Research Article

Nuclear Receptor Unfulfilled Regulates Axonal Guidance and Cell Identity of Drosophila Mushroom Body Neurons

  • Suewei Lin,

    Affiliation: Department of Neurobiology, University of Massachusetts Medical School, Worcester, Massachusetts, United States of America

  • Yaling Huang,

    Affiliation: Janelia Farm Research Campus, Howard Hughes Medical Institute, Ashburn, Virginia, United States of America

  • Tzumin Lee mail

    Affiliations: Department of Neurobiology, University of Massachusetts Medical School, Worcester, Massachusetts, United States of America, Janelia Farm Research Campus, Howard Hughes Medical Institute, Ashburn, Virginia, United States of America

  • Published: December 22, 2009
  • DOI: 10.1371/journal.pone.0008392


Nuclear receptors (NRs) comprise a family of ligand-regulated transcription factors that control diverse critical biological processes including various aspects of brain development. Eighteen NR genes exist in the Drosophila genome. To explore their roles in brain development, we knocked down individual NRs through the development of the mushroom bodies (MBs) by targeted RNAi. Besides recapitulating the known MB phenotypes for three NRs, we found that unfulfilled (unf), an ortholog of human photoreceptor specific nuclear receptor (PNR), regulates axonal morphogenesis and neuronal subtype identity. The adult MBs develop through remodeling of γ neurons plus de-novo elaboration of both α′/β′ and α/β neurons. Notably, unf is largely dispensable for the initial elaboration of γ neurons, but plays an essential role in their re-extension of axons after pruning during early metamorphosis. The subsequently derived MB neuron types also require unf for extension of axons beyond the terminus of the pruned bundle. Tracing single axons revealed misrouting rather than simple truncation. Further, silencing unf in single-cell clones elicited misguidance of axons in otherwise unperturbed MBs. Such axon guidance defects may occur as MB neurons partially lose their subtype identity, as evidenced by suppression of various MB subtype markers in unf knockdown MBs. In sum, unf governs axonal morphogenesis of multiple MB neuron types, possibly through regulating neuronal subtype identity.


The brain consists of neurons that are wired in specific patterns, and establishing a complex brain involves multiple tightly regulated developmental processes. In Drosophila, it starts with birth of neuroblasts (Nbs) with specific fates that are largely acquired through spatial patterning [1]. The Nbs then proliferate to produce multiple neuron types often in an invariant sequence [2], [3]. Post-mitotic neurons subsequently undergo extensive morphogenesis and some neurons remodel to form the circuitry [4], [5]. Although these basic processes are known, the detailed mechanisms that govern each step of brain development remain only partially resolved. Identifying more genes required for the various aspects of brain development is essential for elucidating further how the complex brain develops.

Nuclear receptors (NRs) are ligand-regulated transcription factors that play important roles in key metabolic and developmental pathways, including lipid and glucose homeostasis, aging, and cell fate determination [6], [7], [8], [9], [10], [11], [12], [13], [14],[15],[16]. NRs have also been shown to govern diverse aspects of neural development, such as the maintenance of neuronal precursors [17], [18], [19], neuronal cell death [20], axon guidance [21], [22], [23] and neuronal remodeling [24], [25]. Furthermore, mutations in NRs have been implicated in several neurodegenerative diseases [11], [15], [26], [27], [28], [29], [30], [31], [32]. Determining the functions of NRs will provide ample opportunities for better understanding brain development and neuron degeneration.

NR genes are highly conserved across the animal kingdom. The 48 NR genes identified in mammals can be categorized into six subfamilies according to their protein structure similarity [33]. The neural functions of most NRs have not been explored at all. Compared to the 48 NR genes in mammals, the Drosophila genome contains only 18 NR genes, though all six NR subfamilies are represented [33]. The smaller number of NR genes makes it easy to survey NR functions in brain development. Four fly NRs have been shown to regulate various aspects of neural development. Seven-up (svp), which encodes a homolog of human COUP-TF orphan NR, represses the expression of a temporal identity gene hunchback (hb) in embryonic neuroblasts after the first mitosis to ensure the subsequently produced neurons acquire proper cell fates [8]. Moreover, in the late larval stage, a burst expression of svp is required for neuroblasts to exist the cell cycle or undergo apoptosis [34]. Tailless (tll), the Drosophila homolog of human Tlx orphan NR, is robustly expressed in the larval brain in certain neuroblasts and ganglion mother cells (GMCs) to promote cell cycling and prevent apoptosis [17]. Ecdysone receptor (EcR) and its heterodimeric partner ultraspiracle (usp) work together to regulate axonal and dendritic remodeling during metamorphosis [24], [25], [35], [36], [37]. EcR have also been shown to regulate programmed neuronal cell death [20], [38] and secondary arbors of adult-specific neurons [39]. However, the functions of the remaining 14 NRs in neural development have not been studied.

The Drosophila mushroom body (MB), the olfactory learning/memory center, has been shown as an excellent model to study gene functions in neural development (e.g. [24], [40], [41], [42]), and thus could also be good for studying NR functions. The adult MB consists of three major types of neurons- γ, α′/β′ and α/β [43]. These neuron types are specified according to their date of birth by temporal identity factors, such as chinmo [44]. The γ neurons are born before the mid-3rd instar larval stage [around 4 days after larval hatching (ALH)]; their axons initially form bifurcated branches in the larval brain. Later, these axons are pruned through fragmentation and glia engulfment at early pupal stage, followed by the re-extension of adult-specific axons that only elaborate horizontally toward the midline of the adult brain [43], [45]. The α′/β′ neurons are born between the mid-3rd instar and puparium formation, and α/β neurons are born at the pupal stage. Both α/β and α′/β′ axons form bifurcated branches without remodeling during development, and thus may only be functional in the adult brain [43]. As to roles of NRs in MB development, EcR, USP and TLL have been shown playing important roles in regulating MB remodeling and neurogenesis, respectively (see below) [17], [24].

In this paper, we systematically silenced each of the 18 Drosophila NR genes in MBs using miRNA-based RNA interference [46], [47]. In addition to EcR, usp, and tll, we isolated unfulfilled (unf), the fly NR2E3 subfamily ortholog of C. elegans fax-1 and human photoreceptor-specific nuclear receptor (PNR) [48], [49]. unf has a novel function in MB development, and unf knockdown caused a severe MB lobe extension defect. Further analysis revealed that unf was required for proper axonal guidance of all three major types of the MB neurons. Without unf, the MB axons were misguided and wandered around at the end of peduncle. Interestingly, unf was mainly required for adult-specific axonogenesis. In later-born adult-specific α′/β′ and α/β neurons, unf played a critical role in initial neurite extension; in contrast, for early-born γ neurons, unf was largely dispensable to the establishment of larval projections, but absolutely required for their re-elaboration of axons during early metamorphosis.

