ABSTRACT
The gut-innervating stomatogastric nervous system of Drosophila, unlike the central and the peripheral nervous system, derives from a compact, single layered epithelial anlage. Here we report how this anlage is initially defined during embryogenesis by the expression of proneural genes of the achaete-scute complex in response to the maternal terminal pattern forming system. Within the stomatogastric nervous system anlage, the wingless-dependent inter-cellular communication system adjusts the cellular range of Notch-dependent lateral inhibition to single-out three achaete-expressing cells. Those cells define distinct invagination centers which orchestrate the behavior of neighboring cells to form epithelial infoldings, each headed by an achaete-expressing tip cell. Our results suggest that the wingless pathway acts not as an instructive signal, but as a permissive factor which coordinates the spatial activity of morphoregulatory signals within the stomatogastric nervous system anlage.
INTRODUCTION
Neurons are generated by the proliferation of progenitor cells that develop from undifferentiated and pluripotent ectodermal cells at an early stage of embryogenesis. In insects, recent work has focused on the origin of the central (CNS) and the periph-eral nervous system (PNS) (reviewed by Campos-Ortega, 1993; Jan and Jan, 1993). Cell ablation experiments in the grasshop-per embryo highlighted two important features of neural precursor formation in insects (Doe and Goodman, 1985). First, distinct groups of ectodermal epithelial cells display the potential to form the precursor cells, and second, cell interactions result in the selection of a single cell within each cell group that can potentially form a neural precursor. Mutagenesis screenings in Drosophila identified two classes of genes that regulate each of these processes. The ‘proneural genes’ (Romani et al., 1987; Ghysen and Dambly-Chaudiere, 1989) control competence of ectodermal cells to become neural precursor cells, and the activ-ities of the ‘neurogenic genes’ (Poulson, 1937; Lehmann et al., 1981) prevent more than one among the competent group of cells from adopting a neural fate. Once identified, the neural pre-cursors, termed neuroblasts (CNS) or sensory organ mother cells (PNS), delaminate from the ectodermal epithelium and give rise to the characteristic and stereotyped patterns of neurons and axonal connections observed during late stages of embryogene-sis (reviewed by Campos-Ortega, 1993; Goodman and Doe, 1993; Jan and Jan, 1993).
Proneural genes include the three transcription units of the achaete-scute complex (AS-C) (Garcia-Bellido, 1979; Romani et al., 1987). The AS-C genes encode HLH-type transcription factors (Villares and Cabrera, 1987) which are thought to control the expression of neural-specific developmental genes. In mutants that fail to express the AS-C genes, neuroectoder-mal cells lose their ability to follow a neural fate and contribute instead to the epidermis (Garcia-Bellido, 1979; Jiménez and Campos-Ortega, 1979; Dambly-Chaudiere and Ghysen, 1987; Jiménez and Campos-Ortega, 1990). The AS-C genes are initially expressed in small clusters of ectodermal cells (Cubas et al., 1991; Skeath and Carroll, 1992), the proneural cell clusters. From this cell cluster, the neural precursor cell is singled-out and continues AS-C gene expression (Cabrera et al., 1987). The singling-out of neural precursor cells within the proneural cell clusters is caused by a process of cell-cell com-munication referred to as lateral inhibition (Simpson, 1990) and results in a concomitant loss of AS-C gene expression in the other cells of the proneural cluster, which thereby restores an epidermal cell fate.
Singling-out by lateral inhibition is mediated by the activity of the neurogenic genes which are integrated into the Notch-mediated signaling pathway (Lehmann et al., 1981; Artavanis-Tsakonas et al., 1991; Fortini and Artavanis-Tsakonas, 1993). Within this intercellular communication system, the gene products of Delta (Dl) and Notch (N) seem to act as ligand and receptor, respectively (Heitzler and Simpson, 1991). Their activities are mediated by downstream components such as Enhancer of split (E(spl)) or mastermind (mam; reviewed by Campos-Ortega, 1993). Loss-of-function mutants in any one of these genes fail to single-out AS-C gene-expressing cells and thereby show an overproduction of neural precursor cells both in the central and the peripheral nervous system (Brand and Campos-Ortega, 1988; Goriely et al., 1991).
Here we describe the early development of the Drosophila stomatogastric nervous system (SNS), a system which coordi-nates ingestion, swallowing and peristaltic movements of the gut (reviewed by Penzlin, 1985). The SNS is composed of four ganglia which derive from a compact epithelial anlage within the dorsal roof of the stomodeal invagination. Once the SNS anlage is formed in response to positional information provided by the maternal terminal system (Perrimon, 1993), three distinct AS-C-expressing cells are singled-out under the control of the N-mediated signaling pathway. The range of N-mediated signaling depends on wingless (wg) activity. Our results suggest that wg acts as a permissive rather than instructive component within the SNS anlage.
MATERIALS AND METHODS
Drosophila strains and mutant embryos
Drosophila strains were kept under standard conditions. Mutant alleles are described in Lindsley and Zimm (1992). Embryos were collected from stocks balanced with FM7 (Notch55e and Df(1)B57,ac− sc− l’sc−), CyO (mastermindIB99 and winglessIG23) or TM3, Sb (Delta9P, Enhancer of splitx1, forkheadXT6, huckebeinA32 and Df(3)taillessG). The balancer chromosomes carried a lacZ reporter gene containing the fushi tarazu (FM7) or the hunchback promoter (CyO and TM3) which allow homozygous mutant embryos to be unambiguously identified on the basis of the lack of hunchback or fushi tarazu staining patterns.
torPM embryos derived from homozygous torPM females. Df(1)B57; Kr1 double mutant embryos were distinguished on the basis of the lack of external sense organs in the dorsal epidermis of the trunk and on the basis of the segmentation phenotype of Kr (Gaul and Jäckle, 1987). dshv26, armXM19 and sggM11-1 mutant embryos derived from females bearing homozygous mutant germ line clones. Germ line clones were generated by the ‘FLP-DFS’ technique as described by Siegfried et al. (1994). These females were mated to FM7, ftz-lacZ males. This allows one to distinguish the respective mutant embryos that have received neither the maternal nor the zygotic wild-type product by the absence of ftz-lacZ reporter gene expression.
