Mechanisms of gap gene expression canalization in the Drosophila blastoderm

Extensive variation in early gap gene expression in the Drosophila blastoderm is reduced over time because of gap gene cross regulation. This phenomenon is a manifestation of canalization, the ability of an organism to produce a consistent phenotype despite variations in genotype or environment. The canalization of gap gene expression can be understood as arising from the actions of attractors in the gap gene dynamical system. In order to better understand the processes of developmental robustness and canalization in the early Drosophila embryo, this study investigated the dynamical effects of varying spatial profiles of Bicoid protein concentration on the formation of the expression border of the gap gene hunchback. At several positions on the anterior-posterior axis of the embryo, attractors and their basins of attraction were analyzed in a dynamical model describing expression of four gap genes with the Bicoid concentration profile accounted as a given input in the model equations. This model was tested against a family of Bicoid gradients obtained from individual embryos. These gradients were normalized by two independent methods, which are based on distinct biological hypotheses and provide different magnitudes for Bicoid spatial variability. It was shown how the border formation is dictated by the biological initial conditions (the concentration gradient of maternal Hunchback protein) being attracted to specific attracting sets in a local vicinity of the border. Different types of these attracting sets (point attractors or one dimensional attracting manifolds) define several possible mechanisms of border formation. The hunchback border formation is associated with intersection of the spatial gradient of the maternal Hunchback protein and a boundary between the attraction basins of two different point attractors. How the positional variability for hunchback is related to the corresponding variability of the basin boundaries was demonstrated. The observed reduction in variability of the hunchback gene expression can be accounted for by specific geometrical properties of the basin boundaries. The mechanisms of gap gene expression canalization in early Drosophila embryos were clarified. These mechanisms were specified in the case of hunchback in well defined terms of the dynamical system theory (Gursky, 2011).

This study presents the dynamical analysis of the simplified model of the gap gene network on the ensemble of early Drosophila embryos. The main goal was to decode the mechanistic basis of the gap gene border formation and stability under the Bcd morphogen variance. The hb border formation mechanisms were described in terms of attracting sets and their attraction basins calculated in the nuclei surrounding the border position (Gursky, 2011).

The results reveal that the border formation can be associated with the event of intersection between a boundary separating the attraction basins of two di®erent point attractors and the initial Hb profile presenting the input from the maternally expressed hb gene. Attracting sets of another type, the unstable manifolds of saddle equilibria, actively participate in the adjustment of the border position. They do so by attracting the solution trajectories in the nuclei surrounding this position. The model predicts that these attracting manifolds can be involved in the border formation for some Bcd profiles (Gursky, 2011).

The hb border correctly forms in the model by the onset of gastrulation for all individual Bcd profiles. For about a half of these profiles, however, the Kr and Gt patterns in the solutions exhibit defects in the anterior part of the spatial domain (solution classes II and III). It turns out that the hb border formation mechanism involving the attracting manifolds is mostly associated with these cases. This may lead to the conclusion about restricted applicability of this mechanism in the case of hb expression. However, this mechanism exists and plays an important role for the gap domain borders in a posterior part of the embryo, where the domains form and vary in time under the control of an unstable manifold. To analyze canalization for the posterior borders, the variation for external inputs from Cad and Tll should be taken into account, where these transcription factors are among the key regulators, and a modified model should be considered including an input from the terminal gene huckebein (Gursky, 2011).

As previously reported, the model exhibits a significant filtration (canalization) of the Bcd positional variability at the level of hb border formation. The results show how this filtration stems from the stable behavior of the attraction basin boundaries. Has been shown that the mutual regulatory repression between the gap genes accounts for the observed variance reduction, thus presenting a buffering mechanism for canalization. This buffering mechanism was translated to the level of attractors and their attraction basins. As the hb border position is well encoded by the intersection between the initial Hb profile and corresponding attraction basin boundaries, the stability of hb border predicted by the model can be explained by inspecting the geometrical properties of these attraction basins (Gursky, 2011).

From this inspection, the following two mechanisms responsible for the observed robustness can be elucidated. First, the initial Hb profile is a monotonously decreasing function of A/P position, while the basin boundary to be crossed is a monotonously increasing one, i.e., these curves have opposite dependencies on the A/P position. This purely geometrical fact evidently prescribes a smaller variation of the intersection point when the basin boundary is changing due to the variance of Bcd concentration, as opposed to the case if the curves would jointly rise or jointly fall along the A/P axis (Gursky, 2011).

The second mechanism is associated with the specific nonlinear form of the response curve. The gap gene cross regulation of hb bends the response line exhibited in absence of this regulation. This bending e®ectively reduces the Hb positional variance by about half. In terms of attractors, this bending is controlled by the fact that a basin boundary responsible for the hb border formation does not change monotonously, but oscillate in the state space with the changing Bcd profile (Gursky, 2011).

The results show that the full range of the hb positional variance is broken down into two almost equal parts, the anterior and posterior ones. These parts are associated with two families of the Bcd individual profiles (Family I and Family II, respectively) and two di®erent mechanisms of hb border formation. The Bcd profiles from Family I lead to the hb border formation as a switch from a hb/ON attractor in a hb-expressing nucleus to a hb/OFF attractor in a hb-nonexpressing nucleus, while for Family II the border forms with the help of an attracting invariant manifold in a hb-nonexpressing nucleus. Since the difference between the two families is in the amplitude of the Bcd profiles, it is concluded that Bcd profiles of high amplitude canalize by a dynamical mechanism different from those of lower amplitude. Each dynamical mechanism provides only half of the full variance for the hb border, but in two adjacent spatial domains. Therefore, the change of the dynamical mechanism that happens with rising Bcd amplitude e®ectively doubles the variance (Gursky, 2011).

The hb border positions from the more posterior range are placed posterior to the spatial position of a bifurcation annihilating attractor A3. This bifurcation position delimits the anterior and posterior dynamical regimes in the model. Therefore, the Bcd profiles from the second family shift the hb border to the posterior dynamical regime, which is characterized by an active role of an attracting invariant manifold in the pattern formation (Gursky, 2011).

The results indicate that the posterior range of hb positional variation is almost equal to the anterior one only due to smaller variation of the Bcd profiles in Family II compared to Family I. This suggests that the solutions in the anterior and posterior dynamical regimes have quite different sensitivity rates to variation of the Bcd concentration. For Family I, the standard deviation for the hb border position is 2.6 times less than for the Bcd threshold position, while it is only 1.4 times less in the case of Family II. This difference can be explained by an observation that Bcd profiles of higher amplitude correspond to the linear part of the response curve, and this is a consequence of specific regulatory interactions in the gap gene circuit as explained further (Gursky, 2011).

The model was used (1) to study the canalization mechanisms based on the assessment that the model provides one of the best spatio-temporal precision for the description of gap gene expression. This model is an approximation to a more general model of gene regulation, which should be grounded on the statistical-mechanical formalism. One possible limitation is the linear approximation for the argument of the nonlinear regulation function g. The canalization mechanisms described in terms of attractors and attraction basins generally depend on the structure of the model that predicts these attracting states. Therefore, an important direction for future investigations should be verification of the proposed mechanisms in a phase space of a more general model (Gursky, 2011).

The nonlinear nature of the Bcd readout by the gap gene circuit is clearly represented in a specific nonlinear form of the response curve showing the Bcd dependence of the hb border position in the model. The nonlinear part of the curve can be explained by the regulatory actions on hb from the other gap genes. In particular, a regulatory analysis in the full model revealed that the regulatory interactions between hb, gt, and Kr underlie the folding part of the response curve. The gap gene cross-regulation also participate in the linear parts of the response curve by tuning the incline of these parts (Gursky, 2011).

It was previously pointed out that the gt and Kr expression borders in the anterior part of the A/P axis show large variation in the model in response to Bcd variation because the model is missing some regulators in this part. For example, these gt and Kr borders are absent in the solutions from class III. This fact raises doubts on the specific folding part that the response curve exhibits in the middle range of the Bcd concentration values. On the other hand, the folding part exists only for the Bcd profiles associated with the solutions from class I, with all expression borders formed correctly, which means that an essential portion of the artificial variation of the gt and Kr borders can be excluded from the consideration without affecting the folding form of the curve (Gursky, 2011).

The model was investigated on the ensemble of Bcd profiles normalized by the alternative method, which provided lower Bcd variance. One used this method as an artificial limit case, in which the ensemble possessing minimal Bcd variance was dealt with, and it was applied for the crosschecking purposes (Gursky, 2011).

No essential discrepancy was found in the mechanisms of hb border formation and canalization for the two normalization methods. A distinct bifurcation structure in the model with the new parameter values does not lead to changes in the solutions during the biologically important time. The model preserves an attracting invariant manifold related to the posterior dynamical regime. The same border formation mechanisms appear except the one associated with the attractor/manifold transition. It is important that, even though the second family of Bcd profiles does not appear in the alternative normalization case, the invariant manifolds still play their role in adjusting the border position. The model also demonstrates an essentially nonlinear response curve for the hb border. Therefore, the conclusions formulated above are robust with respect to the choice of the normalization method, and, in more general terms, they should be valid for different estimates of the actual Bcd variance (Gursky, 2011).

This correspondence can be explained by the fact that the parameters A and l obtained for the alternatively normalized Bcd profiles form a subset in similar parameters obtained in the case of the basic normalization method. Roughly speaking, the alternatively normalized Bcd profiles can be associated with Family I. In particular, this means that the Bcd data rescaled according to the alternative algorithm support the conclusion formulated above about different dynamical mechanisms of canalization for Bcd profiles of different amplitude (Gursky, 2011).

There is an important issue concerning the comparison of the Bcd variance filtration rates. The calculations reveal that, for the basic normalization method, the Hb positional variation of 1.3%EL in the model output follows from the Bcd positional variation of 4.5%EL, thus implying that more than 70% of the positional variance has been filtrated. The same calculations for the alternative normalization method give the filtration rate of approximately 60%. Therefore, the filtration still happens in the model even if Bcd profiles are normalize according to the precisionist hypothesis. This result is quite expected since the reported dynamical mechanisms underlying the processing of the Bcd variation in the model are valid irrespective of the absolute variation range. Whatever actual variation the Bcd morphogen exhibits, the nonlinear model response translates it to a smaller variation of the target gene patterns (Gursky, 2011).

It is concluded that the formation of hb border is coded by the intersection between the maternal Hb gradient and a boundary between attraction basins in the gap gene dynamical system. Small positional variance for hb border can be explained by the geometrical properties of this basin boundary and its nonmonotonic dependence on the Bcd concentration. Main features of the phase portraits underlying the canalization mechanisms do not depend on the normalization method for Bcd (Gursky, 2011).

Earliest zygotic hb expression is in the anterior 45% of the embryo. Maternal hb expression forms an anterior to posterior gradient that disappears by cellularization (Tautz, 1987).

HB is required for the formation and segmentation of two regions of the Drosophila embryo: a broad anterior domain and a narrow posterior domain. Accumulation of HB transcript in the posterior of the embryo occurs in two phases: an initial cap covering the terminal 15% of the embryo followed by a stripe at the anterior edge of this region. Zygotic transcripts are absent from the central domain of Krüppel expression (Tautz, 1987 and Margolis, 1995).

Regulation of POU genes by castor and hunchback establishes layered compartments in the Drosophila CNS

The tightly choreographed neuroblast (NB) expressions of Hb, Pdm-1 and Pdm-2, and Castor suggest temporally integrated processes participate in their formation (see Lateral views of Drosophila CNS). Clonal analysis of ventral cord NB lineages has revealed that many early delaminating NBs produce lineages that span most of the ventral cord's dorsal/ventral axis. For example the NB5-2, one of the first NBs to delaminate, generates a ventral-dorsal column of 17 to 26 cells. The dynamics of Hb, Pdm-1 and Cas expression in NBs indicates that many of the early S1 and S2 delaminating NBs may sequentially express all three and thereby produce lineages spanning all three compartments. Two such ventral cord candidates are the early NB5-2s and NB7-4s. Shortly after their delamination, during early stage 9, they activate Hb expression, while later, after several rounds of GMC divisions, they activate Cas expression. The fact that NBs co-expressing Hb/Pdm-1 or Pdm-1/Cas have been detected, but never Hb/Cas, further suggests that at least some of the early NBs make the Hb->Pdm->Cas transition. However, not all NBs undergo these transitions. This is particularly evident in NBs that enter the proliferative zone during later delamination waves. For example, the first ventral cord NBs to express Cas, the S3 NB6-1s, activate Cas shortly after delaminating from the ectoderm and do not express Hb (Kambadur, 1998).

