Control of midline glia development in the embryonic Drosophila CNS

ELSEVIER

Mechanisms of Development62 (1997) 79-91

Control of midline glia development in the embryonic Drosophila CNS H e n r i k e S c h o l z 1, E v e l i n S a d l o w s k i , A n d r e a K l a e s , C h r i s t i a n Kl~imbt *'2 Universittit zu KOln, lnstitutfar Entwicklungsbiologie, 50923 KOln, Germany Received 13 December 1996; accepted 18 December 1996

Abstract

The midline glial cells are required for correct formation of the axonal pattern in the embryonic ventral nerve cord of Drosophila. Initially, six midline cells form an equivalence group with the capacity to develop as glial cells. By the end of embryonic development three to four cells are singled out as midline glial cells. Midline glia development occurs in two steps, both of which depend on the activation of the Drosophila EGF-receptor homolog and subsequent rasl/raf-mediated signal transduction. Nuclear targets of this signalling cascade are the ETS domain transcription factors pointedP2 and yan. In the midline glia pointedP2 in turn activates the transcription of argos, which encodes a diffusible negative regulator of EGF-receptor signalling. © 1997 Elsevier Science Ireland Ltd. Keywords: Drosophila CNS; Midline glial cells; rasl pathway

1. Introduction Two major cell types make up all complex nervous systems: neurones, which form the neuronal lattice, and glial cells, which are intermingled between. In the embryonic central nervous system (CNS) of Drosophila a remarkable degree of glial diversity has recently been described (KRimbt and Goodman, 1991; Ito et al., 1995; Kl~imbt et al., 1996). Based on the mechanisms controlling their development, however, these glial cells can be grouped into just two classes, the lateral glia and the midline glia. The development of the lateral glia depends on the activity of the gene glial cells missing (gcm). In its absence lateral glial cells are transformed into neurones (Hosoya et al., 1995; Jones et al., 1995; Vincent et al., 1996). The only known glial cells which do not require and do not express gcm are those in the CNS midline. The CNS midline comprises a small group of about 26 morphologically distinct cells, generated by 7 - 8 midline progenitor cells per abdominal segment. The anterior-most

* Correspondingauthor. Fax: +49 221 4705164. Presentaddress: Departmentof Neurology,UCSFGalenCenter, 1001 San Francisco,CA94110, USA. z Presentaddress: Instiffitfur Neurobiologie,48149 MUnster,Germany.

three midline progenitors cells give rise to the midline glia, whereas the remaining progenitor cells give rise to midline neurones. During mid-embryogenesis (stage 13/14) about six midline glial cells are found per neuromere. This number decreases to 3 - 4 midline glial cells per segment by the end of embryogenesis due to apoptosis (Jacobs and Goodman, 1989; KRimbt et al., 1991; Bossing and Technan, 1994; Sonnenfeld and Jacobs, 1995; Zhou et al., 1995). Genetic analyses have shown that all midline cells perform important functions during the establishment of the CNS axon pattern (KRimbt et al., 1991; Seeger et al., 1993). Several genes, spitz, Star, rhomboid and pointed, are known to affect the development of the midline glia (Mayer and Ntisslein-Volhard, 1988; Bier et al., 1990; K1/imbt et al., 1991; Rutledge et al., 1992; Klhrnbt, 1993; Kolodkin et al., 1994). The loss of midline glia function leads to a common axon pattern phenotype of fused commissures. Midline cells are already specified at the blastoderm stage when they start to express the gene single minded (sim), the master regulatory gene of midline development (Crews et al., 1988; Nambu et al., 1990, 1991). Expression of sim is controlled in part by inductive influences possibly mediated by the neurogenic gene Notch (Menne and KRimbt, 1994; Martfn-Bermudo et al., 1995). Notchmediated signalling appears not only to be required for

0925-4773/97/$17.00 © 1997 ElsevierScienceIreland Ltd. All rights reserved PII S0925-4773(96)00652-1

80

H. Scholz et al. / Mechanisms of Development 62 (1997) 79-91

the specification of the mesectodermal anlage, but also for the determination of midline glia cell fate within the mesectoderm, since removal of zygotic Notch function leads to a complete loss of midline glial cells (Menne and K1/imbt, 1994). In a next step the Drosophila EGFreceptor homologue DER is required for development of the midline glia. Within the embryonic CNS, DER is specifically expressed and required for midline glia differentiation in the midline glia (Zak et al., 1990; Raz and Shilo, 1992). DER is one of the most extensively studied receptor tyrosine kinases (RTK) involved in many developmental processes (Xu and Rubin, 1993; Diaz-Benjumea and Hafen, 1994; Gonzalez-Reyes et al., 1995; Kuo et al., 1996; Roth et al., 1996). DER activity is transmitted from the cell membrane to the nucleus through a conserved biochemical pathway, involving rasl and a cascade of Serine/Threonine kinases (raf, MEK, MAPK) (Wassarman et al., 1995). The activation of MAPkinase finally triggers the execution of a specific developmental program by phosphorylation of specific transcription factors (Dickson, 1995). Two known targets of the MAPkinase in Drosophila are the transcription factors encoded by pointed and yan. Two transcripts are generated from the pointed (pnt) locus, encoding two different ETS domain proteins, pointedP1 and pointedP2 (K1/imbt, 1993). pointedP1 acts as a constitutively active transcription factor, whereas pointedp2 requires phosphorylation by MAPkinase to become a potent transcriptional activator (Brunner et al., 1994; O'Neill et al., 1994). yan also encodes an ETS domain protein with eight MAPkinase phosphorylation sites (Lai and Rubin, 1992; Tei et al., 1992). Yan functions as a transcriptional repressor and controls the choice between cell division and differentiation. Phosphorylation of yan by the MAPkinase leads to its translocation from the nucleus to the cytoplasm where it becomes degraded (Rebay and Rubin, 1995; Rogge et al., 1995). Thus, activity of MAPkinase concomitantly activates the pointedP2

transcriptional activator and promotes removal of the inhibitory yan protein from the nucleus. DER is activated by binding of TGFu-like ligands encoded by spitz and gurken (Rutledge et al., 1992; Neuman-Silberberg and Schupbach, 1993; Schweitzer et al., 1995b). An additional component of DER signalling is an inhibitory secreted protein, argos, which is likely to serve as an antagonistic ligand of DER (Freeman et al., 1992; Schweitzer et al., 1995a) Within the ventral ectoderm argos transcription is induced by DER signalling through the activity of the gene pointed, indicating that argos serves as an inhibitory feedback loop (Gabay et al., 1996; Golembo et al., 1996). Both pointedP2 and the DER protein are specifically expressed in the midline glial cells. Below we addressed the question whether DER controls midline glia development via the rasl signalling cascade, and which aspects of midline glia development depend on DER signalling. We have employed the GALA technique (Brand and Perrimon, 1993) and constructed driver lines which allowed targeted expression in the CNS midline. Our results indicate that DER signalling occurs via rasl and is required during at least two steps. First, it is required for the generation of the correct number of midline glial cells, and second, it controls the subsequent differentiation of these cells.

