The labial gene is required to terminate proliferation of identified neuroblasts in postembryonic development of the Drosophila brain

The neural cells of the Drosophila central brain develop from a set of approximately 100 neural-stem-cell-like neuroblasts which derive from the cephalic neuroectoderm in the early embryo (reviewed by Urbach and Technau, 2004; Technau et al., 2006; Hartenstein et al., 2008). During embryogenesis, these neuroblasts divide in an asymmetric manner to self-renew and produce ganglion mother cells which generally give rise two postmitotic neural progeny (reviewed by Skeath and Thor, 2003; Doe, 2008; Knoblich, 2008). This initial phase of embryonic neurogenesis gives rise to the functional brain of the Drosophila larva. Towards the end of embryogenesis, most neuroblasts enter a reversible cell cycle arrest called quiescence, which separates the initial phase from the subsequent secondary phase of neurogenesis (Isshiki et al., 2001; Tsuji et al., 2008; Egger et al., 2008). In response to intrinsic and extrinsic factors involving nutritionally activated mitogens and glial cell-dependent interactions, neuroblasts resume proliferation during early larval stages (Chell and Brand, 2010; Sousa-Nunes et al., 2011). During this postembryonic phase of neurogenesis the majority of the adult-specific neurons of the brain are generated (Truman and Bate, 1988; Prokop and Technau, 1991). The adult-specific neural cells produced postembryonically by each individual neuroblast form a lineage-related cluster of immature neurons which differentiate in the pupal phase and contribute to the functional adult brain circuits (Truman et al., 2004; Pereanu and Hartenstein, 2006; Hartenstein et al., 2008).

Timely, precise and irreversible termination of postembryonic neuroblast proliferation is crucial to ensure that the correct number of neural progeny is generated and to avoid the danger of uncontrolled overgrowth (reviewed by Neumüller and Knoblich, 2009; Weng et al., 2010; Knoblich, 2010; Reichert, 2011). This process varies in temporal and spatial respects in the developing brain, but is largely finished by the end of metamorphosis as no identifiable neuroblasts are present at adult stages (Truman and Bate, 1988; Ito and Hotta, 1992). For most of the neuroblasts of the central brain and thoracic ganglia, termination of proliferation is achieved by series of cellular adjustments, involving shrinkage, lengthening of the cell cycle, expression of nuclear prospero and then cell cycle exit via a symmetric final division (Maurange et al., 2008). In contrast, for neuroblasts in the abdominal ganglia, which cease dividing in larval stages, termination of proliferation involves another mechanism, namely induction of programmed cell death in neuroblasts through expression of Hox gene-encoded transcription factors (reviewed by Pearson et al., 2005; Rogulja-Ortmann and Technau, 2008; Miguel-Aliaga and Thor, 2009; Sousa-Nunes et al., 2010). More specifically, in all neuroblasts of the central abdomen, the Hox gene abdominal-A (abd-A) is expressed in a short pulse during larval development in order to trigger programmed cell death (Bello et al., 2003). This ability of Hox genes to trigger programmed cell death in the abdominal ganglia is tightly regulated by epigenetic mechanisms involving the Polycomb group of genes (Bello et al., 2007).

Hox genes have also been shown to act in the development of the central brain in Drosophila, and notably for the Hox gene labial, loss-of-function has been associated with severe pattering defects in embryonic brain development (Diederich et al., 1989; reviewed by Lichtneckert and Reichert, 2008; Reichert and Bello, 2010). During embryogenesis, labial is expressed throughout the tritocerebrum anlage; all thirteen neuroblasts of the tritocerebrum as well as two neuroblasts of the deutocerebrum are Labial-positive (Younossi-Hartenstein et al., 1996; Urbach and Technau, 2003). If labial is inactivated, postmitotic cells are generated; however, they do not extend neurites and lack the expression of neuronal markers, indicating that labial is required to establish neuronal identity in the embryonic tritocerebrum (Hirth et al., 1998). Interestingly, these defects can be rescued by targeted misexpression by all Hox genes except Abd-B (Hirth et al., 2001). Moreover, expression of labial in the tritocerebrum can be subject to cross-regulatory interactions among Hox proteins during embryonic brain development (Sprecher et al., 2004).

