The pipsqueak-domain proteins Distal antenna and Distal antenna-related restrict Hunchback neuroblast expression and early-born neuronal identity

The generation of neuronal diversity during development is a tightly controlled process in which multiple regulatory factors provide neural stem cells with both spatial and temporal information. Much progress has been made in understanding spatial patterning, in which neural stem cells are topographically organized according to their unique dorsoventral and anteroposterior positions that specify cell fate (Doe and Technau, 1993; Skeath and Thor, 2003; Guillemot, 2005). However, much less is known regarding temporal patterning, a process in which neural stem cells change their transcriptional signatures over time to generate distinct progeny (Pearson and Doe, 2003).

In the developing Drosophila embryo, each of the 30 neuroblasts in a hemisegment can be uniquely identified by its stereotyped position, molecular markers and unique cell lineage (Doe, 1992; Broadus et al., 1995; Bossing et al., 1996; Schmidt et al., 1997; Schmid et al., 1999; Baumgardt et al., 2009; Karlsson et al., 2010). Neuroblasts undergo a series of asymmetric cell divisions that give rise to a self-renewed neuroblast and a smaller ganglion mother cell (GMC), which typically divides one more time to produce two postmitotic daughter cells (Goodman and Doe, 1993). Each neuroblast generates distinct sets of progeny, but always in a stereotyped birth order. This is accomplished by the sequential expression of the zinc-finger transcription factors Hunchback (Hb) and Krüppel (Kr), the redundant POU domain transcription factors Pdm1 (Nubbin – FlyBase) and Pdm2 (henceforth both called Pdm), and the zinc finger transcription factor Castor (Cas) (Isshiki et al., 2001; Novotny et al., 2002). Because daughter cell fate is governed by the transcriptional profile of the neuroblast at the time of its birth, the timing and duration of expression of each temporal identity factor in neuroblasts is crucial for specification of the progeny cell type and its abundance. For example, use of the Gal4-UAS system to extend the length of Hb expression in neuroblasts results in extra neuronal cell types with early-born identities at the expense of those with later-born identities (Isshiki et al., 2001; Novotny et al., 2002). Thus, regulation of the timing and duration of Hb expression is an early and essential step in the generation of neuronal diversity.

Here, we show that the pipsqueak motif nuclear proteins Distal antenna (Dan) and Distal antenna-related (Danr) function in a parallel pathway to that of Svp to limit Hb neuroblast expression and the number of early-born progeny in multiple neuroblast lineages.

We used the following fly stocks to analyze mutant phenotypes at 23°C: dan danr^ex56/TM3 Ubx-lacZ, danr^ex35/TM3 Ubx-lacZ, dan^ems3/TM3 Ubx-lacZ (Emerald et al., 2003), hb^P1 hb^FB/TM3 hb-lacZ to remove Hb CNS expression (Hulskamp et al., 1994; Isshiki et al., 2001), and svp^e22/TM3 Ubx-lacZ (Kanai et al., 2005). svp^e22/TM3 Ubx-lacZ and dan danr^ex56/TM3 Ubx-lacZ were recombined on chromosome 3 to generate triple mutants lacking svp, dan and danr. We used the following gal4 lines: engrailed (en)-gal4 (Harrison et al., 1995; Isshiki et al., 2001; Pearson and Doe, 2003) for expression in the posterior compartment of each segment, inscuteable (insc)-gal4 (1407-gal4, Bloomington Stock Center) on chromosome 2 and scabrous (sca)-gal4 (Cleary and Doe, 2006) for expression in all neuroblasts. For Hb misexpression studies, en-gal4 was crossed to UAS-hb on chromosomes 2 and 3 (Wimmer et al., 2000). For Dan rescue experiments, the following stocks were generated and crossed to UAS-dan: dan danr^ex56/TM3, Ubx-lacZ: en-gal4; dan danr^ex56/TM3, Ubx-lacZ, insc-gal4; dan danr^ex56/TM3, Ubx-lacZ and sca-gal4; dan danr^ex56/TM3, Ubx-lacZ. yw embryos were used for wild-type expression analysis.

