mGCMa is a murine transcription factor that overrides cell fate decisions in Drosophila

Mechanisms of Development 82 (1999) 141–150

mGCMa is a murine transcription factor that overrides cell fate decisions in Drosophila Rita Reifegerste a, Jo¨rg Schreiber b, Sven Gu¨lland b, Anja Lu¨demann b, Michael Wegner b ,* a

Department of Cell Biology, Emory University, 1648 Pierce Drive, Atlanta, GA 30322-3030, USA Zentrum fu¨r Molekulare Neurobiologie, Universita¨t Hamburg, Martinistrasse 52, D-20246 Hamburg, Germany

b

Received 27 January 1999; accepted 5 February 1999

Abstract During neural development of Drosophila melanogaster, Glial Cells Missing (GCM), functions as a binary switch that promotes glial cell fate while simultaneously inhibiting the neuronal fate. Sequence similarities between GCM and the recently identified mouse protein mGCMa are strictly limited to the aminoterminal DNA-binding domain. Here we show that mGCMa efficiently activates transcription in Drosophila cells just as Drosophila GCM activates transcription in mammalian cells. Transactivation potential was present in two separate regions of mGCMa outside the DNA-binding domain. One of them mapped to the carboxyterminal 88 amino acids, a location corresponding exactly to the transactivation domain of GCM. Similarities between GCM and mGCMa were also observed in vivo. Overexpression of mGCMa in the developing nervous system of Drosophila embryos led to an increase in glial-like cells at the expense of neurons. Outside the neurogenic region, mGCMa interfered with epidermal development, as evident from changes in cell morphology and marker expression. Thus, mGCMa function is at least partially independent of a cell’s predisposition to a neural fate. The potent activity of mGCMa in Drosophila and its extensive functional similarities to GCM make mGCMa a candidate for a regulator of mouse glial development.  1999 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Glial Cells Missing; Glia; Transcription; Determination; Neural development

1. Introduction Neurons and glial cells are the two major cell types in the nervous system. Both arise from common neuroectodermal precursor cells throughout the peripheral and the central nervous system (Jan and Jan, 1994). What directs the precursor towards these alternate cell fates has only recently been rendered accessible to analysis with the identification of Glial Cells Missing (GCM) in Drosophila (Hosoya et al., 1995; Jones et al., 1995; Vincent et al., 1996). Whenever GCM was expressed, the precursor adopted a glial fate. In the absence of GCM, precursors underwent neuronal differentiation. Ectopic expression of GCM in the embryonic nervous system led to an increase of glial cells, whereas inactivation of GCM caused a reciprocal increase of neurons with absolute cell numbers in the nervous system * Corresponding author. Tel.: +49-40-42803-6274; fax: +49-40-428036602.

remaining relatively constant (Hosoya et al., 1995; Jones et al., 1995). Thus, GCM has been postulated to function as a binary switch between both cell fate choices in cells of neuroectodermal origin (Anderson, 1995; Pfrieger and Barres, 1996). Recently, this picture has become more complicated. Firstly, GCM was found to be transiently expressed in the scavenger cell lineage, thus showing that GCM might have additional functions in cells outside the nervous system (Bernardoni et al., 1997). Secondly, glial cell differentiation could be induced by GCM outside the nervous system in mesodermal cells, indicating that GCM function does not require neural competence (Bernardoni et al., 1998). We have previously shown that GCM exerts its function as a transcription factor (Schreiber et al., 1997). Its DNAbinding domain has been mapped to the aminoterminal region of the protein and exhibited a strong affinity for the octameric sequence 5′-ATGCGGGT-3′ (Akiyama et al., 1996; Schreiber et al., 1997). The carboxyterminal part of

0925-4773/99/$ - see front matter  1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S09 25-4773(99)000 27-1

142

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

GCM, on the other hand, contained a modular transactivation domain. When tested on simple promoter constructs, GCM’s activity as a transcriptional activator was strongly dependent on the presence of multiple binding sites in the promoter, indicating that it functioned best in a cooperative manner (Schreiber et al., 1997). In the meantime, several proteins have been identified from mouse and man which share significant homology to the DNA-binding domain of GCM (Akiyama et al., 1996; Altshuller et al., 1996). As shown for mGCMa from mouse, these proteins exhibit DNA-binding characteristics virtually identical to the prototypic GCM from Drosophila (Schreiber et al., 1998). Together these proteins constitute the GCM family of transcription factors. Interestingly, sequence similarity outside the DNA-binding domain is absent between different members of the

GCM family. Especially the carboxyterminal transactivation domain of GCM is poorly conserved among mammalian family members. Furthermore, the mammalian GCM proteins have so far proved to be challenging to study with regards to their expression and function during embryonic development. Expression of mGCMa, which has also been described as Gcm1, could not be detected using in situ hybridization on embryos of various stages and neonates (Altshuller et al., 1996). Here, we have undertaken a comparative analysis of the function of mGCMa in various in vitro systems and by ectopic expression in transgenic flies.