2. Results

The major embryonic CNS axon tracts are organised in a ladder like pattern. Two commissures are found within each neuromere (Fig. 2A). Their formation occurs during embryonic stages 12-14 and depends in part on the function of midline glial cells. The 3-4 midline glial cells found in each neuromere can be specifically labelled by /3-galactosidase expression conferred by the enhancer trap insertion AA142 (Fig. 4B). Mutants lacking functional

Fig. 1. Ectopic expression in the CNS midline using the GAlA system. Embryos heterozygous for sim-GAL4 (A-C) or a sli-GAL4 (D-L) and UAS-IacZ have been stained for/3-galactosidase expression. (A-I) show lateral views, (K,L) ventral views, anterior is to the left. (A) In sim-GAL4, UAS-lacZ embryos/3-galactosidase expression can be detected in all midline cells in a stage 10 embryo. (B) In a stage 12. /3-galactosidase expression is also detected in the anterior and posterior midgut invagination. (C) In stage 15 embryos the sim-GAL4 driver line confers expression to all midline cells. Highest levels can be detected in the midline glial cells (arrow heads). The sli-GAL4 driver line confers /3-galactosidase expression from stage 12 onwards (D-I). During stage 12 (D,G) the midline glial cells appear as elongated cells, with their apical side anchored in the epidermis. Following the migration of the midline glia, these processes are retracted from the epidermis. Now thin cellular processes of the posterior and anterior midline glial cells appear to ensheath all midline cells within a given segment (H, arrowheads). In stage 13 embryos/3-galactosidase expression can also be detected in the MP1 neurones, identified by their characteristic axonal processes (K, arrowhead). In stage 15/16 (F,I) embryos the midline glia ensheath anterior and posterior conunissures (asterisks). No processes can be detected surrounding the remaining midline cells. In some segments 1-2 ventral midline cells express/3-galactosidase as well. In addition the salivary glands express/3-galactosidase. Fig. 2. Rescue of the pointed axon pattem phenotype by expression of pointed in the midline glial cells. The figure shows frontal views of dissected embryonic CNS preparations, stage 15/16. The CNS axon pattern is revealed by the monoclonal antibody BPI02 and subsequent HRP immunohisto= chemistry. Anterior is up. (A) In wild type embryos most axon tracts are found in the two longitudinal connectives (lc) and the anterior (ac) and posterior commissure (pc). (B) In mutant pointed zxs8embryos the segmental commissures appear fused. (C) Expression of UAS-pointedP2 in the midline glial cells using the sli-GAL4 driver line rescues the pointed axon pattern phenotype, indicating that indeed the midline glial cells are responsible for the fused commissure phenotype. (D) A similar phenotypic rescue of the mutant pointed phenotype can be obtained by expression ofpointedP1 in the midline glial cells. In addition, connectives appear thinner.

i~i~....

~ i

iii¸~¸¸

I

82

H. Scholz et aL / Mechanisms of Development 62 (1997) 79-91

midline glial cells develop a common embryonic CNS axon pattern phenotype of fused commissures (K1/imbt et al., 1991). 2.1. Directed expression in midline cells

In order to perform a detailed analysis of the function of pointed during the development of midline glial cells, we have employed the G A I A system to direct expression of various transgenes (Brand and Perrimon, 1993). To drive expression in all midline cells we generated a s i m - G A L 4 driver line. Embryos carrying the transgenes s i m - G A L 4 and U A S - l a c Z show/3-galactosidase expression in all midline cells from stage 9 onwards. Later in development simGAL4 drives expression most prominently in the midline glial cells (Fig. 1A-C). To more specifically manipulate the midline glia, we used a slit promoter fragment (see Section 4 for details). To monitor the expression domain conferred by the sliGAL4 driver line we used a UAS-lacZ and a UAS-CD2 transgene (Dunin-Borkowski and Brown, 1995) (Fig. 1). First,/3-galactosidase expression is found at stage 11 when the presumptive midline glial cells appear as elongated cells with their basal side anchored in the ventral ectoderm. The nucleus is found close to the basal cell surface at the dorsal roof of the developing nerve cord (Fig. 1D,G). During germband retraction, two of the midline glial cells start their posteriorward migration. At this early stage of development the six presumptive midline glial cells express /~-galactosidase under the control of the sliGAL4. At stage 13/14 the midline glial cells have migrated on top and below the developing anterior commissure (Fig. 1E,H) and start to ensheath the commissural axons. Concomitant with their migration, the midline glial cells retract their apical end feet which connected them to the

underlying epidermis and form a ring of thin glial processes surrounding all midline cells of a segment (Fig. 1H). At stage 16, 3 - 4 midline glial cells express/3-galactosidase in each abdominal segment. The glia has completely ensheathed anterior and posterior commissures, which appear as unstained areas (Fig. 21, asterisk). From stage 13/14 the sli-enhancer also directs expression in the MP1 neurones, which can be recognised by their characteristic axonal trajectory following expression of the CD2 antigen (Fig. 1K). In addition, ventral cell bodies devoid of obvious cellular processes express the/3-galactosidase in late stage 16 embryos (Fig. 1I). The nature of these cells is unknown. 2.2. PointedP2 function in the midline glia depends on phosphorylation pointedP2 is specifically expressed in the embryonic midline glial cells (K1/imbt, 1993) and has recently been found to act downstream of the sevenless receptor tyrosine kinase signalling cascade (Brunner et al., 1994; O'Neill et al., 1994). Its function during photoreceptor cell development appears to require phosphorylation of the threonine residue at position 151 (Brunner et al., 1994). To analyse pointed function during the development of the midline glia, we first asked whether directed expression ofpointedP2 using the sli-GAL4 driver line is capable of rescuing the fused commissure phenotype of homozygous mutant pointed ass embryos (Fig. 2B). Two recombinants were generated (UAS-pnt P2, pnt ass and sli-GAL, pnta88). All mutant embryos derived from a cross of these two lines expressed pointedP2 in midline glial cells. They could unambiguously be identified with the help of a blue balancer as well as by their characteristic tracheal phenotype (Scholz et al., 1993). In all cases we observed pheno-