Antibody staining was performed according to standard methods (Rothwell and Sullivan, 2000). Primary antibodies, dilutions and sources were: rat anti-Dan 1:400 (kind gift from J. Curtiss, New Mexico State University, NM, USA); mouse anti-Eve 1:50 (3C10-c), mouse anti-Engrailed (4D9) 1:10, mouse anti-Islet 1:200 and mouse anti-Eagle 1:100 (8B11) (Developmental Studies Hybridoma Bank, University of Iowa, IA, USA); rabbit anti-Cut 1:500 (kind gift from Y. N. Jan, UCSF, San Francisco, CA, USA); rabbit anti-Hb 1:200 (Tran and Doe, 2008); rat anti-Zfh2 1:200 (Tran et al., 2010); guinea pig anti-Runt 1:500 and guinea pig anti-HB9 1:500 (East Asian Distribution Center for segmentation Antibodies, Mishima, Japan); rabbit anti-Castor 1:1000 (kind gift from W. Odenwald, Bethesda, MD, USA); anti-β-galactosidase 1:500 (Promega, Madison, WI, USA); mouse anti-Svp 1:500 and rat anti-Svp 1:500 (kind gift from T Isshiki, National Institute of Genetics, Japan) (Kanai et al., 2005); guinea pig anti-Miranda 1:400 (Lee et al., 2006). Alexa Fluor 488- or Alexa Fluor 555-conjugated secondary antibodies were from Molecular Probes (Invitrogen, Eugene, OR, USA); Cy5- or Dylight 549-conjugated secondary antibodies and Cy3- or Dylight 405-conjugated Streptavidin were from Jackson ImmunoResearch (West Grove, PA, USA). Biotinylated secondary antibodies were from Vector Laboratories (Burlingame, CA, USA) and Jackson ImmunoResearch.

Markers used to identify neurons from different neuroblast (NB) lineages were:

dan and danr cDNA were obtained from the Drosophila Gene Collection. To generate RNA probes to detect CG34110 specifically, we used the Superscript One-Step RT-PCR Kit (Invitrogen). We used gene-specific primers to obtain CG34110 cDNA (forward, TCGACAG GATTCATTTGCGAGAC and reverse, TTGAGCAGACTCACA AACCGAGAC) with the T7 promoter sequence on the reverse primer). Digoxygenin-labeled RNA probe was transcribed using the Roche DIG Labeling Kit (Roche Diagnostics, Indianapolis, IN, USA). In situ hybridization was performed as described previously (Grosskortenhaus et al., 2005).

To identify genes that regulate early-born versus late-born temporal identity in Drosophila embryonic neuroblast lineages, we conducted a forward genetic screen of ~100 second and third chromosomal deficiency lines for mutants that had altered numbers of Even-skipped (Eve)⁺ early-born neurons. Eve labels a small number of identifiable early-born neurons from multiple neuroblast lineages: anterior corner cell (aCC) and posterior corner cell (pCC) from NB1-1, RP2 from NB4-2, and the first five motor neurons (U1-U5) from NB7-1 (Fig. 1A). Here, we describe the phenotype of deficiency Df(3R)Exel6201, in which nearly half of all hemisegments showed extra early-born aCC/pCC, RP2 or U1/U2 neurons (46.6%; 99 hemisegments counted from nine embryos). We did not observe any gross morphological deficits in the CNS of the Df(3R)Exel6201 mutant embryos. Phenotypes of other deficiencies will be described elsewhere. Df(3R)Exel6201 is a small chromosomal deletion affecting only three genes: distal antenna (dan), distal antenna-related (danr) and the uncharacterized gene CG34110. To control for any phenotypes in the deficiency caused by background mutations on this chromosome, we obtained a second small deficiency (dan danr^ex56) removing the same three genes (dan, danr and CG34110) that was independently generated on a different genetic background (Emerald et al., 2003). We confirmed that the ectopic Eve early-born neuron phenotype was also present in dan danr^ex56 mutants and in the dan danr^ex56/Def(3R)Exel6201 transheterozygous mutants (Fig. 1B).