We also found that the expression of Trio, a guanine nucleotide exchange factor important for the MB axonal guidance [50], as well as several subtype-specific markers were significantly reduced in the unf knockdown adult MB. This result implies that unf is required for MB neurons to acquire correct gene expression profile (i.e. cell identity) and that improper expression of multiple guidance molecules resulting from the loss of neuron identities might underlie the axonal defects induced by unf knockdown.


Silencing Individual NRs by miRNA Unveils a Role of Nuclear Receptor unf in MB Development

To study NR functions in fly brain development, we generated UAS-miRNA transgenic fly lines against each of the 18 nuclear receptor (NR) genes so far identified in the fly genome. We drove the miRNA transgenes using GAL4-OK107 [51] to knock down individual NRs through development of the MBs. We detected MB abnormalities when the miRNAs against ecdysone receptor (EcR), ultraspiracle (usp), tailless (tll), ftz transcription factor 1 (ftz-f1) or unfulfilled (unf) were induced (Table S1).

EcR and its heterodimeric partner usp have been shown to regulate the axonal pruning of MB γ neurons [24], and tll was recently demonstrated to promote the efficient proliferation of the MB neuroblasts and ganglion mother cells [17]. Our miRNA screening results were consistent with the previous studies. Knocking down EcR or usp blocked pruning of MB γ neurons, which maintained larval-type projections in the adult brain (Fig. 1A–C). In contrast, silencing tll affected MB proliferation, resulting in a tiny MB purely consisting of early-type MB neurons (Fig. 1D,E). This recapitulates the tll mutant MB phenotype [17]. The consistency between the miRNA results and published data suggests that our miRNA approach worked efficiently in knocking down the endogenous nuclear receptors.


Figure 1. unf is required for the proper axonogenesis of the mushroom body neurons.

(A–C) Anti-FasII antibody staining (magenta) revealed the γ and α/β lobe morphology of wild-type (A), EcR- (B) and usp- (C) knockdown adult MBs. The knockdown was induced by GAL4-OK107 dependent induction of miRNAs. Note the EcR- and usp-knockdown MBs contained unpruned γ processes (arrows in [B] and [C]) and lacked a normal-looking γ lobe (e.g. arrowhead in [A]). Scale bar: 20 µm. (D–F) Adult wild-type (D), tll- (E) and unf- (F) knockdown MB, labeled by GAL4-OK107 (green). The knockdown was induced by GAL4-OK107 dependent induction of miRNAs. Note the tll-knockdown MB contained only few neurons (E), and the axon lobes of unf-knockdown MB failed to extend along the correct paths (F).


In addition to the nuclear receptors known to be required for MB development, we found that silencing ftz-f1 and unf caused abnormal MB morphologies. In the MB, distinct sets of MB axon bundles show different levels of FasII expression [52]; but silencing ftz-f1 elicited ectopic FasII-positive bundles in about 30% of MBs (Table S1; data not shown). Ectopic FasII bundles could result from axon guidance defects, misregulation of cell fate, or incomplete γ axon pruning. Since the phenotype was subtle and the penetrance was low, we did not pursue analysis of ftz-f1 in this study. In contrast, silencing unf caused a much stronger MB phenotype (Table S1; Fig. 1F); the five MB-characteristic axon lobes were missing in all unf knockdown MBs (Fig. 1F). While most axons might extend across the brain through the peduncle, they failed to form the lobes that normally project on the anterior surface of the brain hemisphere. Instead, the stereotyped MB lobes were replaced with a ball-like structure around the end of peduncle (Fig. 1F). The MBs were otherwise grossly normal, consisting of many cell bodies residing over the dendrite-based calyx that sometimes appears slightly larger than normal (data not shown).

The unf miRNA-a used in the initial screening targets two sites on the unf coding sequence, one at 289–310 nt and the other at 1130–1151 nt (Fig. 2A). To ascribe the above phenotype to loss of unf, we first learned that the induction of this miRNA by GAL4-OK107 did effectively deplete unf protein in MB neurons (see below). We then tried to rule out off-target effects by generating unf miRNA-b to target an independent site at 1001–1022 nt (Fig. 2A). GAL4-OK107-dependent induction of unf miRNA-b also led to a similar MB lobe extension defect, though the phenotype was weaker with some axons fully extended (Fig. 2B). To confirm that these phenotypes are indeed due to unf knockdown, we induced unf miRNA-b in flies heterozygous for a deficiency [Df(2R)ED2426] that covers the unf locus, and found that the MB axonal lobe defect became as severe as that caused by unf miRNA-a (Fig. 2C,D). Notably, when the unf miRNA-a was induced in the deficiency heterozygous background, the lobe phenotype was not enhanced (data not shown), suggesting that the abnormality caused by unf miRNA-a was a very strong unf loss-of-function phenotype. Taken together, these results show that unf is essential for the extension of the MB axonal lobes.


Figure 2. The MB lobe phenotype was caused by specific knockdown of unf by miRNA.

(A) Illustration of the unf cDNA that contains a 5′ region encoding a putative DNA binding domain (DBD) and a 3′ region encoding a putative ligand binding domain (LBD). *The miRNA target sites for unf miRNA-a. #The miRNA target site for unf miRNA-b. (B–C) unf in MB was knocked down by the GAL4-OK107 induced unf miRNA-b in a wild-type fly (B), or a fly heterozygous for a deficiency Df(2R)ED2426 that covers unf gene (C). Note some MB axons can still project to their correct positions when the unf miRNA-b was induced in the wild-type background (indicated by arrowheads in [B]). Scale bar: 20 µm. (D) Statistic results of the percentage of MB lacking dorsal projection in flies heterozygous for Df(2R)ED2426 (Df), or in flies with OK107-dependent induction of unf miRNA-a(unfi-a), unf miRNA-b (unfi-b), or unf miRNA-b plus one allele of Df(2R)ED2426 (unfi-b/Df).


unf Is Mainly Required for the Formation of Adult-Specific MB Lobes

To determine the pathological mechanisms underlying the lobe defects, we followed unf knockdown MBs through development. We found that most unf knockdown MBs (88%, n = 75) were grossly normal at mid-3rd instar larval stage when the MB primarily consists of γ neurons (Fig. 3A,B), although 12% of unf knockdown MBs had thinner dorsal lobes (n = 75; inset in Fig. 3B). This is in great contrast to the 100% lobe extension defect of unf knockdown adult MBs (e.g. Fig. 1F). This result was not due to the insufficient RNAi knockdown in the early larval stage, because GAL4-OK107 induced unf miRNA can knockdown UNF protein completely within 6 hr ALH (See below). The observations suggest that unf is mainly involved in the formation of the adult MBs.


