Interactive Fly, Drosophila

In the developing Drosophila retina, the proneural gene for photoreceptor neurons is atonal, a basic helix-loop-helix transcription factor. Using atonal as a marker for proneural maturation, the stepwise resolution of proneural clusters was examined during the initiation of ommatidial differentiation in the developing eye disc. In addition, evidence is provided that atonal is negatively regulated by rough, a homeobox-containing transcription factor expressed exclusively in the retina. This interaction leads to the refinement of proneural clusters to specify R8, the first neuron to emerge in the retinal neuroepithelium. Either ectopic expression of atonal or the removal of rough results in the transformation of a discrete 'equivalence group' of cells into R8s. In addition, ectopic expression of rough blocks atonal expression and proneural cluster formation within the morphogenetic furrow. Thus, rough provides retina-specific regulation to the more general atonal-mediated proneural differentiation pathway. Expression of Rough and Atonal is mutually exclusive: as atonal expression resolves from an initial ubiquitous stripe to individual proneural clusters, rough expression emerges in the intervening cells. The opposing roles of atonal and rough are not mediated through the Notch pathway, as their expression remains complementary when Notch activity is reduced. These observations suggest that homeobox-containing genes can serve a function of tissue-specific repression for bHLH factors. Rough is not likely to be a direct negative regulator of Enhancer of split expression since their expression patterns show extensive overlap. Instead, Rough-induced loss of E(spl) expression may be due to loss of atonal expression in a manner analogous to E(spl) requirement for achaete and scute activity. Notch signaling is also presumably required for E(spl) expression in this system (Dokucu, 1997).

The neuropeptide receptor NKD is a Drosophila homologue of the mammalian tachykinin receptors. This receptor is expressed during Drosophila embryonic development and in the adult fly. Use of the NKD promoter region to drive beta-galactosidase expression in transgenic flies reveals a bipartite promoter organization: the distal region controls NKD expression in neurosecretory cells of the central nervous system during late embryogenesis, whereas the proximal region is responsible for transient expression in peripheral nervous system during late embryogenesis, whereas the proximal region is responsible for transient expression in peripheral nervous system precursor cells early in development. This early NKD expression, first restricted to the sensory organ precursor cell, an atonal positive cell, is abolished in the ato1 mutant. The proneural protein Atonal, in association with Daughterless, transactivates the NKD promoter in Schneider S2 cells via the proximal E box NKDE2. Furthermore, heterodimers of Atonal and Daughterless interact with this E box in gel shift assay (Rosay, 1995).

An outstanding model to study how neurons differentiate from among a field of equipotent undifferentiated cells is the process of R8 photoreceptor differentiation during Drosophila eye development. Senseless is a Zn finger transcription factor that is expressed and required in the sensory organ precursors (SOPs) for proper proneural gene expression. In senseless mutant tissue, R8 differentiation fails and the presumptive R8 cell adopts the R2/R5 fate. senseless repression of rough (ro) in R8 is an essential mechanism of R8 cell fate determination and misexpression of senseless in non-R8 photoreceptors results in repression of rough and induction of the R8 fate. Surprisingly, there is no loss of ommatidial clusters in senseless mutant tissue and all outer photoreceptor subtypes can be recruited, suggesting that other photoreceptors can substitute for R8 to initiate recruitment and that R8-specific signaling is not required for outer photoreceptor subtype assignment (Frankfort, 2001).

Since the expression pattern of Sens overlaps that of Ato and sens lies downstream of proneural genes in the PNS, this relationship was confirmed in the developing eye. ato¹ mutant clones were generated in the eye using the FLP/FRT system; Sens is not detected within these clones. Similarly, Sens is not detected in eye discs dissected from ato¹ mutant larvae. UAS-ato was expressed under the control of sevenless-GAL4, which is expressed in all photoreceptor cells except R8, R2, and R5, and it was found that expression of Sens is strongly activated in response to ectopic ato. Taken together, these data place sens downstream of ato in the developing eye (Frankfort, 2001).

Thus a new model is proposed for the genetic regulation of R8 differentiation that includes the relationships among ato, sens, and ro. In this model, ato induces sens within the R8 equivalence group and R8, and sens is in turn required for maintenance of ato expression. Since R8 may transiently differentiate in sens mutant clones, ato is likely sufficient to confer specificity to R8 differentiation, whereas sens is required to 'lock-in' and maintain this program of R8 differentiation, primarily via the repression of ro. Thus, mutual antagonism of ato and ro is likely mediated by sens. sens presumably has a ro-independent role in R8 differentiation as well, because loss of ro function does not completely rescue the sens mutant phenotype (Frankfort, 2001).

Drosophila sensory neurons form distinctive terminal branch patterns in the developing neuropile of the embryonic central nervous system. A genetic analysis of factors regulating arbor position shows that mediolateral position is determined in a binary fashion by expression (chordotonal neurons) or nonexpression (multidendritic neurons) of the Robo3 receptor for the midline repellent Slit. Robo3 expression is one of a suite of chordotonal neuron properties that depends on expression of the proneural gene atonal. Different features of terminal branches are separately regulated: an arbor can be shifted mediolaterally without affecting its dorsoventral location, and the distinctive remodeling of one arbor continues as normal despite this arbor shifting to an abnormal position in the neuropile (Zlatic, 2003).

Two characteristics of sensory neurons that contribute to the final specification of their terminal arborization can be distinguished: position and modality. This study deals specifically with the question of modality-dependent arborization and, as a model for understanding the mechanisms involved, uses the distinctively different projections formed by the ch and md neurons, respectively. The complexity of individual arbors can be reduced by focusing separately on the essential features that give each arbor its characteristic identity; this study initially concentrated on arbor positioning within the mediolateral axis of the CNS. This work was carried out within a framework provided by a prominent set of axon fascicles that are labeled by antibodies to the cell adhesion molecule Fasciclin II (FasII), and this enables the detection of small changes in the position and/or structure of terminal arbors. The axons of the ch neurons branch and terminate on an intermediate axon fascicle, while those of the md neurons terminate on a medial fascicle. There are additional distinguishing features in the two projections, such as their positions in the dorsoventral axis and the late embryonic remodeling of the arbor formed by the dorsal bipolar dendrite (dbd) neuron, an identified member of the md class (Zlatic, 2003).

Using this system, the following four questions were asked about the way in which cell type-specific arborization is patterned in the developing CNS. (1) Since it is known that the Slit/Robo system positions axon fascicles in the mediolateral axis of the CNS, it can be asked whether the same system directly or indirectly controls the positions at which terminal arbors are formed in this axis. (2) Are different aspects of the same arbor, such as dorsoventral as opposed to mediolateral positioning, coordinately controlled or independently specified? (3) Will the likely targets of terminating neurons influence the position at which termination actually occurs? (4) What are the genetic controls that specify the different characteristics of a terminal arbor? It was found that arbor position in the mediolateral axis is determined in a binary fashion by the expression (ch neurons) or nonexpression (md neurons) of the Robo3 receptor for the midline repellent Slit. Robo3 expression is one of a suite of properties characteristic of ch neurons that depends on the expression of atonal in the precursors of these cells. Thus, while ectopic expression of Robo3 in md neurons switches their terminal arbor from a medial to an intermediate fascicle in the mediolateral axis, other properties of the arbor (such as late remodeling of the dbd arbor) are unaffected. In contrast, ectopic expression of the proneural gene ato in the precursors of md neurons transforms all aspects of their central projection to form a ch-like arborization (Zlatic, 2003).

The terminal arbors of sensory neurons in the Drosophila embryo form within a developing neuropile that consists largely of commissural and longitudinal axon fascicles together with the intervening branches and dendrites of other neurons. This mass of axons and dendrites is an organized three-dimensional structure within which there are clear signs of a conserved architecture that recurs throughout the insects and probably more widely in the arthropod phylum. At its most fundamental level, there is a subdivision in which the principal arborizations of the motoneurons form dorsally, whereas sensory terminals are generally found more ventrally. The endings of sensory neurons themselves terminate in distinctive regions of the more ventral neuropile that have an overall similarity across the insects -- the mechanosensory endings in the most ventral sector and the proprioceptive endings more dorsally (Schrader, 2000). This suggests that connections form in a structured environment that imposes an order on the developmental process. It is the nature of this ordering process with which this paper is concerned. The analysis focuses on the distinctive terminal arborizations formed by two different types of sensory neurons, the ch neurons and the dbd neurons (Zlatic, 2003).

The first detailed maps of sensory terminals in the developing CNS of the Drosophila embryo and larva revealed that the arborizations of the sensory axons are as characteristic of particular sense organs as the structures of the sense organs themselves (Merritt, 1995; Schrader, 2000). For example, ch organs have arbors that form at an intermediate and fairly ventral level in the neuropile. In contrast, the arbors of most md neurons develop at a more medial and slightly more ventral location. Within the md class, however, four neurons have specialized projections. In particular, the dbd neuron has a dorsal projection that is uniquely remodeled toward the end of embryogenesis. Interestingly, dbd and the other neurons with specialized arbors correspond to identifiable subsets of cells within the md class that are distinguished both by their peripheral structures and by patterns of gene expression (Zlatic, 2003).

The first finding is that in wild-type embryos, md and ch arborizations are closely aligned with elements of the FasII-positive axon tracts that form a conspicuous set of longitudinal pathways in the embryonic and larval nervous systems. These consistent patterns of termination allow the precise definition, in two dimensions (the dorsoventral and mediolateral axes), of the different coordinates at which ch and dbd terminals will normally form in developing neuropile (Zlatic, 2003).

How are these coordinates specified for each neuron type? Previous work in Drosophila has shown that the Robo receptors for the repellent Slit are responsible for positioning CNS axon fascicles in the medial, intermediate, and lateral tracts on either side of the midline. Robo has also been shown to influence dendritic growth and synaptic connectivity in the giant fiber system of the adult fly. At the time of sensory neuron branch formation, Robo is expressed in dbd/md and ch neurons and acts to confine both kinds of projections to the ipsilateral CNS. In a robo mutant, both sets of neurons form bilateral projections, but the distinctive locations of the terminals persist on both sides of the midline. This distinction depends on the expression of Robo3, which is present in ch neurons but not in dbd/md neurons. When Robo3 is absent or downregulated, ch terminals form at the more medial position characteristic of db/dmd neurons. Similarly, if robo3 is ectopically expressed in dbd/md neurons, their terminals are shifted into the ch region of the neuropile. The action of Robo3 on mediolateral positioning is a direct one for two reasons: (1) either ectopic expression of Robo3 or downregulation of Robo3 function (by expression of comm) in sensory neurons but not their central targets selectively induces lateral shifts of sensory arbors and (2) ch projections can form in an intermediate position independently of their normal substrate for arborization. In robo3 mutants, the intermediate FasII fascicle on which ch arbors normally form shifts medially, as do the ch projections. However, when Robo3 function is selectively restored in the sensory system of robo3 mutants, the ch projections form in the intermediate region of the neuropile, as in wild-type (Zlatic, 2003).

These experiments also show that the locations at which terminals form in the mediolateral and dorsoventral axes are separately specified and one can be manipulated independently of the other. Thus, in all cases where dbd/md or ch terminals are shifted in the mediolateral axis by changing levels of Robo3, the dorsoventral position remains unchanged: a medially shifted ch terminal is ventral; a laterally shifted dbd terminal is dorsal. This finding reinforces the distinction between two alternative ways of determining the position of sensory axon terminals in the embryonic neuropile. The simplest model would suggest that growing axons are attracted to a specific central target: in this case, the position of the target (however determined) fixes the three-dimensional location of the terminal arbor of the sensory neuron. These results suggest that this model does not apply: in these experiments, the position of the terminal arbor is regulated by factors that separately specify its position in two dimensions. It is the response of the sensory axon to these different factors rather than to their targets that determines where the terminal arbor will form (Zlatic, 2003).