To determine which of these three missing genes caused the observed CNS phenotype, we assayed their gene expression patterns and performed rescue experiments. Both dan and danr were highly expressed in neuroblasts and neurons of the developing CNS, whereas CG34110 was expressed in a subset of sensory neurons at late stages but was expressed at low or undetectable levels in the CNS (see Fig. S1 in the supplementary material). Based on expression patterns, we focused on the dan and danr genes. Dan and Danr are two closely related nuclear proteins containing a pipsqueak motif DNA-binding domain; this family of proteins is conserved from fungi to vertebrates (Lehmann et al., 1998). Dan and Danr were previously reported to have overlapping expression patterns and redundant functions in antennal development (Emerald et al., 2003) and have been shown to regulate cell specification in the retina (Suzanne et al., 2003; Curtiss et al., 2007). We used the Gal4-UAS system to misexpress dan in all neuroblasts and found that it could partially rescue the dan danr^ex56 phenotype (see Fig. S2 in the supplementary material), so we conclude that loss of dan and/or danr causes the observed CNS phenotype (described in more detail below).

Dan and Danr act redundantly in antennal development (Emerald et al., 2003) and had similar expression in the embryonic CNS (see Fig. S1 in the supplementary material), raising the possibility that they act redundantly in the CNS. We found that the hypomorphic dan^ems3 mutant (Emerald et al., 2003; Curtiss et al., 2007) has no Eve CNS phenotype, but the presumptive null danr^ex35 mutant (Emerald et al., 2003) has an Eve CNS phenotype in the NB1-1 lineage similar to that of the dan danr double mutant and a weak phenotype in the NB7-1 lineage. These observations suggest that endogenous levels of Dan are not sufficient for normal CNS development (see Fig. S2A in the supplementary material). Importantly, overexpression of Dan alone rescued nearly all aspects of the dan danr double mutant phenotype (see Fig. S2B in the supplementary material), suggesting that elevated levels of Dan alone can compensate for Danr loss. We propose that Dan and Danr have redundant or similar functions in the CNS, but both proteins are required for normal development.

To determine the identity of the ectopic Eve neurons in the dan danr mutant, we used molecular markers that distinguish early-born progeny of the NB7-1 and NB1-1 lineages. In the NB7-1 lineage, the first five divisions produce GMCs that generate the Eve⁺ U1-U5 motor neuron and their Eve⁻ siblings. The following markers can distinguish all five U neurons: Hb (U1 and U2), Zfh2 (U2-U5), Cut (U3-U5), Runt (U4 and U5) and Cas (U5 only) (Isshiki et al., 2001) (Fig. 2A). In the NB1-1 lineage the first GMC generates the Eve⁺ aCC/pCC neurons that are also Hb⁺. The aCC/pCC and U1/U2 neurons are adjacent, but can be distinguished because only aCC/pCC are Cut⁺. Using these markers, we found that both NB1-1 and NB7-1 generated extra early-born neurons in dan danr mutants. In the NB1-1 lineage, dan danr mutants showed ectopic aCC/pCC neurons in 34.3±6.2% of hemisegments (ranging from one to eight extra corner cells), with the average number of ectopic aCC/pCC neurons being 2.6±0.2 (n=110 hemisegments from 11 embryos; Fig. 2). In the NB7-1 lineage, dan danr mutants showed ectopic U1 neurons in 24.3±5.3% of hemisegments (ranging from one to three extra early-born U neurons; n=98 hemisegments from ten embryos; Fig. 2). Later-born U4-U5 neurons were affected, but to a lesser degree than early-born progeny (only 4.8±2.4% of n=42 hemisegments from three embryos had ectopic U4/5 neurons). We conclude that Dan/Danr act to limit the number of early-born neurons generated in both the NB1-1 and NB7-1 lineages.