Staining and embedding procedures
Antibody stainings were performed as previously described (Macdonald and Struhl, 1986). Antibodies were diluted at the following concentrations: 1:50 mAb22C10 (Fujita et al., 1982), 1:50 anti-fasciclin II (Grenningloh et al., 1991), 1:50 anti-fkh (Weigel et al., 1989b), 1:500 anti-Kr (Gaul and Jäckle, 1987), 1:3 anti-ac (Skeath and Carroll, 1992), 1:20 anti-crb (Tepass and Knust, 1993), 1:1000 rabbit anti-β-galactosidase (Cappel) and 1:200 mouse anti-β-galac-tosidase (Promega). Whole-mount in situ hybridizations using digoxigenin-labeled probes for l’sc, sc and wg cDNAs were performed according to the method of Tautz and Pfeifle (1989). After antibody staining or whole-mount in situ hybridization, embryos were embedded in glass capillaries to allow rotation under the compound microscope (Prokop and Technau, 1993).
RESULTS
Morphology and origin of the Drosophila SNS
The SNS is located in the head region close to the embryonic brain. It consists of four prominent interconnected ganglia (Fig. 1A; overview), i.e. the frontal ganglion (FG), the esophageal ganglia (EG1 and EG2) and the proventricular ganglion (PG) (González-Gaitán et al., 1994). They can be visualized by mAb22C10 (Fujita et al., 1982) (Fig. 1B-D,G,H) or by anti-fasciclin II (Grenningloh et al., 1991) antibody staining (Fig. 1E,F) at late stages of embryogenesis (González-Gaitán et al., 1994; Hartenstein et al., 1994).
The SNS of late Drosophila embryos. (A) Diagram illustrating the components of the SNS (red) and their relative positions with respect to morphological landmarks such as different portions of the gut and brain as seen by mAb22C10 or anti-fasciclin II antibody stainings (B-H); anterior is left. (B) Dorsal view of the SNS (mAb22C10 staining). The frontal ganglion (FG) is formed by two groups of cells which are interconnected by the frontal commisure (fcm). From a mid-dorsal position in the fcm, the frontal nerve (fn) projects anteriorly and innervates the two palisades of dorsal pharyngeal muscles. The recurrent nerve (rn) projects posteriorly and passes below the supraesophageal commisure (sec) connecting the brain hemispheres. At the posterior wall of the pharynx, the rn splits into two short branches connecting, to the right, with the esophageal ganglion 1 (EG1) and, to the left, with the esophageal ganglion 2 (EG2). (C) Lateral view of the embryo shown in B. The EG1 is formed by a cluster of 10 neurons in the posterior-most part of the pharynx (in focus in B) which extends into a row of seven pairs of neurons along the esophagus and a pair of neurons slightly separated from the others.(D) Different lateral focal plane showing that the rn splits to target the EG1 and EG2 at the posterior wall of the pharynx. EG2 connects to the proventricular ganglion (PG) via the proventricular nerve (pn). The PG is located on top of the proventriculus at a position where it connects to the esophagus (not shown). (E) Lateral view of an earlier stage 16 embryo (anti-fasciclin II antibody staining) showing that each cell group of the FG connects ipsilaterally to the subesophageal ganglion via the frontal connective (fcn). In the ventral cord of the CNS, a total of six fascicles within each longitudinal connective can be observed (Grenningloh et al., 1991; not shown). A dorsal fascicle and the most ventral one fasciculate together (arrowhead) at the subesophageal ganglion; they split again anteriorly to contribute to the brain neuropile (bn) and the fcn, respectively. (F) Different lateral focal plane showing that from a mid-dorsal position in the fcn, the rn and two posterior frontal connectives (pfc) project posteriorly. Each pfc runs on top of the pharynx and turns ventrally after passing below the sec to connect ipsilaterally to the subesophageal ganglion in a position near the meeting point of the fcn with the longitudinal connective at the subesophageal ganglion (SEG). In a dorsal position, neural elements of the ring gland (RG) can also be seen with anti-fasciclin II antibody staining. (G) Close-up of the proventriculus of a late stage 17 embryo (mAb22C10 staining). Two groups of nerves project from the PG. The first group is composed of three internal nerves (in) innervating the proventriculus. The proventriculus is made up of three layers as a consequence of the refolding of the gut during development: an external midgut layer (m) and two internal, recurrent (r) and esophageal (e), layers (Skaer, 1993). The three internal nerves run internally between r and e. The second group of nerves projecting from the PG is formed by three midgut nerves (mn) that are apposed externally to the proventriculus. Magnification is 4-fold higher as compared to the specimen shown in B-F. (H) Close-up of the midgut epithelium in a late stage 17 embryo (mAb22C10 staining). The mn projects on top of the midgut epithelium where they branch. Varicosities in the mn suggest the existence of synapses that coincide with the position of mAb22C10 stained cells (as well as other neural markers; not shown) integrated in the midgut epithelium (nc). Magnification is 4-fold higher as compared to the specimen shown in B-F. Orientation of embryos is anterior to the left; in C-F, dorsal side is up.