In contrast to the extensive information on the role of the labial gene in embryonic brain (tritocerebrum) development, virtually nothing is known about the expression and function of labial, or any other Hox gene, in postembryonic brain development of Drosophila. Hence, it is unclear if Hox genes have any influence on the development of the adult-specific, secondary neurons that make up the bulk of the neuronal circuitry in the adult brain. Here we show that the Hox gene labial is expressed in late L3 larval stage brain in four neuroblasts that give rise to the identified labial expressing neuroblast lineages BAlp4, BAlv, TRdm and TRdl. Moreover, we demonstrate that two additional labial expressing neuroblasts are present in the late L2 stage – but not in the early L3 stage. Remarkably, while MARCM-based clonal mutation of labial in the BAlp4, BAlv, TRdm and TRdl neuroblast lineages does not result in any obvious mutant phenotype, a striking effect of clonal labial loss-of-function does occur, namely the formation of two ectopic neuroblast lineages that are not present in wildtype brains. These two ectopic neuroblast lineages are also observed following MARCM-based block of cell death and, significantly, these ectopic lineages are Labial-positive. Since both clonal cell death block and clonal labial inactivation result in the formation of the same two ectopic neuroblast lineages, these findings imply that labial is required in these two postembryonic brain neuroblast lineages for termination of proliferation through programmed cell death. This analysis of labial function reveals a novel cell autonomous role of a Hox gene in shaping the lineage architecture of the brain during postembryonic development.

Unless otherwise stated fly stocks were obtained from the Bloomington Stock center. Wildtype was w¹¹¹⁸. Gal4-Lines that were used: GAL4^Mz1407, UAS-mCD8::GFP^LL5 (Luo et al., 1994; Betschinger et al., 2006) (B.C.B., unpublished); w; wor-GAL4, UAS-mCD8::GFP^LL5/CyO, ActGFP^JMR1 (Albertson et al., 2004). One recombinant chromosome was constructed: FRT82B, lab¹⁴. For labial mutant and wildtype MARCM analysis (Lee and Luo, 1999; Lee and Luo, 2001), FRT82B, lab¹⁴/TM3, actGFP^JMR2; or FRT82B males were crossed to y,w,hsFLP¹²²; tubP-GAL4, UAS-mCD8::GFP^LL5/CyO, actGFP^JMR1; FRT82B, tubP-GAL80^LL5/(TM6,Tb,Hu) females (B.C.B., unpublished). For labial mutant “rescue” analysis UAS-labial; FRT82B, lab¹⁴/TM3, actGFP^JMR2 males were crossed to y,w,hsFLP¹²²; tubP-GAL4, UAS-mCD8::GFP^LL5/CyO, actGFP^JMR1; FRT82B, tubP-GAL80^LL5/(TM6,Tb,Hu) females. For H99 and wildtype control MARCM analysis, w; FRT2A, Df(3L)H99, kni^ri-1/TM6 [w+] or w; FRT2A were crossed to y,w,hsFLP¹²²; tubP-GAL4, UASmCD8::GFP^LL5/CyO; FRT2A,tubPGAL80^LL9/(TM6,Tb,Hu) females (B.C.B., unpublished). For labial misexpression analysis UAS-labial; FRT82B males were crossed to y,w,hsFLP¹²²; tubP-GAL4, UAS-mCD8::GFP^LL5/CyO, ActGFP^JMR1; FRT82B, tubP-GAL80^LL5/(TM6,Tb,Hu) females. For 24 hour ALH heatshock MARCM experiments, embryos were collected on standard medium over a 4 hour time window, raised at 25°C for 48 hours and then heat-shocked for 1 hour at 37°C. For embryonic heatshock MARCM experiments, embryos were collected for an 8 hour time window on standard medium at 18°C, raised at 18°C for 15 hours and afterwards heatshocked for 1 hour at 37°C. For RNAi experiments, UAS-labRNAi²⁹⁹⁰ (obtained from VDRC) was crossed to GAL4^Mz1407, UAS-mCD8::GFP^LL5 and wor-GAL4, UAS-mCD8::GFP^LL5/CyO, actGFP^JMR1. For p35 cell death block experiments, UAS-p35^BH2 was crossed to GAL4^Mz1407, UAS-mCD8::GFP^LL5.