Fig. 3. EGF receptor activity is required for midline glia development. (A,B) Dissected stage 12 embryos stained for/3-galactosidaseexpression directed by the AA142 enhancer trap insertion. (C,D) Whole mount stage 15 embryos stained for CNS axon tracts using MAb BP102. Anterioris to the left. (A) In wild type embryos the AA142 enhancer directs/8-galactosidase expression in 3-4 midline glial cells per segment (arrowheads). In addition, the salivary gland as well as a few cells lateral to the CNS (asterisks) express the AA142 reporter gene. (B) In top3c81embryos 1-2 midline ceils express the AA142 reporter gene. Expression in the salivary glands appears unchanged; however, the lateral cells express increased levels of/3-galactosidase. (C) In a wild type embryo the major CNS axon tracts are organised in a regular ladder-like pattern consisting of two longitudinal connectives and two axon commissures, per segment. (D) Expression of a dominant negative form of the EGF receptor in all midline cells using the sim-GAL4 driver line results in a partial fusion of commissures pointing towards midline glial cell defects. Fig. 4. rasl signalling is required for midline glia development. (A-F) are frontal views of dissected embryonicCNS preparations, stage 15/16. The CNS axon pattern is revealed by the monoclonalantibody BP102 and subsequent HRP immunohistochemistry(brown). (B-E) Midline glial cells are labelled by/~-galactosidase expression directed by the AA142 enhancer trap insertion./~-galaetosidase is detected through alkaline phosphatase irnmunohistochemistry (purple). Anterior is up. (G-I) show whole mount in situ hybridisations. Anterior is to the left. (A) Expression of DN-ras1 in all midline cells using the sim-GAL4 driver line in a Df(rasl)/+ embryo results in a partial fusion of the axon commissures, indicating a function of rasl in midline glia development. (B) In wild type stage 15 embryos 3-4 midline glial ceils express the AA142 enhancer. Similar ~-galactosidase expression is mediated by the argoswl~enhancer trap insertion. (C) Followingexpression of activated ras I in the midline glia using the sli-GAL4 driver line, six midline cell express the AA142 reporter gene. (D) Similarly, 5-6 midline cells express the AA142 reporter gene followingexpression of activated raf using the sli-GAIA driver line. (E) argos encodes a negative regulator of EGF-receptoractivity. In argosa7 mutant embryos 4-5 midline cells are labelled by the AA142 enhancer trap; in addition, two cells posterior to the posterior commissure express low levels of/~-galactosidase activity. (F) In mutant pointed~88 embryos no argoswll expression is detected in the midline glial ceils, indicating that aos expression is dependent on pointedP2. Note that expression of argosw~ is still detected in the ectoderm. (G) pointedP2 is expressed in the midline glial cells. (H) Within the CNS argos is expressed in the same cells as pointedP2. (I) In mutant rasl embryos reduced argos expression can be detected in the midline glial cells.

H. Scholz et al. /Mechanisms of Development 62 (1997) 79-91

typic rescue of the fused commissure phenotype of otherwise mutant pointed as8 embryos (Fig. 2C). To test whether phosphorylation might be required for pointed function in the midline glia we expressed the pntT~5~Aprotein which cannot be phosphorylated by MAPkinase any more (Brunner et al., 1994; O'Neill et al., 1994). Expression of this variant does not rescue the pointed ~ axon pattern phenotype (data not shown). The failure of rescue by pnt T151A shows that post-translational activation of pointedP2 is required to allow normal development of the midline iiii~!~;v~.71!i~i~i~

~ i ~

¸¸ ~ ~

ii!~

i ~ i ~

x~ ~

3

/

83

glia. Thus, expression of an activated pointedP2 protein should also lead to a rescue of the mutant pointed a88 commissure phenotype. The pointedP1 protein, which shares the ETS DNA binding domain with pointedP2, has been shown to act as a constitutive active transcription factor (Kl~imbt, 1993; O'Neill et al., 1994). As with expression of pointedP2, expression of pointedP1 in the midline glia results in the rescue of the fused commissure phenotype of mutant pointed AS~ embryos (Fig. 1D). These data indicate that in the midline glial cells poin-

84

H. Scholz et aL / Mechanisms of Development62 (1997) 79-91

tedP2 has to be activated by phosphorylation, possibly through MAPkinase, to induce transcription of its target genes. 2.3. EGF-receptor signalling is required for midline glial cell development

Within the CNS the Drosophila homologue of the EGF receptor (DER) is specifically expressed in the midline glial cells (Zak et al., 1990) (Fig. 3E-G). To follow midline glia development in embryos lacking zygotic DER expression, we analysed the expression of the AA142 midline glia marker in homozygous top 3C81 embryos (Fig. 3A,B). Loss of DER function leads to reduction of the number of AA142-positive cells during stage 12. Instead of four, only 1-2 midline glial cells are labelled by the AA142 enhancer trap. In older mutant embryos (stage 15/ 16) these midline cells apparently die. Next, we expressed a dominant negative (DN) form of DER (kindly provided by B.-Z. Shilo) using the GAL4 system. Expression of DN-DER under the control of sli-GAL4 in wild type embryos did not result in an abnormal CNS phenotype. However, when we expressed DN-DER in all midline cells using the sim-GAL4 driver, we obtained a weak fused commissure phenotype, indicative of defects in midline glia development (Fig. 3D). Thus, DER appears to be required for the initial determination of the correct midline glial cell number as well as for further midline glia differentiation. This is in agreement with temperature shift experiments (Raz and Shilo, 1992). 2.4. rasl signalling during midline glia development