It is concluded that there are several properties of ch neurons that dictate the position and shape of their terminal arbors, of which Robo3 expression is one. Robo3 expression is specified by the proneural gene ato. Thus, when ato is expressed in other embryonic sensory neurons, these cells too express Robo3. In addition, misexpressing ato can completely transform dbd projections into ch-like projections (although the cell body remains dbd-like). In such cases, both mediolateral and dorsoventral positions are changed and the late stage remodeling does not take place. Thus, ato not only regulates Robo3 expression but also controls the whole suite of properties that dictates the position-specific termination of ch neurons. The mechanism of this regulation, at least for Robo3 expression, is likely to be an indirect one. (1) Transformation (including Robo3 expression) of md into ch-like neurons by ectopic ato requires that ato be expressed early, during the period of sensory organ specification and not as the arbors form. (2) Only primary (not secondary) ch organs express ato, yet all ch neurons express Robo3 and project their terminal arbors into a common region of the neuropile (Zlatic, 2003).

Previous studies have shown that ectopic ato can transform adult es organs into ch organs and that those transformations are achieved by an active downregulation of cut, which is required for es organ formation but is also expressed in a subset of md neurons. If ato upregulated Robo3 by downregulating Cut, then in cut mutant embryos all normally cut-expressing md neurons should be found to express Robo3. This is not the case and in cut mutants Robo3 is expressed only in the ch neurons, as in wild-type. It is also found that md neurons of the dorsal cluster, which normally express Cut, retain their wild-type projections in 15 hr cut mutant embryos (Zlatic, 2003).

While ato regulates ch-like characteristics, what specifies the unique features of the dbd arbor? Alone among md neurons, the dbd neuron projects to the dorsomedial FasII fascicle and undergoes late stage remodeling. The md neurons are a heterogenous population with respect to their expression of proneural transcription factors and identity genes such as cut. Interestingly, the SOPs of the dbd (and ddaE) neurons do not express cut or the AS-C but a different proneural gene, amos. By analogy with the role of ato in determining the ch-like projection pattern, amos may determine aspects of the dbd-specific projection pattern. It is also observed that ectopic expression of ato in dbd can generate intermediates, halfway transformations between dbd and ch, both with respect to mediolateral positioning and late stage remodeling. These intermediates may reveal competitive interactions between ch- and dbd-specific transcription factors. Thus, proneural genes could conceivably act in a combinatorial fashion to specify diverse shapes and positions of terminal branches (Merritt, 1995). Such a view is reinforced by the fact that it is those sensory neurons, which express a combination of proneural genes in their SOPs, that have distinctive terminal arbors (Merritt, 1995). However, there are likely to be additional factors that contribute to specify individual neuronal arbors, in particular in those instances in which terminal projection varies as a function of position. For example, in the cockroach the transcription factor Engrailed controls axon projections and synaptic choice of identified sensory neurons. Similarly, the prepattern genes araucan and caupolican play a role in establishing the difference in the projection pattern between lateral and medial es neurons in the adult Drosophila notum (Zlatic, 2003 and references therein).

Working from first principles, it might be supposed that different kinds of sensory neurons would locate the distinctive positions appropriate for their terminal arbors by seeking out their target interneurons in the developing CNS. However, in its simplest form, this model must be incorrect: in one axis at least, terminating axons detect and are positioned by their response to a cue that is produced not by their targets but by cells on the midline. This might suggest that sensory terminals are located at appropriate positions within the neuropile by factors that are quite independent of the neurons with which they will ultimately form connections. Indeed, an early experiment in the zebrafish embryo showed that if the target Mauthner cell is removed, the conspicuous cap of commissural axon terminals that normally encloses the Mauthner axon hillock continues to form at its proper location in the neuropile. It cannot necessarily be concluded from these experiments that sensory terminals in Drosophila are similarly positioned by factors entirely independent of the target. In the periphery, motoneuron axons are guided to their proper muscles by a hierarchy of cues, starting with a transcription factor code that delivers them to particular regions of the muscle field within which they then seek out and synapse with their targets. In the case of the embryonic sensory neurons, a hierarchy of cues, including target-derived signals, could also be envisaged that would contribute to the final projection pattern and certainly to synaptogenesis (Zlatic, 2003).

When approaching the apparently complex issue of how appropriate connections are formed between neurons within the meshwork of alternatives presented in the developing neuropiles, there may be two simplifying processes that should be considered. The first is that initial patterns of termination may be determined by a combination of cues that separately dictate distinct features of the arbor. The second is that the combination of such cues coupled with specific patterns of receptor expression may lead to the local arborization of pre- and postsynaptic neurons on a common substrate. Such patterns of growth would build a coherent platform on which detailed connectivity could then be established (Zlatic, 2003).

Evolution of neural precursor selection: functional divergence of proneural proteins

How conserved pathways are differentially regulated to produce diverse outcomes is a fundamental question of developmental and evolutionary biology. The conserved process of neural precursor cell (NPC) selection by basic helix-loop-helix (bHLH) proneural transcription factors in the peripheral nervous system (PNS) by atonal related proteins (ARPs) presents an excellent model in which to address this issue. Proneural ARPs belong to two highly related groups: the ATONAL (ATO) group and the NEUROGENIN (NGN) group. A cross-species approach was used to demonstrate that the genetic and molecular mechanisms by which ATO proteins and NGN proteins select NPCs are different. Specifically, ATO group genes efficiently induce neurogenesis in Drosophila but very weakly in Xenopus, while the reverse is true for NGN group proteins. This divergence in proneural activity is encoded by three residues in the basic domain of ATO proteins. In NGN proteins, proneural capacity is encoded by the equivalent three residues in the basic domain and a novel motif in the second Helix (H2) domain. Differential interactions with different types of zinc (Zn)-finger proteins mediate the divergence of ATO and NGN activities: Senseless is required for ATO group activity, whereas MyT1 is required for NGN group function. These data suggest an evolutionary divergence in the mechanisms of NPC selection between protostomes and deuterostomes (Quan, 2004).

ATO and NGN proteins share 47% identity in the bHLH domain including eight out of 12 amino acid residues in the DNA binding basic domain and are expressed in both the Drosophila and vertebrate PNS. However, differences in their usage for early NPC specification in the PNS between vertebrates and invertebrates have been noted. NGN proteins do not act early in NPC specification in invertebrates, whereas ATO proteins do not act early in NPC specification in vertebrates. Does this switch in the use of proneural proteins reflect a mechanistic difference or an inert change of expression pattern of otherwise functionally equivalent genes? To address this issue, a comparative analysis was initiated using Drosophila and Xenopus as model systems. To assay the proneural activity of mouse NGN1 and fly ATO in vertebrates, the mRNA of each was injected into a single blastomere of a two cell-stage Xenopus embryo. Neuronal induction was detected at stage 15 via whole-mount in situ hybridization for N-tubulin, an early marker of neuronal differentiation. Compared with uninjected embryos, injection of Ngn1 mRNA induces a large number of ectopic neuronal precursors on the injected side. By contrast, injection of Ato mRNA does not induce a significant increase in N-tubulin expression. Similarly, the injection of mRNA for Math1, a mouse ortholog of ato does not significantly increase N-tubulin expression at stage 15 but very few scattered N-tubulin positive cells can be seen at stage 19. Similar observations have been made for Xath1 which has been shown to be a much weaker inducer of neuronal precursors than NGN1. These data suggest that the Xenopus ectoderm responds robustly to NGN group proteins, but very weakly to ATO group proteins, to induce neurogenesis (Quan, 2004).

One explanation for the weakness of ATO and MATH1 activity in Xenopus is that NGN proteins are more potent neural inducers than ATO proteins and that, in parallel, stronger induction is needed in the vertebrate neuroectoderm than in the Drosophila neuroectoderm. To test this possibility, ATO proteins and NGN proteins were misexpressed in Drosophila using the UAS/Gal4 system and neural induction was assayed by counting the number of sensory bristles produced. Consistent with the fact that ato and Math1 completely rescue each other's loss of function, the two genes show very similar phenotypes and they were used interchangeably throughout the study. Expression of ngn2 with four different wing disc Gal4 drivers (C5-Gal4, 71B-Gal4, 32B-Gal4 and dpp-Gal4) showed no neural induction. Sixteen out of 23 ngn1 transgenic lines showed no neural induction. The other seven showed very weak induction with the strongest Gal4 driver, dppGal4. Therefore, the combination of dppGal4 and the strongest uas-ngn1 transgenic line were used in the rest of this study to determine the genetic and molecular basis of the difference in activity between ATO proteins and NGN proteins. The dppGal4 driver in Drosophila is used to induce genes of interest along the anteroposterior (AP) axis of the wing disc. Wild-type flies have no external sensory bristles or chordotonal organs (CHOs) on the AP axis of the wing blade. By contrast, a large number of sensory bristles is found along the AP axis of the wing with 100% penetrance when either MATH1 or ATO is expressed using dppGal4. In addition, both ATO and MATH1 induce CHOs. Expression of the strongest NGN1 transgenic line results in very few bristles in only 70% of the flies examined and no detectable CHOs. Quantitative analysis reveals that the number of sensory bristles induced by MATH1 is sixfold more than induced by NGN1. Identical observations were made in the few surviving flies under the same conditions using strong UAS-ATO lines. In the vertebrate PNS, NGN1 and NGN2 are sometimes co-expressed, and activate the expression of NeuroD group proteins. Therefore, it is possible that the weak neural induction of mouse NGN1 is due to the lack of homologs of NeuroD proteins in flies. However, co-expression of NGN1 and NGN2 or NGN1 and MATH3, a NeuroD group protein, failed to enhance the proneural activity of NGN1 in Drosophila (Quan, 2004).

One explanation for the very small number of bristles obtained after strong expression of NGN1 may be that the protein is able to induce NPCs, but most of these NPCs fail to differentiate properly and do not give rise to sensory organs. To test this possibility, NPC formation was examined directly upon expressing NGN1, ATO and MATH1 with dppGal4 in A101-lacZ flies. A101-lacZ is an NPC specific enhancer trap. The normal pattern of NPCs is revealed by anti-ß-GAL staining in third instar larval (L3) wing discs. Misexpression of ATO along the AP axis of the wing disc results in the induction of ectopic NPCs within the domain of ATO expression. By contrast, despite high levels of NGN1 expression, no detectable increase in NPCs is observed upon expression of NGN1. Similarly, ATO, but not NGN1, induced asense expression, another marker of NPC specification (Quan, 2004).

Is the weak activity of NGN1 specific to ectopic expression in the wing disc? Wide expression of NGN1 in embryos using da-Gal4 does not result in ectopic neurons. Finally, attempts were made to rescue the loss of ato in the eye imaginal disc using Gal4-7 and uas-ngn1. Gal4-7 induces expression anterior to the morphogenetic furrow and has been used to restore photoreceptors to ato mutant eye discs using scute and Math1. Expression of NGN1 in ato mutant discs did not result in any rescue nor did it induce ectopic R8 cells when expressed in control discs. For simplicity, the number of bristles was used as a quantitative assay for NPC formation for the remainder of the study (Quan, 2004).