Programmed transformations in neuroblast gene expression during Drosophila CNS lineage development

During Drosophila embryonic CNS development, the sequential neuroblast (NB) expression of four proteins, Hunchback (Hb), Pou-homeodomain proteins 1 and 2 (referred to collectively as Pdm), and Castor (Cas), identifies a transcription factor network regulating the temporal development of all ganglia. The Zn-finger proteins Hb and Cas, acting as repressors, confine Pdm expression to a narrow intermediate temporal window; this results in the generation of three panneural domains whose cellular constituents are marked by expression of Hb, Pdm, or Cas. Seeking to identify the cellular mechanisms that generate these expression compartments, the lineage development of isolated NBs in culture was studied. The Hb, Pdm, and Cas expression domains are generated by transitions in NB gene expression that are followed by gene product perdurance within sequentially produced sublineages. These results also indicate that following Cas expression, many CNS NBs continue their asymmetric divisions and generate additional progeny, which can be identified by the expression of the bHLH transcription factor Grainyhead (Gh). Gh appears to be a terminal embryonic CNS lineage marker. Taken together, these studies indicate that once NBs initiate lineage development, no additional signaling between NBs and the neuroectoderm and/or mesoderm is required to trigger the temporal progression of Hb followed by Pdm and then Cas, and subsequently Gh expression during NB outgrowth (Brody, 2000).

Underpinning the formation of NB lineages are spatially and temporally regulated transcription factor networks that play pivotal roles in establishing the unique cellular identities of NBs and their progeny. Prior to NB delamination, during the initial specification of NBs, two spatially regulated transcription factor networks subdivide the ventral neuroectoderm along its anterior/posterior (A/P) x axis and dorsal/ventral (D/V) y axis. Later, during NB lineage development, at least one additional network, acting over several hours, gives rise to sequentially formed multilayered basal (inner or dorsal) to apical (outer or ventral) neuronal subpopulations. Along the basal/apical z axis, neuronal subpopulations in all ganglia can be identified by their expression of the transcription factors Hb, Pdm and Cas. Hb marks a deeper, basally distributed population of neurons that are born early, Cas marks a superficial, apically distributed population of neurons that are born late, and Pdm marks an intermediate population arrayed between the Hb- and the Cas-expressing cells. Both genetic and molecular analysis indicates that two Zn-finger proteins, Hb and Cas, act as repressors to silence pdm expression. By restricting pdm expression primarily to intermediate-born neuronal precursors these structurally different Zn-finger proteins help establish three pan-CNS neural subpopulations whose cellular constituents are marked by the expression of Hb, Pdm, or Cas (Brody, 2000).

To what extent are the z axis expression domains generated successively by invariant gene expression programs, maintained in different NBs, versus sequential gene expression programs within sublineages of single NBs? To better understand the nature of the temporal components regulating the CNS z axis network, attempts were made to determine if the sequential expression of Hb, Pdm-1, and Cas occurs during NB lineage development in vitro. In order to analyze the capacity of individual NBs to generate a full repertoire of Hb-, Pdm-, or Cas-expressing sublineages, overnight cultures were simultaneously immuno-stained for all three factors and the percentage of cells within a clone that were positive for each factor was subsequently determined. Not all clones contained cells expressing each of the transcription factors. The majority of clones containing Cas-expressing cells also contain additional NB descendants marked by the expression of Hb or Pdm-1. Triple-immunolabeling studies have revealed that clones expressing only Cas are the exception. Taken together the results indicate that many isolated S1 and S2 NBs, when maintained in culture, will generate neuronal descendants that are marked by Hb, Pdm, or Cas expression. Given that Hb and Cas are repressors of pdm gene NB expression, these observations also suggest that the overlapping Hb/Pdm and Pdm/Cas expressions, both in vivo and in culture represent transition states in NB gene expression. In other words, NBs undergo sequential transitions in gene expression, thus generating the multiple cell layers seen in vivo (Brody, 2000).

Triple-immunolabeling studies have revealed that many of the overnight NB clones contain a subset of cells that do not contain detectable levels of Hb, Pdm-1, or Cas. In many of these in vitro lineages the putative NB is also unstained. The bHLH transcription factor Gh is known to be expressed in CNS NBs but only after stage 14. In view of the late onset of Gh expression in NBs and the triple-staining results identifying cells in o/n clones that do not express Hb, Pdm-1, or Cas, it was hypothesized that these negative cells may represent an additional late NB expression window marked by Gh expression. To test this hypothesis, the spatial/temporal expression dynamics of Gh were compared to other members of the z axis network. Similar to its late activation during in vivo development, Gh expression was observed only in overnight cultures; when more than one Gh-positive cell was detected in a clone they were consistently found clustered together. Two-thirds of the Cas+ clones had at least one Gh+ cell and the average number of Gh+ cells in all clones was 2.3. Approximately 2/3 of the Gh+ clones also contained Hb-immunopositive cells. While no Hb-Gh coexpressing cells were observed, approximately 20% of the Gh+ cells also expressed Cas. Given the late onset of Gh expression in both the embryo and the cultured NB clones and the overlapping Cas and Gh expression, it is likely that Gh marks a fourth temporal window for NB transcription factor expression. In addition, because there was an average of more than one cell in an o/n clone that was immunopositive for Gh, it is likely that Gh is also expressed/maintained in a sublineage(s) born after the one marked by Cas expression (Brody, 2000).

The principle finding of this study is that built on top of the x and y axis neural identity systems is an additional temporal network that defines successive stages of lineage maturation in an apical/basal z axis. This global CNS network, identified by the temporal cascade of Hb followed by Pdm and subsequently Cas NB expression, most likely ensures in part that each NB generates a column of uniquely specified neuronal subtypes. The shared transcription factor expression within a given temporal layer also suggests that the cellular constituents of these expression domains may also have similar patterns of downstream target gene expression (Brody. 2000).

The following model for the origin of the layer sublineages marked by these transcription factors has been suggested. As each NB divides, generating a succession of GMCs, it undergoes multiple transitions in transcription factor expression. In succession, the NBs express Hb, Pdm, Cas, and Gh. The first progeny generated by the early S1 and S2 NBs express Hb, and the presence of Hb protein persists in their neural progeny. These early S1 and S2 NBs go on to activate the expression of the Pdms that, like Hb, persist in neural sublineages generated during this temporal window. Subsequently Cas is activated in NBs, represses Pdm transcription, and likewise persists in neural sublineages. After Cas expression, a fourth neural subpopulation, generated by dividing NBs, expresses Gh. This Gh subpopulation most likely represents the terminal sublineage of the embryonic NB. The data also reveal that not all NBs generate cells that occupy all four layers, a result that reflects the unique set of lineages, generated by each NB. Most likely, each NB has a preprogrammed time of delamination, but the timing of transitions is synchronized in a global fashion. The model further suggests that late delaminating NBs can be distinguished from early NBs by their inability to activate Hb. Although Hb is activated shortly after the S1s and S2s have delaminated, Hb is never seen in the proliferative zone during late delaminations (Brody, 2000).

What mechanism drives transitions in transcription factor expression in NBs and in their GMC progeny? It has been shown that Hb, Cas, and Pdm are involved in a regulatory circuit in which Hb and Cas repress Pdm in a cooperative, nonoverlapping fashion both early and late within NB lineages. In addition, Pdm is also required for the proper expression of Cas. It is likely, therefore, that this Hb to Pdm followed by Cas network is responsible for temporal transitions in transcription factors, related to the generation of multiple cellular layers. This conclusion must be tempered by the observation that less than 50% of the cells in clones and, by implication, in the CNS, are positive for even one of these transcription factors. There must be other factors involved in sublineage determination related to CNS layering. If the transitions observed are not caused by the partitioning of mRNA and proteins between NBs and their GMC, but by regulatory interactions within the cells themselves, then there must be additional mechanisms that are involved in the rapid disappearance of these molecules. Expression of transcription factors restricted to one or two generations of NB development could be accomplished if these transcription factors were autoregulatory, repressing their own expression in NBs and in their progeny (Brody. 2000).

For more information on Drosophila neuroblast lineages, see Linking neuroblasts to their corresponding lineage, a site carried by Flybrain, an online atlas and database of the Drosophila nervous system.

Molecular markers for identified neuroblasts in the developing brain of Drosophila

The Drosophila brain develops from the procephalic neurogenic region of the ectoderm. About 100 neural precursor cells (neuroblasts) delaminate from this region on either side in a reproducible spatiotemporal pattern. Neuroblast maps have been prepared from different stages of the early embryo (stages 9, 10 and 11, when the entire population of neuroblasts has formed), in which about 40 molecular markers representing the expression patterns of 34 different genes are linked to individual neuroblasts. In particular, a detailed description is presented of the spatiotemporal patterns of expression in the procephalic neuroectoderm and in the neuroblast layer of the gap genes empty spiracles, hunchback, huckebein, sloppy paired 1 and tailless; the homeotic gene labial; the early eye genes dachshund, eyeless and twin of eyeless; and several other marker genes (including castor, pdm1, fasciclin 2, klumpfuss, ladybird, runt and unplugged). Based on the combination of genes expressed, each brain neuroblast acquires a unique identity, and it is possible to follow the fate of individual neuroblasts through early neurogenesis. Furthermore, despite the highly derived patterns of expression in the procephalic segments, the co-expression of specific molecular markers discloses the existence of serially homologous neuroblasts in neuromeres of the ventral nerve cord and the brain. Taking into consideration that all brain neuroblasts are now assigned to particular neuromeres and individually identified by their unique gene expression, and that the genes found to be expressed are likely candidates for controlling the development of the respective neuroblasts, these data provide a basic framework for studying the mechanisms leading to pattern and cell diversity in the Drosophila brain, and for addressing those mechanisms that make the brain different from the truncal CNS (Urbach, 2003).

The cephalic gap genes are expressed in large domains of the procephalon and play a crucial role not only in the patterning of the peripheral ectoderm, but also in regionalizing the brain primordium. The segmental organization of the Drosophila brain is based on the expression pattern of segment polarity and DV patterning genes. To see whether the cephalic gap genes respect the neuromeric boundaries segment polarity and DV patterning genes, and to provide a basis for studying their potential role in the formation or specification of brain precursor cells, the expression was studied of orthodenticle, empty spiracles, sloppy paired 1, tailless, huckebein, and hunchback in the developing head ectoderm, as well as in the entire population of identified NBs during stages 9-11 (Urbach, 2003).

hunchback (hb) expression in the anterior half of the embryo falls below the limit of detection at the beginning of germ band extension, but accumulates during the extended germ band stage in the CNS, where it is transiently expressed in early, fully delaminated, trunk NBs (S1 and S2) and their progeny. Antibody staining reveals that, from stage 8 onwards, Hb protein is not detected in the head neuroectoderm, but is very dynamically expressed in brain NBs. At stage 9, only about half of the identified deuto- and protocerebral NBs show Hb protein at a detectable level, suggesting that Hb is not a general marker for early NBs. Correspondingly, it is found that Hb protein is also lacking in particular S1 and S2 NBs of the trunk. In some of the early brain NBs, Hb first becomes detectable at stage 10, after their delamination. For example, the early NBs Pcv9 and Pcd6 delaminate at late stage 8 but do not start Hb expression before stage 10. By stage 10, Hb is expressed in about 26 brain NBs, most of which delaminate between stage 9 and 10. In most of these NBs, Hb expression is progressively lost, but is observed in an increasing amount of progeny cells. At stage 11, it is confined to a small subpopulation of about five tritocerebral and four to six protocerebral NBs. Thus, as opposed to the trunk, hb expression in the brain is not limited to early NBs. Hb is expressed in the brain until stage 15, when it is detected in a few cells of the protocerebrum (Urbach, 2003).