The EGF-receptor often transduces a signal through a conserved rasl/raf signalling cascade. Embryos lacking zygotic rasl expression showed reduced numbers of midline glial cells (Fig. 4H,I). To further analyse the function of rasl we expressed a dominant negative rasl (DN-rasl) (Lee et al., 1996) in the midline. Expression of DN-rasl under control of sli-GAL4 resulted in no abnormal CNS phenotype. However, expression of DN-rasl under the control of the sim-GAL4 driver led to a slight fusion of commissures in heterozygous rasl embryos (Fig. 4A). This, as well the fact that fewer midline glial cells are found in mutant rasl embryos, indicates that rasl function

is required for the determination and differentiation of the midline glia. In addition to the loss of function experiments we expressed activated rasl and raf in the midline glia to determine the gain of function phenotypes. In wild type embryos 3 - 4 midline glial cells express the AA142 enhancer trap (Fig. 4B). Following expression of activated rasl in the midline glial cells using the sli-GAL4 driver line, we observed 6 - 8 midline cells expressing the AA142 marker compared to the normal set of about four midline glial cells (Fig. 4B,C). A similar result was obtained following expression of activated raf in all midline cells (Fig. 4D). Despite the increased numbers of midline glial cells the axonal scaffold is unaffected. BrdU labelling experiments indicated that neither expression of activated rasl nor the expression of activated raf induced extra cell divisions in the midline (data not shown). The above data indicate that rasl signalling indeed controls the number of midline glial cells. Reduced levels of rasl lead to fewer midline glial cells whereas elevated levels of rasl activity lead to the formation of extra midline glial cells. It is likely that the activation of rasl occurs via the EGF-receptor. Activation of the EGF-receptor is initiated by binding of its ligand, spitz (Schweitzer et al., 1995a). In spitz mutants midline glia development is impaired (K1/imbt et al., 1991). On the other hand, EGFreceptor signalling is negatively regulated by the argos protein, which is expressed specifically in the midline glial cells (Fig. 4H) (Freeman et al., 1992; Schweitzer et al., 1995a). Thus, in argos mutants elevated levels of EGFreceptor signalling might be expected at the CNS midline. Mutant argos embryos show no abnormal CNS axon pattern phenotype; however, we observed an increase in the number of midline glial cells (Fig. 4D), similar to, yet not as extreme as, that observed following expression of activated rasl in the midline glia (Fig. 4C). Analogously, an increase of lateral chordotonal organs was observed in the PNS of mutant argos embryos (Okabe et al., 1996). Thus, activation of the EGF-receptor is likely to be responsible for the activation of rasl in the midline glial cells. A final nuclear target of receptor tyrosine kinase signalling is the transcription factor pointed (Brunner et al., 1994; O'Neill et al., 1994; Gabay et al., 1996). pointedP2 requires phosphorylation to become an active transcriptional activator, pointedP2 in turn activates argos tran-

Fig. 5. yan function during midline development.(A-D) are whole mount in situ hybridisationof digoxygenin-labelledyan cDNA to wild type embryos. Anterior is up. (E,F) Expressionof the AA142 reportergene is detected by alkalinephosphataseimmunohistochemistry.Anterioris to the left. (G,H)show frontal views of dissected CNS preparations from stage 15 embryos. Anterior is up. (A-D) yan is expressed very dynamicallyduring embryonic development. (A) During gastrulation yan expression is excluded from the mesoderm as well as from the presumptive CNS midline cells. During germ band extension at stage 9/10 first expression in the CNS can be detected in a few cells flanking the midline (B, arrowhead). During stage 11 expression of yan becomesfirst detectablein the CNS midline. In addition lateral cells, possiblycorrespondingto the lateral glioblasts, transientlyexpress yan (arrow). (D). yan expression in the midline becomesconfined to the midline glial cells during stage 13. (E) In wild type embryos3-4 midline glial cells are labelled by/3-galactosidaseexpressiondirected by the AA142 enhancer trap insertion. (F) In homozygousyan mutant embryos,which can be identified by their anterioropenphenotype(arrowhead)only 2-3 midline glial cells are labelled by the AA142 enhancertrap insertion. (G) Expressionof activated yan in all midline glial cells leads to fused commissuresand to disruptions in the longitudinal connectives. (H) Expressionof activatedyan in the midline glial cells using the sli-GAL4 driver line leads to a moderate fusion of the segmentalcommissures.

H. Scholz et aL / Mechanisms of Development 62 (1997) 79-91

scription since no argos RNA nor activity of the argos wll enhancer trap could be detected in the midline glial cells of mutant pointed embryos (Fig. 4F). In addition, argos R N A expression is activated following ectopic pointedP1 or

85

pointedP2-VP16 expression (data not shown) (Gabay et al., 1996). To further address the question whether activation of pointed via the rasl pathway is required for the transcriptional activation of argos we assayed argos

86

H. Scholz et aL / Mechanisms of Development 62 (1997) 79-91

expression in mutant rasl embryos. Since maternal expression is still present, rasl signalling appears to be acting at reduced levels in these embryos. In correlation we see a reduced expression o f argos R N A in the midline glial cells, whereas the lateral expression of argos is unaffected (Fig. 4H,I).

in the midline glial cells, just as it does in the developing photoreceptor ceils (Brunner et al., 1994; O ' N e i l l et al., 1994).

2.6. Function of D-jun during midline development

2.5. Function of yan during midline glia development

Expression of activated pointed (pointedP1 or a VP16pointedP2 fusion) in all midline cells directs the ectopic