Embryos were dechorionated, fixed and labeled according to standard protocols (Patel, 1994). Larval brains were fixed and immunostained as previously described (Bello et al., 2007). The following antibodies were used: guineapig-anti-Dpn (1:1000) (J. Skeath), mouse-anti-BP016 Neurotactin (1:20) (DSHB), mouse-anti-nc82 Bruchpilot (1:20) (DSHB), rabbit-anti-Labial (1:200) (F. Hirth and H.R., unpublished), rat-anti-Labial (1:200) (F. Hirth and H.R., unpublished). Alexa fluorescence-conjugated secondary antibodies (Molecular Probes) were used at 1:200.

All fluorescent images were recorded using a Leica SP confocal microscope. Optical sections range from 1–2 µm with a pictures size of 1024×1024 pixels. Collected images were arranged and processed using ImageJ, Adobe Photoshop and Adobe Illustrator. Cell counts were performed with the ImageJ plugin “cell counter”. For highlighting specific MARCM clones, cell bodies and neurites from other clones in the vicinity were removed in every single optical section. 3D models (superposition) were generated with the Fiji 3D viewer tool by selecting structures of interest such as MARCM clones or the neurite scaffold. Schemes were drawn in Adobe Illustrator.

To investigate the expression of the Hox gene labial in postembryonic brain development, we performed an immunocytochemical analysis of whole-mount brains of wandering third-instar (late L3 stage) larvae using a Labial-specific antibody (LAB) in combination with an anti-Bruchpilot (NC82) antibody to visualize neuropile structures. Expression of the labial gene was detected in two bilaterally symmetrical groups of cells located posterior to the antennal lobe and adjacent to the SOG (subesophageal ganglion) in the general region of the posterior central brain that corresponds to the developing tritocerebrum (Fig. 1). In confocal single optical sections, these labial expressing neural cells were observed in spatial association with a small number of secondary axon tracts labeled by MZ¹⁴⁰⁷-Gal4 suggesting that these neurons might correspond to a small set of neuroblast lineages (Luo et al., 1994; Betschinger et al., 2006). In addition to the labeled neuron groups, labial expression was also observed in four larger cells which co-expressed the marker deadpan (DPN) indicating that they were neuroblasts (San-Juán and Baonza, 2011). These four neuroblasts were also located in the same posterior central brain region and were invariably associated with the labial-expressing cell clusters (Fig. 2). Several other Deadpan-positive neuroblasts were located in the vicinity of these four labial-expressing neuroblasts, but none of these were Labial-positive.

To identify the postembryonic lineages generated by the four labial expressing neuroblasts, we performed a MARCM-based clonal analysis with an ubiquitous tub-Gal4 driving UAS-mCD8::GFP (Lee and Luo, 1999; Lee and Luo, 2001). Clones were induced at random at 24 hours after larval hatching (ALH) and recovered at the late L3 larval stage and, therefore, only secondary (adult-specific) lineages of individual neuroblasts were labeled. MARCM-labeled clones were co-labeled with the anti-Labial antibody and with an anti-Neurotactin (NRT; BP106) antibody that is specific for secondary lineages. We recovered four neuroanatomically distinct neuroblast lineages that had Labial-immunopositive neuroblasts.