The SNS of late Drosophila embryos. (A) Diagram illustrating the components of the SNS (red) and their relative positions with respect to morphological landmarks such as different portions of the gut and brain as seen by mAb22C10 or anti-fasciclin II antibody stainings (B-H); anterior is left. (B) Dorsal view of the SNS (mAb22C10 staining). The frontal ganglion (FG) is formed by two groups of cells which are interconnected by the frontal commisure (fcm). From a mid-dorsal position in the fcm, the frontal nerve (fn) projects anteriorly and innervates the two palisades of dorsal pharyngeal muscles. The recurrent nerve (rn) projects posteriorly and passes below the supraesophageal commisure (sec) connecting the brain hemispheres. At the posterior wall of the pharynx, the rn splits into two short branches connecting, to the right, with the esophageal ganglion 1 (EG1) and, to the left, with the esophageal ganglion 2 (EG2). (C) Lateral view of the embryo shown in B. The EG1 is formed by a cluster of 10 neurons in the posterior-most part of the pharynx (in focus in B) which extends into a row of seven pairs of neurons along the esophagus and a pair of neurons slightly separated from the others.(D) Different lateral focal plane showing that the rn splits to target the EG1 and EG2 at the posterior wall of the pharynx. EG2 connects to the proventricular ganglion (PG) via the proventricular nerve (pn). The PG is located on top of the proventriculus at a position where it connects to the esophagus (not shown). (E) Lateral view of an earlier stage 16 embryo (anti-fasciclin II antibody staining) showing that each cell group of the FG connects ipsilaterally to the subesophageal ganglion via the frontal connective (fcn). In the ventral cord of the CNS, a total of six fascicles within each longitudinal connective can be observed (Grenningloh et al., 1991; not shown). A dorsal fascicle and the most ventral one fasciculate together (arrowhead) at the subesophageal ganglion; they split again anteriorly to contribute to the brain neuropile (bn) and the fcn, respectively. (F) Different lateral focal plane showing that from a mid-dorsal position in the fcn, the rn and two posterior frontal connectives (pfc) project posteriorly. Each pfc runs on top of the pharynx and turns ventrally after passing below the sec to connect ipsilaterally to the subesophageal ganglion in a position near the meeting point of the fcn with the longitudinal connective at the subesophageal ganglion (SEG). In a dorsal position, neural elements of the ring gland (RG) can also be seen with anti-fasciclin II antibody staining. (G) Close-up of the proventriculus of a late stage 17 embryo (mAb22C10 staining). Two groups of nerves project from the PG. The first group is composed of three internal nerves (in) innervating the proventriculus. The proventriculus is made up of three layers as a consequence of the refolding of the gut during development: an external midgut layer (m) and two internal, recurrent (r) and esophageal (e), layers (Skaer, 1993). The three internal nerves run internally between r and e. The second group of nerves projecting from the PG is formed by three midgut nerves (mn) that are apposed externally to the proventriculus. Magnification is 4-fold higher as compared to the specimen shown in B-F. (H) Close-up of the midgut epithelium in a late stage 17 embryo (mAb22C10 staining). The mn projects on top of the midgut epithelium where they branch. Varicosities in the mn suggest the existence of synapses that coincide with the position of mAb22C10 stained cells (as well as other neural markers; not shown) integrated in the midgut epithelium (nc). Magnification is 4-fold higher as compared to the specimen shown in B-F. Orientation of embryos is anterior to the left; in C-F, dorsal side is up.
The FG is composed of two groups of neurons, one on either side of the pharynx (Fig. 1A,B). They are connected by axon bundles contributing to the frontal commisure (Fig. 1B). They also connect to the subesophageal ganglion of the CNS via the frontal connectives (Fig. 1E,F). EG1 and EG2 are associated with the esophagus (Fig. 1A,D,F). EG1 extends as a row of seven pairs of neurons apposed to the esophagus (Fig. 1C); EG2 is composed of about 10 tightly clustered neurons (Fig. 1B,D). Both ganglia are connected to the FG via the recurrent nerve (Fig. 1B,D,F). The PG is formed by a cluster of about 10 neurons which are apposed to the proventriculus (Fig. 1A,D); Skaer, 1993). The PG is connected to the EG2 via the proventricular nerve (Fig. 1D). Axons projecting from different SNS ganglia innervate distinct portions of the gut, e.g. the FG innervates the dorsal pharyngeal muscles via the frontal nerve (Fig. 1B), the EG1 innervates the esophagus (Fig. 1C), and the PG innervates both the proventriculus and the midgut (Fig. 1G,H).
The developmental origin of the SNS ganglia can be traced back to a compact, single-layered epithelial anlage at the dorsal roof of the stomodeal invagination (Fig. 2A; Poulson, 1937; Schoeller, 1964; Campos-Ortega and Harten-stein, 1985). During stage 10 (stages according to Campos-Ortega and Hartenstein, 1985), the stomodeal invagination is characterized by the expression of forkhead (fkh; Fig. 2B), a gene involved in gut development (Weigel et al., 1989a,b). At this stage, the SNS anlage is spatially defined by the localized expression of the AS-C proneural genes achaete (ac), scute (sc) and lethal of scute (l’sc) (Romani et al., 1987), and by the expression of the gap gene Krüppel (Kr; Gaul and Jäckle, 1987) (Fig. 2C-G).
Gene expression in the SNS neural anlage. Anti-crumbs (A), anti-forkhead (B), anti-Krüppel (C,D) and anti-achaete (F) antibody staining and lethal of scute (E) and scute (G) in situ hybridization in head regions of wild-type embryos at stage 10. Arrows point to the stomodeal invagination; arrowheads point to the SNS anlage. Except in D (ventral view), lateral views are shown; anterior is left, dorsal up.
Gene expression in the SNS neural anlage. Anti-crumbs (A), anti-forkhead (B), anti-Krüppel (C,D) and anti-achaete (F) antibody staining and lethal of scute (E) and scute (G) in situ hybridization in head regions of wild-type embryos at stage 10. Arrows point to the stomodeal invagination; arrowheads point to the SNS anlage. Except in D (ventral view), lateral views are shown; anterior is left, dorsal up.
SNS development becomes morphologically distinct by the appear-ance of three evenly spaced dorsal invaginations (Schoeller, 1964; Campos-Ortega and Hartenstein, 1985; Hartenstein et al., 1994; Fig. 3A), which can be visualized by anti-crumbs (crb) antibody staining (Tepass and Knust, 1993), by fkh and by Kr expression (Fig. 3C-F). At stage 14, the invaginations detach from the developing esophagus, and each forms a vesicular structure (Fig. 3B; see also Hartenstein et al., 1994) which continues to express fkh (Fig. 3C-F) until the stereotyped pattern of the SNS ganglia emerges (cf. Fig 1B,C and Fig. 3G,H; see González-Gaitán et al., 1994; Hartenstein et al., 1994).
Invagination and vesiculation of the SNS primordium. (A) Head region of a wild-type embryo stained with anti-crumbs antibody showing three distinct invaginations (1,2,3) in the roof of the stomodeum during stage 11. (B) The invaginations detach from the lumen of the foregut (not shown), and at stage 14, three vesicles containing neural precursors segregate dorsally. (C,D) Head region of a stage 11 wild-type embryo stained with anti-forkhead antibody. (C) Lateral and (D) dorsal view of the same embryo. (E,F) Head region of a stage 11 wild-type embryo stained with anti-Krüppel antibody (E) Lateral and (F) dorsal view of the same embryo. (G,H) Head region of a stage 16 wild-type embryo stained with anti-forkhead antibody. (G) Lateral and (H) dorsal view of the same embryo. Note forkhead expression in the esophagus (es), the proventriculus (out of focus) and the SNS ganglia (FG, EG1 and EG2; PG, out of focus), but not in the neural elements of the ring gland (see Fig. 1F). Orientation of the embryos is anterior to the left; in lateral views, dorsal up. Other abbreviations: ph, pharynx; sp, salivary gland pit.