2. Results 2.1. Transcriptional activities of GCM and mGCMa in tissue culture To directly compare the function of GCM and mGCMa, we analyzed the influence of either protein on the expression of a reporter gene driven by multimerized binding sites for GCM proteins in transiently transfected cells. Similar to previously obtained results (Schreiber et al., 1997), Drosophila GCM caused a 43-fold increase in the expression of a luciferase reporter with three tandemly arranged binding sites upon co-transfection into U138 human glioblastoma cells (Fig. 1A). When GCM was replaced by the mammalian mGCMa, a slightly higher increase in gene expression was observed with activation rates now reaching 49-fold inductions. The transcriptional activity of Drosophila GCM and mGCMa were also compared in Drosophila S2 cells. Using a luciferase reporter with six GCM binding sites (6 × gbs luc), mGCMa led to a significant 190-fold stimulation of reporter gene expression (Fig. 1B). In parallel transfections, Drosophila GCM caused a 102-fold stimulation. Thus, mGCMa and GCM both function in mammalian cells as well as in Drosophila cells as transcriptional activators. 2.2. Dominant negative function of the isolated DNAbinding domains of GCM and mGCMa

Fig. 1. Transcriptional activities of mGCMa and Drosophila GCM. (A) A luciferase reporter plasmid carrying three tandemly arranged GCM binding sites (3 × gbs luc) was transfected into U138 glioblastoma cells together with empty expression plasmid (-) or expression plasmids for full-length (full length) or truncated versions (DBD) of mGCMa and Drosophila GCM. (B) A luciferase reporter with six tandemly arranged GCM binding sites (6 × gbs luc) was transfected into Drosophila S2 cells together with empty expression plasmid (-) or expression plasmids for mGCMa and Drosophila GCM. (C) 3 × gbs luc was transfected into U138 cells in the presence of full-length mGCMa and increasing amounts of expression plasmids for the isolated DNA-binding domains of mGCMa (filled box) or GCM (open box). Luciferase activities in extracts from transfected cells were determined in three independent experiments, each performed in duplicates. Data are presented as fold inductions which were calculated for each reporter plasmid by comparing luciferase activities to values from cells transfected with reporter plasmid and empty expression plasmid (A,B), or as % activities relative to the maximal stimulation obtained for mGCMa which was arbitrarily given a value of 100% (C).

Removal of all sequences carboxyterminal of the DNAbinding domain was sufficient to prevent promoter activation both in the case of Drosophila GCM and mGCMa (Fig. 1A). The truncated proteins were, however, not only transcriptionally inactive but could also interfere with the function of the full-length proteins. When increasing amounts of the DNA-binding domain of mGCMa were co-transfected with mGCMa holoprotein, mGCMa-dependent activation of reporter gene expression was suppressed (Fig. 1C). Interestingly, the DNA-binding domain of Drosophila GCM was also capable of interfering with the function of mGCMa. When full length Drosophila GCM was used as a transcriptional activator, similar effects were observed. Again, both DNA-binding domains decreased reporter gene expression

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

143

in a dose-dependent manner (data not shown). Therefore, both DNA-binding domains possess dominant-negative function for the homologous holoprotein as well as for the respective heterologous holoprotein. 2.3. Transactivation potential of GCM and mGCMa The transactivation potential had previously been mapped to the extreme carboxyterminus of GCM (Schreiber et al., 1997), a region that does not contain any sequence homology to the corresponding part of mGCMa. Nevertheless, Fig. 1 shows that mGCMa also contains transactivation potential in a region carboxyterminal of the DNA-binding domain. To map the transactivation domain of mGCMa, we fused full-length mGCMa to the DNA-binding domain of the yeast transcription factor Gal4 (Gal4 DBD) and tested this fusion protein for its ability to activate a reporter gene driven by a Gal4-dependent promoter. Using this construct, we obtained a six-fold stimulation of reporter gene expression upon transfection into U138 cells (Fig. 2A). We had previously observed for Drosophila GCM, that in the context of Gal4 fusion constructs the presence of the GCM DNA-binding domain decreased the transactivation potential of its carboxyterminal region (Schreiber et al., 1997). When the aminoterminal 169 amino acids of mGCMa which contain the DNA-binding domain were deleted, stimulation rates increased on average from sixfold to 17-fold. Whether this effect is simply due to functional interference between Gal4 DBD and the DNA-binding domain of mGCMa within the fusion protein, or is a consequence of biologically relevant allosteric interactions between DNA-binding and transactivation domains of mGCMa as recently suggested for other transcription factors (Lefstin and Yamamoto, 1998), is unclear at present. Even amino acids 349–436 were still able to elicit a significant 13-fold stimulation of reporter gene expression. Thus, the last 88 amino acids of mGCMa contain a transactivation domain which is active after transfer to a heterologous DNA-binding domain. The localization of this transactivation domain correlates exactly to the localization of the transactivation domain of Drosophila GCM (Schreiber et al., 1997). Thus, although there is no sequence conservation between the extreme carboxyterminal regions of GCM and mGCMa, both regions have analogous functions. A second deletion series of Gal4 fusions was constructed with mGCMa parts starting at amino acid 169 and extending variable distances towards the carboxyterminus (Fig. 2B). A fusion that contained amino acids 169–400 of mGCMa exhibited a 28-fold induction of reporter gene expression. Similar activation rates were observed for constructs which only contained mGCMa sequences up to amino acid 349 (29-fold) or to amino acid 300 (36-fold). Interestingly, the fusion between Gal4 DBD and amino acids 169–300 of mGCMa is completely non-overlapping with the previously described transactivation domain present in the carboxyterminal region of mGCMa.

