy across Drosophila species, which may provide an explanation for why the BMP-LA is so highly conserved. By contrast, other low affinity cis-elements exhibit considerable degeneracy, with the hypothesis being that a high affinity, optimal motif would be highly constrained in nucleotide composition, but there could be a variety of nucleotide substitutions that would result in lower affinity motif (Crocker et al., 2015; Ramos and Barolo, 2013; Crocker et al., 2016; Farley et al., 2016).

Overall, our results show that differential BMP-dependent gene expression in neuronal subtypes is not only conferred by the integration of pMad/Medea into a combinatorial transcription factor code. In addition, sequence variation of the BMP-RE that is bound by pMad/Medea provides additional critical information required to selectively trans-activate gene expression in specific neuronal subtypes. Further work is required to identify transcription factors that bind the highly conserved sequences that flank the BMP-LA motif in the FMRFa Tv4-enhancer, or interact directly with the BMP-LA motif itself. Once such factors are identified, we will be able to directly assess the contribution of pMad/Medea affinity and specific transcription factors to the BMP-LA in order to form subtype-specific, activation-competent complexes.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Aberle H
   2. Haghighi AP
   3. Fetter RD
   4. McCabe BD
   5. Magalhães TR
   6. Goodman CS

   (2002) Wishful thinking encodes a BMP type II receptor that regulates synaptic growth in Drosophila

   Neuron 33:545–558.

   https://doi.org/10.1016/S0896-6273(02)00589-5

   • PubMed
   • Google Scholar
2. 1. Affolter M
   2. Basler K

   (2007) The decapentaplegic morphogen gradient: from pattern formation to growth regulation

   Nature Reviews Genetics 8:663–674.

   https://doi.org/10.1038/nrg2166

   • PubMed
   • Google Scholar
3. 1. Allan DW
   2. St Pierre SE
   3. Miguel-Aliaga I
   4. Thor S

   (2003) Specification of neuropeptide cell identity by the integration of retrograde BMP signaling and a combinatorial transcription factor code

   Cell 113:73–86.

   https://doi.org/10.1016/s0092-8674(03)00204-6

   • PubMed
   • Google Scholar
4. 1. Allan DW
   2. Park D
   3. St. Pierre SE
   4. Taghert PH
   5. Thor S

   (2005) Regulators acting in combinatorial codes also act independently in single differentiating neurons

   Neuron 45:689–700.

   https://doi.org/10.1016/j.neuron.2005.01.026

   • Google Scholar
5. 1. Allan DW
   2. Thor S

   (2015) Transcriptional selectors, masters, and combinatorial codes: regulatory principles of neural subtype specification

   Wiley Interdisciplinary Reviews: Developmental Biology 4:505–528.

   https://doi.org/10.1002/wdev.191

   • PubMed
   • Google Scholar
6. 1. Angley C
   2. Kumar M
   3. Dinsio KJ
   4. Hall AK
   5. Siegel RE

   (2003) Signaling by bone morphogenetic proteins and Smad1 modulates the postnatal differentiation of cerebellar cells

   The Journal of Neuroscience 23:260–268.

   https://doi.org/10.1523/JNEUROSCI.23-01-00260.2003

   • PubMed
   • Google Scholar
7. 1. Ball RW
   2. Warren-Paquin M
   3. Tsurudome K
   4. Liao EH
   5. Elazzouzi F
   6. Cavanagh C
   7. An BS
   8. Wang TT
   9. White JH
   10. Haghighi AP

   (2010) Retrograde BMP signaling controls synaptic growth at the NMJ by regulating trio expression in motor neurons

   Neuron 66:536–549.

   https://doi.org/10.1016/j.neuron.2010.04.011

   • PubMed
   • Google Scholar
8. 1. Barolo S
   2. Carver LA
   3. Posakony JW

   (2000) GFP and beta-galactosidase transformation vectors for promoter/enhancer analysis in Drosophila

   BioTechniques 29:726–732.

   https://doi.org/10.2144/00294bm10

   • PubMed
   • Google Scholar
9. 1. Barolo S
   2. Posakony JW

   (2002) Three habits of highly effective signaling pathways: principles of transcriptional control by developmental cell signaling