In the developing eye as well as during the formation of the ventral ectoderm, yan antagonises pointed function. As pointed, yan encodes an ETS domain protein, whose function is regulated via phosphorylation (Rebay and Rubin, 1995). yan expression is initially confined to the neuroectoderm and no expression is found in the midline anlage or the mesoderm (Fig. 5A). First expression in the CNS can be detected in stage 10 embryos in groups of cells lateral to the midline (Fig. 5B). During stage 12, most midline cells appear to express yan as well as some lateral CNS cells, which, based on their position, could correspond to the longitudinal glioblasts (Fig. 5C). During stage 13/14 yan expression becomes confined to the midline glia (Fig. 5D), in a pattern similar to the expression ofpointedP2 (Fig. 4). In mutant yan embryos commissures form normally; the number of AA142-positive midline glial cells, however, is reduced to 2 - 3 (Fig. 5E,F). Expression of a mutant yah protein, which cannot be phosphorylated any more (yan act) and presumably blocks D N A binding sites of ETS domain proteins (Rebay and Rubin, 1995), results in a fused commissure phenotype similar to the pointed phenotype. Again we observed a stronger phenotype following expression of yanact using the sim-GAL4 driver line (Fig. 5G) compared to using the sli-GAL4 driver line (Fig. 5H). This demonstrates that yan antagonises pointedP2 function

expression of glia-specific markers in other midline cells (Fig. 6A) (Klaes et al., 1994). No such phenotypes were observed when we expressed pointedP2 in all midline cells. During eye development, pointedP2 co-operates with D-jun in the transcriptional regulation o f target genes (Treier et al., 1995). Therefore, we asked whether D-jun could perform similar functions during midline glia development and analysed the consequences of ectopic Djun expression and coexpression of both transcription factors. As with expression of pointedP1, expression o f activated D-jun in the midline glia did not lead to any CNS axon pattern defects (data not shown). However, expression of activated D-jun in all midline cells resulted in a CNS phenotype different from the one observed following pointedP1 expression (Fig. 6A,B). In many segments commissural axon tracts are missing (Fig. 6B). Interestingly, the midline glial cells, which appeared unchanged in number, stayed at the midline in segments devoid of any commissures. This is particularly remarkable since axonal contact is believed to control survival o f midline glial cells (Sonnenfeld and Jacobs, 1995). In mutants where the formation of commissures is impaired (commissureless (Seeger et al., 1993); schizo (T. Hummel, unpublished); fasI/abl (Elkins et al., 1990)), midline glial cells die or move away from the midline and associate with the con-

Fig. 6. Expression of activated D-jun in all midline cells. The figure shows frontal views of dissected CNS preparations from stage 15/16 embryos. The CNS axon pattern is detected by the monoclonal antibody BP102 and subsequent HRP immunohistochemistry.Midline glial ceils are labelled by/3galactosidase expression driven by the AA142 enhancer trap line and subsequent alkaline phosphatase immunohistochemistry. Anterior is up. (A) Expression of pointedP1 in all midline glial cells using the sim-GAL4 driver line leads to an increase of cells expressing the AA142 enhancer trap marker. In many segments the posterior commissure does not form. (B) Embryos expressing activated D-jun in all naidline cells using the sim-GAL4 driver line display an axon pattern phenotype reminiscent of the phenotype shown by homozygous commissureless embryos (C). However, it is interesting to note that in embryos expressing activated D-jun the midline glial cells stay in the midline lacking obvious axonal contacts (B, arrowheads), whereas in commissurelessmutant embryos the midline glial cells migrate to the longitudinal connectives where they finally degenerate (C, arrowheads). Following expression of activated D-jun the number of midline glial cells as detected by the AA142 enhancer trap appears unchanged to wild type. (D) Expression of activated D-jun leads to a decrease in the number of X55-expressing midline neurones. The severity of the loss of midline neurones correlates with the extent of the commissureless-like phenotype. In the anterior-most segment shown the anterior commissure has formed and the X55positive cell cluster at the midline appears wild type. In the commissureless-like segments the X55-positive cell cluster contains fewer cells than in segments with a normal commissural axon pattern. (E) Lateral view of a stage 16 wild type embryo expressing the X55 marker in the VUM cell cluster. (F) Lateral view of a stage 16 embryo expressing activated D-jun in all midline cells. A reduced number of X55-positive cells can be detected. Fig. 7. Model of midline glia development. Schematic lateral views of the CNS midline illustrate the different steps during midline glia development. For simplicity only three midline glial cells and one midline neurone are shown. (A) As soon as the midline cells are specified, which is manifested in their expression of single minded, the midline glia equivalence group is determined. This process involves the neurogenic gene Notch. (B) In the next step a graded activation of DER, shown in red, occurs via (a) differential expression of rhomboid, which promotes DER signalling and (b) through a preferential expression of DER itself in the midline glia equivalence group. (C) DER activity is transmitted to the nucleus via the ras l/raf pathway. One of the nuclear targets of this pathway is pointedP2 which is symbolised by blue nuclei. Activation of pointedP2 is required to allow the survival of some of the midline glial cells of the initial equivalence group. Activated pointedP2 then induces argos transcription (green arrows). (D) argos encodes a diffusible negative regulator of DER signalling and restricts DER signalling to a few cells of the midline glia equivalence group.

D

C~

CO

C~

I

88

H. Scholzet aL / Mechanismsof Development62 (1997) 79-91

nectives (Fig. 6C). Thus, expression of activated D-jun apparently allows survival of the midline glial cells despite the lack of axonal contact but does not influence the number of midline glial cells. Correlating with the strength of the commissureless-like axon pattern phenotype, we observed a reduction in the number of midline neurones labelled by the enhancer trap insertion X55 (Fig. 6D -F). Since glial cells are present but midline neurones appear reduced in number, we suggest that the midline neurones perform important functions during the formation of commissures. Expression of activated D-jun did not activate the expression of argos, a pointed target gene. Neither did coexpression of pointedP2 and D-jun result in such a transcriptional activation, suggesting that D-jun and pointedP2 do not co-operate during midline glia development.

3. Discussion

In the present paper we have examined the role of the DER-rasl-pointed/yan signalling cascade in midline glia development. Since many of the gene products acting in this pathway are supplied maternally, embryonic phenotypes are rare. We did not generate germ line clones since the complete loss of rasl or raffunction is expected to lead to severely disorganised embryos, hindering a proper analysis of the CNS midline. To circumvent the lack of mutant phenotypes we initially attempted to generate cell-specific gene knock outs by directed expression of antisense constructs. Although similar experiments have been successful in the past (Rosenberg et al., 1985; Schmucker et al., 1992; Volk and VijayRaghavan, 1994), we failed to obtain any phenotypes using this technique. The data Obtained using dominant negative or constitutively active versions of the various proteins as well as some mutant analyses support a three step model on the molecular mechanisms directing the development of the midline glia (see Fig. 7). In a first step, a midline glia equivalence cell cluster is defined. Secondly, within this cluster 3-4 midline glial cells are selected which, as a last step, differentiate and survive until pupariation (Stollewerk et al., 1996). The last two steps are under the control of the EGF-receptor signaling. 3.1. The midline glia equivalence group In a first step six midline cells in each segment, presumably originating from the anterior-most three midline progenitor cells, are determined as the midline glia equivalence group. The generation of the midline glia equivalence group process involves Notch function (Menne and K1/~mbt, 1994) and segmentation genes. It might also depend on the function of single minded, the master regulatory gene of midline development, single minded transcription accumulates in the midline glia