Fig. 2. Transactivation potential of mGCMa. The Gal4-responsive luciferase reporter (3 × UAS luc) was transfected into U138 cells together with expression plasmids for Gal4 DBD or for Gal4 fusions carrying either fulllength mGCMa (aa 1–436) or various parts of mGCMa as listed on the left. Expression and correct subcellular localization was confirmed for all Gal4/ mGCMa fusions by Western blot analysis of nuclear extracts from transfected cells using a monoclonal antibody directed against Gal4 DBD. Luciferase activities were determined in three independent experiments, each performed in duplicates. Data are presented for each Gal4 fusion as fold induction above the level of luciferase activity obtained in transfections with an expression plasmid for the Gal4 DNA-binding domain, which was given an arbitrary value of one. DBD, DNA-binding domain; TA1, TA2, transactivation domains 1 and 2.

When the mGCMa sequences were further deleted so that only amino acids 169–260 were present, transactivation potential was lost (Fig. 2B). A Gal4 construct which contained amino acids 220–300 of mGCMa, on the other hand, still exhibited a significant ten-fold transactivation, which was only abolished upon further trimming to amino acids 260–300 (Fig. 2C). It therefore seems that the 131 amino acids between positions 169 and 300 contain a second transactivation domain with important parts of the domain still present between positions 220 and 300. Thus, it has to be concluded that mGCMa contains two transactivation domains which are centered around amino acids 220–300 and 349–436, respectively, and can function separately from each other. 2.4. Ectopic expression of mGCMa in the nervous system of transgenic Drosophila embryos Our tissue culture experiments indicate a striking functional resemblance between Drosophila GCM and mGCMa. To test whether such a conservation could also be observed in vivo, we ectopically expressed mGCMa in Drosophila using the Gal4 system (Brand and Perrimon, 1993). To this end we introduced a Gal4-responsive transgene (UASmGCMa) by germline transformation into flies. The resulting UAS-mGCMa line was crossed with a scabrous-Gal4 line (Klaes et al., 1994). The scabrous-Gal4 transgene allowed overexpression of mGCMa in neuroectoderm as well as neuroblasts and their progeny. As previously shown, similar overexpression of Drosophila GCM in the embryonic nervous system led to the ectopic expression of the glial marker Repo in presumptive neurons (Hosoya et

144

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

al., 1995; Jones et al., 1995). Repo is a homeodomain protein expressed specifically in all glial cells of the embryonic nervous system with the exception of midline glia (Campbell et al., 1994; Xiong et al., 1994; Halter et al., 1995). Ectopic mGCMa expression under the control of the scabrous-Gal4 transgene is lethal to the animal. Nerve cord condensation did not reach the same extent in scabrous-Gal4/UAS-mGCMa embryos as in age-matched wildtype embryos. Repo expression was observed by immunostaining in many more cells of the central and peripheral nervous system (brown labeling in Fig. 3B,D,F) than in wild-type embryos (black labeling in Fig. 3C,E). Most of these Repo-positive cells exhibited the elongated or irregular shape, typical of glial cells, instead of the round neuronal shape. Formation of these glial-like cells was at the expense of neurons as evident from double-staining using antibodies against Elav and Repo (compare Fig. 3G and H; a-Elav in red, a-Repo in green). Anti-Elav antibodies stain all neuronal nuclei of the developing central and peripheral nervous system (Bier et al., 1988; Robinow and White, 1988). A quantitative analysis revealed a 60%–80% (n = 35) reduction of the number of neurons per hemisegment (61 ± 20 Elav-positive cells per hemisegment as compared to the expected 200 neurons in wild-type) and a similar increase in glial cells. This situation is comparable to the one described in gain-of-function studies for Drosophila GCM (Hosoya et al., 1995; Jones et al., 1995). Embryos were also double-labeled with anti-Repo antibody and Mab 22C10 (shown for scabrous-Gal4/UASmGCMa embryos in Fig. 3B,D). Mab 22C10 stains all neurons of the peripheral nervous system, as evident from Fig. 3A, for wild-type embryos. Again, it became obvious that the excess of glial cells in scabrous-Gal4/UAS-mGCMa embryos is paralleled by a concomitant decrease in neuronal cells. Like GCM, overexpression of mGCMa therefore promotes glial cell differentiation in the developing Drosophila nervous system, while repressing neuronal differentiation at the same time. Both staining with the anti-Elav antibody and with Mab 22C10 showed that the phenotypic penetrance of the mGCMa transgene was not complete. Most likely, those neurons that formed had not accumulated adequate mGCMa protein levels early enough, so that mGCMa did not prevail over the default neuronal differentiation program. A good example for the extent of penetrance obtained in scabrous-Gal4/UAS-mGCMa embryos is shown in Fig. 3D. The arrowheads point to the position of the five Mab 22C10positive neurons of the pentascolopidial lateral chordotonal organ (LCH5) (Hertweck, 1931), which are easily identified due to their five prominent dorsal dendrites, and which are found in each abdominal segment in wild-type embryos (inset in Fig. 3A). These five neurons of LCH5 are mostly missing from abdominal segments after ectopic expression of mGCMa (open arrowheads in Fig. 3D), but can form occasionally (filled arrowhead in Fig. 3D).