Regulation of temporal identity transitions in Drosophila neuroblasts

Temporal patterning is an important aspect of embryonic development, but the underlying molecular mechanisms are not well understood. Drosophila neuroblasts are an excellent model for studying temporal identity: they sequentially express four genes (hunchback -> Krüppel -> pdm1 -> castor) whose temporal regulation is essential for generating neuronal diversity. hunchback -> Krüppel timing is regulated transcriptionally and requires neuroblast cytokinesis, consistent with asymmetric partitioning of transcriptional regulators during neuroblast division or feedback signaling from the neuroblast progeny. Surprisingly, Krüppel -> pdm1 -> castor timing occurs normally in isolated or G2-arrested neuroblasts, and thus involves a neuroblast-intrinsic timer. Finally, Hunchback potently regulates the neuroblast temporal identity timer: prolonged Hunchback expression keeps the neuroblast 'young' for multiple divisions, and subsequent downregulation allows resumption of Krüppel -> pdm1 -> castor expression and the normal neuroblast lineage. It is concluded that two distinct 'timers' regulate neuroblast gene expression: a hunchback -> Krüppel timer requiring cytokinesis, and a Krüppel -> pdm1 -> castor timer which is cell cycle independent (Grosskortenhaus, 2004).

It is concluded that hb is regulated at the transcriptional level in neuroblasts, based on strong correlation with active transcription (intron probe) and protein levels (antibody probe). In addition, hb transcription in GMCs and differentiated neurons, but at this point it cannot be determine if the correlation between protein and transcription is as tight as in the neuroblasts. This does not rule out a role for posttranscriptional regulation, however, to ensure a very short half-life of both hb mRNA and protein. There are predicted miRNA binding sites in the hb 3'UTR and protein degradation (PEST) motifs in the Hb protein that may be necessary to restrict Hb protein to the early portion of neuroblast lineages. There is ample precedent for posttranscriptional regulation of hb in both Drosophila early embryos and C. elegans, but only for translational repression. In Drosophila, Nanos represses hb translation in the early embryo via binding to its 3'UTR. In C. elegans, the hb ortholog hbl-1 regulates temporal identity as part of the heterochronic pathway, and hbl-1 is a target of microRNA regulation through its 3'UTR. It is concluded that precise regulation of hb transcription, coupled with a short half-life of hb mRNA and protein, leads to the observed restriction of Hb protein to the initial cell cycles of neuroblast lineages. Identification of the hb cis-regulatory sequences necessary for proper hb CNS expression has been initiated, and it will be interesting to determine the associated factors that positively and negatively regulate hb transcription in neuroblasts (Grosskortenhaus, 2004).

Cell cycle-arrested neuroblasts maintain hb expression. However, a direct role of the cell cycle (e.g., counting S phases) or an indirect role (e.g., generation of a GMC which could signal back to the neuroblast) has not been distinguished. This study shows that hb transcription is maintained in pebble mutant neuroblasts, which lack cytokinesis but nevertheless go through repeated cell cycles including DNA replication, nuclear envelope breakdown, chromosome condensation, and spindle assembly. Thus, the timely downregulation of hb transcription requires cytokinesis. The requirement for cytokinesis is consistent with two quite different mechanisms: (1) feedback signaling from the GMC to the neuroblast to repress hb transcription, and (2) asymmetric partitioning of an hb transcriptional activator into the GMC to halt hb transcription (Grosskortenhaus, 2004).

Which mechanism is used has not yet been distinguished. Two candidate transcription factors have been tested for a role in hb regulation: Hb and Prospero. The Hb protein does not positively regulate its own transcription in the CNS -- neither hb mRNA nor protein is partitioned into the GMC during neuroblast cell division. The Prospero transcription factor is known to be partitioned into the GMC during neuroblast division, but Prospero protein is cytoplasmic in neuroblasts, and thus unlikely to positively activate hb transcription in this cell type. In addition, misexpression of Prospero in neuroblasts is unable to extend the window of hb transcription and prospero mutants have normal hb expression in neuroblasts, although there is reduced Hb protein in GMCs and neurons by stage 13 and beyond. Thus, Prospero may have a role in maintaining hb transcription in GMCs and neurons, consistent with its nuclear localization in these cell types, but it is not required for timing of hb transcription in neuroblasts (Grosskortenhaus, 2004).

To investigate the role of feedback signaling from the GMC, it would be ideal to do GMC ablations and assay for extended hb transcription in the parental neuroblast, but this experiment is technically very demanding, and even short GMC-neuroblast contact might be enough for the signaling to occur. Whether the feedback signal is mediated by the Notch pathway, which is active in all neuroblasts and GMCs examined to date was tested: blocking the pathway with a sanpodo mutant has no effect on the timing of hb -> Kr -> pdm1 -> cas neuroblast expression. The identification of trans-acting factors that associate with the hb cis-regulatory DNA may be the best approach to distinguish between feedback signaling and transcription factor partitioning mechanisms (Grosskortenhaus, 2004).

Previous work provided strong hints that global extrinsic signals are not required for timing neuroblast temporal identity transitions. (1) Neuroblast lineages are asynchronous, with later-forming neuroblasts expressing hb at the same time adjacent early-forming neuroblasts are expressing cas, making it unlikely that global extrinsic signals trigger gene expression transitions. (2) In vitro culture experiments reported differentiated neuronal clones containing nonoverlapping populations of Hb+, Pdm1+, and Cas+ neurons, consistent with a normal progression of gene expression in the parental neuroblast over time, although gene expression timing was not assayed in neuroblasts. These observations have been confirmed and extended. Isolated neuroblasts progress from Hb+ to Kr+ to triple negative (presumptive Pdm1+) to Cas+ over time in culture, and clones in which the GMCs expressed a later gene than the neuroblast (e.g., Hb+ or Kr+ neuroblasts never had Pdm1+ or Cas+ GMCs). Thus, Hb -> Kr -> Pdm1 -> Cas neuroblast gene expression timing occurs normally in isolated neuroblasts, demonstrating that lineage-extrinsic factors are not required for neuroblast temporal identity transitions. It is possible that extrinsic cues may still override or entrain an intrinsic program, however, which could be tested by heterochronic neuroblast transplants. In summary, in vitro and in vivo data show that timing of temporal identity transitions is regulated by a neuroblast lineage-intrinsic mechanism. For the latter genes in the cascade, it appears that the mechanism is actually intrinsic to the neuroblast itself (Grosskortenhaus, 2004).

All available data suggest that Kr and Cas timing are regulated at the transcriptional level. Kr mRNA and protein are both detected in neuroblasts during embryonic stage 10 and subsequently maintained in a subset of neurons. Similarly, cas mRNA and protein are both widely detected in neuroblasts only at stage 12, and maintained in a subset of late-born neurons. In the future, it will be important to do mRNA/protein double labels for Kr, pdm1, and cas to determine the extent to which mRNA/protein levels are correlated at the single cell level. Unfortunately, it is not easy to assay for active transcription of Kr or cas due to the lack of large introns (Grosskortenhaus, 2004).

Surprisingly, it was found that cell cycle-arrested neuroblasts that lack Hb still express Kr -> Pdm1 -> Cas with the same timing as in wild-type embryos. What mechanism might time Kr -> Pdm1 -> Cas expression in the absence of cell division? Extrinsic cues can be ruled out, because isolated neuroblasts still undergo normal Kr -> Pdm1 -> Cas gene expression timing. A change in nucleo-cytoplasmic ratio, known to time certain early embryonic events, can be ruled out, because wild-type neuroblasts increase their nucleo-cytoplasmic ratio over time, but G2-arrested neuroblasts decrease their nucleo-cytoplasmic ratio as they enlarge without dividing (Grosskortenhaus, 2004).

The most attractive model for Kr -> Pdm1 -> Cas in G2-arrested neuroblasts is a cascade of transcriptional regulation between Kr, Pdm1, and Cas. Misexpression studies have shown that each gene can activate expression of the next gene in the series, and repress the 'next + 1' gene, which could account for the sequential activation of each gene. If each transcription factor can also repress its activator, similar to the known ability of Cas to negatively regulate pdm1 expression, it could explain the sequential downregulation of each gene as well. Currently, all misexpression data are consistent with this simple model. However, analysis of hb and Kr mutants reveals additional complexity. hb mutants show relatively normal Kr -> Pdm1 -> Cas timing, and Kr mutants show relatively normal Pdm1 -> Cas timing. Thus, there must be at least one unidentified input that can activate Kr in the absence of Hb, and pdm1 in the absence of Kr. Regulation of Hb -> Kr -> Pdm1 -> Cas appears to be primarily at the transcriptional level, and thus identification of the relevant cis-regulatory DNA and associated transcription factors should provide insight into the 'timer' mechanism that controls sequential gene expression in neuroblasts (Grosskortenhaus, 2004).

Hb seems to have a special role in advancing the temporal identity timer. It is the only factor in the cascade whose downregulation requires cytokinesis, and as long as it is present (either because of cell cycle arrest or misexpression) the timer is unable to advance. Misexpression of Hb beyond its normal expression window leads to generation of extra early cell types and blocks Kr -> Pdm1 -> Cas progression. However, these experiments do not reveal whether Hb generates these early fates by overriding Kr -> Pdm1 -> Cas neuronal identity while the temporal timer is advancing or if it arrests progress of the temporal timer. The results show that continuous expression of Hb blocks the advancement of the temporal identity timer, keeping the neuroblast in a 'young' state that is fully capable of resuming its normal cell lineage following downregulation of Hb. The ability of Hb to keep the neuroblast in a 'young' multipotent state, despite repeated rounds of cell division, raises the interesting question of how Hb acts at the mechanistic level. Transcriptional targets of Hb in the CNS are so far unknown. A mammalian homolog, Ikaros, is associated with chromatin and remodeling proteins and Drosophila Hb is thought to regulate chromatin-mediated heritable expression of homeotic genes. Thus, Hb might modulate chromatin structure in neuroblasts to prevent expression of later temporal identity genes, or to maintain plasticity of gene expression necessary for maintaining the multipotent state of the neuroblast (Grosskortenhaus, 2004).

Different combinations of gap repressors for common stripes in Anopheles and Drosophila embryos

Drosophila segmentation is governed by a well-defined gene regulation network. The evolution of this network was investigated by examining the expression profiles of a complete set of segmentation genes in the early embryos of the mosquito, Anopheles gambiae. There are numerous differences in the expression profiles as compared with Drosophila. The germline determinant Oskar is expressed in both the anterior and posterior poles of Anopheles embryos but is strictly localized within the posterior plasm of Drosophila. The gap genes hunchback and giant display inverted patterns of expression in posterior regions of Anopheles embryos, while tailless exhibits an expanded pattern as compared with Drosophila. These observations suggest that the segmentation network has undergone considerable evolutionary change in the dipterans and that similar patterns of pair-rule gene expression can be obtained with different combinations of gap repressors. The evolution of separate stripe enhancers in the eve loci of different dipterans is discussed (Goltsev, 2004).

In Drosophila, different levels of the Hunchback and Knirps gap repressor gradients define the limits of eve stripes 3, 4, 6, and 7, while Giant and Kruppel establish the borders of stripes 2 and 5. In situ hybridization probes were prepared for Anopheles orthologues of all four of these gap genes, as well as a fifth gap gene, tailless. hunchback displays a broad band of expression in the anterior half of the Anopheles embryo, encompassing both the presumptive head and thorax. This pattern is similar to that observed in Drosophila, although there are a few notable deviations: (1) there is no obvious maternal expression seen in early Anopheles embryos, whereas maternal hunchback mRNAs are strongly expressed throughout early Drosophila embryos; (2) there is a significant change in the posterior staining pattern. The Drosophila gene displays a strong posterior stripe of expression that is comparable in intensity to the anterior staining pattern. In Anopheles, this staining is significantly weaker than that of the anterior domain, and the posterior pattern is shifted anteriorly into the presumptive abdomen (Goltsev, 2004).