   Genes & Development 16:1167–1181.

   https://doi.org/10.1101/gad.976502

   • PubMed
   • Google Scholar
10. 1. Baumgardt M
    2. Miguel-Aliaga I
    3. Karlsson D
    4. Ekman H
    5. Thor S

    (2007) Specification of neuronal identities by feedforward combinatorial coding

    PLOS Biology 5:0050037.

    https://doi.org/10.1371/journal.pbio.0050037

    • Google Scholar
11. 1. Baumgardt M
    2. Karlsson D
    3. Terriente J
    4. Díaz-Benjumea FJ
    5. Thor S

    (2009) Neuronal subtype specification within a lineage by opposing temporal feed-forward loops

    Cell 139:969–982.

    https://doi.org/10.1016/j.cell.2009.10.032

    • PubMed
    • Google Scholar
12. 1. Benveniste RJ
    2. Thor S
    3. Thomas JB
    4. Taghert PH

    (1998)

    Cell type-specific regulation of the Drosophila FMRF-NH2 neuropeptide gene by Apterous, a LIM homeodomain transcription factor

    Development 125:4757–4765.

    • PubMed
    • Google Scholar
13. 1. Benveniste RJ
    2. Taghert PH

    (1999) Cell type-specific regulatory sequences control expression of the Drosophila FMRF-NH2 neuropeptide gene

    Journal of Neurobiology 38:507–520.

    https://doi.org/10.1002/(SICI)1097-4695(199903)38:4<507::AID-NEU7>3.0.CO;2-X

    • PubMed
    • Google Scholar
14. 1. Berke B
    2. Wittnam J
    3. McNeill E
    4. Van Vactor DL
    5. Keshishian H

    (2013) Retrograde BMP signaling at the synapse: a permissive signal for synapse maturation and activity-dependent plasticity

    Journal of Neuroscience 33:17937–17950.

    https://doi.org/10.1523/JNEUROSCI.6075-11.2013

    • PubMed
    • Google Scholar
15. 1. Berndt AJ
    2. Tang JC
    3. Ridyard MS
    4. Lian T
    5. Keatings K
    6. Allan DW

    (2015) Gene regulatory mechanisms underlying the spatial and temporal regulation of Target-Dependent gene expression in Drosophila Neurons

    PLOS Genetics 11:e1005754.

    https://doi.org/10.1371/journal.pgen.1005754

    • PubMed
    • Google Scholar
16. 1. Campbell G
    2. Tomlinson A

    (1999) Transducing the dpp morphogen gradient in the wing of Drosophila: regulation of dpp targets by brinker

    Cell 96:553–562.

    https://doi.org/10.1016/s0092-8674(00)80659-5

    • PubMed
    • Google Scholar
17. 1. Charbonnier E
    2. Fuchs A
    3. Cheung LS
    4. Chayengia M
    5. Veikkolainen V
    6. Seyfferth J
    7. Shvartsman SY
    8. Pyrowolakis G

    (2015) BMP-dependent gene repression cascade in Drosophila eggshell patterning

    Developmental Biology 400:258–265.

    https://doi.org/10.1016/j.ydbio.2015.02.004

    • PubMed
    • Google Scholar
18. 1. Chayengia M
    2. Veikkolainen V
    3. Jevtic M
    4. Pyrowolakis G

    (2019) Sequence environment of BMP-dependent activating elements controls transcriptional responses to dpp signaling in Drosophila

    Development 146:dev176107.

    https://doi.org/10.1242/dev.176107

    • PubMed
    • Google Scholar
19. 1. Chou HJ
    2. Lai DM
    3. Huang CW
    4. McLennan IS
    5. Wang HD
    6. Wang PY

    (2013) BMP4 is a peripherally-derived factor for motor neurons and attenuates glutamate-induced excitotoxicity in vitro

    PLOS ONE 8:e58441.

    https://doi.org/10.1371/journal.pone.0058441

    • PubMed
    • Google Scholar
20. 1. Crocker J
    2. Abe N
    3. Rinaldi L
    4. McGregor AP
    5. Frankel N
    6. Wang S
    7. Alsawadi A
    8. Valenti P
    9. Plaza S
    10. Payre F
    11. Mann RS
    12. Stern DL