2.5. Expression of mGCMa outside the nervous system Recently, it was shown in Drosophila that GCM is also able to induce glial development in cells of mesodermal origin (Bernardoni et al., 1998). These cells normally do not express GCM, and are even derived from a different germ layer as the neuroectodermal cells which are usually targeted by GCM (Bernardoni et al., 1998). The most striking phenotype observed by ectopic expression of GCM under the control of a twist-Gal4 transgene was the transformation of somatic muscles into Repo-positive cells with glial morphology suggesting that glial differentiation in flies does not depend on a neural default state. To test whether mGCMa is also able to override endogenous developmental programs outside the neurogenic region, we expressed UAS-mGCMa under the control of an armadillo-Gal4 transgene (arm-Gal4, (Sanson et al., 1996)). When arm-Gal4 is provided paternally in armGal4/UAS-mGCMa trans-heterozygotes, mGCMa is expressed ubiquitously from stage nine of embryonic development onwards. The segment polarity gene armadillo plays a dual role as it is involved in the function of adherens junctions in all cells and in the transduction of wingless signaling during the anterior-posterior patterning of the embryonic epidermis of Drosophila (Peifer and Wieschaus, 1990) (reviewed in Willert and Nusse, 1998)). Cells receiving wingless signaling are arranged as segmentally repeated stripes in the embryonic epidermis. Armadillo accumulates to high levels in the nuclei of these cells where it mediates transcription of wingless downstream genes (Orsulic and Peifer, 1996; Pai et al., 1997; van de Wetering et al., 1997). Overexpression of mGCMa under the control of armGal4 was as detrimental to the Drosophila embryo as expression under the control of sca-Gal4. As a consequence of arm-Gal4 directed expression of mGCMa, ectopic formation of Repo-positive cells was observed in a segmentally repeated, striped pattern that resembled the winglessdependent domain of Armadillo protein accumulation (Fig. 4A,B). These Repo-positive cells were characterized by an irregular and elongated shape typical of glial cells. To determine whether these cells had lost their epidermal identity, we stained arm-Gal4/UAS-mGCMa heterozygotes with an antibody against Fasciclin III (FasIII, Fig. 4D). In stage-13 wild-type embryos fasIII expression was detected in all epidermal cells, with higher levels of expression seen in segmentally repeated stripes (Fig. 4C) (Patel et al., 1978). In arm-Gal4/UAS-mGCMa heterozygote embryos FasIII staining in the epidermis was greatly reduced (Fig. 4D). The mGCMa-expressing, Repo-positive cells no longer expressed the epidermal marker. We also investigated expression of tramtrack p69 (ttk p69) and pointed P1 (pnt P1) following arm-Gal4 driven induction of mGCMa in the ectoderm. These genes have been reported to act genetically downstream of gcm in directing proper development of lateral glia in the central nervous system (Klaes et al., 1994; Giesen et al., 1997).

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

145

Fig. 3. Overexpression of mGCMa in neuroblasts and their progeny. (A,C) Ventro-lateral view of stage-15 wild-type embryos stained with Mab 22C10 (A) and anti-Repo antibodies (C) showing the full complement of peripheral neurons and glial cells at the end of embryonic development. The inset in (A) shows two examples of neuronal clusters forming the lateral chordotonal organ (LCH5). (B,D) sca-Gal4/UAS-mGCMa embryo double-labeled with Mab 22C10 in black and anti-Repo antibody in brown. (B) A ventral and slightly lateral view of a stage-15 embryo. The embryo appears younger as judged by the extent of nerve-cord condensation. (D) Magnified lateral view of three neighboring abdominal segments. The arrowheads points to LCH5 neurons, which occasionally formed (filled arrowhead) but were missing in general (open arrowhead). (E,F) Whole-mount and dissected ventral nervous system of wild-type (E) and scaGal4/UAS-mGCMa (F) embryos (stage 15) stained with anti-Repo antibody. (E) is focused on the level of longitudinal glial cells (LG). Note in (F) the massive overexpression of Repo in cells of the central nervous system after ectopic expression of mGCMa in all neuroblasts and their progeny. (G,H) Confocal images of wild-type (G) and sca-Gal4/UAS-mGCMa embryonic nervous systems (H) double stained with antibodies against Repo in green and a neuronal nuclear antigen Elav in red. Anterior is to the right in all panels and ventral is down in (A,B,D). Scale bar is 80 mm in (A,B,C), 25 mm in (D) and 45 mm in (E,F,G,H).

Therefore, they might be expected to be up-regulated upon mGCMa expression in ectoderm. We used the lacZ-P-element insertion lines ttk10556 (Karpen and Spradling, 1992) and pntrM254 (Klaes et al., 1994) in which b-galactosidase-

staining faithfully mimics the expression of ttk p69 and pnt P1, respectively. When embryos which carried all three constructs (UAS-mGCMa; arm-Gal4; pntrM254 or ttk10556) were double-labeled with antibodies against LacZ and

146

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

Fig. 4. Overexpression of mGCMa in ectodermal cells and neuroblasts. (A,B) anti-Repo stain of arm-Gal4/UAS-mGCMa embryos. (A) Ventro-lateral view, (B) Ventral view. Ectopic repo-expression was detected in stripes of ectodermal cells corresponding to the Wingless regulated armadillo expression domain (A), in cells of the ventral epidermis, and in cells underlying the ventral nervous system (B). arm-Gal4/UAS-mGCMa embryos appeared to be arrested in germ-band retraction stage 12–13. (C,D) anti-FasciclinIII staining of stage-13 wild-type (C), and arm-Gal4/UAS-mGCMa (D) embryos. In wild-type, FasciclinIII was expressed in all cells of the epidermis with elevated expression seen in segmental stripes of cells. Note the considerable reduction of FasIII staining in arm-Gal4/UAS-mGCMa embryos in (D) indicating transformation of a large number of cells into a different cell fate. (E,F) Dissected ventral nervous system of stage-13 wild-type (E) and arm-Gal4/UAS-mGCMa (F) embryos stained with anti-Eve antibody. Shown are three different Eve-positive neuronal lineages RP2, aCC/pCC and EL. The arrowheads in (F) point to one example of a missing pair of RP2 neurons in arm-Gal4/UAS-mGCMa embryos. Anterior is to the right in all panels and ventral is down in (A,C,D). Scale bar is 80 mm in (A,B), 60 mm in (C,D) and 45 mm in (E,F).

