Deadpan Contributes to the Robustness of the Notch Response

Notch signaling regulates many fundamental events including lateral inhibition and boundary formation to generate very reproducible patterns in developing tissues. Its targets include genes of the bHLH hairy and Enhancer of split [E(spl)] family, which contribute to many of these developmental decisions. One member of this family in Drosophila, deadpan (dpn), was originally found to have functions independent of Notch in promoting neural development. Employing genome-wide chromatin-immunoprecipitation we have identified several Notch responsive enhancers in dpn, demonstrating its direct regulation by Notch in a range of contexts including the Drosophila wing and eye. dpn expression largely overlaps that of several E(spl) genes and the combined knock-down leads to more severe phenotypes than either alone. In addition, Dpn contributes to the establishment of Cut expression at the wing dorsal-ventral (D/V) boundary; in its absence Cut expression is delayed. Furthermore, over-expression of Dpn inhibits expression from E(spl) gene enhancers, but not vice versa, suggesting that dpn contributes to a feed-back mechanism that limits E(spl) gene expression following Notch activation. Thus the combined actions of dpn and E(spl) appear to provide a mechanism that confers an initial rapid output from Notch activity which becomes self-limited via feedback between the targets.


Introduction
Notch signaling is a well-conserved pathway across metazoans involved in the regulation of many fundamental events including lateral inhibition and boundary formation. For example, in the Drosophila wing, activity of Notch is essential for restricting the width of the veins, the strut-like structures that confer rigidity, and for generating and maintaining the stable boundary between the dorsal and ventral cells, which acts as an organizer for patterning and growth of the whole wing. There the functions of Notch are implemented in part through the activities of wingless (wg) and cut (ct), which are expressed in a stripe at the D/V boundary in response to Notch activity.
Simple at first glance, Notch signaling is initiated by the activation of the Notch single-pass transmembrane receptor by the ligand Delta (Dl) or Serrate (Ser), which leads to its proteolysis and the release of the intracellular domain (NICD). NICD then enters the nucleus and affects transcriptional regulation by directly interacting with DNA-binding proteins of the CSL family [mammalian CBF1, Drosophila Suppressor of Hairless (Su(H)) and C.elagans LAG-1] [1,2]. The downstream targets involved in implementing these actions include genes of the bHLH Hairy and Enhancer of split (HES) gene family, which in Drosophila are located within the 60 kb E(spl) complex. However, activity of the known targets cannot account for all of the functions of Notch, prompting us to perform genome wide chromatin immunoprecipitation (ChIP with anti-Su(H) antibody) to identify other Notch regulated genes [3,4]. One gene which emerged from these analyses was deadpan (dpn), itself a member of the Hairy/Enhancer of Split (HES) subclass of repressor basic Helix-Loop-Helix (bHLH) proteins, but one that had been primarily linked to the specification of neural cells. More recently it has become evident that dpn is Notch regulated in some contexts although the extent to which it is a direct target and its relevance to Notch functions have remained enigmatic [5].
Until recently, studies of dpn have predominantly focused on its roles during Drosophila neurogenesis [6], where its ability to repress transcription of specific genes in the precursor cell is important in promoting neurogenesis [7,8]. Its expression persists in the neural stem cells, so-called neuroblasts, where its functions are partially redundant with those of the conventional Drosophila E(spl) genes [5,9]. In this context, it remains unclear whether or not dpn is a Notch-regulated target because its expression is unaffected in conditions of compromised Notch signaling although it responds to high levels of Notch activity in the mature intermediate neural precursor cells (INPs) of type II neuroblast (NB) lineage [5,9,10]. Furthermore, the regulation and functional relevance of dpn in other developmental processes is only just starting to emerge. Fundamental questions that remain are whether dpn is a direct target of Notch in other contexts, outside the nervous system, and how its function relates to that of the other E(spl) genes.
In this study, we show that there are multiple Notch responsive enhancers (NREs) in the dpn genomic region, which direct Notch dependent expression in a variety of tissues. Expression of dpn therefore is a widespread response to Notch activation. In addition, we demonstrate that it has a role in regulating the establishment of stable expression of cut at the D/V boundary in the wing discs and that it functions in a unidirectional feedback loop with the E(spl) genes, helping to self limit the Notch response.