To explore whether the differential activities of NGN proteins and ATO proteins can be understood at the level of the proteins themselves, a comparative analysis of the amino acid sequence of the basic domain was performed. Several studies have shown that important information is encoded by the basic domain, or specific residues therein. In addition, the 12 amino acids in the basic domain are sufficient to phylogenetically delineate ATO proteins and NGN proteins, arguing that sequence differences within the basic domain are of functional significance. However, previous studies did not investigate the genetic basis or address the evolutionary implications of the variation in basic domain sequence. ATO proteins and NGN proteins share eight residues out of 12 in the basic domain. One is variable, and the other three residues show almost absolute group specificity: they are highly conserved within each group but are essentially never the same between the two groups. To investigate whether this sequence specificity can explain the species-specific activities of ATO proteins and NGN proteins, a chimeric protein was created, exchanging the three group-specific amino acids in the basic domain of NGN1 for those present in ATO, named NGN^bATO. Expression of NGN^bATO induces the appearance of bristles along the AP axis of the wing in all transgenic lines examined. Strong UAS-NGN^bATO lines mimic strong UAS-ATO lines and result in significant lethality and more than 60 bristles per wing in the few surviving flies. Moderate UAS-NGN^bATO lines behave like moderate UAS-ATO lines and induce an average of 33 bristles per fly along the AP axis when compared with an average of seven for strongest UAS-NGN1 lines. Conversely, a chimeric protein was created exchanging the three group-specific amino acids in the basic domain of ATO to NGN1, named ATO^bNGN. Whereas the injection of Ato mRNA in Xenopus embryos has no significant effect on the N-tubulin expression pattern, the injection of Ato^bNGN mRNA induces N-tubulin expression, indistinguishable from that caused by the injection of NGN1. Therefore, the NGN^bATO mutant recovers the NPC inducing activity of ATO in Drosophila, and the ATO^bNGN mutant recovers the NPC inducing activity of NGN1 in Xenopus (Quan, 2004).

It is worthwhile to notice that only some of residues in the basic domain are directly contacting to DNA. The specific activities of ATO and NGN1 are unlikely to depend on differential DNA-binding activity, since ATO proteins and NGN proteins have identical DNA contact amino acids and can activate the NeuroD promoter via the same E-boxes in P19 cells. Interestingly, biophysical and DNA-binding studies comparing MASH1 and MyoD have shown that they display similar binding preferences leading to the conclusion that their different target specificities cannot be explained solely by differential DNA binding. Similar conclusions were made comparing ato and sc activities in neural subtype specification (Quan, 2004).

To investigate whether other functionally specific motifs exist in the bHLH domain of ARP proteins, the evolutionary trace (ET) analysis method was used. ET tracks residues whose mutations are associated with functional changes during evolution. This approach has been used to identify novel functional surfaces, and has recently been shown to be widely applicable to proteins. In practice, ET relies on the phylogenetic tree of a protein family and identifies residues of the alignment that are invariant within branches but variable between them. These positions are called 'class specific'. The smallest number of branches at which a position first becomes class specific defines its rank. The top ranked positions (1) do not vary. Very highly ranked positions (2-8) are such that they vary little and, whenever they do, there is also a major evolutionary divergence. By contrast, poorly ranked positions vary more often, and their variation does not seem to correlate with divergence. Thus, highly ranked positions tend to be functionally important, while poorly ranked ones tend not to be. When examining ARP bHLH domains, ET identified a number of positions that are jointly important in different bHLH domains, yet that undergo significant variation between them. These residues varied in rank from 2 to 7, suggesting that they can undergo non-conservative mutations that are likely to correspond to functional divergence events. These positions tend to be most conserved between NeuroDs and NGN proteins and then undergo variations in ATO proteins, suggesting that they are important for an activity shared by NGN proteins and NeuroDs, but absent in ATO proteins. The data above show that the ability to induce NPCs in vertebrates is precisely such an activity. To investigate further the role of these group-specific residues on functional specificity, a chimeric protein, named NGN^H2ATO (exchanging amino acids 37, 39, 43, 44 and 46 in Helix2 of NGN1 with those present in ATO), was created and tested in Drosophila. Expression of the strongest NGN^H2ATO transgenic line induces a maximum of two bristles along the AP axis of the wing per fly in 50% of the flies. Quantitative analysis shows that, unlike ATO, NGN^H2ATO induces an average of 0.8 bristles along AP axis per fly. These data indicate that the group-specific motif in Helix2 of ATO does not encode proneural activity in Drosophila. Conversely, a chimeric protein, named ATO^H2NGN, was generated exchanging the same five amino acids in Helix2 of ATO to those found in NGN1. Injection of ATO^H2NGN mRNA causes ectopic N-tubulin expression, indistinguishable from the injection of NGN1. Therefore, ATO^H2NGN recovers the activity of NGN1 in Xenopus. Taken together, the mutational analysis results agree with the predictions of the ET analysis indicating that the identified residues in Helix2 mediate the activity of NGN proteins but not of ATO proteins (Quan, 2004).

The data support the hypothesis that ATO proteins and NGN proteins act via different genetic pathways to specify NPCs in different species. What might those pathways be? One simple explanation may be that NGN1 is not able to form heterodimers with fly Daughterless (DA), a required partner protein for NPC specification. In order to test this possibility, co-IP experiments were performed, in which ³⁵S-labeled ATO, MATH1 or NGN1 were co-precipitated with DA-Myc using anti-Myc antibodies. In the presence of DA, mouse MATH1, fly ATO and mouse NGN1 are co-precipitated. Only background levels of NGN1 are detected in the absence of DA. These results suggest that mouse NGN1 can bind physically to fly DA in vitro. To test if DA and NGN1 can interact genetically in vivo, NGN1 was expressed in the absence of one copy of da. The number of sensory bristles produced by NGN1 along the AP axis is greatly decreased in a heterozygous da background. Therefore, mouse NGN1 can physically and genetically interact with fly da in Drosophila in a dose-sensitive manner (Quan, 2004).

Next, the possibility that mouse NGN1 does not respond to the Drosophila Notch signaling pathway was examined. To test this, neural induction by NGN1 was examined in the absence of one copy of Notch (N^+/–) or with the co-expression of Notch pathway genes. The proneural activity of NGN1 is enhanced in a N^+/– background. Conversely, NGN1 activity is completely inhibited by co-expression of a constitutively active form, N^intra or members of the E(Spl) complex, m8 and mdelta. These data demonstrate that mouse NGN1 can be regulated by the Notch signaling pathway in Drosophila. It should be noted that overexpression of ATO in a N heterozygous background results in almost complete lethality and in extremely deformed wings, owing (in part) to a very large number of bristles in the few surviving flies. Since both ATO and NGN proteins can respond to levels of Notch signaling but only ATO proteins can efficiently specify NPCs, it is possible that ATO proteins and NGN proteins use different mechanisms to interact with the Notch signaling pathway. One possibility is that NGN proteins are more sensitive than ATO proteins to levels of transcriptional inhibitors of proneural activity encoded by the E(spl) genes because NGN1 activity, like that of ATO, can be repressed by ectopic expression of E(spl) proteins. However, in contrast to what is observed with Notch, removing a copy of the E(spl) complex does not alter NGN1 activity, suggesting that NGN1 is not more sensitive to levels of transcriptional inhibitors of proneural activity (Quan, 2004).

NPC formation in Drosophila requires the Zn-finger protein Senseless (SENS). Fly proneural proteins first induce sens expression and then synergize with it in a positive feedback loop. This appears to enhance the ability of proneural genes to downregulate Notch signaling in the presumptive NPC. In vertebrates, Senseless-like proteins appear not to act in NPC formation, although they are expressed in the PNS. To test the possibility that SENS shows group specific interactions with bHLH proteins during NPC selection, the abilities of ATO and NGN1 to induce SENS were examined. SENS expression in wild-type L3 wing discs marks NPC formation. Ectopic SENS induction is detected along the AP axis of wing discs when ATO is misexpressed. However, SENS expression is not induced by NGN1. These data suggest that unlike ATO, NGN1 does not efficiently induce SENS expression. Whether lowering endogenous levels of Notch would allow NGN1 to induce SENS was examined. Expression of NGN1 in Notch heterozygous animals, although significantly increasing the number of induced bristles, fails to induce SENS expression when compared with N^+/– controls, arguing that NPCs induced by NGN proteins are specified via a different mechanism not normally used in Drosophila. Although NGN1 does not induce SENS, it is possible that synergy might occur if the requirement for SENS induction is bypassed. Therefore the ability of NGN1 and MATH1 to synergize with SENS in vivo was compared by co-expressing either NGN1 or MATH1 with SENS using a moderate scutellar Gal4 driver (C5-Gal4). Neural induction was examined by counting the ectopic bristles induced on the scutellum. Wild-type flies have four large bristles, or macrochaete, on their scutella. Expression of SENS or MATH1 alone with C5-Gal4 induces a number of ectopic microchaete, or small bristles, on the scutellum. No ectopic sensory bristles were found when NGN1 was expressed alone. Co-expression of NGN1 and SENS has the same effect on the scutellum as the misexpressing SENS alone. Co-expression of MATH1 and SENS, however, causes the appearance of a large number of both micro- and macrochaete. Finally, NGN1 or MATH1 were co-expressed in the absence of one copy of sens. No effect on NGN1 activity in a sens^+/– background was observed. By contrast, the average number of sensory bristles produced by MATH1 along the AP axis was reduced by 42% if a single copy of sens was removed suggesting dose-sensitive interactions. Thus, neither by loss nor gain of function criteria does NGN1 appear to interact with SENS, thus explaining its weak proneural activity and inability to efficiently antagonize Notch signaling in Drosophila. Therefore, SENS is a key extrinsic difference in how ATO proteins and NGN proteins regulate NPC selection (Quan, 2004).

In Xenopus, the C2HC-type Zn-finger protein X-MyT1 is expressed in primary neurons and can be induced by NGN proteins. In addition X-MyT1 has been suggested to play a role in NPC formation and to synergize with NGN proteins. In order to test if X-MyT1, like SENS, shows specificity in its interaction with ARP proteins, its ability to interact with NGN1 and ATO in Xenopus was compared. X-MyT1 mRNA was injected alone or co-injected with either Ngn1 or Ato mRNA. As expected, the injection of X-MyT1 increases the number of N-tubulin-expressing cells in the neural plate domains where neurons normally form, while the injection of Ngn1 mRNA alone leads to induction of N-tubulin expression. Co-injection of Ngn1 and X-MyT1 mRNAs results in very strong N-tubulin induction, pointing to a synergistic interaction between the two proteins. By contrast, co-injection of Ato and X-MyT1 mRNAs does not cause a detectable increase in N-tubulin expression compared with the injection of X-MyT1 mRNA alone. Similarly, the few ectopic N-tubulin-expressing cells observed when Math1 mRNA is injected are not increased by co-injection of Math1 and X-MyT1. Thus, X-MyT1 interacts specifically with NGN1 and not with ATO or MATH1. The data above demonstrate that the correct combination of ARP protein and Zn-finger protein is necessary for NPC induction (Quan, 2004).

Does the coding sequence difference mediate the divergence in the genetic interactions of ARPs? To test this, whether the chimeric proteins recover the ability to interact with the respective Zn-finger proteins was examined. Indeed, expression of NGN^bATO in Drosophila results in the induction of SENS, and the number of bristles induced by NGN^bATO in absence of one copy of sens (sens^+/–) is reduced by ~44%. In addition, strong synergy was observed by co-expression of NGN^bATO and SENS using the dppGal4 driver. Therefore, NGN^bATO is able to induce and interact with SENS in Drosophila. In Xenopus, just like Ngn1, co-injection of Ato^bNGN and X-MyT1 mRNAs results in synergy and very strong ectopic N-tubulin expression when compared with the injection of X-MyT1 or Ato^bNGN alone. Similarly, co-injection of ATO^H2NGN and X-MyT1 mRNAs results in synergy and very strong induction of N-tubulin expression suggesting that ATO^H2NGN and ATO^bNGN use the same mechanism of action as NGN1 (Quan, 2004).