For further identification of theses neuroblast lineages, we determined the projection patterns of each of their secondary axon tracts relative to the ensemble of secondary axon tracts in the late larval brain based on anti-NRT immunolabeling and compared these patterns to those documented in previous lineage mapping studies (Pereanu and Hartenstein, 2006; Spindler and Hartenstein, 2010). Since all of these neuroblast lineages had an invariant and unique projection pattern of their secondary axon tracts (SAT), we were able to unambiguously assign the four labial expressing neuroblast lineages to four previously identified postembryonic lineages, namely BAlp4, BAlv, TRdm and TRdl (Fig. 3).

For a more detailed characterization of these four identified neuroblast lineages, we performed cell counts on the corresponding Labial-immunolabeled MARCM clones. The BAlp4 lineage contained an average of 64 cells (s.d. = 8, n = 12) of which an average of 62 cells (s.d. = 7, n = 12) were Labial-immunopositive, the BAlv lineage contained an average of 81 cells (s.d. = 7, n = 6) of which an average of 19 cells (s.d. = 4, n = 6) were Labial-immunopositive, the TRdm lineage contained an average of 59 cells (s.d. = 6, n = 5) of which an average of 12 cells (s.d. = 8, n = 5) were Labial-immunopositive, and the TRdl lineage contained an average of 86 cells (s.d. = 4, n = 2) of which an average of 31 cells (s.d. = 3, n = 2) were Labial-immunopositive. This analysis indicates that the BAlp4 (basoanterior lineages, posterolateral subgroup) lineage expresses labial in most if not all cells while the BAlv (basoanterior lineages, ventrolateral subgroup), TRdm (dorsomedial tritocerebral lineage) and TRdl (dorsolateral tritocerebral lineage) lineages express labial only in subset of their cells including the neuroblast.

To investigate the role of labial in the development of the BAlp4, BAlv, TRdm and TRdl lineages, we compared the wildtype and labial mutant MARCM clones, induced at 24 hours ALH and recovered at late L3 larval stages, for each of these neuroblast lineages. Mutant GFP-labeled clones were homozygous for lab¹⁴, an embryonic lethal loss-of-function allele of labial (Merrill et al., 1989). All of the recovered labial mutant clones of the BAlp4, BAlv, TRdm and TRdl lineages were similar in their general neuroanatomical features to the respective wildtype clones. They all comprised a single large cell corresponding to the neuroblast as well as an associated cluster of labeled cells corresponding to the secondary neurons, and the secondary axon tracts formed by the secondary neurons had an appropriate, wildtype-like projection pattern in all cases (supplementary material Fig. S1).

To determine if the number of cells in the labial mutant clones was comparable to that of the corresponding wildtype clones, we performed cells counts for each of the four lineages. For all four lineages, the total cell number was not significantly different in wildtype versus labial mutant clones (supplementary material Fig. S1). Thus, average cell counts for wildtype versus mutant were 66 versus 66 (BAlp4), 70 versus 71 (BAlv), 65 versus 62 (TRdm) and 80 versus 88 (TRdl). We conclude that clonal mutation of labial does not alter cell number and secondary axon tract projection in the BAlp4, BAlv, TRdm and TRdl lineages.

In contrast to the lack of overt mutant phenotype in labial mutant BAlp4, BAlv, TRdm and TRdl lineages, a striking and unexpected effect of clonal labial loss-of-function was observed in the developing L3 larval brain, namely the formation of ectopic neuroblast lineages that were not present in wildtype brains. Ectopic lineages were recovered in about 50% of all brains containing randomly induced lab¹⁴ mutant clones. These ectopic neuroblast lineages could be unambiguously identified based on the projection patterns of their ectopic secondary axon tracts within the ensemble of secondary axon tracts of late larval brains (Fig. 4A,B). Morphologically they could be assigned to two different types, which we refer to as Ectopic1^lab (Ect1^lab) and Ectopic2^lab (Ect2^lab) lineages. Ect1^lab was located between the BAlp4 and the BAlv lineages, had an average cell number of 107 cells (s.d. = 24, n = 3) and formed several secondary axon tract projections, of which one always followed an axon tract of the BAlc lineage (Fig. 4C–D′). Ect2^lab was located close to the TRdm and TRdl lineage, had an average cell number of 25 cells (s.d. = 1, n = 3) and projected its secondary axon tract towards the midline (Fig. 4E–F′). These ectopic lineages were only seen in the late larval (L3) brain. Moreover, they were never observed in MARCM-based genetic rescue experiments (clone induction: 24 hours ALH) in which a UAS-labial transgene under the control of the tub-GAL4 driver was used to express the labial gene in labial loss-of-function mutant clones (n = 16).