Figure 3. unf is mainly required for the formation of adult-specific MB lobes.

Wild-type (A, C–F) and unf knockdown (B, G–J) MB at the indicated developmental stages were labeled by GAL4-OK107 (A–J). The insect in [B] is an example of a small portion (12%; n = 75) of the unf -knockdown larval MB that displayed a thinner dorsal lobe phenotype. Brains in [E], [F], [I] and [J] were counterstained with anti-FasII antibody (magenta). unf was knocked down by GAL4-OK107 induced unf miRNA-a. Scale bars: 20 µm.


The adult MBs develop through remodeling of γ neurons plus de-novo elaboration of both α′/β′ and α/β neurons. These dynamic changes in the MB structure can be closely examined during early pupal development. In wild-type MBs, the bifurcated larval γ lobes were completely pruned around 18 hr APF, and at the same time the α′/β′ lobes that were initially wrapped by the larval γ lobes can be clearly seen (Fig. 3C,D). Around 24 hr APF when the adult-specific γ lobes started to extend, the nascent α/β axonal bundles derived from the pupal-born α/β neurons can be detected with anti-FasII mAb 1D4 (Fig. 3E). By 48 hr APF, the adult-specific γ lobe is fully extended and the α/β bundles become much thicker (Fig. 3F).

Examination of unf knockdown MBs through early pupal development revealed multiple abnormalities. First, mutant MBs at 6 hr APF displayed no obvious morphological defect (Fig. 3G). By contrast, at 18 hr APF when the larval γ lobes were largely pruned, there did not exist α′/β′ lobes (Fig. 3H). Second, no axon has extended beyond the peduncle end at 24 hr APF, indicating defects in both γ axon re-extension and α/β bundle formation (Fig. 3I). Third, by 48 hr APF, instead of seeing all five MB lobes, we detected a ball-like structure bulging around the terminus of the peduncle (Fig. 3J). It appears that MB axons were lost at the peduncle end. They elaborated locally, but failed to project into discrete domains based on their subtype identity. These observations reveal that 1) unf is largely dispensable for the initial larval-specific axonal morphogenesis of the γ neurons, but absolutely essential for the extension of the adult-specific γ axons during remodeling; 2) unf is necessary for the de-novo formation of α′/β′ and α/β lobes.

unf Acts in Mature γ Neurons to Govern Axon Re-Extension, while Supporting Initial Axonal Morphogenesis in Later Types of MB Neurons

Given its pleiotropic functions in MB morphogenesis, we sought to determine the role of unf in γ neuron remodeling independent of its effects on other MB neuronal morphogenetic processes. We selectively knocked down unf in mature larval γ neurons using GAL4-201Y [53], resulting in missing adult γ lobes despite presence of other MB lobes (Fig. 4A,B). Phenotypic analysis through development revealed normal pruning of the larval-specific γ lobes (Fig. 4C–F) but no formation of the adult γ lobe (Fig. 4B). This result suggests that unf acts in mature γ neurons to promote axon re-extension during MB remodeling. In addition, lacking the γ lobe did not affect formation of MB α/β lobes (Fig. 4B).


Figure 4. unf acts in mature γ neurons to govern axon re-extension.

Wild-type (A,C,E,G) and unf knockdown (B,D,F,H) MB at the indicated developmental stages, labeled by GAL4-201Y (A–F) or GAL4-MB247 (G,H). Brains in [A], [B], [G] and [H] were counterstained with anti-FasII antibody (magenta). unf was silenced by GAL4-201Y (B,D,F) or GAL4-MB247 (H) induced unf miRNA-a. The arrows indicate the position of normal (A) and truncated γ lobes (B,H). Scale bars: 20 µm.


Notably, the selective loss of the adult γ lobe was evident even when unf was silenced in most, if not all, post-mitotic MB neurons via induction of unf miRNA-a using GAL4-MB247 (Fig. 4G,H). GAL4-MB247 drives UAS-transgene expression in MB neurons that have undergone extensive morphogenesis [44]. This again indicates the requirement for unf in mature γ neurons, and further suggests that α′/β′ and α/β neurons need unf during their initial development. Consistent with these notions, silencing of unf in newborn neurons using asense-GAL4 [44] did not affect larval γ axon projections (Fig. 5A,B) but drastically arrested the nascent α′/β′ (Fig. 5C,D) or α/β axon bundles (Fig. 5E,F) around the peduncle. Taken together, unf acts in mature γ neurons to govern axon re-extension while supporting axonal morphogenesis in newly derived α′/β′ and α/β neurons.


Figure 5. unf is required for the initial axonal morphogenesis in later types of MB neurons.

Newly generated neurons of wild-type (A,C,E) and unf knockdown (B,D,F) MB labeled by asense-GAL4 (green) at the indicated developmental stages. unf was knocked down by asense-GAL4 dependent induction of unf miRNA-a. The arrowheads in [A]–[C] and [E] indicate the normal axonal bundles of the larval MB. The arrows in [D] and [F] indicate the axonal bundles that failed to extend beyond the MB peduncle. Scale bar: 20 µm.


Direct Involvement of unf in Later MB Morphogenetic Processes

The serially-derived MB neurons, made concurrently by four MB progenitors, have undergone morphogenesis in sequence. It has been shown that the projections of specific MBs could be affected by the trajectories of others [50]. Notably, all MB axons stalled around the peduncle end when unf was depleted through development of the MBs, raising the possibility that some later morphogenetic defects may occur as a consequence of earlier axons' failings. The above observation that the α/β lobes could form normally in the absence of the adult γ lobe (Fig. 4B,H) partially rules out such possibility. However, the α′/β′ lobes remained in those γ-lobe-missing MBs (Fig. 4B,H) and might play an essential role in guiding the later derived α/β axons at the peduncle end, a critical choice point to all MB axons.