NB3-1 sequentially gives rise to four GMCs that subsequently generate four RP motor neurons (and siblings) in the order RP1/sib→RP4/sib→RP3/sib→RP5/sib (Tran and Doe, 2008). All four RP motor neurons are Islet⁺ Hb9⁺ (Tup and Exex, respectively – FlyBase), but only the early-born RP1 and RP4 neurons are Hb⁺ (schematic, top panel of Fig. 3A). dan danr mutants showed ectopic RP1/RP4 neurons in 34.5% of hemisegments (ranging from one to three extra early-born RP cells; n=58 hemisegments from 12 embryos; Fig. 3B). We conclude that Dan and Danr function to limit the number of early-born neurons in the NB3-1 lineage, similar to their role in the NB1-1 and NB7-1 lineages. Additionally, similar to the NB7-1 lineage, we found at low frequency extra Hb⁻ later-born neurons (12.1% of the hemisegments, n=58), suggesting that although Dan and Danr might be primarily required for limiting the number of early-born progeny, they might also have a broader role in regulating temporal identity throughout the Hb→Cas expression window (see Discussion).

NB7-3 generates a short lineage: the first GMC makes the EW1 interneuron and its GW motor neuron sibling, followed by GMCs that generate EW2/sib and then EW3 (Isshiki et al., 2001). All four neurons in this lineage are Eagle⁺, but only the first-born EW1/GW neurons are Hb⁺ (schematic, top panel of Fig. 3C). In the NB7-3 lineage, dan danr mutants showed ectopic EW1/GW neurons; however, the phenotype was quite rare, observed in only 3.6±1.6% of hemisegments (n=136 hemisegments from eight embryos; Fig. 3D). As per NB7-1 and NB3-1 lineages, we also infrequently observed hemisegments with extra late-born EW2/EW3 neurons (in two of the 136 hemisegments quantified, e.g. Fig. 3D) and, more frequently, examples in which there were fewer or a total loss of cells in this lineage (21.3% of hemisegments; n=136 from eight embryos). This latter phenotype might arise in part from the failure of NB7-3 to form (data not shown). We conclude that Dan and Danr limit the early-born neurons in all four lineages examined (1-1, 7-1, 3-1 and 7-3). However, the fact that there was lineage-specific variability in how many extra cells were generated, how many hemisegments per embryo had a phenotype, as well as in the specific type of phenotype (i.e. how frequently extra late-born neurons were observed) suggests that additional lineage-specific regulatory components interact with Dan and/or Danr (see Discussion).

One mechanism by which dan danr mutants could generate an extra early-born GMC in multiple lineages is by a slight extension of Hb neuroblast expression. We focused our attention on NB3-1, which is easy to identify owing to its relatively late delamination and persistent superficial position. In wild-type embryos, the Hb⁺ NB3-1 formed at early stage 10 and downregulated Hb by late stage 11 (Fig. 4B,E). In wild type, Hb downregulation in NB3-1 was tightly correlated with the formation of NB7-3, so we used this as a temporal reference point. In dan danr mutants, neuroblasts delaminated at their normal developmental stages, including NB3-1 at early stage 10 (Fig. 4C), and NB7-3 at late stage 11 (Fig. 4D) but NB3-1 maintained Hb expression past the time of NB7-3 formation. This was most apparent in the abdominal segments, in which 70.8% of the 3-1 neuroblasts still maintained Hb expression (n=24 from three embryos) compared with only 2.0% in wild type (n=42 hemisegments from five embryos) (Fig. 4D,E). NB3-1 in thoracic segments were also similarly affected, though to a lesser degree (Fig. 4E). These data indicate that Hb expression in NB3-1 is prolonged in dan danr mutant embryos, consistent with the presence of ectopic RP1/4 early-born neurons in this lineage, and suggest that the same occurs in other neuroblasts that produce extra early-born neurons in the dan danr mutant embryos. However, misexpression of Dan (using en-gal4 UAS-dan) did not result in premature loss of hb expression or lack of early-born neurons in the NB7-1 lineage (data not shown), suggesting that the timely downregulation of Hb in neuroblasts does not occur through direct repression of hb by Dan.