To further confirm that the formation of ectopic neuroblast lineages was indeed due to labial loss-of-function, we performed genetic knockdown experiments in which worniu-Gal4 and MZ¹⁴⁰⁷-Gal4 were used to drive UAS-labRNAi²⁹⁹⁰ in all developing neuroblasts (Albertson et al., 2004). Ectopic neuroblast lineages comparable to those induced by lab¹⁴ mutant clones resulted (Fig. 5A,B). These ectopic lineages were recovered in 50% of the late larval brains for the worniu-Gal4 driver (n = 23) and in 20% of the late L3 larval brains for the MZ¹⁴⁰⁷-Gal4 driver (n = 14). In accordance with the lab¹⁴ mutant clonal analysis (loss-of-function and genetic rescue), these findings indicate that the appearance of ectopic lineages is a labial-specific loss-of-function effect. Moreover, since the targeted knockdown of labial driven by worniu-Gal4 and MZ¹⁴⁰⁷Gal4 is largely neuroblast-specific, these findings also suggest that the ectopic lineage phenotype was due to the absence of Labial protein in the neuroblasts themselves rather than in their neural cell progeny. This assumption is supported by the observation that both types of ectopic lineages recovered in lab¹⁴ clonal MARCM experiments invariably contained a single large Deadpan-positive neuroblast (Fig. 5C–D′).

The cell-autonomous induction of MARCM-based mutant neuroblast clones is only possible in mitotically active progenitor cells (Lee and Luo, 1999; Lee and Luo, 2001). This implies that additional labial-expressing neuroblasts must be present and mitotically active during early larval brain development when the lab¹⁴ mutant ectopic clones were induced. To investigate this, we determined the number of labial-expressing neuroblasts in the wildtype second larval instar stage (L2) by double immunolabeling with anti-Labial and anti-Deadpan. These experiments revealed the presence of six double-labeled cells indicating that six labial-expressing neuroblasts are present at the L2 stage (Fig. 6A–H″). These six neuroblasts were arranged in the L2 brain in a spatial pattern which is comparable to that of the BAlp4, BAlv, TRdm, TRdl, Ect1^lab and Ect2^lab neuroblasts in the labial-mutant late larval brain. Interestingly, an average of six labial-expressing neuroblasts were also present in the late embryonic brain implying that the number of labial-expressing neuroblasts does not change from the late embryonic stage to the second larval instar stage (supplementary material Fig. S2). Given that only four labial-expressing neuroblasts (the BAlp4, BAlv, TRdm, and TRdl neuroblasts) are present in the wildtype late L3 larval brain, these findings suggests that two of the six neuroblasts present in the L2 larval stage are missing in the L3 stage. In accordance with this assumption, double immunolabeling experiments with anti-Labial and anti-Deadpan at the early L3 stage (immediately after the L2/L3 molt) revealed only four labial-expressing neuroblasts, and these were arranged in a spatial pattern corresponding to the BAlp4, BAlv, TRdm, TRdl neuroblasts characterized above in late (wandering) L3 larval stages (Fig. 6I–O″).