To demonstrate that unf is directly involved in the axonogenesis of α/β neurons, we sought to knock down unf after morphogenesis of γ and α′/β′ neurons to selectively deplete unf from newborn α/β neurons. We controlled the timing of targeted RNAi using a temperature-sensitive GAL4 repressor, GAL80[ts] [54]. Organisms with the genotype of tubp-GAL80[ts],UAS-mCD8GFP/UAS-unf-miRNA​-a;tubp-GAL80[ts]/+;GAL4-OK107/+ were initially cultured at a permissive temperature (18 degree Celsius) and shift to a restrictive temperature (29 degree Celsius) to induce the expression of unf miRNA-a at desired developmental stages. Besides labeling all five MB lobes by GAL4-OK107, we counterstained MB with anti-Trio antibody that strongly labels α/β lobes and weakly labels γ lobe [52] to make the five MB lobes distinguishable. Induction of unf RNAi from mid-3rd instar and prior to birth of most α′/β′ neurons disrupted all MB lobes (Fig. 6A,B). In contrast, shifting the temperature after adult eclosion left the gross MB morphology intact (Fig. 6C,D). Intriguingly, when the temperature was shifted around puparium formation after the production of α′/β′ neurons, the α′/β′ lobes were grossly normal but the FasII-positive α/β lobes were malformed (Fig. 6E,F). These results demonstrate that the unf knockdown phenotype in α/β neurons was not due to the defects in α′/β′ lobes. Thus, unf directly governs axonal morphogenesis in all three major types of MB neurons.


Figure 6. Direct involvement of unf in later MB morphogenetic processes.

GAL80[ts] was used to control the timing of GAL4-OK107 dependent induction of unf miRNA-a to knock down unf in MB after mid-3rd instar (A–B), after adult eclosion (C–D) or after puparium formation (E–F). MBs were labeled by GAL4-OK107 (green; A,C,E), anti-FasII (magenta; A–F) and anti-Dac (magenta; A,B,E,F) immunostaining. Scale bar: 20 µm.


unf Regulates Axon Pathfinding of MB Neurons

MB axons uniformly stalled around the peduncle end, a common choice point to the migrating axons, raising the possibility that unf may regulate axon pathfinding rather than simply promoting axon extension. To elucidate the mechanism(s) underlying the axonal lobe defect of the unf knockdown MBs, we followed single axons to determine their trajectories in the lobe defective MBs. Single MB neurons were labeled using flip-out strategy [55], [56]. In wild-type MBs, the flip-out clones consistently projected their axons directly into the MB lobes (Fig. 7A). In contrast, the axonal processes of flip-out clones in the unf knockdown MBs wandered around in the ball-like truncated lobes (Fig. 7B). These wandering axons were apparently lost, and often made unusual back turns or loops (Fig. 7C–F). These phenomena ascribe the lobe defect of the unf knockdown MB to misrouting of axons.


Figure 7. unf regulates axon pathfinding of MB neurons.

(A–F) GAL4-OK107 labeled flip-out clones (green in [A] and [B]; black in [C]–[F]) induced in a wild-type (A) or a unf knockdown (B–F) MB. Brains in [A] and [B] were co-labeled with anti-FasII Ab (magenta). [C]–[F] are high magnification views of some axons in the MB lobe in [B]. The arrows in [C] and [E] mark the axons with an unusual back turn(C), or even form a loop (E). [D] and [F] are duplicate images of [C] and [E] with the abnormal axonal paths highlighted by black lines for better visualization. Scale bars: 20 µm. (G–H) Wild-type (G) and unf knockdown (H) MARCM clones labeled by GAL4-OK107 (green). unf was knocked down by GAL4-OK107 induced unf miRNA-a. Brains were counterstained with anti-FasII Ab (magenta) to mark MB γ and α/β lobes. The arrow in [H] indicates a misrouted axon. Scale bar: 20 µm. (I–J) The calyx region of a unf knockdown adult MB labeled by GAL4-OK107 (green; I) and anti-FasII immunostaining (magenta; I–J). unf was knockdown by GAL4-OK107 induced unf miRNA-a. The arrows indicate a misrouted FasII-positive axonal bundle. Scale bar: 20 µm. (K–L) A illustration of the axonal projections in wild-type (K) and unf knockdown (L) MB.


We further determined if unf is cell-autonomously required in individual MB neurons for proper axon guidance, by knocking down unf in single MB neurons. We generated isolated single-cell clones of γ neurons by MARCM (Mosaic Analysis with a Repressible Cell Marker) [57]. Such uniquely labeled cells were the only MB neurons that lack the GAL4 repressor, GAL80, and thus actively expressed unf miRNA-a. This allowed us to knock down unf in single MB neurons within otherwise unperturbed MBs. Notably, the lone unf knockdown neurons exhibited abnormal axon trajectories in the grossly normal MBs (Fig. 7G,H). The misguided axons were not stalled at the peduncle end, possibly because their surrounding wild-type axons may somehow steer the unf knockdown axons into the lobes. Additional evidence for the involvement of unf in axon guidance came from the observation that about 10% of the strongest unf knockdown MBs showed ectopic FasII-positive axon bundles extending through the calyx into abnormal targets rather than migrating along the peduncle (Fig. 7I,J). Taken together, these observations suggest that unf acts cell-autonomously to regulate individual axons' pathfinding (Fig. 7K,L).

unf Is Expressed in MB Neurons Continuously to Regulate Neuron Subtype Identity

In the MB γ neurons, unf is required in a stage-specific manner. Genes required for remodeling of MBs may dynamically express in response to ecdysone signaling. To determine the expression of unf, we generated a rabbit polyclonal antibody against a short peptide (DVTNDNEEPHA) characteristic of unf. The antibody recognized abundant unf expression in the MB neuronal cell bodies (Fig. 8). Depleting unf by targeted RNAi greatly suppressed the immunostaining signal (Fig. 8A–D), confirming the specificity of the antibody. Notably, unf is enriched in all MB neurons through different developmental stages (Fig. 8). No dynamic changes in its expression could be detected during MB remodeling, and high-level expression continues in adult MB neurons (Fig. 8). This expression pattern provides no clue as to why unf is essential for γ neuron remodeling but largely dispensable for their initial morphogenesis. However, its enrichment in the MBs strongly supports our observations that unf controls some MB-characteristic aspects of neural development.