Results
Multiple Notch Responsive Enhancers Associated with dpn dpn was originally reported as a pan-neural gene [7] with little evidence to suggest that it could be regulated by Notch despite the fact that it is a member of the HES family. More recently it has emerged that there is direct input of Notch into a neuroblast enhancer, although the relevance of this regulation remains unclear [5]. dpn expression is also detected in a wide range of other tissues including the wing, eye and leg imaginal discs, where it overlaps with well-characterised Notch target genes ( [7,11,12] and Fig. 1). Whether or not Dpn is directly regulated by Notch in these tissues has remained unclear, as the characterized neuroblast enhancer does not direct expression in the imaginal discs. However, our genome-wide analysis of Notch targets has indicated that dpn is likely to be directly regulated by Notch in other tissues besides the central nervous system (CNS) since we detect both Su(H) binding (by chromatin immunoprecipitation; ChIP) and changes in mRNA expression in several cells and tissues including wing imaginal discs (4.6 fold upregulated after Notch activation) and DmD8 cells (muscle progenitor related; 4.8-fold upregulated after Notch activation) (Fig. 1A).
Depending on the cellular origin, different locations for Su(H) binding at the dpn locus were detected, suggesting the existence of multiple Notch responsive enhancers. The major bound region in wing imaginal discs (dpn[a]) was located within the second intron of dpn, overlapping a pair of conserved Su(H) binding motifs. Of the two other more minor Su(H) bound regions detected in wing discs, one (dpn[b]) coincided with a region that was also occupied in DmD8 cells and that overlapped with the neuroblast enhancer identified previously [5]. A further prominent Su(H) bound region (dpn[c]) was detected in DmD8 cells but flanked the adjacent peanut gene, so its relationship to dpn was less clear. All the bound regions contained sequences that have good match to Su(H) binding motifs ( Fig. 1A; grey bar graphs).
To assess whether these Su(H) occupied regions identify Notch regulated enhancers we inserted genomic fragments encompassing the bound regions upstream of reporter genes and assessed whether they directed dpn-like expression patterns. First, consistent with the results from the ChIP, dpn[b]GFP, but not dpn[a]GFP ( Fig. 1B99 and C99), recapitulated the endogenous Dpn ( Fig. 1B999 and C999) expression in the adult muscle precursors (AMPs), which have similar characteristics to DmD8 cells. Second, dpn[a]GFP and dpn[b]GFP recapitulated key aspects of dpn expression in other tissues. dpn[a], the intronic enhancer bound by Su(H) in chromatin from wing discs, directed strong expression at the D/V margin and intervein regions, similar to endogenous dpn expression (Fig. 1D). Expression was also detected in the optic lobes of the brain, the leg joints and cone/support cells in the eye discs but not in neuroblasts or photoreceptors (Fig. S1). In contrast, dpn[b] gave weak but detectable expression at the D/V boundary and interveins in the wing, but it exhibited much stronger activity in the eye discs where it was detected in R3/R4 and R7 photoreceptors like endogenous Dpn, and was also strongly expressed in brain and leg discs (Fig. 1E99, F99 and Fig.S1). dpn[c] did not show any significant expression in any of the tissues analyzed so its function remains unclear (Data not shown). These experiments demonstrate therefore that two of the Su(H) bound regions faithfully recapitulate endogenous dpn expression in specific tissues. Furthermore, their expression patterns resemble those of several E(spl) genes, which highlight known sites of Notch pathway activity, such as the D/V boundary and interveins in the wing as well as R4 and R7 photoreceptors in the eye [13,14]. However, there are also some notable differences. For example, neither endogenous dpn nor the dpn enhancers exhibit strong expression in proneural territories unlike E(spl)m7 and E(spl)m8 [13]. In this respect the dpn expression appears most similar to E(spl)mb, which is detected at low levels broadly through the thoracic proneural regions.