What is the fate of the two labial-expressing neuroblasts that are present in L2 but are no longer observed in L3 wildtype larval brains? While it is conceivable that these two neuroblasts are still present in L3 but have terminated their proliferative activity and at the same time ceased to express labial, a simpler explanation is that they are eliminated by programmed cell death at late larval stages. To investigate this possibility, we performed a MARCM clonal analysis of neuroblast lineages in the general region of the developing tritocerebral region using H99, a deficiency removing the proapoptotic genes reaper, grim and head involution defective, in an otherwise wildtype background (White et al., 1994). Homozygous H99 mutant clones were induced at 24 hours ALH and recovered in late L3 larval brains. In these experiments, a number of supernumerary ectopic lineages were observed in the corresponding region linking the central brain and the SOG. Among these, we consistently recovered two ectopic lineages that were comparable in terms of location and secondary axon tract projection pattern to the Ect1^lab and Ect2^lab lineages recovered in the clonal lab¹⁴ mutant assays. We refer to these lineages as Ectopic1^H99 (Ect1^H99) and Ectopic2^H99 (Ect2^H99). Ect1^H99 was located between the BAlp4 and the BAlv lineage, had several secondary axon tracts of which one always projected in a straight medial direction and manifested an average cell number of 80 cells (s.d. = 11, n = 9) of which an average of 70 (s.d. = 12, n = 9) were Labial-positive (Fig. 7A,C–D′). Ect2^H99 was located posterior-laterally to the TRdm and TRdl lineage, extended several secondary axon tracts that projected medially and had an average cell number of 62 cells (s.d. = 8, n = 12) of which an average of 31 were Labial-positive (Fig. 7B,E–F′). Importantly, both ectopic lineages, Ect1^H99 and Ect2^H99, also consistently expressed labial in their neuroblast of origin (Fig. 7G–H′). Comparable results were obtained by targeted apoptosis block in experiments in which MZ¹⁴⁰⁷-Gal4 was used to drive UAS-p35^BH2 in larval brain neuroblasts; ectopic labial-expressing neuroblast lineages that strongly resemble Ect1^H99 and Ect2^H99 in terms of location and secondary axon tract projection pattern were observed (data not shown).

These findings indicate that two of the six labial-expressing neuroblasts present in early larval brain development are eliminated by programmed cell death in the late larval brain. Moreover, they indicate that blocking programmed cell death results in two (labial-expressing) ectopic neuroblast lineages which are comparable in neuroanatomical terms to the two ectopic neuroblast lineages recovered in labial loss-of-function mutant neuroblast clones. This in turn implies that labial is required cell autonomously in these two neuroblast lineages to terminate their proliferation through programmed cell death during late larval development.

Previous studies of Hox gene action in ventral nerve cord development have shown that the Hox genes Antp, Ubx and abd-A are able to trigger programmed cell death in neuroblasts in which they are not normally expressed (Bello et. al., 2003). To determine if the Hox gene labial is also able to induce programmed cell death in central brain neuroblast lineages other than Ect1 and Ect2, we performed a clonal MARCM misexpression assay of labial. GFP-labeled labial mutant clones were induced at embryonic stage 12–15, recovered in late L3 larval brains, and were co-labeled with the neuroblast marker anti-Deadpan. No effect of labial misexpression on neuroblast survival was observed in the following lineages of the central brain: TRvl (ventrolateral tritocerebral lineage) (n = 11), BAlp2 (n = 7), BAlp3 (n = 4), BAmv1 (n = 9), BAmv2 (n = 3), BAmas1 (n = 3), BAmas2 (n = 2), PG5 (n = 7). Similarly, no effect of labial misexpression on neuroblast survival was seen in the labial expressing lineages BAlp4 (n = 9), BAlv (n = 8), TRdl (n = 8). Thus, the ability of labial to terminate neuroblast survival is likely to be restricted to the two neuroblast lineages Ect1 and Ect2.