The Kruppel and knirps staining patterns are similar in Anopheles and Drosophila embryos. In both cases, the principal sites of expression are seen in the presumptive thorax and abdomen, respectively. However, the remaining two gap genes, giant and tailless, exhibit distinctive staining patterns. In Anopheles, giant exhibits a continuous band of staining in anterior regions, whereas the Drosophila gene is excluded from the anterior pole. Moreover, there is a prominent band of staining in the presumptive abdomen of Drosophila embryos that is not seen in Anopheles. Finally, tailless is expressed in a narrow stripe in the posterior pole of Drosophila embryos, whereas Anopheles embryos display a dynamic pattern that (transiently) extends throughout the presumptive abdomen (Goltsev, 2004).

These observations document significant changes in the expression patterns of maternal determinants and gap genes in flies and mosquitoes, although the dynamic eve pattern is quite similar in the two systems. The most notable differences were seen for the gap genes hunchback and giant. Additional in situ hybridization assays were done in an effort to obtain a more comprehensive view of these changing patterns; hunchback is initially expressed in the anterior half of Anopheles embryos, with no staining detected in posterior regions. Weak posterior staining is detected by the onset of gastrulation, but expression appears to be localized within the presumptive abdomen rather than the posterior pole as seen in Drosophila. This shift was confirmed by costaining with eve. In Drosophila, the anterior hunchback pattern is lost except for a stripe of staining in the thorax, and this stripe persists along with the posterior pattern during gastrulation. In Anopheles, the early hunchback expression pattern gives way to localized expression in the presumptive serosa. Drosophila lacks a comparable staining pattern, although similar patterns have been documented in Tribolium, and mothmidges. It is conceivable that the late hunchback pattern is responsible, directly or indirectly, for the repression of eve stripes in the presumptive serosa (Goltsev, 2004).

As seen for hunchback, there is no detectable expression of giant in posterior regions of early Anopheles embryos. Weak staining appears in the posterior pole by the onset of gastrulation. This staining is clearly posterior to the hunchback pattern in the presumptive abdomen. Thus, the posterior hunchback and giant patterns are reversed in Anopheles as compared with Drosophila. The anterior giant pattern encompasses the entire anterior half of Anopheles embryos and extends into the anterior pole. The staining pattern is refined at gastrulation, including the loss of expression in the presumptive serosa and the formation of discrete bands. Nonetheless, unlike the situation in Drosophila, expression persists in the anterior pole, thereby raising the possibility that different mechanisms are used to establish the anterior border of eve stripe 2 in flies and mosquitoes (Goltsev, 2004).

The altered patterns of hunchback and giant expression in posterior regions raise the possibility that different combinations of gap repressors are used to establish eve stripes 5, 6, and 7 in Anopheles and Drosophila. It is unlikely that Giant establishes the posterior border of eve stripe 5 and that Hunchback delimits the posterior border of stripe 7, as seen in Drosophila. The expression profiles of additional gap genes were analyzed in an effort to identify potential repressors for these stripe borders. The most obvious candidates are huckebein and tailless, since both are expressed in the posterior pole of Drosophila embryos. No expression of huckebein was seen in early embryos, although strong staining appears after germband elongation (Goltsev, 2004).

The gap gene tailless is initially detected at the anterior and posterior poles, with roughly equivalent levels of staining at the two sites. At slightly later stages, the anterior domain is lost, and the posterior pattern expands throughout the presumptive abdomen. The tailless transcripts detected in posterior regions exhibit a graded distribution, with peak levels at the posterior pole and progressively lower levels in more anterior regions. During cellularization, staining is reduced in posterior regions and reappears near the anterior pole. This broad and dynamic staining pattern is consistent with the possibility that the Tailless repressor specifies the posterior borders of one or more posterior eve stripes (Goltsev, 2004).

Torso signaling was examined in the Anopheles embryo in an effort to understand the basis for the expanded tailless expression pattern. In Drosophila, tailless is activated by the Torso signaling pathway, which can be visualized with an antibody against diphospho (dp)-ERK. The antibody detects localized staining in the terminal regions of early Drosophila embryos. A similar staining pattern is detected in Anopheles, although staining may be somewhat broader in Anopheles than Drosophila. It is therefore conceivable that the expansion of the posterior tailless expression pattern seen in Anopheles might be due to an expanded activation of the Torso signaling pathway (Goltsev, 2004).

The combinations of gap repressors that define the borders of eve stripes 2 to 7 are known in Drosophila. Stripes 2 and 5 are formed by the combination of Giant and Kruppel repressors, while distinctive borders for stripes 3, 4, 6, and 7 are established by the differential repression of the stripe 3/7 and stripe 4/6 enhancers in response to distinct concentrations of the Hunchback and Knirps repressor gradients. Double-staining assays provide immediate insights into the likely combination of gap repressors that are used for any given stripe. For example, the giant and Kruppel expression patterns abut the borders of eve stripes 2 and 5. Double-staining assays were done to determine the potential regulators of the Anopheles eve stripes. These experiments involved the use of digoxigenin-labeled hunchback, Kruppel, knirps, and giant hybridization probes along with an FITC-labeled eve probe. Different histochemical substrates were used to separately visualize the two patterns (Goltsev, 2004).

The anterior hunchback pattern extends through eve stripe 2 and approaches the anterior border of stripe 3. While the posterior pattern extends through stripes 6 and 7, this pattern is quite distinct from the posterior hunchback pattern seen in Drosophila, which abuts the posterior border of eve stripe 7. The anterior giant pattern extends from the anterior pole to eve stripe 2, while the posterior pattern abuts the posterior border of eve stripe 7. In Drosophila, the posterior giant pattern extends from eve stripe 5 to stripe 7. The Kruppel pattern may be somewhat narrower in Anopheles than Drosophila. It encompasses eve stripe 3 in Anopheles but includes both stripes 3 and 4 in Drosophila. Finally, knirps exhibits the same limits of expression in Anopheles as Drosophila. In both cases, the staining pattern extends from eve stripes 4 to 6. In Anopheles, the anterior knirps pattern straddles the anterior border of eve stripe 1. Some of the eve stripes are associated with the same combinations of gap repressors in flies and mosquitoes (e.g., stripes 2, 3, and possibly 4), whereas others show distinctive combinations of gap repressors (e.g., stripes 5, 6, and 7 (Goltsev, 2004).

The systematic comparison of segmentation regulatory genes in Anopheles and Drosophila suggests that the segmentation gene network has undergone considerable evolutionary change among dipterans despite highly conserved patterns of eve expression. Three particular changes in the network are discussed: the localization of maternal determinants, the formation of the anterior border of eve stripe 2, and the formation of the posterior borders of eve stripes 5, 6, and 7 (Goltsev, 2004).

In Drosophila, hunchback contains two promoters, and the maternal promoter leads to the ubiquitous distribution of hunchback mRNAs throughout early embryos. No hunchback mRNAs were detected in early Anopheles embryos. This apparent absence of maternal transcripts raises the possibility that localized Nanos products are not required for inhibiting the synthesis of Hunchback proteins in posterior regions of Anopheles embryos. In Drosophila, the embryonic lethality caused by nanos mutants can be suppressed by the removal of maternal Hunchback products. This nanos-hunchback interaction is ancient and probably operates in basal insects, and possibly basal arthropods. However, the potential absence of this interaction in Anopheles is consistent with the idea that nanos has an additional essential function. Indeed, a recent study suggests that Nanos is required for maintaining stem cell populations of germ cells in Drosophila (Goltsev, 2004).

Anopheles lacks bicoid and contains a lone Hox3 gene that is more closely related to zen and specifically expressed in the serosa. How is hunchback activated in the presumptive head and thorax in Anopheles? The homeobox gene orthodenticle can substitute for bicoid in Tribolium. However, orthodenticle does not appear to be maternally expressed in Anopheles, but instead, staining is strictly zygotic and restricted to anterior regions, similar to the pattern seen in Drosophila. Sequential patterns of orthodenticle, giant, and hunchback expression are established by differential threshold readouts of the Bicoid gradient in Drosophila. It is possible that an unknown maternal regulatory gradient emanating from the anterior pole is responsible for producing similar patterns of expression in Anopheles. It is proposed that this unknown regulatory factor may be localized to the anterior pole by Oskar. Oskar coordinates the assembly of polar granules and is essential for the localization of Nanos in the posterior plasm. It might also localize one or more unknown determinants in anterior regions of Anopheles embryos (Goltsev, 2004).

The eve stripe 2 enhancer is the most thoroughly characterized enhancer in the segmentation gene network. It can be activated throughout the anterior half of the embryo by Bicoid and Hunchback, but the Giant and Kruppel repressors delimit the pattern and establish the anterior and posterior stripe borders, respectively. Removal of the Giant repressor sites within the stripe 2 enhancer in cis or removal of the repressor in trans causes an anterior expansion of the stripe 2 pattern. However, ectopic expression does not extend to the anterior pole, suggesting that an additional anterior repressor regulates the stripe 2 enhancer. Recent studies identified Sloppy-paired as the likely anterior repressor. The limits of the giant and Kruppel expression patterns seen in Anopheles suggest that they might define the eve stripe 2 borders, just as in Drosophila. However, at the critical time when eve stripe 2 is formed in Anopheles, the giant staining pattern extends to the anterior pole, while the corresponding Drosophila gene is repressed in these regions. It is therefore possible that Giant is sufficient to form the anterior border in Anopheles and that repression by Sloppy-paired represents an innovation in Drosophila (Goltsev, 2004).

There are numerous differences in the patterns of gap gene expression in Drosophila and Anopheles. In Drosophila, the posterior stripe of hunchback expression is the source of a repressor gradient that specifies the posterior borders of eve stripes 6 and 7. Anopheles exhibits a distinct posterior staining pattern, with expression extending through stripes 6 and 7. It is therefore unlikely that Hunchback regulates these stripes as seen in Drosophila. Instead, the location of the posterior hunchback pattern suggests that it regulates the posterior border of eve stripe 5 in Anopheles. In Drosophila, this border is formed by Giant, but in Anopheles, the posterior giant expression pattern is restricted to the posterior pole where it abuts stripe 7. Thus, a combination of Kruppel and Giant defines the eve stripe 5 borders in Drosophila, whereas Kruppel and Hunchback might be used in Anopheles (Goltsev, 2004).

In Drosophila, eve stripes 6 and 7 are regulated by different concentrations of Knirps and Hunchback. Low levels of Knirps define the anterior border of stripe 7, while higher levels are needed to repress eve stripe 6. Conversely, low levels of Hunchback establish the posterior border of eve stripe 6, while higher levels regulate stripe 7. The position of the knirps expression pattern is consistent with the possibility that it defines the anterior limits of stripes 6 and 7, just as in Drosophila. However, the posterior borders of these stripes are probably not regulated by Hunchback. The expanded pattern of tailless expression seen in Anopheles might permit it to establish the posterior border of eve stripe 6 and possibly stripe 7. An alternative candidate for the posterior stripe 7 border is giant, which is expressed in a tight domain within the posterior pole. Consistent with this possibility is the observation that the posterior giant pattern comes on relatively late, and the posterior stripe 7 border is the last to form among the seven eve stripes. The reversal of the posterior hunchback and giant expression patterns, along with the expanded tailless pattern, strongly suggests that different combinations of gap repressors are used to define eve stripes 5, 6, and 7 in Drosophila and Anopheles (Goltsev, 2004).