Invagination and vesiculation of the SNS primordium. (A) Head region of a wild-type embryo stained with anti-crumbs antibody showing three distinct invaginations (1,2,3) in the roof of the stomodeum during stage 11. (B) The invaginations detach from the lumen of the foregut (not shown), and at stage 14, three vesicles containing neural precursors segregate dorsally. (C,D) Head region of a stage 11 wild-type embryo stained with anti-forkhead antibody. (C) Lateral and (D) dorsal view of the same embryo. (E,F) Head region of a stage 11 wild-type embryo stained with anti-Krüppel antibody (E) Lateral and (F) dorsal view of the same embryo. (G,H) Head region of a stage 16 wild-type embryo stained with anti-forkhead antibody. (G) Lateral and (H) dorsal view of the same embryo. Note forkhead expression in the esophagus (es), the proventriculus (out of focus) and the SNS ganglia (FG, EG1 and EG2; PG, out of focus), but not in the neural elements of the ring gland (see Fig. 1F). Orientation of the embryos is anterior to the left; in lateral views, dorsal up. Other abbreviations: ph, pharynx; sp, salivary gland pit.
SNS anlage formation requires activity of the maternal terminal system
The functional significance of fkh, Kr and AS-C gene expression for SNS development is apparent from the mutant phenotypes (Figs 4,5). fkh mutant embryos lack the SNS (Fig. 4A,B) and fail to express Kr and the AS-C gene expression in the SNS anlage (Fig. 5A,B). In the absence of AS-C gene activities, remnants of SNS structures can be observed (Fig. 5C). In the absence of Kr, SNS ganglia are formed (Fig. 5D). Thus, AS-C gene expression is required for the normal SNS development, whereas Kr activity is not necessary. However, the AS-C mutant phenotype is strongly enhanced in embryos lacking both Kr and AS-C gene activities (Fig. 5E). This observation suggests that Kr and AS-C genes may function in a common developmental pathway, implying a proneural-like function of Kr during an early stage of SNS development.
Maternal and zygotic components of the terminal system are necessary for SNS development. (A,B) fkhXT6, (C) torPM, (D)hkbA32 and (E)tllG mutant embryos stained with mAb22C10 at stage 17. (A) Dorsal and (B) lateral view of the same fkhXT6 mutant embryo where SNS ganglia fail to develop. Occasionally a single cell stained with mAb22C10 (arrow) can be detected anterior to the supraesophageal commisure (sec) where the FG is normally located. Dorsally, the ring gland (rg) remains unaffected, confirming its origin outside the stomodeal SNS anlage (cf. Fig. 3G,H). (C) Embryos deriving from homozygous torPM females fail to develop SNS ganglia, while in hkbA32 (D) and tllG (E) homozygous embryos remnants of the SNS (arrowheads) are observed. (C,D) Dorsal view; (E) lateral view. Orientation of the embryos is anterior to the left and dorsal up (lateral views).
Maternal and zygotic components of the terminal system are necessary for SNS development. (A,B) fkhXT6, (C) torPM, (D)hkbA32 and (E)tllG mutant embryos stained with mAb22C10 at stage 17. (A) Dorsal and (B) lateral view of the same fkhXT6 mutant embryo where SNS ganglia fail to develop. Occasionally a single cell stained with mAb22C10 (arrow) can be detected anterior to the supraesophageal commisure (sec) where the FG is normally located. Dorsally, the ring gland (rg) remains unaffected, confirming its origin outside the stomodeal SNS anlage (cf. Fig. 3G,H). (C) Embryos deriving from homozygous torPM females fail to develop SNS ganglia, while in hkbA32 (D) and tllG (E) homozygous embryos remnants of the SNS (arrowheads) are observed. (C,D) Dorsal view; (E) lateral view. Orientation of the embryos is anterior to the left and dorsal up (lateral views).
Krüppel and the achaete-scute complex genes act synergistically during SNS development. (A) Anti-Kr and (B) anti-ac antibody stainings of fkhXT6 mutant embryos at stage 10. No expression of these genes is detected in the roof (arrows) of the stomodeum which is substantially reduced. (A,B) Orientation of the embryos is anterior to the left and dorsal up (lateral views). (C) mAb22C10 staining at stage 17 of Df(1)B57 mutant embryo, lacking the AS-C genes. Remnants of the SNS ganglia can be detected in Df(1)B57 homozygous mutant embryos. Most frequently the EG1 (arrowhead) is seen in a position apposed to the esophagus (es); remnants of the other SNS ganglia are occasionally detected. Orientation of the embryo is anterior to the left and dorsal up (lateral view). (D) mAb22C10 staining at stage 17 of Kr2 mutant embryo. Note that the SNS structures are present (FG, EG1, EG2, fn; other nerves and the PG, out of focus). The abnormal appearance of some structures is caused by the Kr segmentation defect, which affects the head involution. Orientation of the embryo is anterior to the left (dorsal view). (E) mAb22C10 staining at stage 17 of Df(1)B57; Kr1 double mutant embryo. Note that all SNS ganglia are lacking when both Kr and the AS-C genes are absent. Remnants of what appear to be SNS structures were observed infrequently. Orientation of the embryo is anterior to the left (dorsal view). Abbreviation: bh, brain hemisphere.