To address whether dpn[a] and dpn[b] respond to Notch activity, we tested the consequences of manipulating Notch activity in wing and eye imaginal discs. First, expression of NICD in the wing (driven by patched-Gal4 (ptc-Gal4)) was sufficient to elicit ectopic expression of dpn[a] and dpn[b], resembling the response of the endogenous dpn gene ( Fig. 2A and data not shown). Conversely, when Notch was down-regulated by RNAi the expression of dpn[a] and dpn [b] were ablated along with that of dpn itself (Fig. 2B, and data not shown). Similar results were obtained with dpn[b] enhancer in the developing eye. NICD driven by sevenless-Gal4 (sev-Gal4, driving expression at R1, R3, R4, R6, and R7) resulted in ectopic dpn[b] expression so that firstly similar levels were detected in R3 and R4 (orange arrows, Fig. 2C0; compare to wildtype, blue arrows, Fig. 2D) and secondly other photoreceptor cells started to express the reporter (orange circle, Fig. 2C0; compare to wild-type, blue circle, Fig. 2D) correlating with the ectopic Dpn that was also detected (orange circle, Fig. 2C9). Together these data demonstrate that dpn[a] and dpn[b] can mediate the Notch regulation of dpn in the wing, while only dpn[b] responds in photoreceptors. We note that the dpn[b] fragment is larger than the previous neuroblast enhancer that was identified in this region, containing additional Su(H) sites and flanking sequences that confer additional aspects of Notch regulated expression.

Relationship
between dpn and E(spl) gene functions. The expression of dpn in response to Notch activity, and the evidence showing direct binding of Su(H) in several contexts, supports the recent results indicating that dpn has a role in implementing Notch function in regulating proliferation [12]. This is exemplified by the fact that knock-down of dpn enhances the wing phenotype seen in Notch heterozygotes (Fig 3A-F,  Table S1 and see [12]). Twenty percent of the wings from N[55e11]/+ female flies had a single nick in the margin at the distal tip (Fig. 3D). In combination with dpn alleles both the extent and the frequency of wing nicks were enhanced. Thus, 97% of N[55e11]/+; dpn [6]/+ wings contained wing nicks and these frequently extended into the posterior of the wing margin ( Fig. 3F). Slightly lower frequency was detected in combinations with dpn [1] where 78% of wings contained nicks, but again this was significantly enhanced compared to the N[55e11]/+ alone ( Fig. 3E comparing to 3D). In contrast, reductions in E(spl) function did not enhance the wing notching phenotype of Notch heterozygotes under these conditions (Table S1). These genetic interactions therefore support the hypothesis that dpn specifically contributes to Notch function at the D/V boundary in the wing.
Despite the genetic interactions with Notch, remarkably few defects were detected in dpn mutant clones or in tissue expressing RNAi targeting dpn (e.g. Fig 3J and Fig 4). For example, the majority of adult flies showed no defects under conditions where dpn-RNAi was expressed throughout the posterior of the wing using engrailed-Gal4. At best there was a low penetrance of wing nicks (19%) with one RNAi line, consistent with dpn having a subtle role in this process (Fig. 3H). Similar lack of phenotype was seen when E(spl) function was eliminated using a deficiency that spans the entire locus [15]. One possibility, given their shared Notch regulation, is that these HES genes have related functions as recently suggested [9,12], so that the combined activity is required for the robust establishment of Notch function. Elimination of all the E(spl) genes causes embryonic lethality [15]. However, a newly generated deficiency, E(spl)mc-mb [DK33], that removes only two E(spl)bHLH genes (E(spl)mc and E(spl)mb) has a mild wing vein phenotype (Fig. 4A). We therefore assessed whether this phenotype could be modified by a reduction in the levels of dpn, by making combinations with heterozygous dpn alleles ( Fig. 4A-C, Table S2). Wings from homozygous E(spl)mc-mb [DK33] flies had mild thickening in the posterior cross-vein (pcv) and tip of L5 (Fig. 4A). This mild phenotype was dominantly enhanced by removing a single copy of dpn, giving rise to prominent vein thickening at the tips of L4 and L5 (Fig. 4B) as well as to ectopic veins flanking L5 in 26% of wings (Fig. 4C). This genetic interaction suggests that dpn and E(spl) function together during wing vein development.