To gain better insight into how Dan and Danr might be regulating Hb expression in neuroblasts, we analyzed the timing of Dan neuroblast expression relative to that of Hb and Svp, a previously identified negative regulator of Hb expression (Fig. 5; see Fig. S3 in the supplementary material). We found that Dan was broadly and robustly expressed in all newly delaminated Hb⁺ neuroblasts, and its expression extended well beyond the Hb temporal window, fading at early stage 12 (Fig. 5B,C) at the beginning of the Cas window (see Fig. S4 in the supplementary material). In situ hybridization using probes specific to dan or danr revealed that the two genes have similar expression patterns: robust expression by neuroepithelia at late stage 8 and young neuroblasts at stages 9-10, and expression by a subset of neurons at late embryonic stages (see Fig. S1 in the supplementary material, in situ images for neurons not shown). We cannot rule out the possibility that there might be subtle differences in the levels of dan versus danr expression in specific NB lineages, yet our in situ data indicate that dan and danr have remarkably similar expression patterns in the CNS. In contrast to the temporally broad expression of both dan and danr, we confirmed previous observations (Kanai et al., 2005) that Svp is only transiently expressed in neuroblasts just as Hb is being downregulated (Fig. 5). Such distinct temporal expression of Dan and Svp in neuroblasts suggests they have very different functions in limiting the Hb expression window (see Discussion).

Interestingly, Dan and Svp both displayed a second wave of neuroblast expression that peaked at stage 15 (Fig. 5E,F); the function of this later expression is unknown. Dan expression was also observed in post-mitotic neurons (see Fig. S5 in the supplementary material), predominantly the early-born neurons in the deeper layers of the CNS (Isshiki et al., 2001). Most early-born Hb⁺ neurons were also Dan⁺, although there were a few Hb⁺ neurons that did not express Dan (see Fig. S5 in the supplementary material). By contrast, Dan expression did not overlap with the superficially localized later-born Cas⁺ neurons. Because Dan expression overlaps with that of Hb in neuroblasts and post-mitotic neurons, we conclude that Dan and Danr are likely to act indirectly, rather than by direct repression, to limit Hb expression in neuroblasts.

The difference in Dan and Svp neuroblast expression patterns suggests that they have very different functions in specifying temporal identity. Here, we test whether Dan and Svp act in a linear pathway or parallel pathways to limit the number of early-born neurons. We generated a dan danr svp triple mutant and observed an additive phenotype that showed significantly more early-born neurons in the NB7-1 lineage than that observed in either dan danr or svp mutants alone. In the wild-type NB7-1 lineage, there were 2.0 Hb⁺, early-born neurons. In the dan danr mutant there was an average of 2.4 early-born neurons, in the svp mutant there was an average of 5.6 early-born neurons, whereas in dan danr svp triple mutants there were 6.9 of these neurons (Fig. 6A). In addition, Dan expression was not lost in svp^e22 mutants, and Svp expression was not lost in dan danr mutants (Fig. 6B). We conclude that Dan/Danr and Svp function in independent pathways that are both required for the timely downregulation of Hb in neuroblasts.

Dan and Hb are coexpressed in the neuroepithelia just prior to neuroblast delamination, and Svp expression is observed immediately following Hb in neuroblasts, raising the possibility that Hb might activate Dan, Danr and/or Svp. We observed no change in Dan expression in hb mutants in both young neuroblasts (see Fig. S6B in the supplementary material) and in postmitotic neurons (see Fig. S6E in the supplementary material). However, at the stage when Dan is normally no longer expressed by neuroblasts (stage 13; see Fig. S7, top left panel and Fig. S7B in the supplementary material) dan transcription could be strongly upregulated following Hb overexpression in all neuroblasts (Fig. 7A′) or a subset of neuroblasts (Fig. 7C). Dan expression was also activated by Hb misexpression in neuroectoderm, although this was only detectable before stage 12 (data not shown). By contrast, Hb did not activate Dan expression in the dorsal non-neural ectoderm or in any epithelial cells past stage 12 (Fig. 7C, arrows), suggesting Hb activation of Dan is dependent on cell type and developmental stage. Similarly, Hb could induce Svp expression in neuroblasts (compare Fig. S7 in the supplementary material, top right panel and Fig. 7D with 7E), but not in neuroectoderm or epidermis (Fig. 7E, arrows). Thus, Hb can induce neuroblast expression of both Dan and Svp, suggesting that Hb itself participates in a feedback loop to help determine the precise timing with which the neuroblast generates later-born fates.