At the developmental level, the data presented in this study can be explained by two possibilities. The first is that Drosophila and vertebrates use different bHLH proteins with divergent mechanisms for selecting similar cell types: the earliest born neural progenitors. Alternatively, NGN proteins may be involved in selecting neuronal (versus glial) rather than earliest born neural progenitors in vertebrates. This is certainly the case in the mammalian inner ear and it should be determined whether it is a more generally applicable rule, at least in the PNS. These two models for NGN function are not mutually exclusive. It is possible that in different lineages, NGN proteins select first neural, and then neuronal, precursors. This would be compatible with data from both flies and vertebrates showing that Notch signaling, in addition to having anti-neural effects, has also anti-neuronal and pro-glial effects during neural lineage development. Analysis of the fly NGN protein, TAP, may shed some light on this issue. At any rate, a comparative approach should provide a powerful tool for the systematic analysis of the pathways which program neural stem cells (Quan, 2004).

Regardless of the precise developmental step at which ATO proteins and NGN proteins act, it is clear that the genetic and molecular mechanisms by which they act are different, suggesting that the functions of ATO proteins and NGN proteins are regulated by different factors. Furthermore, it is clear that the group-specific amino acids underlie these molecular differences. At this point it is difficult to interpret the precise role of the group specific residues in molecular terms. Nonetheless, three possibilities seem reasonable. The first is that currently unknown proteins bind to these residues. The second is that these residues are sites of differential post-translational modifications which in turn influence the choice of target gene specificity. Finally, it is possible that while these residues do not bind to DNA themselves, they influence the three dimensional structure or the conformational changes which DNA binding residues assume upon contacting DNA. In this scenario, these residues do ultimately influence the choice of the binding site without themselves contacting it. The data illustrate the power of a comparative approach in identifying not only conserved, but also divergent, developmental mechanisms, and suggest a platform for screening for the genes mediating the divergence. It is noteworthy that NGN1, on the one hand, and XATH1 and MATH1, on the other, seem to have retained a type of proneural activity which is largely no longer needed in flies and vertebrates, respectively (Quan, 2004).

Finally, genes common to protostomes and deuterostomes (including atonal, ngn genes, Notch signaling genes, sens and X-MyT1) most probably derive from the last common bilaterian ancestor. This implies that such an ancestor already possessed all the tools to specify a large diversity of neural cell types and lineages, suggesting a structurally, and consequently behaviorally, complex animal (Quan, 2004).

The proneural proteins Atonal and Scute regulate neural target genes through different E-Box binding sites

For a particular functional family of basic helix-loop-helix (bHLH) transcription factors, there is ample evidence that different factors regulate different target genes but little idea of how these different target genes are distinguished. The contribution was investigated of DNA binding site differences to the specificities of two functionally related proneural bHLH transcription factors required for the genesis of Drosophila sense organ precursors (Atonal and Scute). The proneural target gene, Bearded, is regulated by both Scute and Atonal via distinct E-box consensus binding sites. By comparing with other Ato-dependent enhancer sequences, an Ato-specific binding consensus that differs from the previously defined Scute-specific E-box consensus was defined, thereby defining distinct EAto and ESc sites. These E-box variants are crucial for function: (1) tandem repeats of 20-bp sequences containing EAto and ESc sites are sufficient to confer Atonal- and Scute-specific expression patterns, respectively, on a reporter gene in vivo; (2) interchanging EAto and ESc sites within enhancers almost abolishes enhancer activity. While the latter finding shows that enhancer context is also important in defining how proneural proteins interact with these sites, it is clear that differential utilization of DNA binding sites underlies proneural protein specificity (Powell, 2004).

Ato and Sc must interact with distinct DNA binding sites in vivo. In the target genes analyzed in this study, residues immediately flanking the 6-bp core E box allow the definition of distinct EAto and ESc consensus binding sites for Ato/Da and Sc/Da, respectively. It can be deduced that these variant E boxes consist of half sites, with Sc, Ato, and Da contacting GCAG, AWCAK, and STGK, respectively (proneural core sites underlined). Striking affirmation of binding site differences is provided by the common target gene Brd, which is regulated by Ato and Sc in a modular fashion via distinct E boxes in different enhancers. These E-box variations are crucial for function. They are sufficient to confer proneural protein-specific expression patterns on a GFP reporter gene in isolation from other enhancer sequences. Moreover, interchanging ESc and EAto sites within proneural enhancers almost abolishes enhancer activity and so is almost equivalent to destroying the E box. This shows that the correct proneural regulation of target genes requires the presence of a specific E-box binding site in combination with the selective ability to interact with factors bound to other sites within these enhancers (Powell, 2004).

In vivo DNA binding site differences underlie proneural specificity. Yet, paradoxically, evidence from misexpression and protein structure-function studies has strongly suggested that the target gene specificities of Ato and Sc (and their vertebrate homologues) result from specific interactions with protein cofactors and not from intrinsic differences in how proneural proteins contact DNA. For instance, DNA-contacting residues are shared between the proteins, and consequently, as show in this study, Ato/Da and Sc/Da have identical DNA binding properties in vitro. One way to reconcile these observations is that protein-protein interactions with cofactors may induce DNA binding specificity in proneural proteins by modifying the conformation of the DNA-contacting residues of the bHLH domain. It is also possible that DNA binding itself is not selective in vivo but that occupancy of different DNA binding sites triggers productive or unproductive conformational changes in the proneural proteins that influence their interaction with cofactor proteins. The favored view is that specific cofactor interactions will induce distinctive DNA binding affinities; the conformational changes induced by each may be subtle individually but may be interdependent and mutually reinforcing (Powell, 2004).

Significantly, ato can rescue mutations of its mouse orthologue, Math1, and vice versa, suggesting that DNA site preferences will be conserved among vertebrate orthologues. A number of functional E boxes have been characterized in vertebrate neural-specific genes, and in two cases the interacting bHLH protein is likely to be an Ato orthologue: an autoregulatory site in the Math1 promoter (TCAGCTGG) and a proposed Xath5 site in the 3 nAChr promoter (ACAGCTGG). Thus, in these cases the E boxes match EAto in the 5 flanking base (Powell, 2004).

For correct enhancer function, proneural proteins must interact differentially with other DNA binding factors. Although E-box consensus differences underlie specificity, enhancer context is usually also crucial for this specificity to be manifest. In swapping EAto and ESc sequences between the sc and ato autoregulatory enhancers, in only one case was a corresponding 'swap' in enhancer specificity observed: Ato could be made to regulate the sc-SMC-E enhancer via an EAto site. Otherwise, alteration of E-box flanking bases resulted in a severe loss of enhancer activity. This suggests that recruiting a different proneural protein cannot alone change the function of an enhancer. Correct proneural target enhancer function requires a combination of the correct E-box sequence and the ability to interact with other factors bound to the enhancer. This is reminiscent of the cooperative interaction between MyoD and MEF2 in myogenesis and of interaction between Sc/Da and Pannier/Chip to activate ac in a specific part of the thorax. For the Ato enhancer a requirement for cooperative interaction between Ato/Da and the ETS protein Pointed, bound to a site adjacent to the EAto site has been shown (zur Lage, submitted, cited in Powell, 2004). Similarly, neurogenin 2 interacts with LIM factors during the activation of subtype-specific target genes. The finding that EAto and ESc sites encode much specificity in artificial enhancers suggests that tandem E boxes remove the requirement for interaction with factors bound to other DNA sites, perhaps because cooperative binding between proneural proteins themselves is then sufficient and may even allow the recruitment of cofactors by protein interactions alone. Interestingly, the converse situation may also occur: for both the ato and sc enhancers, there is a low level of expression remaining after swapping of E boxes. This suggests that the original bHLH protein can be recruited to the 'wrong' E-box sequence, inefficiently, by interaction with cofactors. A basis for this can be found with MyoD, where interaction with Sp1 allows MyoD to bind to a nonideal site in the human cardiac alpha-actin promoter (Powell, 2004 and references therein).

Another indication of the importance of enhancer context is that parent enhancers support patterns different from those of the isolated E boxes, at least in the case of Ato. [ato-E1]7-GFP is widely expressed in Ato-specific regions in the embryo, whereas the parent ato-FCO-E enhancer is limited to a small subset of chordotonal SOPs (zur Lage, submitted cited in Powell, 2005). In discs, ato-E1 drives expression relatively poorly in FCO precursors compared with ato-FCO-E. TAKR86C is even more extreme: the TAKR86C enhancer is normally active only in a single embryonic chordotonal precursor (the P cell), but the TAKR86C-E2 site drives Ato-dependent expression in the larval and adult eye and not in the P cell. Clearly, the parent enhancers must have other regulatory inputs that restrict expression (Powell, 2004).

Despite the importance of enhancer context and interaction with other factors, the ESc and EAto sequences support strikingly specific expression patterns when taken out of their enhancers. All tandem repeat E-box constructs tested were activated almost exclusively during PNS neurogenesis, despite the presence of some 24 class A factors in Drosophila. None were activated during CNS neurogenesis, myogenesis, or mesoderm formation, even though AS-C proteins function during the former two processes. In the case of two sites, sc-E1 and ato-E1, expression is remarkably consistent, with regulation solely by Ato or Sc, respectively, in PNS neurogenesis; the sites alone must contain all of the information necessary for specific recognition. It is remarkable that ato-E1 does not respond in vivo to the Ato-related protein Cato or Amos, even though the latter has a basic region almost identical to that of Ato and might be expected to have the same DNA binding properties. The main exception to this specificity is the presence of expression in embryonic ectodermal stripes. These resemble muscle attachment sites, suggesting recognition by the Ato superfamily member Delilah. The conclusion is that tandem duplications can overcome the need for DNA binding sites for other factors. Cooperative binding of proneural proteins may negate the need for cofactor interactions, or, as suggested above, cooperative binding may allow the recruitment of cofactors directly (Powell, 2004).

There are dramatic differences between the two Ato sites tested. Unlike ato-E1, the TAKR86C-E2 site drives expression in only a subset of Ato locations; it appears to be photoreceptor specific despite containing a good class A core E-box match (CAGGTG). This opens up the possibility that there may be different subtypes of Ato binding sites. The spatially restricted recognition of TAKR86C-E2 also implies that cellular context is important in how different sites are recognized. One may speculate, for instance, that eye-specific DNA binding properties of Ato may be conferred by interaction with PAX6 proteins. Interestingly, diversity of E-box expression patterns correlates with variability in the consensus sequences. The ESc consensus sequence (based on some 23 sites) is less variable than the Ato/Da consensus, even though the latter is based on only three sites. It is suggested that regulatory fine-tuning by E-box variation is more important for Ato target genes than for Sc target genes (Powell, 2004).

In summary, the E-box sequences and their flanking bases contain impressively sufficient information for regulation by specific proneural proteins. However, there is further complexity: at least the two Ato sites tested support different patterns and have a different relationship with their parent enhancers. Subtle variations in regulation by proneural proteins may therefore contribute to variations in target gene expression; indeed, there may be no such thing as a typical target site or target gene. This may also be true for common target genes: despite the modular regulation of Brd, the possibility is not ruled out that within the spectrum of proneural E boxes there are some sites that are jointly recognized by Sc and Ato in vivo and that this would be another mechanism for regulating common target genes (Powell, 2004).