As there are few specific Notch targets known in the developing veins, we extended our analysis to Notch function at the D/V boundary in the wing, where there are several known targets of Notch activity including ct. In larvae, combined knock-down of E(spl)mb and dpn expression in the posterior compartment of the third instar wing disc led to significant reduction of Cut level in comparison to E(spl)mb alone, although we note that the latter caused a subtle decrease (compare Fig. 4D-D9 to 4E-E9). To further assess whether dpn made a contribution to the regulation of ct expression, dpn function was eliminated in clones of cells that completely lacked all E(spl) functions using a deletion of the complex, Df(3R)E(spl)[b32.2], which has been combined with a gro+ transgene to rescue any effects on the neighbouring gro gene [15]. Such dpnRNAi; Df(3R)E(spl)[b32.2] double mutant clones had lost Cut staining, although knock down of either alone failed to significantly compromise ct expression under these conditions (Fig 4F;data not shown). Similar analysis in the eye disc also suggested that the genes have related functions. Knock down of dpn alone produced few if any defects (some rotational defects but no fate changes observed) while knock down of dpn and E(spl)md 2] as this leads to early defects in patterning; data not shown). Thus these data suggest that the dpn and E(spl) genes have overlapping functions in eye and wing development.
One surprising result emerged when we analyzed the consequences of ablating dpn in early wing discs (Fig. 5). We found that the RNAi mediated knock down of dpn in the posterior compartment of the wing resulted in down-regulation of Cut at the D/V margin in mid third instar discs but not in older discs (Fig. 5C). Similar results were seen using independent RNAi lines,  all produced defects at early but not at later stages (Fig. S2).
Together the results suggest that dpn may make a contribution to Notch function at the D/V boundary, but that its absence can be compensated by another mechanism, to explain the recovery of cut expression at later stages, and the lack of phenotype in the adult wings. It is also striking that the phenotype is one of reduced Cut expression, because dpn and the E(spl)bHLH genes have characteristics of transcriptional repressors. This suggests that there may be an intermediary factor that is repressed by Dpn to enable the expression of Cut (see discussion). We therefore drew up a list of candidates, that are expressed in the relevant domain of the wing disc, and tested whether they were inhibited by Dpn overexpres- sion or up-regulated in conditions of dpn-RNAi. None of those tested (e.g. msh, dve, others) showed any discernable change in response to changes in dpn (Fig. S3).
Regulation of E(spl)bHLH genes by dpn. dpn and E(spl)bHLH genes are all regulated by Notch and have broadly similar expression patterns. This raises the possibility that there could be cross-regulatory interactions between these different HES genes, similar to what has been observed in somitogenesis [16]. We therefore tested whether ectopic expression of dpn or of E(spl)bHLH genes was able to modify the other's expression. First we found that overexpression of dpn (using enGal4 tubGal80ts), but not a control GFP, was sufficient to suppress the expression of two E(spl) reporters, E(spl)mb1.5-CD2 and E(spl)m8-lacZ ( Figure 6A-H). This reduction was even detected in late stage wing discs (Fig 6C and 6G). As there was no effect on a generic Notch reporter, NRE-GFP [17], under equivalent conditions (Fig 6J-J9), it is likely that dpn regulates the E(spl) enhancers directly rather than acting indirectly through Notch. In contrast, over-expression of E(spl)m8, E(spl)mb, E(spl)m7 or E(spl)m5 (Fig. 6I, Fig S4) all failed to modify the dpn expression at the D/V boundary. The data suggest therefore that there is a unidirectional feedback loop between Notch targeted HES genes, with dpn regulating E(spl)bHLH and not vice versa.

Discussion
HES genes are well-known targets of Notch activity. However, in Drosophila only the bHLH genes within the E(spl) Complex were originally thought to be directly downstream of Notch. The expression of another HES genes, dpn, appeared independent of Notch and indeed was associated with cells where Notch activity is considered to be down-regulated (embryonic neuroblasts). More recently it has emerged that dpn expression is under Notch regulation in some contexts [5,9]. Our results extend these findings by demonstrating that dpn is directly bound by Su(H) in vivo. As the Su(H) occupied regions differ according to the tissue-type, it appears that dpn contains several Notch responsive enhancers and our results demonstrate that these direct Notch-dependent expression in different subsets of tissues. Nevertheless, it is striking that a single dpn enhancer, dpn[b] exhibits Notch related expression in both the eye and the wing discs yet these patterns are characteristic of distinct genes/enhancers from the E(spl) Complex [13].