Figure 8. unf is continuously expressed in the MB neurons throughout development.

Wild-type brains (A,B,E–P) at the indicated development stages and a 6 hr-old larval brain with GAL4-OK107 dependent induction of unf miRNA-a (C–D) were labeled with GAL4-OK107 (green; A, C, E, G, I, K, M and O) and UNF immunostaining (magenta; A–P). Scale bars: 20 µm.


Human PNR and C. elegans fax-1, the orthologs of fly unf, have been shown to regulate neuron identity [11], [22], [58], [59]. To investigate if unf controls MB neuron identity and subsequently governs the subtype-specific axon projections, we examined the neuronal cell fate in unf knockdown MBs. We achieved this by knocking down unf using asenase-GAL4 in combination with different Pan-MB or MB subtype-specific markers, including several GAL4s. Given that asense-GAL4 expresses in MB precursors and developing young MB neurons [44], it allows us to knock down unf throughout MB development; and because ansense-GAL4 does not express in adult MB, it will not interfere with the expression of the other MB-specific GAL4s in the adult brain. None of the pan-MB markers, including dachshund (dac) [60], [61], GAL4-OK107 [51] and GAL4-MB247 [62], was affected by unf knockdown (data not shown), suggesting that the MB lineage identity was not regulated by unf. However, all the subtype-specific markers we examined lost their expression in the unf knockdown adult MBs. These include the γ and α′/β′-specific marker Trio in MB cell bodies (some Trio protein can be detected in MB lobes, which might be the residual product of earlier trio expression; see below) [50] (Fig. 9A,B), the γ-specific maker GAL4-NP21 [63] (Fig. 9C,D), the α′/β′-specific marker GAL4-c305a [64] (Fig. 9E,F), the pioneer α/β-specific marker GAL4-c708 [44](data not shown) and the α/β-specific marker GAL4-c739 [53] (Fig. 9G,H). Thus, MB neurons require unf to acquire their subtype identity.


Figure 9. Subtype-specific markers were lost in unf knockdown adult MBs.

The cell body region of wild-type (A, C, E, G, I and K) or unf knockdown (B, D, F, H, J and L) adult (A–H) or wandering larval (I–L) MBs labeled with various MB specific markers, including Trio (A, B, I and J; magenta), GAL4-NP21 (C and D; green), GAL4-c305a (E and F; green), GAL4-c739 (G and H; green) and EcR-B1 (K and L; magenta). Brains in [A], [B], [I] and [J] were co-labeled with GAL4-OK107. The unf miRNA knockdown in [A], [B] and [I]–[L] was induced by GAL4-OK107, and the unf knockdown in [C]–[H] was induced by asense-GAL4 together with GAL4-NP21 (C and D), GAL4-c305a (E and F), or GAL4-c739 (G and H). Note the Trio expression was normal in unf knockdown larval MB cell bodies (J), but became undetectable in unf knockdown adult MB cell bodies (B). Um8G: UAS-mCD8::GFP. Scale bar: 20 µm.


Notably, at the wandering larval stage, when the morphology of the unf knockdown γ neurons was mostly normal, the expression of the γ-specific genes such as trio or EcR-B1 [24], [50] was unaffected (Fig. 9I–L). The loss of the γ-specific markers during metamorphosis suggests that there is a second wave of cell fate specification and/or consolidation occurring at the larva-to-pupa transition for the γ neurons to acquire their final fate; and unf is possibly required for this process. In sum, loss of neuron subtype identity may underlie misguidance of axons in unf knockdown MBs.


Silencing individual NRs through development of the MBs by transgenic miRNAs has allowed us to identify unf as another NR (in addition to EcR, usp, and tll) that regulates MB development. Previously, unf was shown to be essential for the wing expansion and fertility of adult flies, and abundantly expressed in the developing MBs [49]. Here we studied the function of unf in further detail and learned that unf acts in all three major types of MB neurons to promote proper neuron subtype identity and axon guidance. Comparable axon stalling defects of adult MB neurons, as well as missing larval MB dorsal axonal branches were observed in unf mutant organisms (Bates et al., submitted), validating our study of unf's mechanism of action by targeted RNAi.

The involvement of unf in cell fate determination is evident by the loss of subtype-specific markers in the unf knockdown MB. Similar mechanisms have been shown in C. elegans and human. In human, the unf ortholog PNR is specifically expressed in rod photoreceptor cells to promote the rod-cell identity by repressing the expression of S-cone cell specific genes [11], [58]. Mutations in PNR leads to enhanced S-cone syndrome (ESCS) which is an inherited disease causing hypersensitivity to short-wave light due to increased numbers of S-cone cells at the expense of rod photoreceptor cells [11], [58]. However, in the present study, we did not observe any reciprocal cell number change in unf knockdown MBs. Thus, unlike PNR in human, unf in flies is not used to repress a default cell fate. Given that unf is required for the proper gene expression in all three major types of MB neurons, it may play a general role in assisting the temporal identity factors, such as chinmo, to diversify neuronal cell fates [3], [44].

Loss of proper identity might underlie the axon guidance defect observed in the unf knockdown MB. The C. elegans ortholog of unf, fax-1, has been suggested to regulate cell identities of 18 neurons, including both motor- and interneurons; in fax-1 mutant, some neurotransmitters and synaptic proteins are not properly expressed in the neurons normally expressing fax-1 [22], [59], [65]. Notably, like unf in fly MB neurons, fax-1 is also required for axonal pathfinding for several C. elegans neurons, indicating a functional conservation among unf orthologs in different species [22], [59], [65].

unf has been hypothesized to act as a transcriptional repressor, because its ligand-binding domain (LDB) failed to activate gene expression [66]. The function of the human ortholog PNR to repress S-corn specific genes supports this hypothesis [58]. Contrary to the human PNR, in both flies and worms, loss of unf or fax-1 leads to the down-regulation of many neural genes [22], [59]. However, since there is no evidence for gene activation by direct binding of unf or fax-1, it is possible that unf and fax-1 regulate these neural genes indirectly by repressing other repressors.