Two-step selection of a single R8 photoreceptor: a bistable loop between senseless and rough locks in R8 fate following Ato activation of senseless

Patterning of sensory organs requires precise regulation of neural induction and repression. The neurocrystalline pattern of the adult Drosophila compound eye is generated by ordered selection of single founder photoreceptors (R8s) for each unit eye or ommatidium. R8 selection requires mechanisms that restrict R8 potential to a single cell from within a group of cells expressing the proneural gene atonal (ato). One model of R8 selection suggests that R8 precursors are selected from a three-cell 'R8 equivalence group' through repression of ato by the homeodomain transcription factor Rough (Ro). A second model proposes that lateral inhibition is sufficient to select a single R8 from an equipotent group of cells called the intermediate group (IG). This study provides new evidence that lateral inhibition, but not ro, is required for the initial selection of a single R8 precursor. In ro mutants ectopic R8s develop from R2,5 photoreceptor precursors independently of ectopic Ato and hours after normal R8s are specified. Ro directly represses the R8 specific zinc-finger transcription factor senseless (sens) in the developing R2,5 precursors to block ectopic R8 differentiation. These results support a new model for R8 selection in which lateral inhibition establishes a transient pattern of selected R8s that is permanently reinforced by a repressive bistable loop between sens and ro. This model provides new insight into the strategies that allow successful integration of a repressive patterning signal, such as lateral inhibition, with continued developmental plasticity during retinal differentiation (Pepple, 2008).

Ro is a homeodomain-containing protein and has been shown to bind DNA at two sites in its own enhancer containing an ATTA core sequence. To explore the possibility that Ro directly represses sens, the R8 specific sens enhancer was identified and the mechanisms regulating sens expression was characterized. A 645 bp fragment within the second intron of the sens genomic locus named F2 was identified that is sufficient to drive reporter expression specifically in photoreceptors of the developing eye-antennal imaginal disc. To test whether the F2 region is necessary for R8-specific sens expression, the 645 bp region was specifically deleted from the sens-L genomic rescue construct generating DeltaF2. In sens-null mutants, one copy of DeltaF2 rescues the null phenotype in all tissues except the eye. Thus, F2 is the sens eye enhancer and is necessary and sufficient for R8-specific sens expression (Pepple, 2008).

F2 contains two potential Ro-binding sites known as H1 and H2, for homeodomain 1 and 2. To test for a direct interaction, electrophoretic mobility shift assays (EMSAs) were performed. A probe containing H1 and H2 is bound specifically by Ro protein in vitro. Complete loss of binding occurs with mutation of H2. Mutation of H1 does not prevent Ro binding, but there may be a mild decrease in binding compared with the wild-type probe. To test the in vivo significance of these interactions, each site was mutated in a reporter generated with the minimal R8-specific enhancer, B-short-GFP, and the effect on GFP was evaluated. Although H1 is not required for Ro binding in vitro, mutation of H1 in B-short (termed H1^*) leads to consistent expression of GFP in two extra cells per ommatidium. These two cells were identified as the R2,5 photoreceptor pair by co-localization of GFP with β-galactosidase from the R2,5-specific enhancer trap RM104. GFP expression is also expanded into the R2,5 pair with the H2 mutation (H2^*). Mutation of both H1 and H2 (H1,2^*) results in a GFP expression pattern indistinguishable from H2^*. To test whether the loss of ro function has the same effect on B-short-GFP expression as does mutation of the Ro-binding sites, ro^X63 clones were generated. In the absence of ro function, both Sens and B-short-GFP expression are detected in two to three cells per ommatidium. Together with the in vitro binding data, these in vivo results suggest that Ro directly represses sens expression in R2,5 photoreceptors (Pepple, 2008).

Therefore, this work shows that Ro directly represses sens in developing R2,5 cells and that de-repression of Sens is sufficient to initiate R8 cell fate in the absence of ectopic Ato. Although there are a small number of ectopic Ato-expressing cells in column 3 in ro^x63 mutants, it is not likely that the additional 'R8' cells are due to misregulation of Ato since the great majority of ectopic 'R8s' never express detectable Ato protein after the intermediate group stage. It is more likely that the extra Ato-positive cells are due to secondary Sens activation of proneural gene expression, a previously reported phenomenon (Pepple, 2008).

sens is required for R8 differentiation to occur through repression of Ro in R8, and that ectopic Sens is sufficient to repress endogenous Ro expression. Thus, in the absence of sens, three R2,5 cells develop and in the absence of ro up to three R8 cells form per ommatidium. This reciprocal phenotype supports the existence of the three cell R8 equivalence group and a mechanism of mutual repression between sens and ro that specifies opposite cell types. Although one mechanism regulating this mutual repression is the direct repression of sens by Ro, other roles for Ro may exist. The Ro-binding site mutations do not produce the same level of GFP reporter protein expression elevation in R2,5 precursors that would be predicted from the level of GFP expressed in ro mutants. This suggests that Ro may also regulate sens by repressing an activator of sens expression in R2,5 precursors (Pepple, 2008).

Regardless of the mechanism, the negative-feedback loop between sens and ro is secondary to the initial force driving R8 selection in which Ato and Sens are transiently repressed by lateral inhibition in all but one cell within an IG. Thus, lateral inhibition transiently represses neural differentiation in the eye, establishing the patterned array of precisely spaced ommatidia while retaining the potential for later recruitment of undifferentiated cells to the photoreceptor cell fate. If the effects of lateral inhibition were to repress permanently the potential for neuronal differentiation, further retinal development would be blocked. Therefore, the effects of lateral inhibition must be limited, and the data indicate that column 3 is the boundary of its influence. Since the effects of lateral inhibition diminish, the negative-feedback loop between sens and ro reinforces the pattern of selected R8s and ensures that only one Sens-expressing cell from the R8 equivalence group develops as an R8. This simple bistable loop translates the transient developmental signal of lateral inhibition into a committed irreversible fate (Pepple, 2008).

In later R8 differentiation, another bistable loop is used to specify the 'pale' or 'yellow' subtypes of R8 photoreceptors. During this late developmental step, the bias for the 'pale' R8 fate is provided by a signal from a 'pale' R7. It is proposed that the bias signal that tips the fate decision in the sens-ro loop is provided by resolution of Ato to a single cell by lateral inhibition. Ato then directly activates Sens expression and biases that cell to the R8 cell fate. It is not yet known what activates Ro expression and thereby establishes the R2,5 cell fates. However, it has been suggested that epidermal growth factor receptor (EGFR) or Hedgehog signaling may be required for Ro expression. As a result, after the R8 bias is established, a signal such as the EGFR ligand Spitz could be sent from R8 to the two neighboring cells that bias their sens-ro loop towards Ro expression and the R2,5 fate. Once Ro expression is initiated in the R2,5 pair, the pattern of a single Sens-expressing R8 per ommatidium becomes irreversible (Pepple, 2008).

Proper patterning of the Drosophila eye requires precise selection of R8 precursors in a highly ordered array. Previously, the potential to assume the R8 fate was generally believed to reside in the single cell that achieved the highest balance of proneural induction by ato and escaped repression by lateral inhibition. This concept has influenced the interpretation of mutants that exhibit multiple R8 phenotypes, such as ro, by linking the extra R8s that form to cells that inappropriately maintain Ato expression. However, the data show that the expression pattern of Ato and Sens in a ro-null mutant is not altered in a manner consistent with this model. This re-evaluation of the ro phenotype suggests the intriguing possibility that undifferentiated cells posterior to the furrow retain the developmental plasticity to develop as R8s even in the absence of ongoing Ato expression (Pepple, 2008).

The ro phenotype demonstrates that, despite initial repression of the R8 cell fate by lateral inhibition, at least two additional cells have the potential to develop as R8s starting in column 3 if Sens expression is de-repressed. One of the subfragments of the sens eye enhancer, fragment C-GFP, is expressed in nearly all cells posterior to the MF, suggesting that sens could be de-repressed in cells other than the R2,5 cell precursors and initiate R8 development. The widespread expression of fragment C-GFP suggests that it lacks an important negative regulatory region distinct from Ro repression. One potential mechanism that may explain the fragment C-GFP expression pattern is that the stripe of Ato expression in the MF confers R8 potential to all cells and that this potential is only transiently repressed by lateral inhibition during patterning. Then, as the effects of lateral inhibition fade, secondary mechanisms repress sens expression and R8 differentiation in cells posterior to the MF. This model, demonstrated by the function of Ro and suggested by fragment C-GFP expression, is distinct from the previous concept that R8 cell fate is limited to cells of the intermediate group (Pepple, 2008).

The minimal eye specific enhancer of sens, fragment B-long, contains at least four potentially discreet regulatory elements that balance the positive and negative inputs required to specify a single R8 precursor per ommatidium. The first positively acting element is under the direct control of Ato/Da heterodimers and contains E-boxes 1 and 4. This element is required for Ato-dependent sens expression in the IGs and in columns 1-3. Although ato is at the top of the genetic cascade required for eye differentiation, sens is only the third direct target identified in the eye after bearded (brd) and dacapo (dap). Ato/Da heterodimers bind to two E-boxes (E1 and E4) to drive early sens expression in R8. This is in contrast to the previously described direct regulation of sens in SOPs of the embryonic and developing adult PNS by Ato and Scute at a single E-box in their common enhancer (Pepple, 2008).

The second positively acting regulatory element resides within the boundaries of fragment E1^*, although the minimal necessary sequence was not specifically identified. This element responds to an Ato-independent mechanism that is sufficient to maintain Sens expression in selected R8 cells after column 3. Sens is known to respond to Ato-independent inductive cues much later in R8 development (48 hours after pupation) when Sens expression requires the spalt genes. However, larval expression of Sens is not disrupted in spalt mutants, suggesting the existence of yet another unidentified positive regulator (Pepple, 2008).

In addition to these two positively acting elements, there are also at least two negative regulatory elements. The Ro-binding element H2, that is responsible for repressing Sens expression in R2,5 cells, was specifically identified. The second element was not specifically identified, but its presence is suggested by the nearly ubiquitous expression of fragment C-GFP. Together these positive and negative regulatory elements outline an elegant strategy for the multi-staged selection of a single R8 per ommatidium and highlights a model where blocking R8 cell fate potential with sequential, independent, repressive mechanisms is an important strategy for patterning and cell fate development in the Drosophila eye (Pepple, 2008).

Robust target gene discovery through transcriptome perturbations and genome-wide enhancer predictions in Drosophila uncovers a regulatory basis for sensory specification

A comprehensive systems-level understanding of developmental programs requires the mapping of the underlying gene regulatory networks. While significant progress has been made in mapping a few such networks, almost all gene regulatory networks underlying cell-fate specification remain unknown and their discovery is significantly hampered by the paucity of generalized, in vivo validated tools of target gene and functional enhancer discovery. This study combined genetic transcriptome perturbations and comprehensive computational analyses to identify a large cohort of target genes of the proneural and tumor suppressor factor Atonal, which specifies the switch from undifferentiated pluripotent cells to R8 photoreceptor neurons during larval development. Extensive in vivo validations of the predicted targets for the proneural factor Atonal demonstrate a 50% success rate of bona fide targets. Furthermore it was shown that these enhancers are functionally conserved by cloning orthologous enhancers from Drosophila ananassae and D. virilis in D. melanogaster. Finally, to investigate cis-regulatory cross-talk between Ato and other retinal differentiation transcription factors (TFs), motif analyses and independent target predictions were performed for Eyeless, Senseless, Suppressor of Hairless, Rough, and Glass. These analyses show that cisTargetX identifies the correct motif from a set of coexpressed genes and accurately predicts target genes of individual TFs. The validated set of novel Ato targets exhibit functional enrichment of signaling molecules and a subset is predicted to be coregulated by other TFs within the retinal gene regulatory network (Aerts, 2010).