    (2015) Low affinity binding site clusters confer hox specificity and regulatory robustness

    Cell 160:191–203.

    https://doi.org/10.1016/j.cell.2014.11.041

    • PubMed
    • Google Scholar
21. 1. Crocker J
    2. Noon EP
    3. Stern DL

    (2016) The soft touch: low-affinity transcription factor binding sites in development and evolution

    Essays on Developmental Biology Part B 117:455–469.

    https://doi.org/10.1016/bs.ctdb.2015.11.018

    • Google Scholar
22. 1. da Silva S
    2. Wang F

    (2011) Retrograde neural circuit specification by target-derived neurotrophins and growth factors

    Current Opinion in Neurobiology 21:61–67.

    https://doi.org/10.1016/j.conb.2010.07.007

    • PubMed
    • Google Scholar
23. 1. DuVal MG
    2. Gilbert MJ
    3. Watson DE
    4. Zerulla TC
    5. Tierney KB
    6. Allison WT

    (2014) Growth differentiation factor 6 as a putative risk factor in neuromuscular degeneration

    PLOS ONE 9:e89183.

    https://doi.org/10.1371/journal.pone.0089183

    • PubMed
    • Google Scholar
24. 1. Eade KT
    2. Fancher HA
    3. Ridyard MS
    4. Allan DW

    (2012) Developmental transcriptional networks are required to maintain neuronal subtype identity in the mature nervous system

    PLOS Genetics 8:e1002501.

    https://doi.org/10.1371/journal.pgen.1002501

    • PubMed
    • Google Scholar
25. 1. Eade KT
    2. Allan DW

    (2009) Neuronal phenotype in the mature nervous system is maintained by persistent retrograde bone morphogenetic protein signaling

    Journal of Neuroscience 29:3852–3864.

    https://doi.org/10.1523/JNEUROSCI.0213-09.2009

    • PubMed
    • Google Scholar
26. 1. Esteves FF
    2. Springhorn A
    3. Kague E
    4. Taylor E
    5. Pyrowolakis G
    6. Fisher S
    7. Bier E

    (2014) BMPs regulate msx gene expression in the dorsal neuroectoderm of Drosophila and vertebrates by distinct mechanisms

    PLOS Genetics 10:e1004625.

    https://doi.org/10.1371/journal.pgen.1004625

    • PubMed
    • Google Scholar
27. 1. Farley EK
    2. Olson KM
    3. Zhang W
    4. Brandt AJ
    5. Rokhsar DS
    6. Levine MS

    (2015) Suboptimization of developmental enhancers

    Science 350:325–328.

    https://doi.org/10.1126/science.aac6948

    • PubMed
    • Google Scholar
28. 1. Farley EK
    2. Olson KM
    3. Zhang W
    4. Rokhsar DS
    5. Levine MS

    (2016) Syntax compensates for poor binding sites to encode tissue specificity of developmental enhancers

    PNAS 113:6508–6513.

    https://doi.org/10.1073/pnas.1605085113

    • Google Scholar
29. 1. Gao S
    2. Steffen J
    3. Laughon A

    (2005) Dpp-responsive silencers are bound by a trimeric Mad-Medea complex

    Journal of Biological Chemistry 280:36158–36164.

    https://doi.org/10.1074/jbc.M506882200

    • PubMed
    • Google Scholar
30. 1. Gao S
    2. Laughon A

    (2007) Flexible interaction of Drosophila smad complexes with bipartite binding sites

    Biochimica Et Biophysica Acta (BBA) - Gene Structure and Expression 1769:484–496.

    https://doi.org/10.1016/j.bbaexp.2007.05.006

    • Google Scholar
31. 1. González A
    2. Jafari S
    3. Zenere A
    4. Alenius M
    5. Altafini C

    (2019) Thermodynamic model of gene regulation for the Or59b olfactory receptor in Drosophila

    PLOS Computational Biology 15:e1006709.

    https://doi.org/10.1371/journal.pcbi.1006709

    • PubMed
    • Google Scholar
32. 1. Grieder NC
    2. Nellen D
    3. Burke R
    4. Basler K
    5. Affolter M