In the MBs, unf primarily governs axonal morphogenesis during later larval and pupal development when different MB neuron types need to make distinct projections. For instance, in early pupae, α/β neurons undergo de-novo axonogenesis to form the α/β lobes while γ axons regenerate to make up the adult-specific γ lobe. Given the notion that unf promotes MB axonogenesis possibly through regulating neuron subtype identity, its selective involvement in late MB morphogenesis could simply reflect the importance of neuron subtype identity in ensuring diverse subtype-specific axonal morphogenesis. However, the larval γ neurons of unf knockdown MBs show normal cell fate as evidenced by proper expression of EcR-B1. This argues for a stage-specific function of unf in MB development. Notably, despite its stage-specific requirement, unf is enriched in the MBs through different developmental stages and into the adult, raising the possibility that its dynamic activity is patterned through temporal control of ligand availability. Recently, an in vitro study suggested that UNF is a heme binding protein [48]. Given the known relationship between heme and lipid metabolism [6], [12], [67], [68], heme levels can serve as an indicator for energy resource and developmental progress. Perhaps by detecting heme levels, unf may potentially coordinate the timing of the unf-medicated adult-specific cell fate determination and axonogenesis. However, in vivo evidence for the interaction between heme and unf remains lacking.

In conclusion, patterned NR activities govern various temporally regulated neural developmental processes of interest. UNF and its orthologs probably promote subtype neuronal differentiation in temporally controlled manners. Elucidating the regulation of NR activities and their control of neuron subtype identity should shed additional light on how diverse neuron types undergo differential morphogenesis and acquire different subtype-specific projections to construct the complex brain.

Materials and Methods

Generation of miRNA Fly Lines

The UAS-miRNA constructs were generated as described in the previous study [46]. The miRNA target sequences of 18 NR genes were shown in Table S1. The transgenic flies were generated by inserting UAS-miRNA constructs into the attp-16 site on the second chromosome using the integration system as previous described [69].

Fly Strains

Beside the UAS-miRNA transgenic fly lines, the fly strains used in this study includes: (1) asense-GAL4 [44]; (2) GAL4-OK107 [51]; (3) GAL4-MB247 [62]; (4) GAL4-201Y [53]; (5) GAL4-NP21 [63]; (6) GAL4-c305a [64]; (7) GAL4-c708 [44]; (8) GAL4-c739 [53]; (9) tubp-GAL80[ts];tubp-GAL80[ts],UAS-mCD8::​GFP;OK107 [70]; (5) Df(2R)ED2426/SM6a (stock#9064, Bloomington stock center); (6) hs-FLP,UAS-mCD8::GFP,FRTG13,tubp-GAL80/C​yO;OK107; (7) hs-FLP[122];Sp/CyO;UAS>rCD2,y+>mCD8::GFP [55], [56].

Induction of Flip-Out Clones and MARCM Analysis

To generate flip-out clones, the newly hatched larvae with proper genotype were applied 30-minute heat-shock in 32°C water bath. For MARCM studies, larvae with proper genotype were collected in 2 hours after hatching, and cultured at the density of 100 larvae per food vial at 25°C. MARCM clones were induced by applying 1 hour heat-shock in 38°C water bath at various developmental stages.

Immunohistochemistry and Confocal Microscopy

Fly brains were dissected, stained and mounted as described in our previous study (Lee and Luo, 1999). Primary antibodies used in this study include rabbit anti-UNF Ab (1:4000); rat anti-mCD8 mAb (1:100; Caltag), mouse anti-FasII mAb (1:100; DSHB), mouse anti-Dac2-3 mAb (1:200; DSHB), and rabbit anti-Trio Ab (1:2000). FITC and Cy3 conjugated secondary antibodies (Jackson ImmunoResearch) were used at the dilution of 1:200 and 1:400, respectively. Immunofluorescent signals were collected by Zeiss LSM confocal microscope and processed using Osirix and Adobe Photoshop.

Supporting Information

Table S1.


(0.06 MB PDF)


We would like to thank S. Robinow for sharing unpublished results, K. Bates for sharing her unf manuscript prior to submission. We would also like to thank S. Robinow, K. Bates, and JM. Knapp for critical reading of the manuscript, and UIUC Immunological Resource center for making anti-UNF antibody.

Author Contributions

Conceived and designed the experiments: SL TL. Performed the experiments: SL YH. Analyzed the data: SL. Contributed reagents/materials/analysis tools: SL YH. Wrote the paper: SL TL.