Identifying target genes for any TF through genome scanning remains a significant challenge because any given consensus sequence has 10³-10⁶ instances throughout the genome. For example, there are more than 600,000 matches to the canonical E-box motif CANNTG in the genome and ~10,000 to ~200,000 single matches to the more specific Ato motif, depending on the similarity threshold employed. To solve this problem a method called cisTargetX was developed to predict motif clusters across the entire genomes of 12 Drosophila species and determine significant associations between motifs and subsets of coexpressed genes. Validation of cisTargetX on publicly available gene sets identifies the correct motif and targets for nearly all tested TFs, demonstrating the general utility of approach. Therefore a cisTargetX Web tool was developed available freely at http://med.kuleuven.be/cme-mg/lng/cisTargetX (Aerts, 2010).

cisTargetX is conceptually similar to the PhylCRM/Lever and ModuleMiner methods for vertebrate genomes and allows determining whether a set of candidate genes, for example a mixture of direct and indirect target genes, is enriched for direct targets of a certain TF or combination of TFs. Compared to other motif discovery methods, such as Clover, PASTAA, PSCAN, and oPOSSUM, cisTargetX integrates motif clustering, cross-species comparisons, and whole-genome backgrounds in the discovery process. Additionally, and unlike the vertebrate methods mentioned above, cisTargetX focuses on homotypic CRMs and therefore allows separating the motif scoring (performed offline) from the gene set enrichment analysis (performed online), yielding a computationally efficient method that can be used as an online Web application. A second difference from PhylCRM/Lever is that once a predicted motif is selected, cisTargetX determines the optimal subset of direct TF targets from the input set (Aerts, 2010).

cisTargetX was applied to Ato downstream genes identifying novel E-box motifs together with a significant enrichment of predicted direct targets. Although both GOF and LOF analysis yielded significant enrichment of E-boxes in misregulated genes, the significance was higher in the GOF analysis. This higher significance is likely because GOF of Ato results largely in the ectopic gain of one particular cell type, namely the R8 photoreceptor precursor, while the LOF condition results in the loss of all cell types and hence the downregulation of a larger set of genes across numerous cell types (Aerts, 2010).

In the third step several predicted Ato target enhancers were tested in vivo. This procedure identified 20 bona fide Atonal target enhancers out of 39 tested predictions, of which 17 are novel. This relatively high success rate almost certainly represents the lower limit of the true enhancer discovery rate because of false negative experimental results such as cases where the isolated enhancer is insufficient or requires its endogenous proximal promoter. Generally, demonstration of in vivo binding of the TF to a target enhancer that has been shown to be functional would be ideal. However, this is often not feasible, either due to lack of reagents or due to spatially and temporally sparse expression patterns of the TF in question. The data suggest that cisTargetX is a cheap, simple, fast, and high-confidence approach for CRM discovery for any TF (Aerts, 2010).

Finally, it is important to note that 11 of the 20 Ato target genes are known to act in sensory organ development or function, indicating that this approach identifies biologically relevant target genes and that the other nine genes are also players in this process (Aerts, 2010).

A significant portion of the Ato target genes encodes signaling molecules regulating most of the known key developmental pathways such as Notch, EGFR, Wnt, and JNK. Ato activates targets that modulate signaling pathways; thus far no evidence exists that Ato (or any other proneural TF) directly activates terminal differentiation genes. Even for molecules like Fas2, long thought to exclusively mediate adhesion during synaptic targeting, recent evidence reveals a role in regulating the precision of EGFR signaling during early photoreceptor specification. While the possibility cannot be excluded that such target genes were missed in this analysis because no approach can be certain of identifying all possible target genes, it is highly unlikely that a specific set of molecular functions would be selected against in an expression analysis approach. Therefore the idea is favored that the terminal differentiation genes are activated by other TFs, or by the TFs downstream of the Ato-regulated signaling pathways. It is noteworthy that the pathways regulated by Ato target genes, as well as many of the target genes themselves or their mammalian homologues, such as sens, dap, Traf4, and Mmp2 are implicated in cancer. It is suggested that Ato's functions in cancer is implemented via the regulation of some or all of the targets identified in this study (Aerts, 2010).

A remarkable finding is that none of the Ato target enhancers is active in a single sensory organ. Instead, Ato activates a unique combination of targets in each sensory organs it specifies. What kind of target genes can, in a combinatorial fashion, lead to differential morphological and functional development? On the basis of the analysis of the diversity of the beak sizes of Darwin's finches, it has been speculated that evolutionary changes in enhancers of signaling molecules have switch-like effects on a developmental gene regulatory network. The data suggest that variation of the proneural target set driven by changes in the cis-regulatory sequences of target genes shapes a unique regulatory state defined by a particular combination of signaling molecules. Interestingly, the Ato response elements within the regulatory sequences of target genes are evolutionarily conserved and their absence appears to alter the expression of these sequences. This observation leads to the hypothesis that a largely common genetic program induces different sensory organs, and that developmental and evolutionary variation of these organs occurs via subtle variations in the cis-regulatory sequences of signaling regulators. It is proposed that similar principles underlie diversification of most, if not all, developmental programs (Aerts, 2010).

The encouraging results for Atonal lead to the prediction of a large set of target genes for multiple TFs involved in retinal differentiation and they furthermore show that expression studies combined with computational predictions are a powerful tool of regulatory network discovery. The identification of Glass and Su(H) targets from wild-type eye versus wing comparisons of gene expression shows that genetic perturbations of TFs are not a prerequisite to find enriched direct targets in a set of candidate genes, at least for tissue specific TFs. Therefore, from wild-type comparative gene expression experiments meaningful results can be obtained (Aerts, 2010).

The cisTargetX analyses in this study compare the enrichment of predicted targets for single motifs (i.e., homotypic enhancer models) within sets of coexpressed genes. The most important advantage of homotypic clusters is that no a priori knowledge of cooperative factors is needed. An additional advantage is that theoretically the predictions can be more specific than 'free' heterotypic clusters in which binding sites for any combination of TFs is allowed (the 'OR' rule), and more sensitive than the 'constrained' class of heterotypic clusters in which all input TFs are required to have binding sites (the 'AND' rule). Tests with heterotypic enhancer models, consisting of motif combinations, generally showed lower enrichment than homotypic models. Genes that are activated in the same temporal and spatial patterns do not necessarily share the same cis-regulatory code, and the performance of genome-wide predictions may not necessarily benefit from heterotypic enhancer models, mainly because of sensitivity problems, at least in approaches similar to cisTargetX that are based on enrichment of direct targets in a candidate gene set. In other words, if many different combinatorial codes exist, then the presence of cofactor sites in only a few enhancers does not yield statistical over-representation and hence does not emerge from the noise. Moreover, coregulation might also occur through different enhancers of the same target genes and many potential examples of this were observed by predicting targets for multiple TFs independently. The important point is that whether coregulation occurs through shared or distinct enhancers, homotypic cluster predictions using cisTargetX, followed by comparisons of the targets between the TFs can discover these relationships (Aerts, 2010).

The putative early retinal differentiation network reconstructed from cisTargetX predictions shows waves of combinatorial regulation orchestrating spatial and temporal gene expression accuracy. Two feed-forward loops were found, namely Ey-Ato and Ato-Sens. These features are similar to the reconstructed regulatory networks underlying early embryonic processes. This finding indicates that exploiting motif predictions in conjunction with expression perturbations allows discovering similar regulatory networks as with ChIP-chip or ChIP-Seq approaches, where more material (e.g., large embryo collections) and specific reagents (e.g., high-quality antibodies) are required. Finally, these predictions represent a useful resource for future experiments aimed at dissecting the mechanistic basis of sensory specification (Aerts, 2010).

Atonal, senseless, and abdominal-A regulate rhomboid enhancer activity in abdominal sensory organ precursors

The atonal (ato) proneural gene specifies different numbers of sensory organ precursor (SOP) cells within distinct regions of the Drosophila embryo in an epidermal growth factor-dependent manner through the activation of the rhomboid (rho) protease. How ato activates rho, and why it does so in only a limited number of sensory cells remains unclear. A rho enhancer (RhoBAD) has been identified that is active within a subset of abdominal SOP cells to induce larval oenocytes and it has been shown that RhoBAD is regulated by an Abdominal-A (Abd-A) Hox complex and the Senseless (Sens) transcription factor (Li-Kroeger, 2008). This study shows that ato is also required for proper RhoBAD activity and oenocyte formation. Transgenic reporter assays reveal RhoBAD contains two conserved regions that drive SOP gene expression: RhoD mediates low levels of expression in both thoracic and abdominal SOP cells, whereas RhoA drives strong expression within abdominal SOP cells. Ato indirectly stimulates both elements and enhances RhoA reporter activity by interfering with the ability of the Sens repressor to bind DNA. As RhoA is also directly regulated by Abd-A, a model is proposed for how the Ato and Sens proneural factors are integrated with an abdominal Hox factor to regulate region-specific SOP gene expression (Witt, 2010).

This study found that the Atonal proneural factor is required for both normal rho enhancer function and the proper specification of abdominal oenocytes. In addition, it was determined that two distinct regions of the RhoBAD enhancer contribute to gene activity within the C1 SOP cells. The RhoA element preferentially drives gene expression within abdominal SOP cells, whereas RhoD drives weaker gene expression within the C1 SOP cells of both the thoracic and abdominal segments. Using a combination of genetic and biochemical analyses, it was found that the Ato, Sens, and Abd-A inputs contribute to proper rho enhancer activity. In particular, it was shown that RhoA, but not RhoD, is directly responsive to the Abd-A Hox factor. In addition, Ato was found to indirectly stimulate RhoBAD activity through both the RhoA and RhoD elements. Although it is currently not understood how Ato stimulates RhoD, it was found that Ato limits the DNA binding activity of the Sens repressor protein to RhoA. Coupled with other recent findings on proneural gene function, these results have two major implications: 1) A model is described for how Ato and Sens inputs are integrated to differentially regulate gene expression during SOP cell lineage development, and 2) How proneural input (Ato) and a Hox factor (Abd-A) cooperate to regulate Rho enhancer activity, at least in part, by limiting Sens-mediated repression is discussed (Witt, 2010).

Sens and the proneural factors are intricately linked during PNS development in Drosophila. Loss-of-function mutations in proneural genes disrupt sens expression resulting in a decrease in sensory organ formation and sens mutations result in decreased proneural gene expression and widespread sensory organ deficits. While both encode transcription factors required for PNS development, they have opposite effects on gene expression when bound to DNA. Proneural factors bind E-box DNA sequences with Daughterless to activate gene expression, whereas Sens binds a distinct DNA sequence to repress gene expression. However, recent data revealed that proneural proteins can convert Sens from a transcriptional repressor to a co-activator. Three different proneural factors (Ac, Sc, and Ato) interact with Sens in GST-pulldown and/or co-immunoprecipitation assays. In addition, cell culture assays showed that Sens stimulates the activation potential of proneural factors bound to E-Box sequences. Thus, Sens is a transcriptional repressor when directly bound to DNA through its zinc finger motifs whereas it is a potent co-activator when recruited to DNA by proneural proteins (Witt, 2010).

This study provides two pieces of information that add to understanding of how Sens and proneural factors regulate gene expression. First, purified Sens and Ato/Da proteins were used to show that Ato decreases the ability of Sens to bind the RhoA enhancer element. As RhoA contains a relatively low affinity Sens site, a parallel experiment was performed using a high affinity Sens site (SensS), and it was found that Ato does not significantly alter Sens binding to an optimized site. This data reveals that Ato's ability to interfere with Sens binding to DNA is site-specific and dependent upon binding affinity. How might Ato interfere with Sens binding to DNA? It has been shown that Ato, Ac, and Sc all directly interact with Sens through the second and third Sens zinc finger motifs. Since Sens requires these motifs to bind DNA, it is likely that the proneural factors compete with DNA for the same zinc fingers. Thus, the following model is proposed: if the binding affinity of Sens to DNA is high, Ato cannot interfere with Sens-mediated repression. However, if the binding affinity of Sens to DNA is low, Ato binds Sens and interferes with its ability to repress gene expression (Witt, 2010).