    (1995) Schnurri is required for Drosophila dpp signaling and encodes a zinc finger protein similar to the mammalian transcription factor PRDII-BF1

    Cell 81:791–800.

    https://doi.org/10.1016/0092-8674(95)90540-5

    • PubMed
    • Google Scholar
33. 1. Groth AC
    2. Fish M
    3. Nusse R
    4. Calos MP

    (2004) Construction of transgenic Drosophila by using the site-specific integrase from phage phiC31

    Genetics 166:1775–1782.

    https://doi.org/10.1534/genetics.166.4.1775

    • PubMed
    • Google Scholar
34. 1. Guha U
    2. Gomes WA
    3. Samanta J
    4. Gupta M
    5. Rice FL
    6. Kessler JA

    (2004) Target-derived BMP signaling limits sensory neuron number and the extent of peripheral innervation in vivo

    Development 131:1175–1186.

    https://doi.org/10.1242/dev.01013

    • PubMed
    • Google Scholar
35. 1. Gurdziel K
    2. Lorberbaum DS
    3. Udager AM
    4. Song JY
    5. Richards N
    6. Parker DS
    7. Johnson LA
    8. Allen BL
    9. Barolo S
    10. Gumucio DL

    (2015) Identification and validation of novel Hedgehog-Responsive enhancers predicted by computational analysis of ci/Gli binding site density

    PLOS ONE 10:e0145225.

    https://doi.org/10.1371/journal.pone.0145225

    • PubMed
    • Google Scholar
36. 1. Halfon MS

    (2002) New fluorescent protein reporters for use with the Drosophila Gal4 expression system and for vital detection of balancer chromosomes

    Genesis : The Journal of Genetics and Development 138:135–138.

    https://doi.org/10.1002/gene.10136

    • Google Scholar
37. 1. Hamaratoglu F
    2. Affolter M
    3. Pyrowolakis G

    (2014) Dpp/BMP signaling in flies: from molecules to biology

    Seminars in Cell & Developmental Biology 32:128–136.

    https://doi.org/10.1016/j.semcdb.2014.04.036

    • PubMed
    • Google Scholar
38. 1. Heinz S
    2. Benner C
    3. Spann N
    4. Bertolino E
    5. Lin YC
    6. Laslo P
    7. Cheng JX
    8. Murre C
    9. Singh H
    10. Glass CK

    (2010) Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities

    Molecular Cell 38:576–589.

    https://doi.org/10.1016/j.molcel.2010.05.004

    • PubMed
    • Google Scholar
39. 1. Henríquez JP
    2. Krull CE
    3. Osses N

    (2011) The wnt and BMP families of signaling morphogens at the vertebrate neuromuscular junction

    International Journal of Molecular Sciences 12:8924–8946.

    https://doi.org/10.3390/ijms12128924

    • PubMed
    • Google Scholar
40. 1. Hobert O
    2. Carrera I
    3. Stefanakis N

    (2010) The molecular and gene regulatory signature of a neuron

    Trends in Neurosciences 33:435–445.

    https://doi.org/10.1016/j.tins.2010.05.006

    • PubMed
    • Google Scholar
41. 1. Hodge LK
    2. Klassen MP
    3. Han BX
    4. Yiu G
    5. Hurrell J
    6. Howell A
    7. Rousseau G
    8. Lemaigre F
    9. Tessier-Lavigne M
    10. Wang F

    (2007) Retrograde BMP signaling regulates trigeminal sensory neuron identities and the formation of precise face maps

    Neuron 55:572–586.

    https://doi.org/10.1016/j.neuron.2007.07.010

    • PubMed
    • Google Scholar
42. 1. Hudson JB
    2. Podos SD
    3. Keith K
    4. Simpson SL
    5. Ferguson EL

    (1998)

    The Drosophila Medea gene is required downstream of dpp and encodes a functional homolog of human Smad4

    Development 125:1407–1420.