and, depending on the context, can act either as transcriptional activator or as transcriptional repressor (Crews et al., 1988; Thomas et al., 1988; Nambu et al., 1990, 1991; Mellerick and Nirenberg, 1995). 3.2. Determination of the correct number of midline glial cells By the end of embryogenesis the final number of midline glial cells is about 3-4. Thus, the correct number of midline glial cells has to be selected from the initially defined equivalence group in a second step. DER mutants show a reduced number of midline glial cells and argos mutants, which possibly show an increased activation of DER in the midline, show an increased number of midline glial cells. Furthermore, expression of activated rasl (or activated ra30 in the midline results in the appearance of extra midline glial cells. Since we failed to detect any cell division following expression of activated ras by BrdU labelling experiments, we assume that DER activity is required to prevent cell death in the midline glia equivalence group. However, we cannot exclude the possibility that the extra midline glial cells are recruited from neighbouring CNS cells. Our model suggests that activation of rasl signalling in the entire midline glial equivalence group promotes survival of all cells in this cluster. Thus, in wild-type about 2-3 Cells in this group down-regulate DER signalling and are destined for cell death, Indeed, it has previously been shown that mutations blocking cell death lead to an increase in the number of midline glial cells (Sonnenfeld and Jacobs, 1995; Zhou et al., 1995). In order to escape programmed cell death, the DER/rasl/raf signalling cascade has to be activated. 3.3. rhomboid and argos control activation of DER during midline glia development DER is expressed in the midline glial cells, whereas its ligand, spitz, is expressed relatively uniformly (Zak et al., 1990; Rutledge et al., 1992). We assume that a graded activation of DER is brought about by the activity of rhomboid, which is thought to promote EGF receptor signalling, possibly by cell autonomous activation of the DER ligand spitz (Sturtevant et al., 1993). rhomboid is expressed in a graded manner throughout the midline with highest levels of expression in some of the presumptive midline glial cells (Bier et al., 1990; Sturtevant et al., 1996). General ectopic expression of rhomboid, which presumably results in an increased activation of DER, leads to the appearance of extra midline glial cells (Sonnenfeld and Jacobs, 1995). Thus, activation of DER is likely most intense in only a few midline glial cells. Via the rasl/rafpathway the DER signal is transduced to the nucleus, where pointedP2 becomes activated through phosphorylation, pointedP2 in turn activates the transcription of argos, argos encodes a diffusible protein which negatively regulates DER sig-

1-1.Scholz et al. / Mechanisms of Development 62 (1997) 79-91 nailing non-cell autonomously and thus competes with spitz function (Freeman et al., 1992; Schweitzer et al., 1995a). Midline cells exposed to high spitz concentrations probably maintain DER activity. In contrast, in midline cells exposed to low levels of active spitz, argos can successfully compete with spitz and will downregulate DER signalling. This process eventually leads to a restriction of DER signalling into just the later midline glial cells. Since argos depends on pointed function, pointed mutant embryos also lack argos function. This would result in an increased activation of the EGF-receptor and correspondingly in the formation of extra glial cells in mutant pointed embryos. This appears to be the case and in stage 13 mutant pointed embryos, six instead of four midline glial cells are labelled by the AA142 enhancer trap line (Kl~imbt, 1993). In older embryos, however, these midline glial cells die due to the requirement for pointed function during later differentiation.

3.4. Additional factors influencing midline glia survival Thus, activation of DER signalling, e.g. activated rasl, promotes survival of glial cells. A similar promotion of cell survival by stimulation of the EGF-receptor has been shown in vertebrate oligodendrocytes (Barres et al., 1992). Just as in oligodendrocyte development (Barres et al., 1993), the survival of midline glial cells can be stimulated by additional signals. One such signal appears to be conveyed via the contact with commissural axons. In mutants which lack commissural axons the midline glial cells die (Sonnenfeld and Jacobs, 1995). It is interesting to note that one can bypass the requirement of axonal contact for midline glia survival by the expression of activated D-jun. Despite the lack of axonal contact the midline glia survives; furthermore, DER signalling appears intact in these cells, as we observed a normal number of midline glial cells per segment. In the midline neurones the expression of activated D-jun apparently leads to a slight reduction in the number of midline neurones, possibly indicating cell death. This would be in agreement with data obtained in vertebrate neuronal tissue culture experiments. Expression of a dominant negative form of jun protects sympathetic neurones from the normally occurring cell death follOwing withdrawal of NGF, whereas expression of wild-type jun itself or the activation ofjun-kinase is sufficient to induce apoptosis in these cells (Ham et al., 1995; Xia et al., 1995). It remains to be tested how D-jun exerts the observed opposing functions during glial and neuronal development. Mutant analyses suggest that DER/rasl signalling is probably also involved in the control of midline glia differentiation through the activation of pointedP2 (Klfirnbt, 1993; Klaes et al., 1994). The control of midline glia development appears to involve a very similar set of proteins as the control of photoreceptor development (the genes Sos, drk, Gapl, raf, rl are all expressed in the devel-

89

oping midline; data not shown). Extensive genetic screens have led to the identification of a number of new genes acting in the rasl pathway during Drosophila eye development, some of which might be specific for eye development (Kafim et al., 1996). One of the genes required for a specific cellular response to a common signal might be Djun. It has been suggested that pointedP2 and D-jun are cooperating in the activation of target genes during photoreceptor development (Treier et al., 1995). In the midline this does not appear to be the case. Expression of activated pointed leads to the activation of glia-specific pointed target genes like argos (Gabay et al., 1996). Expression of pointedP2 or coexpression of pointedP2 and activated Djun does not lead to such a transcriptional activation. In addition, different negative transcriptional regulators encoded by the tramtrack gene are required during development of the different cell types (Xiong and Montell, 1993; Yamamoto et al., 1996; Giesen et al., 1997). However, further analyses of these genes are required to understand how different cells respond differently to a signal delivered by a common signal transduction cascade.