Secondly, expression analysis revealed that cells of the C1 SOP lineage differentially express Ato and Sens during their maturation. The initial SOP cell (SOPI) expresses both Ato and Sens during sensory organ specification. However, Ato protein is rapidly extinguished and no longer detectable once the SOP cell divides, whereas Sens persists into the SOPII cells. The rapid loss of Ato, even when it is expressed using a Gal4 driver, is consistent with recent findings that proneural proteins activate an E3 ubiquitin ligase pathway to trigger their own degradation. Thus, these findings suggest that the early SOP cell expresses both Ato and Sens and that Ato can alter Sens function in two ways: 1) by recruiting Sens to E-Box sequences as a co-activator, and 2) by interfering with Sens's ability to bind low affinity DNA sites (Witt, 2010).

It has been reported that rho is initially weakly expressed in C1 SOP cells in both the thorax and abdomen, and is only up-regulated in the abdominal SOP cells by the Abd-A Hox factor. This study found that the RhoBAD-lacZ reporter is also expressed in this pattern; it is proposed that Ato is part of an initiator pathway that allows rho expression in early C1 SOP cells. Ato does so in two ways: 1) by inhibiting Sens binding to RhoA through direct protein-protein interactions, and 2) by indirectly stimulating RhoD through an unknown mechanism. In total, these interactions result in the initiation of rho expression in early C1 SOP cells of both thoracic and abdominal segments. Ato's subsequent degradation releases Sens to bind RhoA and repress gene expression in thoracic SOP cells. Consistent with this idea, mutations that abolish Sens binding (SensM) result in de-repression of Rho reporters in the thorax. In the abdomen, however, an Abd-A complex out-competes Sens for RhoA to allow continued rho expression, subsequent EGF signaling, and the specification of additional cell types. Thus, Ato cooperates with the Abd-A Hox factor to stimulate EGF signaling by up-regulating rho expression via interfering with Sens-mediated repression (Witt, 2010).

While these findings provide insight into how rho is up-regulated in abdominal SOP cells, they uncover an interesting question: why is rho activated at all within thoracic SOP cells? Currently, there is no known function for rho activity within the thorax as rho mutant embryos show no phenotypic defect in cells surrounding the thoracic SOPs. As the lack of oenocyte production within the thorax is solely due to insufficient Spi secretion (oenocytes form in the thorax if rho is ectopically expressed), these data suggest that Rho levels are too low to trigger enough Spi secretion to affect neighboring cell fate. Consistent with this prediction is that the levels of an activated kinase downstream of EGF signaling (phospho-ERK) are very low in cells neighboring the thoracic C1 SOP cells compared to the abdominal SOP cells. So, why is rho activated within the thorax if it has no functional consequences? One interpretation is that Ato may provide competency for rho expression so that an additional positional factor such as Abd-A can fully stimulate rho and trigger Spi secretion and EGF signaling. In support of this idea, the widespread expression of Abd-A within the thorax activates RhoBAD-lacZ expression only within the C1 SOP cells and oenocytes form only in close proximity to these thoracic SOP cells. Thus, weak rho expression downstream of ato may provide a flexible and responsive system for activating Spi secretion in different body regions (Witt, 2010).

The gene regulatory cascade linking proneural specification with differentiation in Drosophila sensory neurons

In neurogenesis, neural cell fate specification is generally triggered by proneural transcription factors. While the role of proneural factors in fate specification is well studied, the link between neural specification and the cellular pathways that ultimately must be activated to construct specialised neurons is usually obscure. High-resolution temporal profiling of gene expression reveals the events downstream of atonal proneural gene function during the development of Drosophila chordotonal (mechanosensory) neurons. Among other findings, this reveals the onset of expression of genes required for construction of the ciliary dendrite, a key specialisation of mechanosensory neurons. It was determined that atonal activates this cellular differentiation pathway in several ways. Firstly, atonal directly regulates Rfx, a well-known highly conserved ciliogenesis transcriptional regulator. Unexpectedly, differences in Rfx regulation by proneural factors may underlie variations in ciliary dendrite specialisation in different sensory neuronal lineages. In contrast, fd3F encodes a novel forkhead family transcription factor that is exclusively expressed in differentiating chordotonal neurons. fd3F regulates genes required for specialized aspects of chordotonal dendrite physiology. In addition to these intermediate transcriptional regulators, it was shown that atonal directly regulates a novel gene, dilatory, that is directly associated with ciliogenesis during neuronal differentiation. This analysis demonstrates how early cell fate specification factors can regulate structural and physiological differentiation of neuronal cell types. It also suggests a model for how subtype differentiation in different neuronal lineages may be regulated by different proneural factors. In addition, it provides a paradigm for how transcriptional regulation may modulate the ciliogenesis pathway to give rise to structurally and functionally specialised ciliary dendrites (Cachero, 2011).

Numerous genetic and misexpression analyses in a range of organisms have shown that proneural factors influence a neuron's ultimate phenotype (including its subtype identity) at an early stage in its development. However, the nature of this influence on the cell biological processes of neuronal differentiation has remained obscure. This study bridges the gap between early specification by the proneural factor, ato, and the differentiation of Ch neurons. The current model in both Drosophila and vertebrates is that proneural factors activate two types of target gene during neural precursor specification: a common target set for shared neuronal properties and a unique target set for subtype-specific properties. The data suggest that such neuronal subtype differences are ultimately controlled by proneural factors in several ways: by the differential regulation of both specific and common intermediate transcription factors, which in turn regulate genes for aspects of neuronal structural and functional differentiation, and by direct regulation of potential differentiation genes (Cachero, 2011).

The proneural factors ato and sc commit cells to similar but distinct neural precursor fates: Ch and ES neurons are evolutionarily related cell types with similar but distinct structural and physiological properties. Notably, both are characterised by the possession of specialised ciliary-based dendrites. Thus, ciliogenesis is a key pathway that must ultimately be activated in sensory neurons subsequent to proneural factor function. However, there are important differences between the dendrites of Ch and ES neurons. Ch dendrites have a more prototypically organised axonemal structure and possess a characteristic ciliary dilation - a specialisation that separates the Ch ciliary dendrite into functionally distinct zones. Moreover, there is evidence for an active 'beat' of Ch cilia during sensory transduction. In general, ES dendrites appear reduced in structure: although a basal body and short axoneme are present, the tip of the dendrite consists of a 'tubular body' of irregularly packed microtubules. Thus the basic ciliogenesis pathway must be modulated differently in Ch and ES differentiation, and ultimately this must reflect a difference in function between ato and sc proneural factors. The ciliogenic regulator Rfx is expressed and required for both ES and Ch lineages, but it is more strongly and more persistently expressed in Ch lineages (the Ch-enriched pattern). This modulation of Rfx expression is at least partly due to differences in its regulation by proneural factors, since it appears to be a direct target of ato but not sc. This study hypothesises that differences in Rfx regulation by the proneural factors lead to differences in implementation of a core cilia biogenesis program, thereby directly linking early proneural factor function with key differences of neuronal morphology. Consistent with this idea, the data show that several known or predicted ciliogenesis genes also exhibit this Ch-enriched pattern, and some of these are predicted or known Rfx targets (Cachero, 2011).

In this view, the subtype differences between Ch and ES neurons are partly produced by quantitative differences in timing or level of expression of a common differentiation process, which ultimately depends on a qualitative difference in Rfx regulation by the proneural factors. A possible example of this is CG6129. This gene is a predicted Rfx target gene and is expressed in a Ch-enriched pattern (Laurenon, 2007). The homologous mouse protein (Rootletin) localises to the ciliary rootlet and is required for its formation. Thus Ch-enriched expression of CG6129 explains the presence of the ciliary rootlet in Ch neurons but not ES neurons. One prediction of this hypothesis is that overexpression of Rfx in ES neurons will upregulate Ch-enriched genes, and this is borne out by preliminary experiments that show an increase in CG6129 expression in ES neurons upon Rfx overexpression. It is notable that differences in IFT activity are proposed to underlie differences in ciliary morphology while RFX class factors have been associated with regulating genes for IFT in a variety of organisms. This work suggests that variations in Rfx expression level and timing should be explored as a possible factor in cilium diversity (Cachero, 2011). fd3F fits the more conventional view of a proneural target gene that implements a subtype-specific program of differentiation. It is expressed downstream of ato uniquely in Ch neurons and regulates genes required for functional specialisation of the Ch ciliary dendrite. It is likely that Forkhead factors regulate specialisation of ciliogenesis in other organisms. In C. elegans, FKH-2 is expressed widely early in development but is also required specifically for ciliary specialisation of one type of sensory neuron. Foxj1 in mice, Xenopus, and zebrafish appears to be required for the motile cilia of the lung airway and embryonic node, but not for primary cilia. It remains to be determined whether fd3F regulates the machinery for the active beat that occurs in Ch dendrites as part of sensory transduction. Together, these studies of Rfx and fd3F extend the previously limited knowledge of the gene regulatory network underlying ciliogenesis and provide insight into how the core program may be modified to produce the highly specialised and diverse morphologies that cilia adopt for different functions (Cachero, 2011).

Previous to this study, little was known about how ato/sc proneural genes control the acquisition of Ch/ES subtype identity, except that regulation of the Cut homeodomain transcription factor is involved. Mutant and misexpression analyses show that cut is a fate selector switch for ES identity downstream of sc, but nothing is known of its mode of action or targets. Whereas Rfx and fd3F functions are likely to be confined to neuronal morphology, cut affects the identity of support cells too. As a fate switch in the entire lineage, it appears likely that cut is involved in high-level fate specification (like proneural genes) rather than regulating aspects of differentiation directly. However, it is also possible that cut may repress ciliogenesis genes in ES neurons, either directly or by repressing Rfx expression. It will be important to integrate cut into the Ch/ES gene regulatory network in the future (Cachero, 2011).

In the temporal expression profiling data, there is a steady increase in the number of known or suspected differentiation genes expressed in developing Ch cells. Many more are not expressed until after the analysis ends. Ciliogenesis is a highly intricate cellular process requiring the coordination of perhaps hundreds of genes and differences in expression onset may indicate prerequisite steps in the process of differentiation and ciliogenesis. A surprising observation was the significant number of ciliogenesis and differentiation genes that are expressed even at the earliest profiling time point. This is unexpected, since the earliest time point is predicted to be not only before differentiation but also even before cell divisions have generated the neurons. It is suggested that further analysis of expression timing may lead to insights into the cell biology of ciliogenesis. The early activation of differentiation genes may reflect the rapid pace of development in the Drosophila embryo. Thus, early expression of ciliogenesis genes may provide components that prime cells for rapid cilium assembly later once differentiation has been triggered. Along these lines, the findings mirror striking observations of retinal ganglion cells, whose rapid differentiation within 15 minutes of the exit from mitosis has been taken to imply that genes required in postmitotic cells must be transcribed before cell division. A more intriguing possibility is that early expression reflects an orderly time course for ciliogenesis that begins many hours before the final cell division. For example, unc is thought to be required for the conversion of the mitotic centriole to ciliogenic basal body, but this stuyd found that the mRNA and fusion protein are expressed even in SOPs, several cell divisions before terminal differentiation. Interestingly, in mammals newly replicated centrioles mature over two cell cycles. It is conceivable that the sensory neuron basal body might similarly need time to mature (Cachero, 2011).