Krüppel and the achaete-scute complex genes act synergistically during SNS development. (A) Anti-Kr and (B) anti-ac antibody stainings of fkhXT6 mutant embryos at stage 10. No expression of these genes is detected in the roof (arrows) of the stomodeum which is substantially reduced. (A,B) Orientation of the embryos is anterior to the left and dorsal up (lateral views). (C) mAb22C10 staining at stage 17 of Df(1)B57 mutant embryo, lacking the AS-C genes. Remnants of the SNS ganglia can be detected in Df(1)B57 homozygous mutant embryos. Most frequently the EG1 (arrowhead) is seen in a position apposed to the esophagus (es); remnants of the other SNS ganglia are occasionally detected. Orientation of the embryo is anterior to the left and dorsal up (lateral view). (D) mAb22C10 staining at stage 17 of Kr2 mutant embryo. Note that the SNS structures are present (FG, EG1, EG2, fn; other nerves and the PG, out of focus). The abnormal appearance of some structures is caused by the Kr segmentation defect, which affects the head involution. Orientation of the embryo is anterior to the left (dorsal view). (E) mAb22C10 staining at stage 17 of Df(1)B57; Kr1 double mutant embryo. Note that all SNS ganglia are lacking when both Kr and the AS-C genes are absent. Remnants of what appear to be SNS structures were observed infrequently. Orientation of the embryo is anterior to the left (dorsal view). Abbreviation: bh, brain hemisphere.
Since fkh expression requires the maternal terminal signal transduction pathway (Gaul and Weigel, 1990; Weigel et al., 1990), we examined the SNS of embryos lacking the activity of torso (tor). tor encodes a membrane-spanning tyrosine kinase receptor activity (Casanova and Struhl, 1989; Sprenger et al., 1989). Upon local activation, tor initiates the terminal signal transduction pathway (reviewed by Perrimon, 1993). Fig. 4C shows that embryos fail to develop SNS structures when tor activity is absent.
The activated tor signaling pathway eventually leads to localized zygotic target gene expression including the terminal gap genes tailless (tll) and huckebein (hkb) (Weigel et al., 1990). The activities of both genes are required for fkh expression (Gaul and Weigel, 1990; Weigel et al., 1990). tll or hkb embryos fail to develop normal SNS ganglia, although remnants can be observed (Fig. 4D,E). This suggests that hkb and tll act through fkh and that the remaining fkh activity observed in their absence (Gaul and Weigel, 1990; Weigel et al., 1990) is sufficient to generate some SNS remnants.
Invagination centers are selected by N-mediated signaling
ac, sc and l’sc are initially expressed in all cells of the SNS anlage (Fig. 2E-G). While l’sc expression continues during the subsequent invagination process (Fig. 6A,B), both ac and sc expression become restricted to three cells (Fig. 6C,E). The singled-out ac- and sc-expressing cells represent invagination centers in the SNS anlage and they characterize the tip cells of each invagination (Fig. 6G,H). Once the folds are maximally extended, ac and sc become re-expressed throughout the epithelial folds and consequently all SNS cells undergo a second round of proneural gene expression (Fig. 6B,D,F).
ac and sc, but not l’sc expression, are restricted to SNS invagination centers and tip cells. (A) l’sc expression throughout the SNS primordium of late stage 10 embryos as revealed by whole-mount in situ hybridization. (B) Overall l’sc expression throughout the SNS primordium persists during stage 12; later on, l’sc expression also continues in the vesicles (not shown). (C) ac expression, which covers the SNS anlage at early stage 10 (see Fig. 2F), becomes restricted to three cells (1,2,3) as revealed by anti-achaete antibody staining at late stage 10. ac expression in these three cells persists during stage 11 when the invaginations occur (see G,H). (D) At stage 12, ac expression is reinitiated in all the cells within the invagination folds. (E) sc expression as revealed by whole-mount in situ hybridization. sc expression is restricted to three cells (1 is out of focal plane) at late stage 10.(F)sc expression is reinitiated at stage 12 in all the cells within the primordium as has been observed with ac expression (see D). (G) Anti-achaete and anti-crumbs double antibody staining showing nuclear staining (ac) and apical membrane staining (crb) in wild-type embryos at early stage 11. Note that the expression of ac is restricted to the three tip cells (1,2,3) of the incipient invagination folds (arrowheads). (H) Restricted tip cell expression of ac (1-3) persists until invagination (arrowheads) is completed late during stage 11. Orientation of embryos is anterior to the left; (A-D, F-H) lateral view (dorsal up), (E) dorsal view.
ac and sc, but not l’sc expression, are restricted to SNS invagination centers and tip cells. (A) l’sc expression throughout the SNS primordium of late stage 10 embryos as revealed by whole-mount in situ hybridization. (B) Overall l’sc expression throughout the SNS primordium persists during stage 12; later on, l’sc expression also continues in the vesicles (not shown). (C) ac expression, which covers the SNS anlage at early stage 10 (see Fig. 2F), becomes restricted to three cells (1,2,3) as revealed by anti-achaete antibody staining at late stage 10. ac expression in these three cells persists during stage 11 when the invaginations occur (see G,H). (D) At stage 12, ac expression is reinitiated in all the cells within the invagination folds. (E) sc expression as revealed by whole-mount in situ hybridization. sc expression is restricted to three cells (1 is out of focal plane) at late stage 10.(F)sc expression is reinitiated at stage 12 in all the cells within the primordium as has been observed with ac expression (see D). (G) Anti-achaete and anti-crumbs double antibody staining showing nuclear staining (ac) and apical membrane staining (crb) in wild-type embryos at early stage 11. Note that the expression of ac is restricted to the three tip cells (1,2,3) of the incipient invagination folds (arrowheads). (H) Restricted tip cell expression of ac (1-3) persists until invagination (arrowheads) is completed late during stage 11. Orientation of embryos is anterior to the left; (A-D, F-H) lateral view (dorsal up), (E) dorsal view.
In order to establish the genetic requirement for the singling-out of invagination centers, we used ac expression as a marker to examine the selection of invagination centers in neurogenic mutant embryos that lack integral components of the N-mediated signaling pathway (Lehmann et al., 1981; Artavanis-Tsakonas et al., 1991; Fortini and Artavanis-Tsakonas, 1993). In N mutant embryos, initial ac expression covers the SNS anlage as in wild type (not shown). During late stage 10, however, when the three ac-expressing cells are normally singled-out, ac expression continues throughout the SNS anlage (Fig. 7A). Instead of forming distinct invagination folds, the N mutant SNS anlage invaginates en masse (Fig. 7E). In Dl, E(spl) or mam homozygous mutant embryos, ac expression is also not restricted (Fig. 7B-D), and the SNS anlage invaginates as observed in the N mutants (Fig. 7F-H). These results indicate that the singling-out of ac-expressing cells in the wild-type SNS anlage is dependent on N-mediated lateral inhibition as observed for early CNS and PNS development. However, instead of delaminating, the ac-expressing cells of the SNS anlage define the position of invagination centers.