Despite the clear regulation by Notch, there is however relatively few phenotypes resulting from loss of dpn in many tissues. For example, both the wing and eye disc exhibit robust expression of dpn but neither exhibit phenotypes when dpn function was ablated. However, genetic interactions demonstrate that dpn function is related to Notch and both our evidence, and that from recent studies [9,12], indicate that it has partially redundant functions with the E(spl) genes. This is exemplified by the fact that absence of dpn or of E(spl) Complex alone has little effect on the D/V boundary, but the combined knock down leads to loss of key gene expression.
What then is the relevance of dpn in these contexts, especially given that there are 7 E(spl)bHLH genes that also appear to have largely redundant functions? We propose two components to dpn function to explain its importance in the Notch response. Clues for the first come from the fact that we could detect subtle phenotypes from reductions in dpn when we analyzed early developmental stages. Thus the absence of dpn led to a delay in the ability of Notch to up-regulate cut. Earlier studies also demonstrated a subtle decrease in cut expression in cells lacking E(spl) genes [18]. These results could be explained if both E(spl) and dpn make a contribution to cut regulation. We suggest that this must be indirect, via the inhibition of a repressor, since both dpn and E(spl)bHLH are thought to be dedicated repressors (Fig. 7). So far, we have not identified another repressor that could act as an intermediary.
The second component of dpn function is suggested by the observation that Dpn can repress the enhancers derived from E(spl)bHLH genes but not vice versa. Futhermore, we observed that cells with high levels of Dpn often had lower levels of E(spl)bHLH on a cell by cell level. We therefore propose that there is a direct regulatory relationship between dpn and E(spl)bHLH, whereby dpn represses E(spl)bHLH expression (Fig. 7). This could set a maximum threshold for E(spl) gene expression since, in previous studies, we have found that dpn shows a less rapid upregulation following Notch activation than the E(spl) genes [19]. This is reminiscent of the differences seen between HES gene responses in the oscillatory clock involved in somitogenesis [20] and suggests that similar HES gene cross-regulatory network may underpin other Notch dependent processes.
Mitotic clones were generated by FLP-mediated mitotic recombination [25]. Clones lacking dpn were obtained by crossing FRT42B dpn [6]/CyO males to hsp-FLP122; FRT42B/CyO virgin females. Control clones were generated using the wild type FRT42B chromosome. The progeny of these crosses were heat shocked at 37uC for 1 h between 48 and 72 h after egg-laying. Discs were dissected and analyzed 3 days after the induction of the clones.
Generation of E(spl)Dmc-mb [DK33-10.1]: Small deficiencies in the E(spl)-Complex were generated in a cross to induce P-element mediated male recombination [27] using the P-element: P{lacW}K33 [28], where the transposon is inserted 61 bp upstream of the transcriptional start site of E(spl)mc, oriented with the 39 terminal repeats close to E(spl)mc. One of the progeny had a deletion of approximately 6.2 Kb, starting inside the P-element (retaining the last 2361 bp of the 59-end of the P{lacW}) deleting all the E(spl)mc gene and most of E(spl)mb coding region, terminating after the first 31amino acids. For

Molecular Cloning
Putative NREs in dpn region: ChIP-enriched regions (putative NREs in dpn) were amplified from Drosophila genomic DNA using primers containing specific restriction enzyme sequences and cloned into pGreenRabbit/ pRedRabbit vectors [17] for in vivo reporter assays. Sequences of primers and restriction enzymes used for cloning dpn reporters are as follows: dpn

Immunostaining
Antibody staining of wing and eye imaginal discs was performed as described previously [21] with the following antibodies: guinea  [29], rabbit anti-Msh [30]. Images were acquired with a scanning confocal microscope (Leica TCS SP2) and processed using Photoshop (Adobe). Figure S1 Reporter [a] expression was detected in the leg joints (A), the optic lobes of the brain (C), and cone/support cells in the eye discs (E) but not in neuroblasts or photoreceptors, while reporter [b] expression was detected in leg discs (B), in brain neuroblast lineages (D) and in R3-R4 and R7 photoreceptors (F). (TIF)