An implication of the preceding arguments is that each of the seven eve stripes is regulated by a separate enhancer in Anopheles. Only five enhancers regulate eve in Drosophila since four of the seven stripes (3, 4, 6, and 7) are regulated by just two enhancers (3/7 and 4/6) that respond to different concentrations of the opposing Hunchback and Knirps repressor gradients. The change in the posterior hunchback pattern virtually excludes the use of this strategy in Anopheles. Thus, stripes 3 and 7 are probably regulated by separate enhancers since different combinations of gap repressors appear to define the stripe borders. Similar arguments suggest that stripes 4 and 6 are also regulated by separate enhancers (Goltsev, 2004).

Why do some enhancers generate two stripes, while others direct just one? Consider the eve stripe 2 and stripe 3/7 enhancers in Drosophila. The stripe 3/7 enhancer is activated by ubiquitous activators, including dSTAT, and the two stripes are 'carved out' by the localized Hunchback and Knirps repressors. Knirps establishes the posterior border of stripe 3 and anterior border of stripe 7, while Hunchback establishes the anterior border of stripe 3 and posterior border of stripe 7. The stripe 2 enhancer directs just a single stripe due to the localized distribution of the stripe 2 activators, particularly Bicoid. In principle, a ubiquitous activator would cause the stripe 2 enhancer to direct two stripes, stripes 2 and 5. Opposing Giant and Kruppel repressor gradients would carve out the borders of the two stripes, similar to the way in which Hunchback and Knirps regulate the stripe 3/7 and stripe 4/6 enhancers. Presumably, the eve stripe 5 enhancer directs a single stripe of expression because it is regulated by a localized activator, possibly Caudal (Goltsev, 2004).

It is suggested that ancestral dipterans contained an eve locus with separate enhancers for every stripe. Anopheles eve might represent an approximation of this ancestral locus. The consolidation of enhancers that generate multiple stripes was made possible by cross-repression of gap gene pairs. In Drosophila, there are mutually repressive interactions between Hunchback and Knirps, as well as between Giant and Kruppel. The use of these interacting gap pairs along with ubiquitous activators permits the formation of two stripes from a single enhancer. It is possible to envision two ways in which mutual cross-repression of these gap genes helps to establish the precise patterns of pair-rule gene expression: (1) it ensures that there are zones free of repressor activity on both sides of Kruppel (for the Kruppel and Giant pair) and Knirps (for the Knirps and Hunchback pair) domains; (2) it protects the patterns of pair-rule gene expression from mutations that could potentially shift the domains of gap gene expression. For example, a mutation that could shift the expression of Kruppel would simultaneously shift the expression of Giant always leaving a repressor-free zone where Eve stripes would be established. Therefore, the evolution of the eve locus depends on the changes in the preceding tier of the segmentation network: refinement in gap gene cross-regulatory interactions (Goltsev, 2004).

Finally, it is easy to imagine that certain dipterans have a single enhancer for stripes 2 and 5, rather than the separate enhancers seen in Drosophila. Perhaps, the symmetric repression of Giant and Kruppel is a relatively recent occurrence, only now creating the opportunity for consolidated expression of stripes 2 and 5 (Goltsev, 2004).

Drosophila long-chain acyl-CoA synthetase acts like a gap gene in embryonic segmentation

Long-chain acyl-CoA synthetases (ACSLs) convert the long chain fatty acids to acyl-CoA esters, the activated forms participating in diverse metabolic and signaling pathways. dAcsl is the Drosophila homolog of human ACSL4 and their functions are highly conserved in the processes ranging from lipid metabolism to the establishment of visual wiring. This study demonstrates that both maternal and zygotic dAcsl are required for embryonic segmentation. The abdominal segmentation defects of dAcsl mutants resemble those of gap gene knirps (kni). The central expression domain of Kni transcripts or proteins was reduced whereas the adjacent domains of another gap gene Hunchback (Hb) were correspondingly expanded in these mutants. Consequently, the striped pattern of the pair-rule gene Even-skipped (Eve) was disrupted. It is proposed that dAcsl plays a role in embryonic segmentation at least by shifting the anteroposterior boundaries of two gap genes (Zhang, 2011).

In Drosophila embryo, a hierarchy of maternal, gap, pair-rule and segment polarity genes which encode transcription factors establish the anteroposterior axis and the embryonic segmentation. The spatially restricted transcription factors determine the complex gene expression patterns in the early embryo. Along with the maternal determinants, the gap gene products specify the boundaries of the adjacent gap gene expression domains and the downstream pair-rule gene stripes. Among them, Knirps (Kni) and Hunchback (Hb) form their expression patterns partly through mutual repression (Zhang, 2011).

The known maternal effectors are not sufficient to establish the gap domains and it is likely that unidentified maternal molecules exist and modulate the gap gene expression. The abundant maternally-deposited lipids in embryos have been recognized as an energy source for early embryo development. These molecules also have important functions in diverse signaling pathways during larval development such as shaping morphogen gradients. However, it remains unclear whether lipids participate in any way in the establishment of embryonic segmentation (Zhang, 2011).

Long chain acyl-CoA synthetase (ACSL) is a family of enzymes which adds Coenzyme A to the long chain (C12-20) fatty acids. As the activated form of fatty acids, the Acyl-CoA participates in various cellular processes including lipid metabolism, vesicle trafficking and signal transduction. ACSL4 is a member of the mammalian ACSL family and its mutations have been associated with non-syndromic X-linked mental retardation (MRX). The Drosophila gene dAcsl encodes the homolog of human ACSL4 and they are functionally conserved ranging from building visual circuitry to lipid homeostasis (Zhang, 2009). However, the developmental function of dAcsl at the embryonic stages remains unexplored (Zhang, 2011).

This report illustrates that dAcsl is required for embryonic segmentation both maternally and zygotically. The impaired segmentation caused by dAcsl mutations is similar to that of gap gene kni mutants. In dAcsl mutants, the domain of Kni transcripts or proteins was reduced whereas the domain of another gap gene Hb protein was correspondingly expanded. Consequently, the pair-rule gene expressions were perturbed in these embryos. It is proposed that dAcsl participates in embryonic segmentation by spatially modulating gap gene expression (Zhang, 2011).

The segmentation defects of dAcsl mutants resemble those of gap gene kni. The posterior domain of Kni transcripts or proteins was narrowed whereas the adjacent domains of another gap gene Hb correspondingly expanded in these mutants. The findings reveal the connection between long-chain acyl-CoA synthetase and embryonic segmentation in Drosophila. It is proposed that dAcsl functions in embryonic segmentation by modulating gap gene expression (Zhang, 2011).

The similarity in mutant phenotypes uncovers the possible link between this enzyme and kni. Although the strong genetic interaction exists between dAcsl and kni, two observations suggest that the function of dAcsl in segmentation seems not limited to kni. Firstly, the anterior Eve stripes were also affected in some mutant embryos where Kni is not expressed. Secondly, dAcsl also genetically interacted with Kr. The alteration of gap gene expression is consistent with the genetic interaction results, in which kni or Kr reduction enhanced dAcsl segmentation defects whereas hb did not. Since the anterior zygotic Hb domain was expanded posteriorly in dAcsl mutants, this Hb shift could affect the anterior boundaries of both Kr and Kni domains. Accordingly, certain degree of rescue of the dAcsl mutant phenotype was expected when hb gene dosage was lowered by half. However, an obvious effect was seen, which could simply be that one zygotic dosage of the Hb products along with the maternal contribution is enough to fulfill its normal function at this stage (Zhang, 2011).

Also, the early zygotic expression of Hb was somehow expanded more posteriorly, indicating a spatial increase in response to Bcd activity. No corresponding increase of Bcd was detected at protein levels though the Bcd gradient seemed less steep in the mutants. Additionally, the effects not limited to kni-like phenotype would have been seen if there were a posterior-ward shift due to a major change in Bcd. Further, because removing zygotic copy of dAcsl contributed ~ 4% more occurrence of segmentation defects than the maternal mutation alone (~ 11%), alteration in the gap gene functions cannot explain the defects developed post-zygotically unless dAcsl is also zygotically activated before cellularization (Zhang, 2011).

How can the gap gene-like phenotype in dAcsl mutants be explained or how does dAcsl act on gap gene expressions/activities? One possibility is that the altered distribution of the upstream maternal factors since kni transcripts were spatially reduced in the dAcsl maternal mutants. There are abundant lipid droplets which participate in the vesicle transport and store maternal proteins in the early embryo. Since dAcsl is predicted as an enzyme mobilizing fatty acid and required for neutral lipids formation in larval tissues (Zhang, 2009), the aberration of lipid droplets formation was anticipated in dAcsl mutant embryos. Consequently, the distribution of certain maternal determinants may be affected because of the compromised membrane trafficking, altered protein localization, etc. If this hypothesis is true, then other mutations such as Lsd2 which disrupt lipid droplets transport and neutral lipids storage in embryo should give similar phenotype as dAcsl mutations. However, only very minor segmentation defects were observed in Lsd2 mutant cuticles. Does the lipid storage decrease more in dAcsl than in Lsd2 mutants? The triglyceride levels were examined in early embryos and no significant difference could be detected between the wild type and dAcsl or Lsd2 mutant embryos. The relationship between the lipid-droplets formation and embryonic segmentation remains elusive. Nonetheless, as a lipid metabolism-related enzyme, dAcsl's effect in segmentation is specific and intriguing. However, the details of the connection between dAcsl and embryonic segmentation require more intensive investigations (Zhang, 2011).

Hunchback and tracheal development

The Drosophila tracheal system, a tubular network, is formed from isolated ectodermal metameres by guided branch outgrowth and branch fusion. Branch outgrowth is triggered by the localized and transient activity of Branchless (Bnl/dFGF). A mesodermal cell has been discovered that links the leading cells of outgrowing main branches 2.5 hr before they fuse. This bridge-cell serves as an essential guidance post and needs Hunchback (Hb) activity to exert its function. The bridge-cell provides cues acting in concert with Bnl/dFGF signaling to mediate directed branch outgrowth that ultimately leads to position-specific branch fusion (Wolf, 2002).

A single cell that is marked by expression of the gene hunchback (hb) is found at the posterior lateral margin of each tracheal metamere. This cell gives rise to daughter cells that maintain hb expression. The more ventrally located daughter cell maintains a round morphology and remains in position, whereas the dorsal daughter cell connects to the posterior bud of the tracheal metamere, termed the dorsal trunk posterior branch. Subsequently, the dorsal daughter cell elongates and extends posteriorly and thereby contacts to the anterior bud, termed the dorsal trunk anterior branch, of the adjacent posterior tracheal metamere. In this way, the dorsal daughter cell bridges the leading cells of the dorsal trunk anterior and posterior branches of two adjacent metameres, which then fuse about 2.5 hr later to form the continuous dorsal trunk. Thus, the dorsal daughter cell is referred to as the bridge-cell. The cell remains at this position until fusion between the dorsal trunk anterior and posterior branches occurs. During this fusion process, the bridge-cell becomes displaced and hb expression starts to fade (Wolf, 2002).

To trace the origin of the bridge-cell, double-staining experiments were performed with tracheal-specific markers and hb. ß-Galactosidase expression in nuclei of dorsal trunk fusion cells and in nuclei of tracheal cells has revealed a lack of colocalization with bridge-cell hb expression. Furthermore, trachealess (trh) mutant embryos, which lack tracheal cell identity, show hb-expressing bridge-cells as found in wild-type embryos. Thus, these results indicate that the bridge-cell is of nontracheal origin. Finally, double-staining of hb and a mesodermal marker has revealed coexpression of hb and the marker in bridge-cell precursors. Therefore, the bridge-cell is a nontracheal cell and of mesodermal origin (Wolf, 2002).