4. Experimental procedures 4.1. Fly strains The following alleles have been used in this study: rasl e2F (Simon et al., 1991); (ra3~ l(1)ph111-29 (Ambrosio et al., 1989); pnt a88 (Scholz et al., 1993); top 3csl (NtissleinVolhard et al., 1984); spi IIA14 (N~isslein-Volhard et al., 1984); To ectopically express different genes we employed the GAL4 system (Brand and Perrimon, 1993). The following constructs were made and used for germ line transformation, sli-GAL4: a 1 kb EcoRI fragment of the slit promoter region conferring expression specifically to the midline glial cells (Wharton and Crews, 1993) was inserted in tandem in a hsp70-GAL4 expression vector kindly provided by E. Knust, K61n. Two independent insertions were obtained following germ line transformation. Subsequent mobilisation of one of these integrations yielded 16 additional P[white, sli-GAL4] insertions (one X-chromosomal; four seCond chromosomal; and 12 third chromosomal insertion), simGAL4: to direct expression in all midline cells we cloned a 3.5 kb EcoRV fragment from the sim promoter (Kasai et al., 1992) into the hsp70-GAL4 expression vector. Following germ line transformation we obtained nine independent insertions (three second chromosomal insertions and six third chromosomal insertions). To monitor GAL4 expression driven in these lines we used UAS-IacZ flies (Brand and Perrimon, 1993) and UAS-CD2 flies (DuninBorkowski and Brown, 1995). Three different antisense constructs were made and used for germ line transformation: UAS-antisense-rasl (nine independent insertions),

90

H. Scholz et al. / Mechanisms of Development 62 (1997) 79-91

UAS-antisense-rolled (20 independent insertions), and UAS-antisense-pointedP2 (five independent insertions). To express an activated rasl (Fortini et al., 1992) we established six independent insertions of a UAS-rasl w2 construct. In addition, the following UAS-lines were used: UAS-dominant negative rasl (Lee et al., 1996), UAS dominant negative DER (B.-Z. Shilo, unpublished), UAS-act-raf (Perrimon, 1994) and UAS-act-jun (Treier et al., 1995). To label midline glial cells we have used the following enhancer trap lines: AA142: a P[lacZ, white] at 66D; X55 P[lacZ, white] at 56F (Klfimbt et al., 1991); Wll a P[lacZ, white] into the argos gene (Freeman et al., 1992); and 1277 aP[lacZ, white] insertion into the pointedP2 gene (Scholz et al., 1993). 4.2. Antibody staining and sectioning Immunohistochemistry was carried out as described previously (Klgmbt et al., 1991). BrdU labelling experiments where performed as described (Bodmer et al., 1989).

Acknowledgements We thank S. Crews, E. Hafen, M. Mlodzik, D. Montell, B.-Z. Shilo and C.S. Goodman for providing DNA, fly stocks and antibodies. The manuscript was improved by comments of J. Campos-Ortega, S. Granderath, U. Hinz and E. Knust. This work was supported by the DFG through SFB 243 and a Heisenberg fellowship to C.K.

References Ambrosio, L., Mahowald, A.P. and Perrimon, N. (1989) Nature 342, 288-291. Banes, B.A., Hart, I.K., Coles, H.S., Burne, J.F., Voyvodic, J.T., Richardson, W.D. and Raft, M.C. (1992) Cell 70, 31-46. Banes, B.A., Schmid, R., Sendnter, M. and Raff, M.C. (1993) Development 118, 283-295. Bier, E., Jan, L.Y. and Jan, Y.N. (1990) Genes Dev. 4, 190-203. Bodmer, R., Carretto, R. and Jan, Y.N. (1989) Neuron 3, 21-32. Bossing, T. and Technau, G.M, (1994) Development 120, 1895-1906. Brand, A.H. and Perrimon, N. (1993) Development 118, 401-415. Brunner, D., Dticker, K., Oellers, N., Hafen, E., Scholz, H. and Kl~nbt, C. (1994) Nature 370, 386-389. Crews, S.T., Thomas, J.B. and Goodman, C.S. (1988) Cell 52, 143-151. Diaz-Benjumea, F.J. and Hafen, E. (1994) Development 120, 569-578. Dickson, B. (1995) Trends Genet. 11, 106-111. Dunin-Borkowski, O.M. and Brown, N.H. (1995) Developmental Biology 168, 689-693. Elkins, T., Zinn, K., McAllister, L., Hoffmann, F.M. and Goodman, C.S. (1990) Cell 60, 565-575. Fortini, M.E., Simon, M.A. and Rubin, G.M. (1992) Nature 355, 559561. Freeman, M., Klfirnbt, C., Goodman, C.S. and Rubin, G.M. (1992) Cell 69, 963-975. Gabay, L., Scholz, H., Golembo, M., Klaes, A., Shilo, B.-Z. and Kl~rnbt, C. (1996) Development 120, 3355-3362.