  1. 1. Urbach R, Technau GM (2003) Segment polarity and DV patterning gene expression reveals segmental organization of the Drosophila brain. Development 130: 3607–3620.
  2. 2. Pearson BJ, Doe CQ (2004) Specification of temporal identity in the developing nervous system. Annu Rev Cell Dev Biol 20: 619–647.
  3. 3. Yu HH, Lee T (2007) Neuronal temporal identity in post-embryonic Drosophila brain. Trends Neurosci 30: 520–526.
  4. 4. Luo L, O'Leary DD (2005) Axon retraction and degeneration in development and disease. Annu Rev Neurosci 28: 127–156.
  5. 5. Truman JW (1990) Metamorphosis of the central nervous system of Drosophila. J Neurobiol 21: 1072–1084.
  6. 6. Chawla A, Repa JJ, Evans RM, Mangelsdorf DJ (2001) Nuclear receptors and lipid physiology: opening the X-files. Science 294: 1866–1870.
  7. 7. Francis GA, Fayard E, Picard F, Auwerx J (2003) Nuclear receptors and the control of metabolism. Annu Rev Physiol 65: 261–311.
  8. 8. Kanai MI, Okabe M, Hiromi Y (2005) seven-up Controls switching of transcription factors that specify temporal identities of Drosophila neuroblasts. Dev Cell 8: 203–213.
  9. 9. Kramer S, West S, Hiromi Y (1995) Cell fate control in the Drosophila retina by the orphan receptor seven-up: its role in the decisions mediated by the ras signaling pathway. Development 121: 1361.
  10. 10. Luo G, Chen Y, Li X, Liu T, Le W (2008) Nr4a2 is essential for the differentiation of dopaminergic neurons during zebrafish embryogenesis. Molecular and Cellular Neuroscience 39: 202–210.
  11. 11. Milam AH, Rose L, Cideciyan AV, Barakat MR, Tang WX, et al. (2002) The nuclear receptor NR2E3 plays a role in human retinal photoreceptor differentiation and degeneration. Proc Natl Acad Sci USA 99: 473–478.
  12. 12. Pardee K, Reinking J, Krause H (2004) Nuclear hormone receptors, metabolism, and aging: what goes around comes around.Transcription factors link lipid metabolism and aging-related processes. Science of aging knowledge environment: SAGE KE 2004: re8.
  13. 13. Tang LS, Alger HM, Pereira FA (2006) COUP-TFI controls Notch regulation of hair cell and support cell differentiation. Development 133: 3683.
  14. 14. Tran PV, Lee MB, Marín O, Xu B, Jones KR, et al. (2003) Requirement of the orphan nuclear receptor SF-1 in terminal differentiation of ventromedial hypothalamic neurons. Mol Cell Neurosci 22: 441–453.
  15. 15. Zetterström RH, Solomin L, Jansson L, Hoffer BJ, Olson L, et al. (1997) Dopamine Neuron Agenesis in Nurr1-Deficient Mice. Science 276: 248.
  16. 16. Zhou C (1999) The Nuclear Orphan Receptor COUP-TFI Is Required for Differentiation of Subplate Neurons and Guidance of Thalamocortical Axons. Neuron 24: 847–859.
  17. 17. Kurusu M, Maruyama Y, Adachi Y, Okabe M, Suzuki E, et al. (2009) A conserved nuclear receptor, Tailless, is required for efficient proliferation and prolonged maintenance of mushroom body progenitors in the Drosophila brain. Dev Biol 326: 224–236.
  18. 18. Liu H, Belz T, Bock D, Takacs A, Wu H, et al. (2008) The nuclear receptor tailless is required for neurogenesis in the adult subventricular zone. Genes & Development 22: 2473–2478.
  19. 19. Shi Y, Shi Y, Lie DC, Lie DC, Taupin P, et al. (2004) Expression and function of orphan nuclear receptor TLX in adult neural stem cells. Nature 427: 78.
  20. 20. Robinow S, Talbot WS, Hogness DS, Truman JW (1993) Programmed cell death in the Drosophila CNS is ecdysone-regulated and coupled with a specific ecdysone receptor isoform. Development 119: 1251–1259.
  21. 21. Armentano M, Filosa A, Andolfi G, Studer M (2006) COUP-TFI is required for the formation of commissural projections in the forebrain by regulating axonal growth. Development 133: 4151–4162.
  22. 22. Much JW, Slade DJ, Klampert K, Garriga G, Wightman B (2000) The fax-1 nuclear hormone receptor regulates axon pathfinding and neurotransmitter expression. Development 127: 703–712.
  23. 23. Ponnio T, Conneely OM (2004) nor-1 Regulates Hippocampal Axon Guidance, Pyramidal Cell Survival, and Seizure Susceptibility. Mol Cell Biol 24: 9070.
  24. 24. Lee T, Marticke S, Sung C, Robinow S, Luo L (2000) Cell-autonomous requirement of the USP/EcR-B ecdysone receptor for mushroom body neuronal remodeling in Drosophila. Neuron 28: 807–818.
  25. 25. Truman JW (1996) Steroid receptors and nervous system metamorphosis in insects. Dev Neurosci 18: 87–101.
  26. 26. Chaturvedi RK, Beal MF (2008) PPAR: a therapeutic target in Parkinson's disease. J Neurochem 106: 506–518.
  27. 27. Chu Y, Le W, Kompoliti K, Jankovic J, Mufson EJ, et al. (2006) Nurr1 in Parkinson's disease and related disorders. J Comp Neurol 494: 495–514.
  28. 28. Culman J, Zhao Y, Gohlke P, Herdegen T (2007) PPAR-γ: therapeutic target for ischemic stroke. Trends in Pharmacological Sciences 28: 244–249.
  29. 29. Jankovic J, Chen S, Le W (2005) The role of Nurr1 in the development of dopaminergic neurons and Parkinson's disease. Prog Neurobiol 77: 128–138.
  30. 30. Le W, Le W, Xu P, Xu P, Jankovic J, et al. (2002) Mutations in NR4A2 associated with familial Parkinson disease. Nat Genet 33: 85.
  31. 31. Le W, Pan T, Huang M, Xu P, Xie W, et al. (2008) Decreased NURR1 gene expression in patients with Parkinson's disease. Journal of the Neurological Sciences 273: 29–33.
  32. 32. Serra H, Duvick L, Zu T, Carlson K, Stevens S, et al. (2006) RORα-Mediated Purkinje Cell Development Determines Disease Severity in Adult SCA1 Mice. Cell 127: 697–708.
  33. 33. King-Jones K, Thummel CS (2005) Nuclear receptors–a perspective from Drosophila. Nat Rev Genet 6: 311–323.
  34. 34. Maurange C, Cheng L, Gould AP (2008) Temporal transcription factors and their targets schedule the end of neural proliferation in Drosophila. Cell 133: 891–902.
  35. 35. Schubiger M, Wade AA, Carney GE, Truman JW, Bender M (1998) Drosophila EcR-B ecdysone receptor isoforms are required for larval molting and for neuron remodeling during metamorphosis. Development 125: 2053–2062.
  36. 36. Schubiger M, Truman JW (2000) The RXR ortholog USP suppresses early metamorphic processes in Drosophila in the absence of ecdysteroids. Development 127: 1151–1159.
  37. 37. Schubiger M, Carre C, Antoniewski C, Truman JW (2005) Ligand-dependent de-repression via EcR/USP acts as a gate to coordinate the differentiation of sensory neurons in the Drosophila wing. Development 132: 5239–5248.