    • PubMed
    • Google Scholar
43. 1. Jafari S
    2. Alenius M

    (2015) Cis-regulatory mechanisms for robust olfactory sensory neuron class-restricted odorant receptor gene expression in Drosophila

    PLOS Genetics 11:e1005051.

    https://doi.org/10.1371/journal.pgen.1005051

    • PubMed
    • Google Scholar
44. 1. Jaźwińska A
    2. Kirov N
    3. Wieschaus E
    4. Roth S
    5. Rushlow C

    (1999) The Drosophila gene brinker reveals a novel mechanism of dpp target gene regulation

    Cell 96:563–573.

    https://doi.org/10.1016/S0092-8674(00)80660-1

    • PubMed
    • Google Scholar
45. 1. Jiang J
    2. Levine M

    (1993) Binding affinities and cooperative interactions with bHLH activators delimit threshold responses to the dorsal gradient morphogen

    Cell 72:741–752.

    https://doi.org/10.1016/0092-8674(93)90402-C

    • PubMed
    • Google Scholar
46. 1. Kamachi Y
    2. Uchikawa M
    3. Tanouchi A
    4. Sekido R
    5. Kondoh H

    (2001) Pax6 and SOX2 form a co-DNA-binding partner complex that regulates initiation of Lens development

    Genes & Development 15:1272–1286.

    https://doi.org/10.1101/gad.887101

    • PubMed
    • Google Scholar
47. 1. Kelly CE
    2. Thymiakou E
    3. Dixon JE
    4. Tanaka S
    5. Godwin J
    6. Episkopou V

    (2013) Rnf165/Ark2C enhances BMP-Smad signaling to mediate motor axon extension

    PLOS Biology 11:e1001538.

    https://doi.org/10.1371/journal.pbio.1001538

    • PubMed
    • Google Scholar
48. 1. Kim J
    2. Johnson K
    3. Chen HJ
    4. Carroll S
    5. Laughon A

    (1997) Drosophila mad binds to DNA and directly mediates activation of vestigial by decapentaplegic

    Nature 388:304–308.

    https://doi.org/10.1038/40906

    • PubMed
    • Google Scholar
49. 1. Kim NC
    2. Marqués G

    (2010) Identification of downstream targets of the bone morphogenetic protein pathway in the Drosophila nervous system

    Developmental Dynamics 239:2413–2425.

    https://doi.org/10.1002/dvdy.22368

    • Google Scholar
50. 1. Leung TH
    2. Hoffmann A
    3. Baltimore D

    (2004) One nucleotide in a kappaB site can determine cofactor specificity for NF-kappaB dimers

    Cell 118:453–464.

    https://doi.org/10.1016/j.cell.2004.08.007

    • PubMed
    • Google Scholar
51. 1. Lopez-Coviella I
    2. Follettie MT
    3. Mellott TJ
    4. Kovacheva VP
    5. Slack BE
    6. Diesl V
    7. Berse B
    8. Thies RS
    9. Blusztajn JK

    (2005) Bone morphogenetic protein 9 induces the transcriptome of basal forebrain cholinergic neurons

    PNAS 102:6984–6989.

    https://doi.org/10.1073/pnas.0502097102

    • PubMed
    • Google Scholar
52. 1. Majdazari A
    2. Stubbusch J
    3. Müller CM
    4. Hennchen M
    5. Weber M
    6. Deng CX
    7. Mishina Y
    8. Schütz G
    9. Deller T
    10. Rohrer H

    (2013) Dendrite complexity of sympathetic neurons is controlled during postnatal development by BMP signaling

    Journal of Neuroscience 33:15132–15144.

    https://doi.org/10.1523/JNEUROSCI.4748-12.2013

    • PubMed
    • Google Scholar
53. 1. Marqués G
    2. Haerry TE
    3. Crotty ML
    4. Xue M
    5. Zhang B
    6. O'Connor MB

    (2003) Retrograde gbb signaling through the bmp type 2 receptor wishful thinking regulates systemic FMRFa expression in Drosophila

    Development 130:5457–5470.

    https://doi.org/10.1242/dev.00772

    • PubMed
    • Google Scholar
54. 1. Marty T
    2. Müller B
    3. Basler K
    4. Affolter M

    (2000) Schnurri mediates Dpp-dependent repression of brinker transcription

    Nature Cell Biology 2:745–749.