To understand the function of bridge-cells in dorsal trunk formation, it was first asked whether bridge-cell development is affected in hb mutant embryos. Homozygous hbFB mutant embryos, which express a nonfunctional Hb protein because of a premature stop codon mutation, express the hb transcript only transiently in bridge-cell precursors, raising the possibility that these cells may die. In fact, TUNEL staining suggests cell death is occurring at positions that correspond to those of bridge-cell precursors in hbFB mutants but not in wild-type embryos. This finding implies that the lack of hb activity causes bridge-cell precursors to undergo apoptosis. To show apoptosis as the underlying event of transient hb expression in bridge-cells more directly, the baculovirus P35 protein, a suppresser of apoptosis in Drosophila was ubiquitously expressed in hbFB mutant embryos. In contrast with hbFB mutants, which lack hb expression in the bridge-cells at stage 12, hbFB embryos expressing P35 protein maintain hb expression in bridge-cells as is the case in wild-type embryos. Thus, expression of hb serves as a marker for bridge-cells, whereas Hb protein is essential for bridge-cell viability. Therefore, analysis of tracheal development in hbFB mutant embryos would allow direct study of bridge-cell function in dorsal trunk formation (Wolf, 2002).

In hbFB mutant embryos initial tracheal development, including primary branch outgrowth, appears normal up to the end of stage 12. Subsequently, the dorsal trunk branches become stalled and misrouted, whereas the other primary branches are formed as in wild-type embryos. Despite the strong dorsal trunk phenotype, the dorsal trunk branches occasionally fuse in hbFB mutant embryos and form dorsal trunk rudiments. This observation and normal expression of the escargot (esg) fusion cell marker in hbFB mutant embryos suggest that the fusion process, required for dorsal trunk formation, is not impaired in hbFB mutant embryos. These results indicate that hb is not necessary for the initial outgrowth but for the subsequent outgrowth of dorsal trunk branches. Thus, the results also suggest that the hb-dependent bridge-cells are involved in the outgrowth of dorsal trunk branches toward their fusion partners (Wolf, 2002).

Bnl/dFGF is necessary for the primary tracheal branching, including the formation of the dorsal trunk. Therefore, it was asked whether the absence of bridge-cells might interfere with bnl expression. The expression pattern of bnl was unaffected in hb mutant embryos. Also, hb expression in the bridge-cells was not affected in bnl mutant embryos and in embryos that lack the activity of breathless (btl), which codes for the Bnl/dFGF receptor. Thus, bridge-cells do not interfere with the proper expression of Bnl/dFGF around the developing tracheal branches, and hb-expression in the bridge-cells is independent of Bnl/dFGF signaling (Wolf, 2002).

Because localized Bnl/dFGF signaling is not necessary for dorsal trunk formation, it was asked whether the bridge-cell mediates the proposed additional guidance mechanism for dorsal trunk branch outgrowth. By use of the Gal4/UAS-system, Bnl/dFGF was ectopically expressed in tracheal cells to impede the spatial cues that are normally derived from the local arrangement of cell clusters expressing Bnl/dFGF. In contrast with wild-type embryos, embryos with ectopic expression of Bnl/dFGF develop complete dorsal trunk structures but lack the other primary branches. However, hbFB mutant embryos that express Bnl/dFGF ectopically had no signs of dorsal trunk branch outgrowth at all. These results indicate that the bridge-cell is necessary and essential for dorsal trunk formation, suggesting that this cell provides guidance cues specifically during the anterior-posterior dorsal trunk branch outgrowth. Thus, the bridge-cell, in combination with Bnl/dFGF signaling, directs outgrowth of the main tracheal tube and may mediate the proposed additional guidance mechanism (Wolf, 2002).

To test the above inference, hb was ectopically expressed via the Gal4/UAS-system in sensory organ precursor (SOP) cells in positions close to the bridge-cells. The outgrowing dorsal trunk anterior branches were seen in contact with the cells that ectopically express hb, even in the presence of the normal bridge-cells. As a consequence of the ectopic hb expression, the dorsal trunk of the embryos show interruptions and abnormal bottleneck-like fusion points. Thus, hb expression in ectopic cells close to bridge-cells triggers a differentiation program that interferes with the directed outgrowth of the dorsal trunk branches suggesting that hb activity is required not only for the viability but also for the identity of the bridge-cell. Whether the differentiation program involves local and short-range signals and/or provides a migration matrix by cell adhesion is unknown. However, the hypothesis that the bridge-cell serves as an adhesion-dependent guiding post is preferred, since tracheal cell extensions are observed along the bridge-cell directly after the initial contact (Wolf, 2002).

The discovery of the bridge-cell and studies on Bnl/dFGF signaling provide a coherent model of how dorsal trunk formation may occur. After invagination of the tracheal placodes, budding of the tracheal metameres is triggered by localized Bnl/dFGF activity. This signal apparently does not always have the necessary precision on its own to guide the leading cells. The bridge-cell provides this precision by serving as a guidance post to properly position the budding dorsal trunk branches. The results also demonstrate an interplay of cells deriving from two different germ layers, mesoderm and ectoderm, that is necessary to establish the interconnected tubular tracheal network during embryogenesis. The identification of a key player in bridge-cell differentiation, namely the transcription factor Hb, provides an entry point to unravel the molecular targets of hb. The analysis of these targets may also contribute to gaining further insights into the function of the bridge-cells during tubular network formation, possibly in organisms other than Drosophila (Wolf, 2002).


The head and thorax are missing in mutant embryos and abdominal segments 7 and 8 are fused (Tautz, 1987).

It is suggested that Hb and Castor act in a cooperative, non-overlapping manner to control POU gene expression during Drosophila CNS development. By silencing pdm expression in early and late NB sublineages, Hb and Cas establish three pan-CNS compartments whose cellular constituents are marked by the expression of either Hb, Pdm, or Cas. Embryos lacking Hb function suffer multiple defects. During CNS development, hb- embryos fail to develop labial and thoracic ganglia and gaps form between the subesophageal maxillary neuromeres and the abdominal ganglia. In addition, the seventh and eighth abdominal segments are fused due to the absence of parasegment 13. To further characterize the phenotypic consequences triggered by the loss of Hb and compare them with defects caused by loss of Cas, the axon fascicle organization was examined in hb null embryos. hb- embryos have severe axon guidance defects. Missing are the highly ordered ventral cord axon scaffolds made up of longitudinal connective and commissural fascicles. However, although axon fascicles do not form properly, significant numbers of neurons still generate axons, albeit misguided, in hb- embryos. Since many axon-guiding glia and pathfinding neurons are born from early NB sublineages, hb function may be essential for establishing correct axon guidance cues in these sublineages. Unlike the hb- phenotype and consistent with its late NB expression, loss of cas does not disrupt the formation of axon connective or commissure fascicles but it does reduce the number of late forming axons that participate in these fascicles (Kambadur, 1998).

The requirements for the multi sex combs (mxc) gene during development have been examined to gain further insight into the mechanisms and developmental processes that depend on the important trans-regulators forming the Polycomb group (PcG) in Drosophila. Although mxc has not yet been cloned, it is known to be allelic with the tumor suppressor locus lethal (1) malignant blood neoplasm [l(1)mbn]. The mxc product is dramatically needed in most tissues because its loss leads to cell death after a few divisions. mxc also has a strong maternal effect. Hypomorphic mxc mutations are found to enhance other PcG gene mutant phenotypes and cause ectopic expression of homeotic genes, confirming that PcG products are cooperatively involved in repression of selector genes outside their normal expression domains. The mxc product is needed for imaginal head specification, through regulation of the ANT-C gene Deformed. This analysis reveals that mxc is involved in the maternal control of early zygotic gap gene expression known to involve some other PcG genes and suggests that the mechanism of this early PcG function could be different from the PcG-mediated regulation of homeotic selector genes later in development (Saget, 1998).

Induction of uncontrolled growth and deregulation of Hox genes are linked in mammals, where Hox products can induce leukemia. In Drosophila, modification of homeotic gene expression causes homeosis, sometimes associated with increased proliferation but not with uncontrolled tumorous growth, possibly because the identity of each segment is specified by a combination of HOM products. Loss or gain of one HOM gene will likely lead to a new combination that is found elsewhere in wild type, and cells expressing this combination could be expected to follow the corresponding developmental pathway and give rise to homeotic transformations. However, because each cellular identity apparently corresponds to a given proliferation rate, loss or ambiguity of identity due to deregulation of several selector genes in a single cell, such as mxc mutations apparently induce, could lead to loss of proliferation control. Identification of mxc partners and targets, as well as of the molecular nature of the mxc product, may help throw light on the genes and mechanisms involved in this process (Saget, 1998).

It has been proposed that certain PcG genes are required for the maintenance of the expression domains of knirps and giant, through a mechanism similar to the regulation of homeotic genes. The regionalization of the Drosophila embryo depends on the maternally supplied products of bicoid (bcd), hunchback (hb), and nanos (nos). Nos represses the translation of the maternal HB mRNA in the posterior embryonic region. This permits the expression of the zygotic gap genes knirps (kni) and giant (gt), which specify posterior identities. These genes would otherwise be repressed by Hb. Embryos from nos/nos mothers form no abdominal segments, but this phenotype can be rescued by a total lack of hb in the maternal germline. It can also be dominantly rescued by the mutation of maternally supplied regulator molecules that normally repress kni and gt in the zygote. Pelegri and Lehmann (1994) have shown that certain mutant products of the PcG genes E(z), Psc, and pleiohomeotic can partially rescue nos by such a maternal effect. To determine if mutation of mxc also affects this regulation, the cuticles of embryos were examined from mxc/+;hb nos/nos mothers that were heterozygous for different mxc mutations. This genetic background was used because a decrease in the amount of maternal hb product can partially rescue the nos phenotype in F1 embryos. Such embryos can differentiate a few abdominal denticle belts and form an adequate background to evaluate increased rescue of nos. Thus loss-of-function PcG mutations should have a strong effect on rescue, and the embryos from hb nos/nos mothers that have two PcG mutations in their genetic background should permit increased rescue of the nos phenotype (Saget, 1998).

Any of three E(z)son (suppressor of nanos) alleles or a hypomorphic pleiohomeotic allele partially rescue the phenotypes of hb nos/nos progeny by a maternal effect; deficiencies covering E(z) or the Psc/Su(z)2 complex also allow some maternal rescue of hb nos/nos progeny, yet the strongest effect is observed with the gain-of-function E(z)son alleles. The EMS-induced allele mxcG48 rescues the hb nos/nos progeny phenotype, whereas a deficiency of mxc does not. Some rescue with the Psc/Su(z)2 complex deletion Df(2)vgB is also observed and strong rescue (consistently >50%) is observed with an EMS-induced pleiohomeotic allele phob, described as amorphic. This suggests that phob and mxcG48 are probably not amorphic alleles, and that maternal rescue of hb nos/nos progeny by a PcG gene is most efficient with a non-null mutation (Saget, 1998).

Segmentation of embryos from transheterozygous mothers was also examined. Because neither a reduction of wild-type PcG product nor two PcG mutations in trans in the hb nos/nos mothers increases nos rescue, these data strongly suggest that, whatever the mechanism of gap gene regulation by these PcG mutations may be, it does not function like the PcG-mediated maintenance of homeotic gene expression in embryos and in imaginal discs. The strong rescue provided by several non-null EMS-induced mutations, which may produce mutant proteins, leads to a proposal that modified PcG proteins are poisoning a normal process. How this process depends on wild-type regulation by PcG products has yet to be established (Saget, 1998).

Neural precursors often generate distinct cell types in a specific order, but the intrinsic or extrinsic cues regulating the timing of cell fate specification are poorly understood. Drosophila neuroblasts sequentially express the transcription factors Hunchback->Krüppel->Pdm->Castor, with differentiated progeny maintaining the transcription factor profile present at their birth. Hunchback is necessary and sufficient for first-born cell fates, whereas Krüppel is necessary and sufficient for second-born cell fates: this is observed in multiple lineages and is independent of the cell type involved. It is proposed that Hunchback and Krüppel control early-born temporal identity in neuroblast cell lineages (Isshiki, 2001).

To begin investigating birth order dependent cell fate specification in the Drosophila CNS, the morphological and molecular features that distinguish early- versus late-born neurons were investigated. The axon projections and cell positions of first-born neurons were examined from datasets of DiI-labeled neuroblast clones, and find that first-born neurons typically occupy the deepest (most internal) position in the clone and have the longest axon projections of any cell in the clone (see also Brody, 2000). Conversely, later-born neurons lie in more superficial positions (nearest the ventral epithelium) and have relatively short projections (Isshiki, 2001).