Here, we show that Dan and Danr are required to limit Hb expression in neuroblasts and restrict the number of early-born neurons generated in multiple neuroblast lineages. The orphan nuclear hormone receptor protein Svp also functions to limit Hb expression in neuroblasts, and our data strongly suggest that Dan/Danr and Svp function in parallel pathways that are each independently required. First, the temporal expression patterns of Dan and Danr versus Svp do not suggest their coordinated activity: Dan and Danr are expressed from the time of neuroblast formation (stage 9), beyond Hb downregulation (stage 10), until the time of strong Castor expression (stage 12). By contrast, Svp protein is very transiently detected in neuroblasts only at the onset of Hb downregulation. Second, Dan/Danr and Svp are not in a linear transcriptional hierarchy: neither mutant affects expression of the other gene. Third, dan danr and svp mutants have distinct phenotypes: for example, compared with the dan danr mutant, the svp mutant has many more early-born neurons in the NB7-1 lineage, whereas it does not have any extra early-born neurons in the NB1-1 lineage. Fourth, the dan danr svp null triple mutant has the summed phenotypes of the dan danr double mutant and the svp single mutant. Fifth, misexpression of Svp, but not Dan, can repress hb transcription in neuroblasts (Kanai et al., 2005) (data not shown). The fact that neither one appears to have an effect on cell fate when misexpressed in postmitotic neurons suggest that both Svp and Dan function at the level of the mitotic precursors. Taken together, it appears that Dan/Danr and Svp are each required to downregulate hb expression in neuroblasts, but do so using separate mechanisms. The data are consistent with Svp directly repressing neuroblast hb transcription (although this has not been shown) whereas Dan and Danr act more indirectly.

Do Dan, Danr and Svp have lineage-specific functions? Despite the widespread expression of Dan and Danr in early neuroblasts (Fig. 5; see Fig. S1 in the supplementary material) and the widespread transient expression of Svp in most neuroblasts (Broadus et al., 1995; Kanai et al., 2005) (Fig. 5), it is likely that each has lineage-specific functions. For example, in the NB1-1 lineage, ectopic early-born neurons are generated in dan danr mutants, but not in svp mutants. Further comparing Dan versus Danr in this lineage, it appears that Danr is more important than Dan, because the danr^ex35 single mutant phenocopies the dan danr^ex56 double mutant in the number of ectopic aCC/pCC neurons generated and the number of hemisegments affected per embryo. In contrast to the NB1-1 lineage, Dan and Danr each appear to be required for limiting the number of early-born neurons in the NB7-1 lineage, as danr^ex35 single mutants had a weaker phenotype than the dan danr double mutant. Additionally, there are more early-born neurons in the NB7-1 lineage in svp mutants than in dan danr mutants, highlighting their lineage-specific differences. These differences might be due to different levels or functions of each protein in distinct neuroblasts. For example, there is variability in dan and danr mRNA levels between neuroblasts (see Fig. S1 in the supplementary material), suggesting that distinct neuroblasts might have different levels of Dan and/or Danr protein (although Dan protein levels appear constant between newly formed neuroblasts), or that they express Dan and/or Danr protein for different durations. Alternatively, or in addition, the lineage-specific variation might be due to unique cofactors present in different neuroblasts. This seems likely, as Hb misexpression in all neuroblasts has varying effects within different lineages. For example, NB1-1 generates only one to three ectopic early-born neurons in response to Hb misexpression, whereas NB7-1 generates ~20 ectopic early-born neurons (Isshiki et al., 2001; Pearson and Doe, 2003). Consistent with the notion that co-factors can alter the functional output of transcriptional regulators, recent evidence shows that the co-regulator CtBP forms complexes with distinct eye specification factors, including Dan and Danr, to regulate proliferation versus differentiation during eye development in Drosophila (Hoang et al., 2010).