Singling-out of ac-expressing invagination centers depends on N-mediated signaling. (A-D) Anti-achaete and (E-H) anti-crumbs antibody stainings of embryos mutant for different components of the N-mediated signaling pathway. (A,E) Notch55e, (B,F) Delta9P, (C,G) Enhancer of splitx1 and (D,H) mastermindIB99 mutant embryos at stage 11. Note that ac expression persists throughout the SNS anlage (A-D) and that the SNS anlage invaginates en masse without forming distinct invagination folds (E-F). Orientation of the embryos is anterior to the left and dorsal up (lateral views).
Singling-out of ac-expressing invagination centers depends on N-mediated signaling. (A-D) Anti-achaete and (E-H) anti-crumbs antibody stainings of embryos mutant for different components of the N-mediated signaling pathway. (A,E) Notch55e, (B,F) Delta9P, (C,G) Enhancer of splitx1 and (D,H) mastermindIB99 mutant embryos at stage 11. Note that ac expression persists throughout the SNS anlage (A-D) and that the SNS anlage invaginates en masse without forming distinct invagination folds (E-F). Orientation of the embryos is anterior to the left and dorsal up (lateral views).
The range of N-mediated signaling in the SNS depends on the wg pathway
When analysing wingless (wg) mutant embryos, we noted an SNS phenotype (see below) which suggested that the range of N-mediated signaling in the SNS anlage may depend on a second intercellular communication system which depends on wg activity (reviewed by Perrimon, 1994). The essential com-ponents of the wg signaling pathway include wg itself, dishevelled (dsh) (Klingensmith et al., 1994), shaggy/zeste white 3 (sgg; Siegfried et al., 1992) and armadillo (arm; Peifer and Wieschaus, 1990). dsh, sgg and arm are expressed in all Drosophila cells, while the expression of the secreted wg protein is spatially and temporally restricted (van den Heuvel et al., 1989; Siegfried et al., 1992; Li and Noll, 1993; Peifer et al., 1993; Klingensmith et al., 1994). Analysis of their epistatic relationships showed that wg activates dsh-dependent inhibition of sgg, which in turn acts negatively on arm (Noordermeer et al., 1994; Perrimon, 1994; Siegfried et al., 1994).
At stage 10, wg expression encompasses the SNS anlage of wild-type embryos (Fig. 8A). In the absence of wg activity, initial ac expression in the SNS anlage is normal (not shown), but only one ac-expressing cell was observed in the center of the SNS anlage of the wg mutant embryos (compare Figs 6C, 8D), and only one corresponding central invagination fold was formed (Fig. 8E). The same observations were made in embryos lacking dsh or arm activity (Fig. 8F-I). In contrast, more than three ac-expressing cells (Fig. 8J) and corresponding numbers of invaginations were observed in sgg mutant embryos (Fig. 8K-M). The opposite effects seen in sgg mutants as compared to the wg, dsh or arm mutants reflect a restriction versus an expansion of the range of N-mediated signaling. This implies that the wg pathway functions to spatially control the range of N-mediated signaling in the SNS anlage to allow for the selection of the three invaginations (Fig. 9).
The number of ac-expressing cells and invagination folds depends on the wg pathway. (A) wg expression encompasses the SNS anlage of stage 10 wild-type embryos as revealed by whole-mount in situ hybridization; lateral view, anterior to the left, dorsal up. (B) Anti-achaete and (C) anti-crumbs antibody staining of wild-type SNS primordia at stage 11; dorsal view, anterior to the left. (D-M) Anti-achaete (D,F,H,J) and anti-crumbs (E,G,I,K-M) antibody stainings of wgIG23 (D,E), dshv26 (F,G), armXM19 (H,I) and sggM11-1 (J-M) mutant embryos at stage 11; (D-I,L,M) lateral view; (J,K) dorsal view. Note a single ac-expressing cell (arrows) and the corresponding single invagination fold (arrowhead) in wgIG23 (D,E), dshv26 (F,G) and armXM19 (H,I). In contrast, five ac-expressing cells (arrows) are found in sggM11-1 (J, compare with the wild-type ac pattern in B). Note that the corresponding number of invagination folds appears in sggM11-1 mutants (K-M). (K)A sggM11-1 embryo with four invagination folds (compare with C). (L,M) Different focal planes of a lateral view of the same embryo to demonstrate the four invagination folds. Note that the number of both ac-expressing cells and the corresponding invaginations in sggM11-1 embryos is variable; there are more than three and up to five. The dshv26, armXM19 and sggM11-1 mutant embryos lack the respective gene activity both maternally and zygotically. For a description of the generation of such embryos see Materials and Methods. Note that during late stage 11, when the invagination process is completed, wild-type wg expression is restricted to a ring in the inner part of the esophagus that will contribute to the proventriculus development (not shown).
The number of ac-expressing cells and invagination folds depends on the wg pathway. (A) wg expression encompasses the SNS anlage of stage 10 wild-type embryos as revealed by whole-mount in situ hybridization; lateral view, anterior to the left, dorsal up. (B) Anti-achaete and (C) anti-crumbs antibody staining of wild-type SNS primordia at stage 11; dorsal view, anterior to the left. (D-M) Anti-achaete (D,F,H,J) and anti-crumbs (E,G,I,K-M) antibody stainings of wgIG23 (D,E), dshv26 (F,G), armXM19 (H,I) and sggM11-1 (J-M) mutant embryos at stage 11; (D-I,L,M) lateral view; (J,K) dorsal view. Note a single ac-expressing cell (arrows) and the corresponding single invagination fold (arrowhead) in wgIG23 (D,E), dshv26 (F,G) and armXM19 (H,I). In contrast, five ac-expressing cells (arrows) are found in sggM11-1 (J, compare with the wild-type ac pattern in B). Note that the corresponding number of invagination folds appears in sggM11-1 mutants (K-M). (K)A sggM11-1 embryo with four invagination folds (compare with C). (L,M) Different focal planes of a lateral view of the same embryo to demonstrate the four invagination folds. Note that the number of both ac-expressing cells and the corresponding invaginations in sggM11-1 embryos is variable; there are more than three and up to five. The dshv26, armXM19 and sggM11-1 mutant embryos lack the respective gene activity both maternally and zygotically. For a description of the generation of such embryos see Materials and Methods. Note that during late stage 11, when the invagination process is completed, wild-type wg expression is restricted to a ring in the inner part of the esophagus that will contribute to the proventriculus development (not shown).