What genes might regulate these birth order-specific neuronal properties? It is known that deep layer neurons are Hb+, middle layer neurons are Pdm+, while superficial layer neurons are Cas+ (Kambadur, 1998). These findings have been confirmed in this study and Kr is identified as a new deep layer transcription factor. Kr is weakly detected in Hb+ neurons, and strongly Kr+ neurons define a deep layer that lies between the Hb+ and Pdm+ layers. Some neurons showing coexpression of Kr/Pdm and Pdm/Cas are also detected. Tests were made to see whether Hb, Kr, Pdm, and Cas are expressed in a temporal order within neuroblasts at the time each layer of neurons is being generated. Indeed, sequential, transient expression of Hb -> Kr -> Pdm -> Cas in neuroblasts was observed that is 'stabilized' in progeny born during each window of gene expression (Kambadur, 1998; Brody, 2000; Isshiki, 2001). In addition, Pdm is transiently expressed in a subset of newborn Hb+ neuroblasts and their first-born GMCs, probably due to persistence of Pdm from the neuroectoderm, but it is usually not maintained in their Hb+ neuronal progeny (Kambadur, 1998; Isshiki, 2001).

Interestingly, the temporal expression pattern of Hb, Kr, Pdm, and Cas within neuroblasts parallels the spatial pattern of these genes during segmentation. Hb, Kr, Pdm, and Cas are detected in progressively more posterior domains at cellular blastoderm, respectively. Thus, the spatial order of these genes during segmentation is the same as their temporal order in neuroblasts, raising the possibility of a conserved gene cassette used in both segmentation and neurogenesis (Isshiki, 2001).

Identified neuroblast lineages were assayed to test the hypothesis that transient Hb -> Kr -> Pdm -> Cas expression in neuroblasts is stably maintained in neuronal progeny born during each window of gene expression. Three model neuroblast lineages were characterized, an early forming neuroblast (7-1) and two late-forming neuroblasts (7-3 and 2-4), where specific neuronal progeny were tracked from birth to differentiation. Although early- and late-forming neuroblasts begin their cell lineages hours apart, all show the same sequential, transient Hb -> Kr -> Pdm -> Cas pattern of expression (Isshiki, 2001).

Neuroblast 2-4 sequentially expresses Hb/Kr, Kr, Kr/Pdm, Pdm, Pdm/Cas, and Cas. GMCs and neurons with most of these expression patterns can be detected in deep to superficial layers of the CNS, respectively, except Kr+/Pdm+ or Pdm+ GMCs were rarely observed. In addition, GMC-1 is transiently Pdm+. To track gene expression patterns at the level of identified neuronal/glial progeny, gene expression was examined in the neuroblast 7-3 and 7-1 lineages (Isshiki, 2001).

Neuroblast 7-3 produces only three GMCs: GMC-1 generates the EW1 interneuron and the GW motoneuron, GMC-2 produces the EW2 interneuron and an EW2 sibling that rapidly undergoes programmed cell death, and GMC-3 differentiates directly into the EW3 interneuron. These neurons are referred to here by the abbreviated names of 1/1G, 2, 3 (respectively) to reflect their birth order. Neuroblast 7-3 sequentially expresses Hb, Hb/Kr, Kr, Kr/Pdm, Pdm; it is never Cas+. In addition, the newborn neuroblasts 7-3 and GMC-1 are transiently Pdm+. GMC-1 is Hb+ Kr+ and generates the Hb+ Kr+ 1/1G neurons; GMC-2 is Kr+ and generates the Kr+ interneuron 2, and GMC-3 is Pdm+ and generates the transiently Pdm+ interneuron 3 (Isshiki, 2001).

Neuroblast 7-1 generates over 20 GMCs, but only the first five GMCs express the nuclear marker Even-skipped (Eve). Each of the five Eve+ GMCs produces one Eve+ motoneuron (named U1, U2, U3, U4, U5, based on birth order) and one Eve- sibling neuron which cannot be tracked. Neuroblast 7-1 sequentially expresses Hb/Kr, Kr, Kr/Pdm, Pdm, and Pdm/Cas. The Hb+ Kr+ neuroblast produces two Hb+ Kr+ GMCs which develop into Hb+ Kr+ U1 and U2 motoneurons; the Kr+ neuroblast generates a Kr+ GMC-3 which produces the Kr+ U3 motoneuron; and GMC-4 and GMC-5 are born during the Pdm+ and Pdm+ Cas+ phases of gene expression, respectively, and produce the Pdm+ U4 motoneuron and the Pdm+ Cas+ U5 motoneuron (Isshiki, 2001).

Three conclusions are drawn from this detailed gene expression analysis: (1) nearly all of the 30 known neuroblasts go through an invariant temporal pattern of Hb -> Kr -> Pdm -> Cas gene expression, including early-forming and late-forming neuroblasts; (2) Hb -> Kr -> Pdm -> Cas gene expression is transient in neuroblasts, but is maintained in differentiated neuronal progeny; (3) Hb -> Kr -> Pdm -> Cas gene expression is correlated with birth order and not a particular cell type. For example, Hb+ progeny are all early-born, but can differentiate as interneurons, motoneurons, or glia depending on their parental neuroblast (Isshiki, 2001).

In the wild-type neuroblast 7-3 lineage, the first-born GMC and its 1/1G neuronal progeny are Hb+ Kr+; later-born interneurons 2 and 3 do not express Hb. In addition, the molecular markers Zfh-1, Zn finger homeodomain 2 (Zfh-2), and the neurotransmitter corazonin label different subsets of these neurons. In embryos specifically lacking Hb in the CNS (hb CNS mutants), either a duplication of interneuron 2 at the expense of the first-born 1/1G neurons is observed (11%), consistent with a duplication of GMC-2 fate, or a specific loss of the first-born 1/1G neurons (89%) is observed (Isshiki, 2001).

It cannot be distinguish whether loss of first-born 1/1G neurons is due to cell death or due to 'skipping' of the GMC-1 fate (i.e., the lineage begins with GMC-2). In contrast, when neuroblast 7-3 is forced to continuously express Hb ('UAS-hb'), there are additional neurons in the lineage (as many as 11 cells -- average of 6.3 cells), and all exhibit first-born 1/1G fates based on molecular markers, neurotransmitter expression, and characteristic axon projections. It is concluded that Hb is necessary for normal GMC-1 development, but not later-born cell fates; that continuous Hb can transform all progeny toward a GMC-1 fate, and that continuous Hb expression leads to an extension of the neuroblast cell lineage (Isshiki, 2001).

In the wild-type neuroblast 7-1 lineage, the first two GMCs and their two Eve+ U1, U2 motoneuron progeny are Hb+; later progeny do not express Hb. In hb CNS mutants, Eve+ GMC-1/GMC-2 or their Eve+ U1, U2 motoneuron progeny were rarely detected; however, later-born GMCs and neurons develop normally based on molecular marker expression. Whether loss of the Eve+ U1, U2 motoneurons is due to cell death or due to skipping of the first two GMCs fates cannot be distinguised. In contrast, when neuroblast 7-1 is forced to continuously express Hb, there are extra Eve+ cells (as many as 19 cells; average of 14.4 cells), and all differentiate as early-born U1/U2 motoneurons based on molecular marker expression. Thus, Hb is necessary for normal GMC-1 and GMC-2 development without affecting later-born cell fates, and continuous Hb can transform many or all progeny toward a GMC-1/GMC-2 fate (Isshiki, 2001).

Is Hb required for the specification of all first-born neurons? To broaden this analysis, the first-born progeny were assayed from the well-characterized 1-1 and 4-2 lineages. Both lineages produce an Eve+ GMC-1; in the 1-1 lineage, the NB generates the Eve+ aCC motoneuron/pCC interneuron siblings, whereas in the 4-2 lineage, the NB produces the Eve+ RP2 motoneuron and its Eve- sibling. In hb CNS mutants, the first-born Eve+ neurons typically survive but abnormally express Zfh-2, a marker for later-born neurons, and the aCC and RP2 motoneurons fail to project to their proper dorsal muscle target; both phenotypes are consistent with a transformation of GMC-1 to a later-born GMC fate. In contrast, when all neuroblasts are forced to continuously express Hb, there are duplications of the first-born aCC/pCC (17%), duplications of the first-born RP2 (6%), or triplications of RP2 (4%). These results show that hb regulates first-born motoneuron and interneuron cell fates in the neuroblast 1-1 and 4-2 lineages (Isshiki, 2001).

If hb controls first-born identity in all neuroblast lineages, then severe motoneuron axon projection defects should be observed because many motoneurons derive from Hb+ first-born GMCs. Wild-type embryos have about 35 motoneurons with a stereotyped projection pattern to ventral and dorsal bodywall muscles, including the Hb+ aCC, RP2, U1, and U2 motoneurons that innervate dorsal muscles. hb CNS mutants show a reduction in the number of motoneuron projections, particularly to the dorsal muscles. Embryos where hb is misexpressed in all neuroblasts show the opposite phenotype: a striking increase in the number of motoneurons, particularly to the dorsal muscles. These results suggest that hb regulates first-born cell fate and/or axon projection patterns in most or all of the many neuroblast lineages that produce first-born motoneurons (Isshiki, 2001).

These results suggest that hb is required for specifying first-born GMC identity in lineages where first-born GMCs produce motoneurons or interneurons. To determine whether hb specifies first-born cell fate in lineages that produce glia, the thoracic 6-4 (6-4T) neuroblast lineage, in which the first-born GMC produces glia, and the 7-4 lineage, in which only later-born GMCs generate glia, were examined. The 6-4T lineage produces a Hb+ GMC-1, which produces the two Hb+ MM-CB glia that migrate to the midline, whereas neuroblast 7-4 first generates Hb+ interneurons, and then produces many Hb- glia, including two dorsoventral channel glia located at the midline. hb CNS mutants lack the first-born MM-CB glia at the midline, but have no change in the number of later-born channel glia at the midline. In contrast, forced misexpression of hb in all neuroblasts results in extra MM-CB glia at the midline and a decrease in the number of midline channel glia. These results, together with the neuronal phenotypes, strongly suggest that Hb regulates first-born temporal identity, rather than cell type identity, within multiple neuroblast lineages (Isshiki, 2001).

In the wild-type neuroblast 7-3 lineage, the first-born GMC-1 and its 1/1G neuronal progeny are Hb+ Kr+, while the second-born GMC-2 and interneuron 2 are Hb- Kr+. In embryos lacking Kr CNS expression (Kr CNS mutants), the first-born 1/1G sibling neurons are variably affected: both can be missing (17%), 1G can be missing (73%), or both can be normal (10%); however, the second-born interneuron 2 is always missing (100%) and the third-born interneuron 3 is almost always normal (90%). Absence of interneuron 2 could be due to cell death or due to skipping of the GMC-2 fate. In contrast, when neuroblast 7-3 is forced to continuously express Kr ('UAS-Kr,'), there are extra cells in the lineage (4-8; average of 5.5), and all but two cells differentiate as GMC-2 derived interneuron 2; the two unaffected cells are the GMC-1 derived 1/1G neurons. It is concluded that Kr contributes to GMC-1 development (where it is expressed with Hb) and is essential for GMC-2 development (where it is expressed without Hb); that continuous Kr can transform all progeny except GMC-1 toward a GMC-2 fate, and that continuous Kr leads to an extension of the neuroblast cell lineage (Isshiki, 2001).