Giesen, K., Hummel, T., Harrison, S., Travers, A. and Klgrnbt, C. (1997) Development, submitted. Golembo, M., Schweitzer, R., Freeman, M. and Shilo, B.Z. (1996) Development 122, 223-230. Gonzalez-Reyes, A., Elliott, H. and St. Johnston, D. (1995) Nature 375, 654-658. Ham, J., Babij, C., Whitfield, J., M. Pfarr, C.M., Lallemand, D., Yaniv, M. and Ruhin, L.L. (1995) Neuron 14, 927-939. Hosoya, T., Takizawa, K., Nitta, K. and Hotta, Y. (1995) Cell 82, 10251036. Ito, K., Urban, J. and Technau, G.M, (1995) Roux's Arch. Dev. Biol. 204, 284-307." Jacobs, J.R. and Goodman, C.S. (1989) J. Neurosci. 9, 2402-2411. Jones, B.W., Fetter, R.J., Tear, G. and Goodman, C.S.G. (1995) Cell 82, 1013-1023. Karim, F.D., Chang, H.C., Therrien, M., Wassarman, D.A., Laverty, T. and Rubin, G.M. (1996) Genetics 143, 315-329. Kasai, Y., Nambu, J.R,, Lieberman, P.M. and Crews, S.T. (1992) Proc. Natl. Acad. Sci. USA 89, 3414-3418. Klaes, A., Menne, T., Stollewerk, A., Scholz, H. and Klambt, C. (1994) Cell 78, 149-160. K1Lrnbt, C. (1993) Development 117, 163-176. KlSanbt, C. and Goodman, C.S. (1991) Glia 4, 205-213. Klambt, C., Jacobs, R.J. and Goodman, C.S. (1991) Cell 64, 801-815. Kl~nbt, C., Hummel, T., Menne, T., Sadlowski, S., Scholz, H. and Stollewerk, A. (1996) Dev. Genet. 18, 40-49. Kolodldn, A.L., Pickup, A.T., Lin, D.M., Goodman, C.S. and Banerjee, U. (1994) Development 120, 1731-1745, Kuo, Y.M., Jones, N., Zhou, B., Panzer, S., Larson, V. and Beckendoff, S.K. (1996) Development 122, 1909-1917. Lai, Z.-C. and Rubin, G.M. (1992) Cell 70, 609-620. Lee, T., Feig, L. and Montell, D.J. (1996) Development 122, 409-418. Martfn-Bermudo, M.D., Carmena, A. and Jim6nez, F. (1995) Development 121,219-224. Mayer, U. and Niisslein-Volhard, C. (1988) Genes Dev. 2, 1496-1511. Mellerick, D.M. and Nirenberg, M. (1995) Dev. Biol. 171, 306-316. Menne, T.V. and Kl~nbt, C. (1994) Development 120, 123-133. Nambu, J.R., Franks, R.G., Hu, S. and Crews, S.T. (1990) Cell 63, 6375. Nambu, J.R., Lewis, J.O., Wharton, K.A. and Crews, S.T. (1991) Cell 67, 1157-1167. Neuman-Silberberg, F.S. and Schupbach, T. (1993) Cell 75, 165-174. Ntisslein-Volhard, C., Wieschaus, E. and Kluding, H. (1984) Roux's Arch Dev. Biol. 193, 267-282. O'Neill, E.M., Rebay, I., Tjian, R. and Rubin, G.M. (1994) Cell 78, 137147. Okabe, M., Sawamoto, K. and Okano, H. (1996) Dev. Biol. 175, 3749. Perrimon, A.H.B.N. (1994) Genes Dev. 8, 629-639. Raz, E. and Shilo, B.-Z. (1992) Development 114, 113-123. Rebay, I, and Rubin, G.M. (1995) Cell 81,857-866. Rogge, R., Green, P.J., Urano, J., Horn-Saban, S., Mlodzik, M., Shilo, B.-Z., Hartenstein, V. and Banerjee, U. (1995) Development 121, 3947-3958. Rosenberg, U.B., Preiss, A., Seifert, E., J~ickle, H. and Knipple, D.C. (1985) Nature 313, 703-706. Roth, S., Neuman-Silberberg, F.S., Barcelo, G. and Schupbach, T. (1996) Cell 81,967-978. Rutledge, B.J., Zhang, K., Bier, E., Jan, Y.N. and Perrimon, N. (1992) Genes Dev. 6, 1503-1517. Schmucker, D., Taubert, H. and J~ickle, H. (1992) Neuron 9, 1025-1039. Scholz, H., Deatrick, J., Klaes, A. and Kl~nbt, C. (1993) Genetics 135, 455-468. Schweitzer, R., Howes, H., Smith, R., Shilo, B.-Z. and Freeman, M. (1995a) Nature 376, 699-702. Schweitzer, R., Shahrabany, M., Seger, R. and Shilo, B.-Z. (1995b) Genes Dev. 9, 1518-11529.

H. Scholz et al. /Mechanisms of Development 62 (1997) 79-91

Seeger, M., Tear, G., Ferres-Marca, D. and Goodman, C.S. (1993) Neuron 10, 409-426. Simon, M.A., Bowtell, D.D.L., Dadson, G.S., Laverty, T.R. and Rubin, G.M. (1991) Cell 67, 701-716. Sonnenfeld, M.J. and Jacobs, J.R. (1995) Development 121,569-578. StoUewerk, A., K1Nnbt, C. and Cantera, R. (1996) Microsc. Res. Tech. 35, 294-306. Sturtevant, M.A., Roark, M. and Bier, E. (1993) Genes Dev. 7,961-973. Sturtevant, M.A., Roark, M., O'Neill, J.W., Biehs, B., Colley, N. and Bier, E. (1996) Dev. Biol. 174, 298-309. Tei, H., Nihonmatsu, I., Yokokura, T., Ueda, R., Sano, Y., Okuda, T., Sato, K., Hirata, K., Fujita, S. and Yamamoto, D. (1992) Proc. Natl. Acad. Sci. USA 89, 6856-6860. Thomas, J.B., Crews, S. and Goodman, C.S. (1988) Cell 52, 133-141. Treier, M., Bohmann, D. and Mlodzik, M. (1995) Cell 83, 753-760. Vincent, S., Vonesch, J.-L. and Giangrande, A. (1996) Development 122, 131-139.

91

Volk, T. and VijayRaghavan, K. (1994) Development 120, 59-70. Wassarman, D.A., Therrien, M. and Rubin, G.M. (1995) Curr. Opin. Genet. Dev., 44-50. Wharton, K.A. and Crews, S.T. (1993) Mech. Dev. 40, 141-154. Xia, Z., Dickens, M., Raingeaud, J., Davis, R.J. and Greenberg, M.E. (1995) Science 270, 1326-1331. Xiong, W.C. and Montell, C. (1993) Genes Dev. 7, 1085-1096. Xu, T. and Rubin, G.M. (1993) Development 117, 1223-1237. Yamamoto, D., Nihonmatsu, I., Matsuo, T., Miyamoto, H., Kondo, S., Hirata, K. and Ikegami, Y. (1996) Roux's Arch. Dev. Biol. 205, 215224. Zak, N.B., Wides, R.J., Schejter, E.D, Raz, E. and Shilo, B.-Z. (1990) Development 109, 865-874. Zhou, L., Hashimi, H., Schwarz, L.M. and Nambu, J.R. (1995) Curt. Biol. 5, 784-790.