    https://doi.org/10.1038/35036383

    • PubMed
    • Google Scholar
55. 1. McCabe BD
    2. Marqués G
    3. Haghighi AP
    4. Fetter RD
    5. Crotty ML
    6. Haerry TE
    7. Goodman CS
    8. O'Connor MB

    (2003) The BMP homolog gbb provides a retrograde signal that regulates synaptic growth at the Drosophila neuromuscular junction

    Neuron 39:241–254.

    https://doi.org/10.1016/S0896-6273(03)00426-4

    • PubMed
    • Google Scholar
56. 1. McCabe BD
    2. Hom S
    3. Aberle H
    4. Fetter RD
    5. Marques G
    6. Haerry TE
    7. Wan H
    8. O'Connor MB
    9. Goodman CS
    10. Haghighi AP

    (2004) Highwire regulates presynaptic BMP signaling essential for synaptic growth

    Neuron 41:891–905.

    https://doi.org/10.1016/S0896-6273(04)00073-X

    • PubMed
    • Google Scholar
57. 1. Miguel-Aliaga I
    2. Allan DW
    3. Thor S

    (2004) Independent roles of the dachshund and eyes absent genes in BMP signaling, axon pathfinding and neuronal specification

    Development 131:5837–5848.

    https://doi.org/10.1242/dev.01447

    • PubMed
    • Google Scholar
58. 1. Miguel-Aliaga I
    2. Thor S
    3. Gould AP

    (2008) Postmitotic specification of Drosophila insulinergic neurons from pioneer neurons

    PLOS Biology 6:e58.

    https://doi.org/10.1371/journal.pbio.0060058

    • Google Scholar
59. 1. Minami M
    2. Kinoshita N
    3. Kamoshida Y
    4. Tanimoto H
    5. Tabata T

    (1999) Brinker is a target of dpp in Drosophila that negatively regulates Dpp-dependent genes

    Nature 398:242–246.

    https://doi.org/10.1038/18451

    • PubMed
    • Google Scholar
60. 1. Moser M
    2. Campbell G

    (2005) Generating and interpreting the brinker gradient in the Drosophila wing

    Developmental Biology 286:647–658.

    https://doi.org/10.1016/j.ydbio.2005.08.036

    • PubMed
    • Google Scholar
61. 1. Müller B
    2. Hartmann B
    3. Pyrowolakis G
    4. Affolter M
    5. Basler K

    (2003) Conversion of an extracellular dpp/BMP morphogen gradient into an inverse transcriptional gradient

    Cell 113:221–233.

    https://doi.org/10.1016/S0092-8674(03)00241-1

    • PubMed
    • Google Scholar
62. 1. Pan Y
    2. Tsai CJ
    3. Ma B
    4. Nussinov R

    (2010) Mechanisms of transcription factor selectivity

    Trends in Genetics 26:75–83.

    https://doi.org/10.1016/j.tig.2009.12.003

    • PubMed
    • Google Scholar
63. 1. Pavelock KA
    2. Girard BM
    3. Schutz KC
    4. Braas KM
    5. May V

    (2007) Bone morphogenetic protein down-regulation of neuronal pituitary adenylate cyclase-activating polypeptide and reciprocal effects on vasoactive intestinal peptide expression

    Journal of Neurochemistry 100:603–616.

    https://doi.org/10.1111/j.1471-4159.2006.04293.x

    • PubMed
    • Google Scholar
64. 1. Pyrowolakis G
    2. Hartmann B
    3. Müller B
    4. Basler K
    5. Affolter M

    (2004) A simple molecular complex mediates widespread BMP-induced repression during Drosophila development

    Developmental Cell 7:229–240.

    https://doi.org/10.1016/j.devcel.2004.07.008

    • PubMed
    • Google Scholar
65. 1. Ramos AI
    2. Barolo S

    (2013) Low-affinity transcription factor binding sites shape morphogen responses and enhancer evolution

    Philosophical Transactions of the Royal Society B: Biological Sciences 368:20130018.

    https://doi.org/10.1098/rstb.2013.0018

    • Google Scholar
66. 1. Rawson JM
    2. Lee M
    3. Kennedy EL
    4. Selleck SB

    (2003) Drosophila neuromuscular synapse assembly and function require the TGF-beta type I receptor saxophone and the transcription factor mad

    Journal of Neurobiology 55:134–150.

    https://doi.org/10.1002/neu.10189

    • PubMed
    • Google Scholar
67. 1. Reichhart JM
    2. Ferrandon D

    (1998)

    Green balancers

    Dros. Inf. Serv 81:201–202.