In the wild-type neuroblast 7-1 lineage, the first two GMCs and their U1/U2 motoneuron progeny are Hb+ Kr+, GMC-3 and its U3 motoneuron progeny are Kr+, and subsequent GMCs do not express Kr. In Kr CNS mutants, one of the U3/U4 motoneurons is frequently missing (73%), although all earlier- and later-born neurons develop normally. It is suspected that the missing neuron is the normally Kr+ U3, based on cell position, but markers to distinguish U3/U4 in Kr mutant embryos are not available. As in the 7-3 lineage, the Kr phenotype may arise through cell death or a skipping of the GMC-3 fate. In contrast, continuous expression of Kr in neuroblast 7-1 results in extra Eve+ neurons (8-14; average of 10.0) with all differentiating as U3 or U4 motoneurons, except the normal pair of early-born U1/U2 motoneurons. It is suspected all neurons have the normally Kr+ U3 fate, but markers are not available to distinguish U3/U4 fates. It is concluded that Kr is necessary for U3 motoneuron development, and that continuous Kr can transform most or all 7-1 progeny, except the first-born U1/U2 neurons, toward a U3 neuron fate (Isshiki, 2001).

Accurate temporal regulation of Hb, Kr, and Cas is critical for proper CNS development, so it is important to determine the mechanisms that regulate sequential gene expression in neuroblasts. Regulatory interactions between Hb, Kr, Pdm, and Cas can be detected using misexpression assays: overexpression of Hb activates Kr and represses Pdm and Cas; overexpression of Kr activates Pdm, represses Cas, but has no effect on Hb expression; and Pdm positively regulates Cas expression (Brody, 2000), leading to the model that each gene can activate the next gene in the pathway and repress the 'next plus one' gene. These interactions are not necessary for driving sequential gene expression, however, as it is observed that hb, Kr, or cas mutations have only subtle alterations in the remaining gene expression profiles, and loss of hb or Kr does not appear to affect the fate of cells born later in the lineage based on existing markers. Thus, an independent pathway must also drive the sequential expression of Hb -> Kr -> Pdm -> Cas in neuroblasts. This mechanism involves cell cycle progression, directly or indirectly, because newly formed neuroblasts remain Hb+ Kr+ if they are cell cycle-arrested before their first division. They rarely if ever make the transition to Kr+ Hb-, Pdm+, or Cas+ (Weigmann, 1995 and Cui, 1995). It is concluded that a cell cycle-dependent 'clock' is required to drive the transitions in Hb -> Kr -> Pdm -> Cas gene expression (Isshiki, 2001).

The data do not support a model in which global temporal cues trigger gene expression transitions simultaneously in all neuroblasts. Most of the 30 neuroblasts, including the earliest neuroblasts to form (e.g., 7-1 and 7-4) and some of the latest neuroblasts to form (e.g., 2-4 and 7-3), go through the same Hb -> Kr -> Pdm -> Cas gene expression cascade. Exceptions are neuroblasts 2-1, 3-3, 5-1, and 6-1 (which start with Kr, Pdm, or Cas). Thus, early-forming neuroblasts can generate Cas+ progeny at the same time that late-forming neuroblasts produce Hb+ progeny (Isshiki, 2001).

A model is favored in which the timing of Hb -> Kr -> Pdm -> Cas expression is regulated primarily by a cell cycle-dependent clock but also by regulation within the Hb -> Kr -> Pdm -> Cas pathway (Kambadur, 1998). Evidence for the latter mechanism is that misexpression studies show that Hb, Kr, Pdm, and Cas typically activate the next gene in the pathway and repress the 'next plus one' gene in the pathway; and that loss of function mutations can result in premature expression of later genes in the pathway and the skipping of GMC fates. Evidence for the cell cycle-dependent clock mechanism is that hb and Kr mutants have relatively subtle changes in hb, Kr, pdm, or cas expression or in later-born GMC fates; that cell cycle arrested neuroblasts remain Hb+ Kr+ and fail to make a transition to Hb- Kr+, Pdm+, or Cas+; and that when neuroblast 1-1 is cell cycle arrested for several hours prior to its first cell division and then triggered to divide, it will produce an Eve+ GMC-1 instead of a later-born Eve- GMC (Weigmann, 1995), highlighting the importance of the cell cycle progression rather than developmental time in regulating GMC identity (Isshiki, 2001).

All early developing neuroblasts, such as 7-1, produce two Hb+ GMCs before downregulating Hb expression, whereas many late developing neuroblasts, such as 7-3, produce just one Hb+ GMC. How do neuroblasts regulate the number of Hb+ GMCs produced? All neuroblasts could express Hb for the same length of time but have different cell cycle rates, or all neuroblasts may have the same cell cycle rate but vary the length of Hb expression (Isshiki, 2001).

Hb, Kr, and Cas are transiently expressed in neuroblasts but maintained in GMC and neural progeny (Pdm can be transient or stable in neuroblast progeny, depending on the lineage) (Kambadur, 1998; Brody, 2000; Isshiki, 2001). By inheriting and maintaining the gene expression profile of their parental neuroblasts, GMCs can 'memorize' their birth order. This seems a powerful and efficient way for stem cells to make a variety of fate-restricted progeny in invariant sequence. A similar mechanism may be used during vertebrate cortical and retinal development, where precursors transiently express genes that are maintained in a subset of differentiated progeny. It is currently unknown what distinguishes transient neuroblast expression from persistent GMC/neuronal expression (Isshiki, 2001).

Loss of Hb or Kr from early-born GMCs results in loss or transformation of neurons normally derived from these GMCs, but later-born neurons develop normally based on nuclear markers, neurotransmitter expression, and axon projections. Neuronal loss in hb and Kr mutants is likely due to multiple mechanisms: (1) Cell death. Necrotic Eve+ or Eg+ neurons can be detected in the 7-1 or 7-3 lineages of Kr CNS mutants, supporting a cell death model. Although necrotic Eve+ or Eg+ neurons in hb mutants have not been detected, more early 7-3 lineages are seen in which three Prospero+ GMCs are born than mature 7-3 lineages with progeny from all three GMCs, suggesting that cell death also occurs in hb mutants. (2) GMC skipping (i.e., the neuroblast skips a GMC fate without duplicating a later-born fate). In Kr mutants, neuroblast 7-3 typically produces two Prospero+ GMCs at the time GMC-1 and GMC-2 are normally born, consistent with a skip of GMC-2 fate. Similarly, hb mutants often produce only two Prospero+ GMCs in the 7-3 lineage, consistent with a skip of the GMC-1 fate. It is not possible to assay for GMC skipping in the 7-1 lineage. (3) GMC transformation. hb mutants clearly show transformation of GMC-1 to GMC-2 fate in the 7-3 lineage, based on the observed duplication of interneuron 2 at the expense of the 1/1G sibling neurons. This phenotype may arise if the endogenous Kr in GMC-1 is sufficient to induce GMC-2 fate in the absence of Hb. In Kr mutants, a similar transformation of GMC-2 into the GMC-3 fate is never observed, perhaps because the Kr- GMC-2 does not prematurely express Pdm. Thus, there is evidence that hb mutants show all three phenotypes, GMC death, skipping, and transformation, while Kr mutants show only GMC death and skipping phenotypes (Isshiki, 2001).

Misexpression of Hb or Kr can transform all GMCs toward a first-born or second-born fate, respectively. This is likely to be a transformation of GMC identity, rather than an increase in the proliferation of early-born GMCs, because the extra early-born neurons are produced at the expense of the later-born neurons in the lineage, and the extra cells are not reduced in size (extra embryonic divisions lead to reduced cell size). Interestingly, Kr misexpression fails to transform early-born Hb+ GMCs into the later-born Kr+ fate. Kr is induced early enough to affect GMC-1 in the 7-3 lineage because GMC transformations are seen in other lineages (e.g., 7-1) prior to the time GMC-1 is born in the 7-3 lineage, yet it has no effect. The model that first-born Hb+ fates are dominant over second-born Kr+ fates is favored. In the future, cell type-specific Hb misexpression studies could be used to determine precisely when birth order-specific cell fates become fixed: in neuroblasts, GMCs, or neurons? It would also be interesting to determine if a pulse of Hb expression midway through a neuroblast lineage is sufficient to induce first-born cell fates, and if so, does the temporal program resume or reset to the beginning of the lineage after the Hb pulse ends (Isshiki, 2001)?

hb, Kr, pdm, and cas are not the only genes controlling temporal identity in neuroblast lineages. Some markers for first-born fate occur normally in hb mutants, such as Eve expression in the 1-1 and 4-2 lineage. Moreover, Hb misexpression may not fully transform every cell in the 7-3 lineage to a first-born fate: a full transformation would produce equal numbers of interneuron 1/motoneuron 1G siblings, but typically only 2-3 motoneurons and 5-6 interneurons are seen, suggesting that only the first 2-3 GMCs are fully transformed to a first-born fate. Finally, in most lineages, there are GMCs produced after Cas expression ends; additional genes such as grainyhead (Brody, 2000) may specify the temporal identity of these GMCs (Isshiki, 2001).

Hb is expressed in virtually all first-born GMCs, and these can differentiate into motoneurons, interneurons, or glia, depending on the neuroblast lineage. Similarly, high level Kr is detected in virtually all second-born GMCs (i.e., the GMCs following the Hb+ GMCs), and they can differentiate into motoneurons, interneurons, or glia. Not only are Hb and Kr expressed in multiple cell types, but they are necessary and sufficient for the proper cell fate specification of motoneurons, interneurons, or glia, depending on the neuroblast lineage (Isshiki, 2001).

An extremely interesting question is how GMC 'temporal identity' (regulated by Hb and Kr) is coordinated with individual 'neuroblast identity' to achieve the proper sequence of cell types that characterizes each neuroblast lineage. The Hb protein, its putative mammalian ortholog Ikaros, and the mammalian Kr-related EKLF protein all associate with chromatin remodeling proteins, and both Hb and Kr are thought to regulate chromatin-mediated heritable expression of homeotic genes. Thus, Hb and Kr may modulate chromatin structure such that different subsets of genes are accessible for transcription in first-born versus second-born GMCs, with the palette of genes expressed by a first-born or second-born GMC, dependent on the neuroblast-specific transcription factors they inherit. In this manner, neuroblast identity might be integrated with GMC temporal identity to create the unique cell types characterizing each neuroblast lineage. This is conceptually similar to homeotic genes and tissue-specific genes working together to uniquely specify distinct cell types in each tissue at different anterior-posterior levels of the body axis (Isshiki, 2001).

When neuroblast 7-3 is forced to continuously express Hb or Kr, it generates an extended lineage of up to ten neurons instead of four neurons and one programmed cell death. The increase does not appear to be due to an extra round of cell division by the normally postmitotic neurons because the extra cells are produced at the expense of later-born cell types and because smaller cells are not seen (as expected, since extra cell divisions in the embryo lead to smaller cell size. It is proposed that Hb or Kr misexpression results in production of extra GMCs, and it is suggested that each neuroblast has an intrinsic mechanism for triggering quiescence that is related to successful transition from Hb and Kr expression to later genes in the hierarchy. These data also show that Hb and Kr can regulate features of neuroblast cell biology (cell cycle control) in addition to regulating GMC temporal identity (Isshiki, 2001).

The temporal gene expression in neuroblasts (early to late: Hb -> Kr -> Pdm -> Cas) mimics the major domains of gene expression at cellular blastoderm (anterior to posterior: Hb -> Kr -> Pdm -> Cas). Additional studies will be needed to discern common and distinct regulatory features between Hb, Kr, Pdm, and Cas expression during segmentation and neurogenesis. The expression of all genes at both stages of development raises the question of which function is ancestral. Hb is detected in the CNS of various arthropod, leech, and C. elegans embryos. In mammals, Hb-related genes of the Ikaros family are best known for regulating immune development, but some also show CNS expression. A mammalian Pdm homolog, SCIP/Oct-6, is expressed in specific cortical layers of the brain, and a mammalian Cas ortholog exists but has not been characterized. It will be interesting to determine whether genes regulating temporal identity in Drosophila neuroblasts have similar functions in the mammalian CNS or immune system (Isshiki, 2001).

hunchback: Biological Overview | Evolutionary Homologs | Regulation | Targets of activity | Protein Interactions | Post-transcriptional Regulation | Developmental Biology | References

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

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