Since Rfx and some ciliogenesis genes are expressed in SOPs, what prevents ciliogenesis from being activated in the non-neuronal support cells? One possibility would be an extension of model recently proposed for the generation of support cell differences, in which Notch signalling between daughter cells confines the function of genes to one branch of the lineage. This would predict that ciliogenesis genes and/or Rfx are Notch target genes. Another possibility is that some of the gene products are asymmetrically segregated. Thirdly, ciliogenesis may not be triggered until one or more key gene products are produced in the neuronal cell (Cachero, 2011).

As a corollary, it will be important to explore further the gene regulatory network underlying the temporal and cell-type differences in ciliogenesis genes. Some early expressed differentiation genes are known or predicted Rfx targets (Laurenon, 2007). This gives a rationale for the early regulation of Rfx by ato in Ch lineages. However, in both C. elegans and D. melanogaster, Rfx regulates only a subset of ciliogenesis genes (notably, it does not regulate IFT-A genes). Further studies on ato target genes and the ciliogenesis regulatory network in sensory neurons will identify other important regulators. It remains to be determined how many differentiation genes are, like dila, direct targets of ato. Interestingly, vertebrate proneural factors are hypothesised to regulate directly the transition from cycling neural progenitor (or neural stem cell) to postmitotic differentiating neuron. Perhaps ato has retained some part of an ancestral proneural factor function in direct regulation of terminal differentiation despite the subsequent evolution of SOPs that must undergo several divisions before differentiating (Cachero, 2011).

Protein Interactions

Like the achaete-scute complex genes, Atonal heterodimerizes with Daughterless protein to bind to E boxes (Jarman, 1993).

Proneural and abdominal Hox inputs synergize to promote sensory organ formation in the Drosophila abdomen

The atonal (ato) proneural gene specifies a stereotypic number of sensory organ precursors (SOP) within each body segment of the Drosophila ectoderm. Surprisingly, the broad expression of Ato within the ectoderm results in only a modest increase in SOP formation, suggesting many cells are incompetent to become SOPs. This study shows that the SOP promoting activity of Ato can be greatly enhanced by three factors: the Senseless (Sens) zinc finger protein, the Abdominal-A (Abd-A) Hox factor, and the epidermal growth factor (EGF) pathway. First, it was shown that expression of either Ato alone or with Sens induces twice as many SOPs in the abdomen as in the thorax, and does so at the expense of an abdomen-specific cell fate: the larval oenocytes. Second, Ato was shown to stimulate abdominal SOP formation by synergizing with Abd-A to promote EGF ligand (Spitz) secretion and secondary SOP recruitment. However, it was also found that Ato and Sens selectively enhance abdominal SOP development in a Spitz-independent manner, suggesting additional genetic interactions between this proneural pathway and Abd-A. Altogether, these experiments reveal that genetic interactions between EGF-signaling, Abd-A, and Sens enhance the SOP-promoting activity of Ato to stimulate region-specific neurogenesis in the Drosophila abdomen (Gutzwiller, 2010).

How proneural pathways that specify sensory precursor cells throughout the body are integrated with region-specific patterning genes to yield the correct type and number of sensory organs is not well understood. This study shows that three factors enhance the ability of Ato to promote ch organ SOP cell fate in the Drosophila abdomen; the EGF pathway mediated by the Spi ligand, the Abd-A Hox factor, and the Sens zinc finger transcription factor (Gutzwiller, 2010).

EGF signaling is used reiteratively throughout development to specify the formation of distinct cell types along the body plan. In the embryonic Drosophila abdomen, EGF signaling initiated by the activation of rhomboid (rho) in a set of ch organ SOP cells induces the formation of both a cluster of abdomen-specific oenocytes as well as a set of 2° ch organ SOP cells. But how does the EGF-receiving cell know whether to become a larval oenocyte that is specialized to process lipids or a ch organ SOP cell that forms part of the peripheral nervous system? Previous studies have shown that oenocyte specification requires at least two inputs: (1) the reception of relatively high levels of EGF signaling and (2) the expression of the Spalt transcription factors. Hence, oenocytes develop in close proximity to the abdominal C1 SOP cells that lie within a Spalt expression domain and express high levels of rho. In contrast, 2° SOP cells require less EGF signaling and form if the receiving cells lack Spalt. Consistent with this model, genetic studies have shown that oenocytes fail to develop and one to two additional ch organ SOP cells are specified in Spalt mutant embryos, whereas ectopic Spalt expression in the ventral ectoderm inhibits the recruitment of 2° SOP cells. Thus, Spalt promotes oenocyte development and antagonizes 2° ch organ specification in the Drosophila embryo (Gutzwiller, 2010).

Evidence that ato has the opposite effect as Spalt: it promotes ch organ SOP cells at the expense of oenocyte specification. Witt (2010) showed that ato loss-of-function results in decreased expression of activity of the rho enhancer, RhoBAD (Witt, 2010), in C1 SOP cells and induces fewer oenocytes. These data are consistent with EGF signaling being compromised in ato mutant embryos and oenocyte specification being dependent upon the reception of high levels of Spi. This study shows that Ato gain-of-function stimulates RhoBAD expression yet results in the inhibition of oenocyte formation. Importantly, the loss of oenocytes is not due to decreased EGF signaling as similar whorls of phospho-ERK-positive cells and even extra phospho-ERK staining are observed in Ato-expressing segments compared with non-expressing segments. In addition, no difference was detected in cell death between Ato-expressing and non-Ato-expressing segments (using an anti-cleaved Caspase3 marker), indicating the oenocyte loss is not due to apoptosis. Instead, Ato promotes the formation of additional ch organ SOP cells in abdominal segments that normally form oenocytes. Moreover, while the broad activation of EGF signaling (PrdG4;UAS-Rho) induces many extra oenocytes and a few scolopodia, the co-expression of Ato and Rho induces many scolopodia and few oenocytes. These data suggest that if the Spi-receiving cell expresses high Ato relative to Salm then ch organ development occurs whereas if the Spi-receiving cell expresses high Salm relative to Ato then oenocytes are formed. Thus, Ato plays a role in both the Spi-secreting (induction of rho expression) and Spi-receiving cell to dictate the choice of cell fate (Gutzwiller, 2010).

The broad expression of Ato within the ectoderm revealed differences in sensory organ competency between the thorax and abdomen. In particular, it was found that Ato induced approximately twice as many ch organ SOP cells in the abdomen as in the thorax. Moreover, the co-expression of Ato with the Abd-A Hox factor induced significantly more ch organ cell formation than expression of either factor alone (none by Abd-A, four by Ato, and eight by Ato/Abd-A). These data suggest that Ato and Abd-A synergize to enhance ch organ SOP formation in the abdomen, an prompted an examination whethere these SOP cells are predominantly 1° or 2° cells. This problem was first addressed by first showing that the co-expression of Ato and Abd-A stimulates Rho enhancer activity (RhoAAA) within additional cells and results in enhanced phospho-ERK staining. Second, it was shown that Ato and Abd-A require the EGF pathway to enhance ch organ development as co-expression of both factors in a spi mutant embryo failed to promote more ch organs than expression of Ato alone. These data indicate that the co-expression of Ato and Abd-A enhances the ability of 1° ch organ SOP cells to activate rho, stimulates Spi secretion and, since the receiving cell expresses Ato, 2° SOPs form instead of oenocytes. The net result is that Ato and Abd-A synergize to activate the EGF pathway to promote region-specific neurogenesis within the Drosophila abdomen (Gutzwiller, 2010).

The Sens transcription factor is essential for the formation of much of the peripheral nervous system in Drosophila and previous studies revealed that Sens can stimulate the sensory bristle-forming activity of the Scute and Achaete proneural factors in the wing disc. Similarly, it was found that Sens stimulates the ability of Ato to generate internal stretch receptors in the embryo and that Ato and Sens promote more sensory organ development in the abdomen than in the thorax. In addition, while the overall number of ch organs formed by Ato and Sens co-expression is decreased in spi mutant embryos, significantly more ch organ SOP cells in the abdomen than in the thorax are observed in this EGF-compromised genetic background. Thus, Ato and Sens can stimulate abdominal ch organ SOP cell development in the presence or absence of Spi-mediated cell signaling (Gutzwiller, 2010).

So, what is the relationship between Ato, Sens, and Abd-A in regulating both EGF signaling and region-specific sensory organ formation? It was previously found that Ato, Sens, and Abd-A control EGF signaling through the regulation of a cis-regulatory element within the rhomboid (rho) locus (RhoBAD) (Li-Kroeger, 2008; Witt, 2010). RhoBAD acts in abdominal C1 SOP cells to induce oenocyte formation, and Ato and Abd-A both stimulate RhoBAD expression, at least in part, by limiting the ability of Sens to repress RhoBAD activity. Moreover, they do so using different mechanisms. An Abd-A Hox complex containing Extradenticle and Homothorax directly competes with the Sens repressor for overlapping binding sites in RhoBAD (Li-Kroeger, 2008). In contrast, Ato does not directly bind RhoBAD but does directly interact with Sens to limit its ability to bind and repress Rho enhancer activity (Witt, 2010). Consequently, SOPs that co-express Ato and Abd-A are likely to limit the ability of Sens to repress Rho and thereby increase the number of ch organ SOP cells that secrete Spi. Consistent with this prediction, the co-expression of Ato and Sens preferentially stimulates Rho enhancer activity within abdominal segments compared to thoracic segments. Each SOP cell that expresses rho would further enhance sensory organ development through the recruitment of 2° SOP cells via Spi-mediated signaling. Hence, the genetic removal of spi results in a significant decrease in the number of ch organ SOP cells that develop in response to Ato and Sens. Thus, the ato-sens genetic pathway, which is used throughout the body to promote SOP formation, interacts with an abdominal Hox factor to stimulate EGF signaling and promote additional cell fate specification in the abdomen (Gutzwiller, 2010).

While the above model fits well with most of the data, two unexpected findings were observed when comparing the ability of Ato-Sens co-expression to induce ch organ development in the presence and absence of spi function: First, it was predicted that Ato-Sens co-expression in the thoracic regions, which lack Abd-A, should predominantly induce the formation of 1° ch organ SOP cells that do not require EGF signaling for their development. However, it was found that significantly fewer ch organs form in the thorax of spi mutants, indicating that EGF signaling can enhance 2° sensory organ formation within thoracic segments that co-express Ato and Sens. Interestingly, previous studies have shown that both rho and the Rho enhancers are weakly active within thoracic C1 SOP cells, but their levels do not reach a high enough threshold to induce oenocyte formation. However, it is possible that the co-expression of Ato and Sens sufficiently sensitizes the receiving cells to respond to low levels of EGF signaling and become ch organ SOP cells. The second unanticipated finding is that Ato and Sens co-expression still induced significantly more ch organ development within the abdomen (5-6 extra SOP cells) relative to the thorax (1-2 extra SOP cells) in the absence of Spi-mediated signaling. This finding suggests that Ato and Sens can genetically interact with the Abd-A Hox factor to promote sensory organ development in an Spi-independent manner. Currently, it is not understood how Abd-A enhances the proneural activity of the Ato-Sens factors in the absence of Spi signaling. One possibility is that Abd-A and Ato use similar mechanisms to limit Sens-mediated repression of additional target genes besides rho to stimulate ch organ development. Alternatively, Abd-A could independently regulate other factors such as those involved in the Notch-Delta pathway to enhance the competency of the ectoderm to respond to the Ato-Sens pathway. Intriguingly, a Hox factor (lin-39) in C. elegans has been shown to directly regulate Notch signaling during vulval development, and the vertebrate Hoxb1 factor regulates neural stem cell progenitor proliferation and maintenance by modulating Notch signaling. Since differential Notch-Delta signaling is a key pathway in deciding neural versus non-neural cell fates, the ability of Hox factors to modify this pathway could result in segmental differences in neurogenesis (Gutzwiller, 2010).

The Interactive Fly resides on the
Society for Developmental Biology's Web server.