Do Dan and Danr function redundantly? We are unable to rigorously test this hypothesis owing to the lack of a dan null single mutant, but the available evidence suggests that they do have redundant functions. First, they have nearly identical expression patterns. But most crucially, overexpression of Dan in the dan danr double mutant can nearly completely rescue the CNS phenotype (see Fig. S2 in the supplementary material), suggesting that high enough levels of Dan can compensate for loss of Danr. However, a danr null single mutant shows a strong phenotype in the NB1-1 lineage and a partial phenotype in the NB7-1 lineage, suggesting that endogenous levels of Dan are insufficient for normal CNS development. The most parsimonious explanation is that each protein has equivalent function, but that both genes are required to generate sufficient levels of Dan/Danr protein.

We have observed that Hb overexpression can activate expression of both Dan and Svp. What is the significance of this activation? Previous work has shown that Hb can function both as a transcriptional activator and a repressor. Although its repressive functions are required for the neuroblast to specify early-born fates and maintain neuroblast competence (Tran et al., 2010), its activator functions remain elusive. One possibility is that Hb-mediated activation of Svp, and the subsequent Svp-mediated downregulation of Hb, create a negative feedback loop to ensure timely progression of the neuroblast to later temporal fates. This is not unlike what has been observed for Cas, which activates feed-forward and feed-back transcriptional cascades to regulate temporal identity in the NB5-6 lineage (Baumgardt et al., 2009). By contrast, Hb activation of Dan expression might be part of the mechanism by which Hb maintains neuroblast competence, because Dan is unlikely to repress hb expression directly (see below).

What might be the mechanism by which Dan and Danr function to restrict the duration of Hb expression in neuroblasts? Some clues might come from the fact that Dan and Danr are found in a subgroup of pipsqueak-domain containing nuclear proteins that have been proposed to regulate higher order chromatin structure by targeting distal DNA elements (Siegmund and Lehmann, 2002). Pipsqueak, the founding member of the family, has been shown to recruit Polycomb group complexes to specific regions of the genome to mediate gene silencing (Huang et al., 2002). Perhaps Dan and Danr modify chromatin structure through recruitment of chromatin remodeling complexes, which indirectly affects hb transcription by changing the accessibility of the hb locus to other transcriptional regulators (Fig. 8). Such a function in modulating chromatin architecture might not be restricted to regulating just hb expression, but can extend to other temporal identity factors as well. Indeed, in NB7-1, the initial Hb→Cas ‘competence window’, during which the U1-U5 motor neurons are generated (Pearson and Doe, 2003; Cleary et al., 2006), matches nearly exactly the window of Dan and Danr expression. This raises the possibility that Dan and Danr might have a more global role in NB temporal progression by stabilizing ‘transition states’ between successive temporal identity factors (e.g. Hb→Kr, Kr→Pdm or Pdm→Cas). Such a function might explain the low frequency misregulation of later-born neuron numbers in several lineages (7-1, 3-1, 7-3), in addition to the extra early-born neurons phenotypes. Future experiments that address the role of Dan and Danr in later temporal identity transitions will provide a better understanding of the mechanisms that control the progression of temporal identity in neuroblasts.

We thank Sen-Lin Lai, Clemens Cabernard, Khoa Tran, Chong Liu and Rui Galvão for helpful discussions and critical reading of the manuscript; Stephen Cohen for providing fly stocks; and Jennifer Curtiss (NMSU), Ward Odenwald (NIH), Takako Isshiki (NIG) and John Reinitz (EADC) for antibodies. This work was supported by the Leukemia Lymphoma Society (M.K.) and NIH grant HD27056 to C.Q.D, who is an investigator of the Howard Hughes Medical Institute. Deposited in PMC for release after 6 months.