    • Google Scholar
68. 1. Ross S
    2. Hill CS

    (2008) How the smads regulate transcription

    The International Journal of Biochemistry & Cell Biology 40:383–408.

    https://doi.org/10.1016/j.biocel.2007.09.006

    • PubMed
    • Google Scholar
69. 1. Ryder E
    2. Ashburner M
    3. Bautista-Llacer R
    4. Drummond J
    5. Webster J
    6. Johnson G
    7. Morley T
    8. Chan YS
    9. Blows F
    10. Coulson D
    11. Reuter G
    12. Baisch H
    13. Apelt C
    14. Kauk A
    15. Rudolph T
    16. Kube M
    17. Klimm M
    18. Nickel C
    19. Szidonya J
    20. Maróy P
    21. Pal M
    22. Rasmuson-Lestander A
    23. Ekström K
    24. Stocker H
    25. Hugentobler C
    26. Hafen E
    27. Gubb D
    28. Pflugfelder G
    29. Dorner C
    30. Mechler B
    31. Schenkel H
    32. Marhold J
    33. Serras F
    34. Corominas M
    35. Punset A
    36. Roote J
    37. Russell S

    (2007) The DrosDel deletion collection: a Drosophila genomewide chromosomal deficiency resource

    Genetics 177:615–629.

    https://doi.org/10.1534/genetics.107.076216

    • PubMed
    • Google Scholar
70. 1. Shi Y
    2. Massagué J

    (2003) Mechanisms of TGF-beta signaling from cell membrane to the nucleus

    Cell 113:685–700.

    https://doi.org/10.1016/S0092-8674(03)00432-X

    • PubMed
    • Google Scholar
71. 1. Siepel A
    2. Bejerano G
    3. Pedersen JS
    4. Hinrichs AS
    5. Hou M
    6. Rosenbloom K
    7. Clawson H
    8. Spieth J
    9. Hillier LW
    10. Richards S
    11. Weinstock GM
    12. Wilson RK
    13. Gibbs RA
    14. Kent WJ
    15. Miller W
    16. Haussler D

    (2005) Evolutionarily conserved elements in vertebrate, insect, worm, and yeast genomes

    Genome Research 15:1034–1050.

    https://doi.org/10.1101/gr.3715005

    • PubMed
    • Google Scholar
72. 1. Torres-Vazquez J
    2. Warrior R
    3. Arora K

    (2000) Schnurri is required for dpp-dependent patterning of the Drosophila wing

    Developmental Biology 227:388–402.

    https://doi.org/10.1006/dbio.2000.9900

    • PubMed
    • Google Scholar
73. 1. Uchikawa M
    2. Ishida Y
    3. Takemoto T
    4. Kamachi Y
    5. Kondoh H

    (2003) Functional analysis of chicken Sox2 enhancers highlights an array of diverse regulatory elements that are conserved in mammals

    Developmental Cell 4:509–519.

    https://doi.org/10.1016/S1534-5807(03)00088-1

    • PubMed
    • Google Scholar
74. 1. Veverytsa L
    2. Allan DW

    (2011) Retrograde BMP signaling controls Drosophila behavior through regulation of a peptide hormone battery

    Development 138:3147–3157.

    https://doi.org/10.1242/dev.064105

    • PubMed
    • Google Scholar
75. 1. Vuilleumier R
    2. Lian T
    3. Flibotte S
    4. Khan ZN
    5. Fuchs A
    6. Pyrowolakis G
    7. Allan DW

    (2019) Retrograde BMP signaling activates neuronal gene expression through widespread deployment of a conserved BMP-responsive cis-regulatory activation element

    Nucleic Acids Research 47:679–699.