Repo, no b-galactosidase could be detected in Repo-positive cells of the epidermal region. Thus, it is unclear at present, whether the cells which had lost FasIII and gained repo expression, were fully transformed to the glial fate. armadillo gene function is also required in the neuroectoderm and in neuroblast rows 4, 5 and 6 for neuroblast formation and identity (reviewed in Bhat, 1998; Bhat, 1999). Development of these neuronal lineages in the armadillo-expression domain might therefore be affected by ectopic mGCMa expression. We focused our attention on one of the best described neuronal lineages, RP2/sib (ChuLaGraff et al., 1995; Bhat, 1996; Bhat and Schedl, 1997). The RP2 neuron is an even-skipped (eve) positive motoneuron which innervates muscle number two of the dorsal musculature. RP2 derives from the first ganglion mother cell (GMC-1) of neuroblast NB4-2. In stage-13 wild-type embryos, the RP2 neurons were reliably identified by their position and by staining with an anti-Eve antibody (Fig. 4E). Missing RP2 neurons were a prominent phenotype in armGal4/UAS-mGCMa trans-heterozygotes (open arrowheads in Fig. 4F). This phenotype was observed in 28% (n = 91)

of hemisegments. In a fraction of cases, RP2 neurons might also be displaced rather than lost. Additionally, the nervous system of these embryos showed a variety of other defects, including occasionally missing U/CQ neurons derived from neuroblast NB6-2 and a minor reduction in the number of aCC/pCC neurons originating from NB1-1 (data not shown). The broad range of effects and the phenotypic variability again suggest that, the amount of accumulated mGCMa protein as well as the time point of expression, play crucial roles in determining cell fate decision.

3. Discussion Since its recent discovery, a lot has been learnt about the function of GCM and its mode of action as a transcription factor during glial development of Drosophila (Hosoya et al., 1995; Jones et al., 1995; Akiyama et al., 1996; Vincent et al., 1996; Bernardoni et al., 1997; Schreiber et al., 1997; Bernardoni et al., 1998). GCM has also been found to be the

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

prototype of a new group of transcription factors as GCMrelated proteins have been identified in mouse and man (Akiyama et al., 1996; Altshuller et al., 1996; Schreiber et al., 1998). Which functions these GCM-related proteins perform during mammalian development is unknown. 3.1. GCM and mGCMa show similar function in the highly conserved DNA-binding domain as well as in less conserved regions of the protein We have recently shown that mGCMa, a GCM-related protein from mouse, has DNA-binding characteristics very similar to Drosophila GCM as evidenced by the fact that both proteins yielded identical methylation interference patterns upon DNA binding and reacted in an almost indistinguishable manner to base pair changes within the octameric DNA recognition motif (Schreiber et al., 1998). We now show that both proteins also behave very similarly in their property as transcription factors when tested either in mammalian or Drosophila cells. Their similar mode of action is also corroborated by transfection experiments with truncated forms, in which sequences outside the DNA-binding domain were removed. These isolated DNA-binding domains were not only transcriptionally inactive, but behaved as dominant negatives in tissue culture experiments, a characteristic which in the future could be exploited for studying GCM function in vivo. Intriguingly, both isolated DNA-binding domains interfered with transcriptional activity of mGCMa and GCM giving further evidence for the similarity between both proteins. Despite the fact that sequence similarity is restricted to the aminoterminal DNA-binding domain, functional similarity could also be observed between the carboxyterminal parts which both harbor the respective protein’s transactivation potential (Schreiber et al., 1997 and this manuscript). mGCMa and GCM contain a potent transactivation domain in their last 88 and 85 amino acids, respectively. The location of one transactivation domain is thus conserved. A second independently functioning transactivation domain between amino acids 220–300 of mGCMa, however, seems to be unique to the mouse protein. 3.2. Overexpression studies of mGCMa in transgenic Drosophila embryos is compatible with a function for mGCMa in mammalian glial development Given the extensive similarities between mGCMa and GCM in various in vitro assays, we also compared both proteins on a functional level in vivo. We introduced a UAS-mGCMa transgene into flies and ectopically expressed the transgene in neurogenic and non-neurogenic regions of the embryos. As similar gain-of-function experiments have been performed on GCM (Hosoya et al., 1995; Jones et al., 1995; Bernardoni et al., 1998), the effects of ectopic mGCMa expression can be directly compared with those