Formation of invagination centers by a wg-dependent range of N-mediated signaling in the SNS anlage. (A) The wild-type SNS anlage (blue cells) is outlined within the stomodeum (grey cells). N-mediated signaling (red arrows) functions to single-out three ac/sc-expressing cells (green nuclei) within the SNS anlage by lateral inhibition (left panel). Three invagination folds are then formed, each headed by an ac/sc-expressing cell (central panel). The range of N-mediated signaling is controlled by the wg pathway (right panel). Active arm protein (associated with cell junctions; square between cells) may decrease the reception and/or emission efficiency of the Notch-dependent signal (N) and hence shorten the range of Notch-dependent lateral inhibition (red arrows). wg activity, which is present throughout the SNS anlage (see Fig. 8A), reduces (but does not abolish) the level of sgg activity via dsh activity, which in turn diminishes arm activity. Thus, the range of N-mediated signaling is critically dependent upon the levels of arm activity and this is determined by wg-adjusted sgg activity. (B) In the absence of N-mediated signaling, the singling-out process does not occur. Thus, ac expression persists throughout the SNS anlage (see Fig. 7A-D) and leads to an uncoordinated invagination of all ac-expressing cells (see Fig. 7E-H).(C) Absence of wg or dsh causes high levels of sgg activity and therefore low levels of active arm (Perrimon, 1994). Thus, wg, dsh or arm mutations lead to a maximal range of N-mediated signaling, which then encompasses the SNS anlage and allows the singling-out of only one ac-expressing cell (Fig. 8D,F,H) that leads the single invagination in the middle of the anlage (see Fig. 8E,G,I). (D) In sgg mutants, arm activity is not decreased. This high level of arm activity causes a stronger reduction of the efficiency of N-mediated signaling than in wild-type which shortens the range of lateral inhibition. As a consequence, more than three ac-expressing cells are singled-out (see Fig. 8J) and each supernumerary cell leads to the formation of an extra invagination fold (see Fig. 8K-M).
Formation of invagination centers by a wg-dependent range of N-mediated signaling in the SNS anlage. (A) The wild-type SNS anlage (blue cells) is outlined within the stomodeum (grey cells). N-mediated signaling (red arrows) functions to single-out three ac/sc-expressing cells (green nuclei) within the SNS anlage by lateral inhibition (left panel). Three invagination folds are then formed, each headed by an ac/sc-expressing cell (central panel). The range of N-mediated signaling is controlled by the wg pathway (right panel). Active arm protein (associated with cell junctions; square between cells) may decrease the reception and/or emission efficiency of the Notch-dependent signal (N) and hence shorten the range of Notch-dependent lateral inhibition (red arrows). wg activity, which is present throughout the SNS anlage (see Fig. 8A), reduces (but does not abolish) the level of sgg activity via dsh activity, which in turn diminishes arm activity. Thus, the range of N-mediated signaling is critically dependent upon the levels of arm activity and this is determined by wg-adjusted sgg activity. (B) In the absence of N-mediated signaling, the singling-out process does not occur. Thus, ac expression persists throughout the SNS anlage (see Fig. 7A-D) and leads to an uncoordinated invagination of all ac-expressing cells (see Fig. 7E-H).(C) Absence of wg or dsh causes high levels of sgg activity and therefore low levels of active arm (Perrimon, 1994). Thus, wg, dsh or arm mutations lead to a maximal range of N-mediated signaling, which then encompasses the SNS anlage and allows the singling-out of only one ac-expressing cell (Fig. 8D,F,H) that leads the single invagination in the middle of the anlage (see Fig. 8E,G,I). (D) In sgg mutants, arm activity is not decreased. This high level of arm activity causes a stronger reduction of the efficiency of N-mediated signaling than in wild-type which shortens the range of lateral inhibition. As a consequence, more than three ac-expressing cells are singled-out (see Fig. 8J) and each supernumerary cell leads to the formation of an extra invagination fold (see Fig. 8K-M).
DISCUSSION
We show that the larval SNS originates from a single-layered epithelium in which distinct invagination centers are selected. The selection process depends on lateral inhibition by wg-dependent N-mediated signaling and results in three singled-out ac-expressing cells from which distinct invagination folds are formed, each headed by an ac-expressing tip cell. Thus, unlike PNS and CNS development, SNS morphogenesis is not initiated by delamination of single neural precursor cells in a scattered pattern but rather resembles the process of vertebrate neurulation from a compact epithelium (reviewed by Gilbert, 1994).
SNS anlage formation and neural fate
The genetic circuitry that establishes the SNS anlage includes the known maternal and zygotic components necessary to control pattern formation in the anterior terminal region of the embryo (Perrimon, 1993). Of those, fkh activity is required for the expression of Kr and the genes of the AS-C. While the SNS is absent in fkh mutant embryos, the lack of the AS-C gene activities causes only a reduction of neurons, as observed during CNS development (Jiménez and Campos-Ortega, 1990), and the absence of Kr activity has no significant effect on SNS development. However, the absence of both Kr and AS-C gene activities causes a dramatic reduction of the SNS which exceeds the degree of reduction in the absence of AS-C gene activities. This observation suggests that Kr provides proneural-like function, which can be compensated for by AS-C gene activities. Since remnants of SNS ganglia can still be detected in the absence of AS-C and Kr gene activities, other components with proneural activity must act in the SNS anlage of these embryos.
Intriguing parallels between CNS/PNS and SNS development are not only apparent with respect to the initial expression patterns of the proneural genes, but also with respect to the singling-out of ac-expressing cells in response to neurogenic gene activities. However, while the selected cells already represent the CNS and PNS neural precursors (reviewed by Campos-Ortega, 1993; Goodman and Doe, 1993; Jan and Jan, 1993), the corresponding three cells of the SNS anlage define the invagination centers from which epithelial folds are generated. Furthermore, l’sc expression is not part of the singling-out process in response to N-mediated signaling and ac and sc expressions become reinitiated. Thus, the cells of the developing SNS combine ac, sc and l’sc activities during a second round of AS-C gene expression. In view of this second round of AS-C gene expression, the first one may not determine neural fate, but rather sets a morphoregulatory process into motion that causes epithelial invagi-nations. The second round of AS-C expression may therefore represent the functional equivalent to the proneural gene expression during CNS and PNS development. At present we do not know how the N-dependent singling-out of l’sc-expressing cells in the SNS anlage is circumvented and how the reinitiation of ac and sc gene expression is controlled.