In the TRvm and PG5 lineages, labial misexpression did result in aberrant secondary axon tract projection patterns. The TRvm lineage normally projects posteriorly and its secondary axon tract terminates close to where the TRco lineage SAT forms a commissure. In the labial misexpression assay, the secondary axon tract of the TRvm lineage projects posteriorly but then turns laterally to terminate close to the secondary axon tract of the BAlv lineage (supplementary material Fig. S3A,B). The PG5 lineage is located medial to the BAla1–4 lineages and its secondary axon tract normally projects medially to terminate in between the bifurcating secondary axon tract of the TRdl lineage. (The PG5 lineage has not been included in previous mapping studies (Pereanu and Hartenstein, 2006; Spindler and Hartenstein, 2010).) In the labial misexpression assay, the PG5 lineage secondary axon tract projects medially, but then turns posteriorly to terminate close to the secondary axon tract of the BAlv lineage (supplementary material Fig. S3C,D). Thus, while labial misexpression does not appear to affect neuroblast survival, it can result in aberrant secondary axon tract projection patterns in central brain lineages during postembryonic development.

Our findings on the role of labial in postembryonic brain development are in accordance with a model in which labial is cell autonomously required for the stage-specific programmed cell death of two of the six postembryonic neuroblasts that express labial during larval stages (Fig. 8). This model is supported by expression studies which indicate that six labial-expressing neuroblasts are present in the developing brain at the late embryonic stage (stage 17) and at the end of the second larval stage (L2), while only four labial-expressing neuroblasts continue to be present during the third larval stage (L3). Moreover, the model is supported by functional studies which indicate that this reduction in labial-expressing neuroblast number is due to Labial-dependent programmed cell death of two neuroblasts during postembryonic brain development, since clonal labial loss-of-function leads to the recovery of two ectopic neuroblast lineages and these two ectopic neuroblast lineages are also recovered (and express labial) following clonal cell death block.

Given that six Labial-positive neuroblasts are present at late L2 stages, and only four Labial-positive neuroblasts are present at early L3 stages, we posit that the labial-dependent apoptosis of the two affected neuroblasts (Ect1, Ect2) is associated with the L2/L3 transition. Since that the L2/L3 transition involves molting that is associated with elevated levels of steroid hormones such as ecdysone, it is possible that ecdysis-triggering endocrine signals participate in the labial-dependent apoptotic event in Ect1 and Ect2 (reviewed by Truman, 2005). Alternatively, transiently expressed temporal transcription factors might regulate the competence of the affected neuroblasts to undergo apoptosis in a labial-dependent manner (Maurange et al., 2008; Chell and Brand, 2008). While the molecular nature of these signals is currently not known, they are apparently not sufficient to elicit programmed cell death in all labial-expressing neuroblasts, since the labial-expressing BAlp4, BAlv, TRdm, and TRdl neuroblasts are not affected.

The neuroblast-specific requirement of the Hox gene labial in programmed cell death during postembryonic brain development reported here is novel and differs in several respects from the type of Hox-gene dependent programmed cell death that occurs in the abdominal ganglia during postembryonic development of the ventral nerve cord (Bello et al., 2003). In the larval abdominal ganglia, the Hox gene abd-A is expressed in a short pulse during the mid-L3 stage and results in the cell autonomous programmed cell death of all neuroblasts that express the abd-A pulse. In contrast, in the larval brain, the Hox gene labial is expressed during early larval development in six larval neuroblasts and this only results in the cell autonomous programmed cell death in two of these neuroblasts around the L2/L3 transition. Moreover, in contrast to the general apoptotic effect of clonal misexpression of abd-A (as well as Antp or Ubx) in larval neuroblasts of the ventral nerve cord as reported by Bello and coworkers (Bello et al., 2003), our studies indicate that the clonal misexpression of labial in larval neuroblasts of the brain does not result in apoptosis. Misexpression of labial does, however, result in axonal projections defects in central brain lineages. Interestingly, genetic misexpression of vertebrate Hox genes, including misexpression of the labial ortholog Hoxb1, has been shown to result in axonal projection defects of developing motoneurons (reviewed by Butler and Tear, 2007; Guthrie, 2007).