    https://doi.org/10.1093/nar/gky1135

    • PubMed
    • Google Scholar
76. 1. Weiss A
    2. Charbonnier E
    3. Ellertsdóttir E
    4. Tsirigos A
    5. Wolf C
    6. Schuh R
    7. Pyrowolakis G
    8. Affolter M

    (2010) A conserved activation element in BMP signaling during Drosophila development

    Nature Structural & Molecular Biology 17:69–76.

    https://doi.org/10.1038/nsmb.1715

    • PubMed
    • Google Scholar
77. 1. Wharton SJ
    2. Basu SP
    3. Ashe HL

    (2004) Smad affinity can direct distinct readouts of the embryonic extracellular dpp gradient in Drosophila

    Current Biology 14:1550–1558.

    https://doi.org/10.1016/j.cub.2004.08.053

    • PubMed
    • Google Scholar
78. 1. Winter SE
    2. Campbell G

    (2004) Repression of dpp targets in the Drosophila wing by brinker

    Development 131:6071–6081.

    https://doi.org/10.1242/dev.01538

    • PubMed
    • Google Scholar
79. 1. Xiao L
    2. Michalski N
    3. Kronander E
    4. Gjoni E
    5. Genoud C
    6. Knott G
    7. Schneggenburger R

    (2013) BMP signaling specifies the development of a large and fast CNS synapse

    Nature Neuroscience 16:856–864.

    https://doi.org/10.1038/nn.3414

    • PubMed
    • Google Scholar
80. 1. Xu X
    2. Yin Z
    3. Hudson JB
    4. Ferguson EL
    5. Frasch M

    (1998) Smad proteins act in combination with synergistic and antagonistic regulators to target dpp responses to the Drosophila mesoderm

    Genes & Development 12:2354–2370.

    https://doi.org/10.1101/gad.12.15.2354

    • PubMed
    • Google Scholar
81. 1. Zandvakili A
    2. Campbell I
    3. Gutzwiller LM
    4. Weirauch MT
    5. Gebelein B

    (2018) Degenerate Pax2 and senseless binding motifs improve detection of low-affinity sites required for enhancer specificity

    PLOS Genetics 14:e1007289.

    https://doi.org/10.1371/journal.pgen.1007289

    • Google Scholar
82. 1. Zhang H
    2. Levine M
    3. Ashe HL

    (2001) Brinker is a sequence-specific transcriptional repressor in the Drosophila embryo

    Genes & Development 15:261–266.

    https://doi.org/10.1101/gad.861201

    • PubMed
    • Google Scholar

Author details

1. Anthony JE Berndt

   Department of Food & Fuel for the 21st Century, University of California San Diego, San Diego, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Katerina M Othonos

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0132-7393

2. Katerina M Othonos

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Conceptualization, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Contributed equally with

   Anthony JE Berndt

   Competing interests

   No competing interests declared

3. Tianshun Lian

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Formal analysis, Validation, Investigation, Visualization, Methodology

   Competing interests

   No competing interests declared

4. Stephane Flibotte

   UBC/LSI Bioinformatics Facility, University of British Columbia, Vancouver, Canada

   Contribution

   Software, Writing - review and editing

   Competing interests

   No competing interests declared

5. Mo Miao

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

6. Shamsuddin A Bhuiyan

   Department of Psychiatry, University of British Columbia, Vancouver, Canada

   Contribution

   Bioinformatics, Writing - review and editing

   Competing interests

   No competing interests declared

7. Raymond Y Cho

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

8. Justin S Fong

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

9. Seo Am Hur

   Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

   Contribution

   Investigation, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-4163-7182

10. Paul Pavlidis

    Department of Psychiatry, University of British Columbia, Vancouver, Canada

    Contribution

    Software, Supervision, Writing - review and editing

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-0426-5028

11. Douglas W Allan

    Department of Cellular and Physiological Sciences, University of British Columbia, Vancouver, Canada

    Contribution

    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

    For correspondence

    doug.allan@ubc.ca

    Competing interests

    No competing interests declared

    "This ORCID iD identifies the author of this article:" 0000-0002-3488-8365

• 1,059

  views

• 144

  downloads

• 3

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.