Singling-out of invagination centers
N-mediated signaling is both necessary and sufficient to provide lateral inhibition throughout the SNS anlage in a manner analogous to that seen in the proneural cell clusters initiating both PNS and CNS development (Campos-Ortega, 1993). This argument is supported by the finding that in the absence of wg activity, one ac-expressing invagination center is selected. The selection of three invagination centers in the wild-type SNS anlage therefore depends upon the functional range of N-mediated signaling, which is controlled by wg activity (Fig. 9A).
Local wg activity causes dsh-dependent inhibition of sgg which, in turn, acts negatively on arm, i.e. wg activity eventually causes the increase of active arm protein at the receiving end of the pathway (Noordermeer et al., 1994; Perrimon, 1994; Siegfried et al., 1994). The finding of only one selected ac-expressing invagination center in wg, dsh and arm mutant embryos indicates that a reduction or loss of arm activity leads to an expansion of the range of N function in the SNS anlage (Fig. 9C). Conversely, the absence of sgg leads to increased arm activity which generate supernumerary ac-expressing cells and the corresponding number of invaginations (Fig. 9D). This means that critical levels of sgg activity must be present at functional levels in the wild-type SNS anlage. wg activity, which covers the entire wild-type SNS anlage, does not repress, but rather reduces its level so that the range of lateral inhibition becomes adjusted for the selection of three centers as outlined in the model presented in Fig. 9A.
How wg and N may interact in the SNS anlage
sgg encodes a cytoplasmic serine/threonine kinase (Bourouis et al., 1990) which had been proposed to act as an integral component of the wg pathway as well as of N-mediated signaling (Simpson et al., 1993). In the notum, clonal analysis of sgg mutant cells suggested the possibility that sgg functions downstream of N during lateral inhibition (Ruel et al., 1993; Simpson et al., 1993). However, our results show that the lack of sgg activity does not cause a N-like SNS phenotype (Fig. 8J). Thus, sgg cannot be essential for N-mediated signaling in the SNS anlage per se. In our view, the selection of supernumerary invagination centers in sgg mutants can be explained by high levels of active arm protein that shortens the range of N-mediated lateral inhibition (Fig. 9D). Likewise, the increased levels of sgg kinase activity (in wg or dsh mutants) should then lower the level of active arm so that N-mediated signaling functions throughout the SNS anlage, generating a single invagination fold in the center (Fig. 9C).
arm encodes a protein similar to β-catenin and placoglobin (Peifer and Wieschaus, 1990), leading to the suggestion that arm regulates cellular junctions to control the reception of signaling molecules (reviewed by Peifer et al., 1993; Martínez Arias, 1994). Since wg is expressed throughout the SNS anlage, it may act to affect arm-dependent cell junctions throughout the epithelium. Our findings are consistent with wg acting autonomously to affect the reception of other molecules by con-trolling a cell adhesion prerequisite for signaling. Moreover, the control of cell adhesion has the potential to modify or prevent the transmission of signals between adjoining cells by either affecting the emission or reception of the signal. This implies that the wg pathway is not necessarily generating an instructive signal in its own right but rather regulates the functional range of other instructive signals (Fig. 9).
Is wg action in the SNS anlage a paradigm?
The proposal that the activated wg pathway modulates instructive signals appears at variance with the orthodox view that wg initiates a signaling cascade to instruct neighboring cell fates nonautonomously (Immerglück et al., 1990; Chu-LaGraff and Doe, 1993; Couso et al., 1993; Li and Noll, 1993; Struhl and Basler, 1993; Couso et al., 1994; Couso and Martínez Arias, 1994; Perrimon, 1994) and with the recent finding that wg may directly interact with the N-encoded transmembrane receptor as suggested by allele-specific combinations of wg and N mutations (Couso and Martínez Arias, 1994). The latter sug-gestion could be integrated into our model only if one assumes that wg-dependent arm function ultimately controls the effects generated by the proposed interaction of wg and N.
Our proposal is consistent with the observation of short-range effects of the wg activity in different embryonic developmental systems such as epidermal segmentation (reviewed by Perrimon, 1994), gut (Immerglück et al., 1990) and CNS development (Chu-LaGraff and Doe, 1993), and with the recent demonstration that the segment polarity function of wg affects engrailed target gene expression only in the adjoining epidermal cells (Vincent and Lawrence, 1994). However, it does not explain the long-range effect of wg activity observed during imaginal development (Struhl and Basler, 1993). This phenomenon could be due to an initial short-range effect of wg, which has been propagated through cell proliferation and by local cascades of cell interactions, as was recently shown for the non-secreted sgg protein (Díaz-Benjumea and Cohen, 1994). Furthermore, the non-diffusible membrane-associated vertebrate homolog of wg, Wnt-1, provides both long-range effects in the Xenopus axis duplication assay and short-range effects on adjoining cells in cell culture transfection assays (Parkin et al., 1993). These findings are consistent with the view that although wg encodes a secreted protein (van den Heuvel et al., 1989), both long-range diffusion and morphogen function may be irrelevant to its action as described here in the SNS anlage. Thus, wg and wg-like molecules of different species may exclusively function to control morphoregulatory signals using an evolutionarily conserved molecular strategy of wg action, i.e. wg activity may serve to regulate morphoregulatory signals and thereby act as a permissive component rather than as an instructive signal (Sampedro et al., 1993).
ACKNOWLEDGEMENTS
We thank N. Patel, E. Knust, G. Panganiban, J. F. de Celis, S. Cohen, N. Perrimon, J. Modolell, C. Goodmann and P. Carreras for molecular probes and fly strains that were essential for this work, and D. Schmucker, J. P. Forjanic, B. Purnell and J. Wittbrodt for their comments and discussion of the manuscript. We thank also U. Schmidt-Ott for sharing his results and preparations of tor and tll mutants. This work was supported by a European Community post-doctoral fellowship (M. G.-G.), by the Max-Planck Society and the Fonds der Chemischen Industrie (H. J.).