The role of labial in terminating proliferation in specific brain neuroblasts during postembryonic development is strikingly different from the function of this Hox gene during embryonic brain development in Drosophila (Hirth et al., 1998; Reichert and Bello, 2010). During embryogenesis, labial is expressed in all tritocerebral neuroblasts and their neural progeny, and functional inactivation of labial does not terminate neuroblast proliferation since postmitotic cells are generated in the mutant domain. However, the generated cells do not express neuronal markers and do not manifest neuronal morphology implying that labial is required to establish regional neuronal identity in the embryonic tritocerebrum. Interestingly, there is a marked decrease in the number of neuroblasts that express labial during embryonic development. At embryonic stage 11, labial is expressed in fifteen neuroblasts of which thirteen are of tritocerebral and two of deutocerebral origin (Urbach and Technau, 2003). In contrast, at the end of embryogenesis (stage 17) only six labial-expressing neuroblasts were present in the developing brain. The fate of the remaining nine embryonic neuroblasts is not known. They may simply cease to express labial and remain present or they may terminate proliferation via cell cycle exit or apoptosis as it has been reported for neuroblasts in the embryonic ventral nerve cord (Abrams et al., 1993; White et al., 1994; Peterson et al., 2002).

The appearance of ectopic neuroblast lineages in the absence of labial during postembryonic brain development is remarkable in several respects. First, the ectopic lineages are identifiable. Only two, morphologically distinct and unique ectopic neuroblast lineages are recovered, and each for each of these, Ect1 and Ect2, neuroanatomical features such as cell number and secondary axon tract projection are reproducibly constant at the end of larval development. Other types of ectopic neuroblast lineages or lineages with variable morphologies were not observed. Second, the ectopic lineages are novel and do not represent “homeotic” transformations into any other wildtype lineages. Notably they form secondary axon tract projections that differ significantly from any other secondary axon tract projection patterns in the larval brain. Nevertheless the ectopic lineages did not distort the other surrounding neuroblast lineages; their ectopic secondary axon tracts integrated into the ensemble of secondary axon tracts of late larval brains in an orderly manner. Third, the existence of ectopic neural lineages in the labial-mutant fly brain bears striking similarities to the ectopic neural assembly formation observed in a study of Hoxa1 mutant mice (Domínguez del Toro et al., 2001). In contrast to previous analyses of mouse Hoxa1 mutants focused on early effects on segmentation and patterning in the developing hindbrain (reviewed by Favier and Dollé, 1997; Lumsden and Krumlauf, 1996), this study shows that during later embryonic development, ectopic groups of neurons in the hindbrain of Hoxa1 mutants derive from ectopic mutant progenitors and establish a supernumerary neuronal assembly that escapes apoptosis and even becomes functional postnatally. Thus, the labial/Hoxa1 gene orthologs in fly and mouse appear to have remarkably similar dual roles in brain development. During early phases of brain development the labial/Hoxa1 genes act in establishing the regional identity of neurons in specific brain neuromeres; during later phases they prevent the formation of ectopic neuronal arrays in these brain neuromeres by terminating progenitor proliferation, thus, effectively sculpting the developing brain.

The observation that brain development in flies and mammals involves not just one but two different functional roles of labial/Hoxa1 genes, both of which appear to be evolutionarily conserved, provides additional support for the notion that comparable and conserved mechanisms operate in brain development of invertebrates and vertebrates (Reichert and Simeone, 1999; Lichtneckert and Reichert, 2005; Lichtneckert and Reichert, 2008). If this is indeed the case then a common and general strategy for generating novel functional features in brain development in bilaterian animals might be based on local changes in the regulation of labial/Hoxa1 (and perhaps other Hox genes), which could result in the evolution of novel neuronal subsets without affecting the function of the neural circuitry already present (Brunet and Ghysen, 1999).

We thank Markus Affolter, Jan Pielage and Simon G. Sprecher for discussions, Volker Hartenstein for helping with lineage identification and Susanne Flister for technical help. Supported by the Swiss NSF.