147

of GCM. Again, mGCMa behaved very similarly to GCM in a number of aspects. When ectopically expressed in the neurogenic region, mGCMa interfered with normal development as evidenced by an increase in glial cells and a concomitant decrease in neurons. The increase in glial cells was measured as a surplus of cells with glial morphology that stained positive for Repo, a homeodomain protein activated in all except midline glia shortly after GCM (Campbell et al., 1994; Xiong et al., 1994; Halter et al., 1995). Because of its expression pattern and because of the presence of 11 GCM-binding sites in its 5′ flanking region (Akiyama et al., 1996), repo very likely is a direct downstream target for GCM in the cascade towards glial differentiation. Intriguingly, mGCMa overrode endogenous cell fate decisions when ectopically expressed in ectoderm outside the neurogenic region. Normal epidermal development was suppressed as measured by a change in cell morphology and a decrease in the expression of the epidermal cell adhesion molecule FasIII. The loss of FasIII staining was equalled by an increase in Repo-staining. Whether these Repo-positive cells are genuine glial cells is difficult to assess at present as there are very few markers available. We used two lacZ-Pelement insertion lines, which faithfully mimic expression of pointed P1 and tramtrack p69. These genes have been reported to be important for development of lateral glia in the central nervous system and to be genetically downstream of GCM (Klaes et al., 1994; Giesen et al., 1997). However, neither one is truly glia-specific. Therefore, there seemed to be a good chance, but no certainty of obtaining ectopic lacZ expression in Repo-positive cells upon epidermal mGCMa expression. Nevertheless, we were not able to detect ectopic expression of either marker. Very recently ectopic expression in cells of mesodermal origin has been reported for Drosophila GCM (Bernardoni et al., 1998). Cells that normally develop into somatic or visceral muscle, instead developed into Repo-positive cells with glial morphology. In this case, glial fate of the cells was confirmed by detecting GCM-induced ectopic lacZ expression in a fraction of the Repo-positive cells in the lacZ-P-element insertion lines AE2 and rC56 which reflect expression of gliotactin and a so far unidentified glial marker (Bernardoni et al., 1998). Thus, there is an apparent difference between the results reported by Bernardoni et al. (1998) and the present study. One reason for this difference could be our use of mGCMa instead of GCM. However, other explanations are equally plausible, as different tissues were targeted in both studies and different lacZ-P-element insertion lines were used for the analysis. Furthermore, expression levels of the UASGCM and the UAS-mGCMa transgenes in both studies cannot be properly compared. The levels, to which the GCM/mGCMa proteins accumulate, are, however, as important as the exact time at which GCM/mGCMa expression occurs. Thus, not all cells with ectopic GCM or mGCMa expression are transformed (Ber-

148

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

nardoni et al., 1998) (and this report). Cell fate is altered only if GCM/mGCMa levels are present above a certain threshold level at the moment of cell fate decision. Threshold activity of GCM/mGCMa could be a consequence of the strong cooperative function which we detected in transient transfection assays (Schreiber et al., 1997). In line with such a cooperative mode of action, multiple binding sites are present within the 5′ flanking region of the putative target gene repo (Akiyama et al., 1996). One way of accumulating high levels of GCM protein under normal in vivo conditions would be a feedback mechanism, in which GCM positively autoregulates its own expression. The fact that both GCM and mGCMa could interfere with normal differentiation even outside the neurogenic region without the existence of a prior neuroectodermal predisposition, also raises important issues regarding the control of GCM/mGCMa expression during normal development. If so potent an activator of glial development, GCM expression has to be tightly regulated in a precise temporal and spatial pattern. It has already been shown that GCM transcript and protein are both destabilized by the presence of an RNA instability signal and a PEST sequence, respectively, and that GCM protein function is redox-dependent (Hosoya et al., 1995; Jones et al., 1995; Akiyama et al., 1996; Schreiber et al., 1998). Most of these mechanisms are also realized in mGCMa (Akiyama et al., 1996; Altshuller et al., 1996; Schreiber et al., 1998). In addition, GCM and mGCMa genes must be tightly regulated on the level of transcriptional initiation. In conclusion, this manuscript provides evidence that ectopic expression of mGCMa influences embryonic development of Drosophila, both within and outside the nervous system in a manner very similar to GCM. Thus, it is tempting to speculate that mGCMa performs similar functions during mouse development as GCM does during Drosophila development. To address this, targeted deletion of mGCMa in mice is needed.

4. Materials and methods 4.1. Plasmids and site-directed mutagenesis The CMV expression plasmids for mGCMa (Akiyama et al., 1996) and Drosophila GCM (Hosoya et al., 1995), the 6 × gbs luc and the 3 × gbs luc reporter plasmid have been previously described (Schreiber et al., 1997; Schreiber et al., 1998). pCMVGal4 contained coding sequences for the DNA-binding domain (DBD) of the yeast Gal4 protein (Schreiber et al., 1997). Insertion of mGCMa sequences led to an in-frame fusion of both open reading frames, with mGCMa following Gal4 sequences. Fragments corresponding to various regions of mGCMa were generated by PCR with flanking EcoRI and SalI sites and similarly inserted into pCMVGal4 (see Fig. 2). The Gal4-responsive luciferase reporter (3 × UAS luc) contained three tandem