  38. 38. Choi YJ, Lee G, Park JH (2006) Programmed cell death mechanisms of identifiable peptidergic neurons in Drosophila melanogaster. Development 133: 2223–2232.
  39. 39. Brown HL, Truman JW (2009) Fine-tuning of secondary arbor development: the effects of the ecdysone receptor on the adult neuronal lineages of the Drosophila thoracic CNS. Development 136: 3247–3256.
  40. 40. Wang J, Ma X, Yang JS, Zheng X, Zugates CT, et al. (2004) Transmembrane/juxtamembrane domain-dependent Dscam distribution and function during mushroom body neuronal morphogenesis. Neuron 43: 663–672.
  41. 41. Boyle M, Nighorn A, Thomas JB (2006) Drosophila Eph receptor guides specific axon branches of mushroom body neurons. Development 133: 1845–1854.
  42. 42. Ng J, Nardine T, Harms M, Tzu J, Goldstein A, et al. (2002) Rac GTPases control axon growth, guidance and branching. Nature 416: 442–447.
  43. 43. Lee T, Luo L, Lee A (1999) Development of the Drosophila mushroom bodies: sequential generation of three distinct types of neurons from a neuroblast. Development 126: 4065–4076.
  44. 44. Zhu S, Lin S, Kao CF, Awasaki T, Chiang AS, et al. (2006) Gradients of the Drosophila Chinmo BTB-zinc finger protein govern neuronal temporal identity. Cell 127: 409–422.
  45. 45. Awasaki T, Ito K (2004) Engulfing action of glial cells is required for programmed axon pruning during Drosophila metamorphosis. Curr Biol 14: 668–677.
  46. 46. Chen CH, Huang H, Ward CM, Su JT, Schaeffer LV, et al. (2007) A synthetic maternal-effect selfish genetic element drives population replacement in Drosophila. Science 316: 597–600.
  47. 47. Shi L, Yu HH, Yang JS, Lee T (2007) Specific Drosophila Dscam juxtamembrane variants control dendritic elaboration and axonal arborization. J Neurosci 27: 6723–6728.
  48. 48. de Rosny E, de Groot A, Jullian-Binard C, Borel F, Suarez C, et al. (2008) DHR51, the Drosophila melanogaster Homologue of the Human Photoreceptor Cell-Specific Nuclear Receptor, Is a Thiolate Heme-Binding Protein. Biochemistry.
  49. 49. Sung C, Wong L, Chang Sen L, Nguyen E, Lazaga N, et al. (2009) The unfulfilled/DHR51 gene of Drosophila melanogaster modulates wing expansion and fertility. Dev Dyn 238: 171–182.
  50. 50. Awasaki T, Saito M, Sone M, Suzuki E, Sakai R, et al. (2000) The Drosophila trio plays an essential role in patterning of axons by regulating their directional extension. Neuron 26: 119–131.
  51. 51. Connolly JB, Roberts IJ, Armstrong JD, Kaiser K, Forte M, et al. (1996) Associative learning disrupted by impaired Gs signaling in Drosophila mushroom bodies. Science 274: 2104–2107.
  52. 52. Crittenden JR, Skoulakis EM, Han KA, Kalderon D, Davis RL (1998) Tripartite mushroom body architecture revealed by antigenic markers. Learn Mem 5: 38–51.
  53. 53. Yang MY, Armstrong JD, Vilinsky I, Strausfeld NJ, Kaiser K (1995) Subdivision of the Drosophila mushroom bodies by enhancer-trap expression patterns. Neuron 15: 45–54.
  54. 54. McGuire SE, Le PT, Osborn AJ, Matsumoto K, Davis RL (2003) Spatiotemporal rescue of memory dysfunction in Drosophila. Science 302: 1765–1768.
  55. 55. Wong AM, Wang JW, Axel R (2002) Spatial representation of the glomerular map in the Drosophila protocerebrum. Cell 109: 229–241.
  56. 56. Zhu S, Chiang AS, Lee T (2003) Development of the Drosophila mushroom bodies: elaboration, remodeling and spatial organization of dendrites in the calyx. Development 130: 2603–2610.
  57. 57. Lee T, Luo L (1999) Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron 22: 451–461.
  58. 58. Chen J, Rattner A, Nathans J (2005) The rod photoreceptor-specific nuclear receptor Nr2e3 represses transcription of multiple cone-specific genes. J Neurosci 25: 118–129.
  59. 59. Wightman B, Ebert B, Carmean N, Weber K, Clever S (2005) The C. elegans nuclear receptor gene fax-1 and homeobox gene unc-42 coordinate interneuron identity by regulating the expression of glutamate receptor subunits and other neuron-specific genes. Dev Biol 287: 74–85.
  60. 60. Martini SR, Davis RL (2005) The dachshund gene is required for the proper guidance and branching of mushroom body axons in Drosophila melanogaster. J Neurobiol 64: 133–144.
  61. 61. Martini SR, Roman G, Meuser S, Mardon G, Davis RL (2000) The retinal determination gene, dachshund, is required for mushroom body cell differentiation. Development 127: 2663–2672.
  62. 62. McGuire SE, Le PT, Davis RL (2001) The role of Drosophila mushroom body signaling in olfactory memory. Science 293: 1330–1333.
  63. 63. Tanaka NK, Awasaki T, Shimada T, Ito K (2004) Integration of chemosensory pathways in the Drosophila second-order olfactory centers. Curr Biol 14: 449–457.
  64. 64. Krashes MJ, Keene AC, Leung B, Armstrong JD, Waddell S (2007) Sequential use of mushroom body neuron subsets during drosophila odor memory processing. Neuron 53: 103–115.
  65. 65. Wightman B, Baran R, Garriga G (1997) Genes that guide growth cones along the C. elegans ventral nerve cord. Development 124: 2571–2580.
  66. 66. Palanker L, Necakov AS, Sampson HM, Ni R, Hu C, et al. (2006) Dynamic regulation of Drosophila nuclear receptor activity in vivo. Development 133: 3549–3562.
  67. 67. Bauer M, Hamm AC, Bonaus M, Jacob A, Jaekel J, et al. (2004) Starvation response in mouse liver shows strong correlation with life-span-prolonging processes. Physiol Genomics 17: 230–244.
  68. 68. Lorenz RT, Parks LW (1991) Involvement of heme components in sterol metabolism of Saccharomyces cerevisiae. Lipids 26: 598–603.
  69. 69. Markstein M, Pitsouli C, Villalta C, Celniker S, Perrimon N (2008) Exploiting position effects and the gypsy retrovirus insulator to engineer precisely expressed transgenes. Nat Genet 40: 476–483.
  70. 70. Shi L, Lin S, Grinberg Y, Beck Y, Grozinger C, et al. (2007) Roles of Drosophila Kruppel-homolog 1 in neuronal morphogenesis. Devel Neurobio 67: 1614–1626.
  71. 71. Honjo K, Furukubo-Tokunaga K (2009) Distinctive neuronal networks and biochemical pathways for appetitive and aversive memory in Drosophila larvae. J Neurosci 29: 852–862.
  72. 72. Marin EC, Watts RJ, Tanaka NK, Ito K, Luo L (2005) Developmentally programmed remodeling of the Drosophila olfactory circuit. Development 132: 725–737.
  73. 73. Honjo K, Furukubo-Tokunaga K (2005) Induction of cAMP response element-binding protein-dependent medium-term memory by appetitive gustatory reinforcement in Drosophila larvae. J Neurosci 25: 7905–7913.