copies of a Gal4 binding site in front of the rat prolactin minimal promoter (Schreiber et al., 1997). An EcoRI/SacI restriction fragment spanning the complete open reading frame of mGCMa was inserted into the pUAST vector (Brand and Perrimon, 1993) after blunting the SacI site, yielding pUAST/mGCMa. 4.2. Transfections, luciferase assays, extract preparation, and Western blots COS cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) containing 10% fetal calf serum and were transfected by the DEAE dextran technique using a concentration of 500 mg/ml DEAE dextran followed by chloroquine treatment (Sock et al., 1996). U138 human glioblastoma cells were propagated in RPMI1640 medium supplemented with 10% fetal calf serum and transfected by the calcium phosphate technique (Renner et al., 1994). Drosophila S2 cells were obtained from Invitrogen, kept in DES medium containing 10% fetal calf serum, and transfected using calcium phosphate precipitates. For extract preparation, cells were transfected with 10 mg of CMV expression plasmid per 100-mm plate. For luciferase assays, cells were transfected with 2 mg of luciferase reporter plasmid and 2 mg of CMV expression plasmid per 60-mm plate unless stated otherwise. The total amount of plasmid was kept constant using empty CMV vector. Cells were harvested 48 h after transfection, and extracts were assayed for luciferase activity (Wegner et al., 1993). Nuclear extracts were prepared and analyzed by Western blots using a mouse monoclonal against the DNA-binding domain of Gal4 (Clontech) as primary antibody at a 1:3000 dilution (Schreiber et al., 1997). 4.3. Fly stocks, P-element-mediated transformation and overexpression of mGCMa White flies were transformed using pUAST/mGCMa (Brand and Perrimon, 1993) by standard transgenic techniques (Rubin and Spradling, 1982). Ten independent transgenic UAS-mGCMa lines were generated and either balanced or made homozygous. Overexpression of mGCMa was performed by crossing these lines to scaGal4 (Klaes et al., 1994), and arm-Gal4 (Sanson et al., 1996) (gift of K. Bhat) transgenic lines, respectively. The UAS-mGCMa line shown in this paper is homozygous viable and was selected according to the most penetrant phenotype seen in overexpression studies. 4.4. Immunohistochemistry Antibody staining was done according to (Doe, 1992). Samples were mounted in 75% glycerol and examined using a Zeiss Axioplan 2 microscope. Primary antibodies were: rat anti-Repo/RK2 (1:2000, (Campbell et al., 1994)), rabbit anti-Repo (1:250, (Halter et al., 1995)), Mab 22C10

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

(1:50, (Fujita et al., 1982)), rat anti-Elav (1:150, Developmental Studies Hybridoma Bank, Iowa), Mab 7G10 antiFasIII (1:2, Developmental Studies Hybridoma Bank, Iowa; (Patel et al., 1978)) and rabbit anti-Eve (1:40, gift of K. Bhat). Secondary antibodies used were: biotinylated goat anti-rat, goat anti-mouse, goat anti-rabbit antibodies (1:200, Vectastain Elite ABC Kit), donkey anti-rat LRSC (1:100, Jackson Lab), and goat anti-rabbit FITC (1:100, Jackson Lab). NiCl (0.03%) and CoCl2 (0.03%) were used in HRP mediated color reactions with diaminobenzidine as substrate to enhance the color.

5. Note added in proof While this manuscript was under revision, results virtually identical to ours were reported for ectopic panneural expression of rGCM1, the rat homolog of mGCMa, in Drosophila embryos using sca-Gal4/UAS-rGCM1 heterozygotes (Kim et al., 1998). Interestingly, transformation of presumptive neurons into glia was only observed for GCMa/GCM1, not however, for the second known mammalian GCM-related protein GCMb/GCM2. Another study showed Repo induction by ectopic expression of the Drosophila GCM in the epidermis using an en-Gal4 driver (Akiyama-Oda et al., 1998), further corroborating the functional similarities between GCM and mGCMa.

Acknowledgements We thank Gerhard Technau, Andrew Tomlinson and Larry Zipursky for antibodies, Christian Kla¨mbt for fly stocks and Krishna Bhat for antibodies, fly stocks and very helpful discussion of the results. This work was supported by a grant from the Deutsche Forschungsgemeinschaft (SFB 444) to M.W.

References Akiyama, Y., Hosoya, T., Poole, A.M., Hotta, Y., 1996. The gcm-motif: a novel DNA-binding motif conserved in Drosophila and mammals. Proc. Natl. Acad. Sci. USA 93, 14912–14916. Akiyama-Oda, Y., Hosoya, T., Hotta, Y., 1998. Alteration of cell fate by ectopic expression of Drosophila glial cells missing in non-neural cells. Dev. Genes Evol. 208, 578–585. Altshuller, Y., Copeland, N.G., Gilbert, D.J., Jenkins, N.A., Frohman, M.A., 1996. Gcm1, a mammalian homolog of Drosophila glial cells missing. FEBS Lett. 393, 201–204. Anderson, D.J., 1995. A molecular switch for the neuron-glia developmental decision. Neuron 15, 1219–1222. Bernardoni, R., Miller, A.A., Giangrande, A., 1998. Glial differentiation does not require a neural ground state. Development 125, 3189– 3200. Bernardoni, R., Vivancos, V., Giangrande, A., 1997. glide/gcm is expressed and required in the scavenger cell lineage. Dev. Biol. 191, 118–130. Bhat, K.M., 1996. The patched signaling mediates repression of goose-

149

berry allowing neuroblast specification by wingless during Drosophila neurogenesis. Development 122, 2921–2932. Bhat, K.M., 1998. Cell-cell signaling during neurogenesis: some answers and many questions. Int. J. Dev. Biol. 42, 127–139. Bhat, K.M., 1999. Segment polarity genes in neuroblast formation and identity specification during Drosophila neurogenesis. BioEssays, (in press). Bhat, K.M., Schedl, P., 1997. Requirement for engrailed and invected genes reveals novel regulatory interactions between engrailed/invected, patched, gooseberry and wingless during Drosophila neurogenesis. Development 124, 1675–1688. Bier, E., Ackerman, L., Barbel, S., Jan, L., Jan, Y.N., 1988. Identification and characterization of a neuron-specific nuclear antigen in Drosophila. Science 240, 913–916. Brand, A.H., Perrimon, N., 1993. Targeted gene expression as a means of altering cell fates and generating dominant phenotypes. Development 118, 401–415. Campbell, G., Goering, H., Lin, T., Spana, E., Andersson, S., Doe, C., Tomlinson, A., 1994. RK2, a glial-specific homeodomain protein required for embryonic nerve cord condensation and viability in Drosophila. Development 120, 2957–2966. Chu-LaGraff, Q., Schmid, A., Leidel, J., Bronner, G., Ja¨ckle, H., Doe, C.Q., 1995. huckebein specifies aspects of CNS precursor identity required for axon pathfinding and target recognition. Neuron 15, 1041–1051. Doe, C.Q., 1992. Molecular markers for identified neuroblasts and ganglion mother cells in the Drosophila central nervous system. Development 116, 855–863. Fujita, S.C., Zipursky, S.L., Benzer, S., Ferrus, A., Shotwell, S.L., 1982. Monoclonal antibodies against the Drosophila nervous system. Proc. Natl. Acad. Sci. USA 79, 7929–7933. Giesen, K., Hummel, T., Stollewerk, A., Harrison, S., Travers, A., Kla¨mbt, C., 1997. Glial development in the Drosophila CNS requires concomitant activation of glial and repression of neuronal differentiation genes. Development 124, 2307–2311. Halter, D.A., Urban, J., Rickert, C., Ner, S., Ito, K., Travers, A., Technau, G., 1995. The homeobox gene repo is required for differentiation and maintenance of glia function in the embryonic nervous system of Drosophila melanogaster. Development 121, 317–332. Hertweck, H., 1931. Anatomie und Variabilita¨t des Nervensystems und der Sinnesorgane von Drosophila melanogaster (Meigen). Z. wiss. Zool. 139, 559–663. Hosoya, T., Takizawa, K., Nitta, K., Hotta, Y., 1995. glial cells missing: a binary switch between neuronal and glial determination in Drosophila. Cell 82, 1025–1036. Jan, Y.N., Jan, L.Y., 1994. Neuronal cell fate specification in Drosophila. Curr. Opin. Neurobiol. 4, 8–13. Jones, B.W., Fetter, R.D., Tear, G., Goodman, C.S., 1995. glial cells missing: a genetic switch that controls glial versus neuronal fate. Cell 82, 1013–1023. Karpen, G., Spradling, A.C., 1992. Analysis of subtelomeric heterochromatin in the Drosophila minichromosome Dp1187 by single P element insertional analysis. Genetics 132, 737–753. Kim, J., Jones, B.W., Zock, C., Chen, Z., Wang, H., Goodman, C.S., Anderson, D.J., 1998. Isolation and characterization of mammalian homologs of the Drosophila gene glial cells missing. Proc. Natl. Acad. Sci. USA 95, 12364–12369. Klaes, A., Menne, T., Stollwerk, A., Scholz, H., Kla¨mbt, C., 1994. The ETS transcription factors encoded by the Drosophila gene pointed direct glial cell differentiation in the embryonic CNS. Cell 78, 149–160. Lefstin, J.A., Yamamoto, K.R., 1998. Allosteric effects of DNA on transcriptional regulators. Nature 392, 885–888. Orsulic, S., Peifer, M., 1996. An in vivo structure-function study of Armadillo, the b-catenin homolog, reveals both separate and overlapping regions of the protein required for cell adhesion and for Wingless signaling. J. Cell. Biol. 134, 1–18. Pai, L.M., Orsulic, S., Bejsovic, A., Peifer, M., 1997. Negative regulation

150

R. Reifegerste et al. / Mechanisms of Development 82 (1999) 141–150

of Armadillo, a Wingless effector in Drosophila. Development 124, 2255–2266. Patel, N.H., Snow, P.M., Goodman, C.S., 1978. Characterization and cloning of Fasciclin III: a glycoprotein expressed on a subset of neurons and axon pathways in Drosophila. Cell 48, 975–988. Peifer, M., Wieschaus, E., 1990. The segment polarity gene armadillo encodes a functionally modular protein that is the Drosophila homolog of human plakoglobin. Cell 63, 1167–1178. Pfrieger, F.W., Barres, B.A., 1996. What the fly’s glia tell the fly’s brain. Cell 83, 671–674. Renner, K., Leger, H., Wegner, M., 1994. The POU-domain protein Tst-1 and papovaviral T-antigen function synergistically to stimulate gliaspecific gene expression of JC virus. Proc. Natl. Acad. Sci. USA 91, 6433–6437. Robinow, S., White, K., 1988. The locus elav of Drosophila melanogaster is expressed in neurons at all developmental stages. Dev. Biol. 126, 294–303. Rubin, G.M., Spradling, A.C., 1982. Genetic transformation of Drosophila with transposable element vectors. Science 218, 348–353. Sanson, B., White, P., Vincent, J.P., 1996. Uncoupling cadherin-based adhesion from wingless signalling in Drosophila. Nature 383, 627–630. Schreiber, J., Enderich, J., Wegner, M., 1998. Structural requirements for DNA binding of GCM proteins. Nucleic Acids Res. 26, 2337–2343.

Schreiber, J., Sock, E., Wegner, M., 1997. The regulator of early gliogenesis Glial Cells Missing is a transcription factor with a novel type of DNA-binding domain. Proc. Natl. Acad. Sci. USA 94, 4739–4744. Sock, E., Enderich, J., Rosenfeld, M.G., Wegner, M., 1996. Identification of the nuclear localization signal of the POU domain protein Tst-1/Oct6. J. Biol. Chem. 271, 17512–17518. van de Wetering, M., Cavallo, R., Dooijes, D., van Beest, M., van Es, J., Loureiro, J., Ypma, A., Hursh, D., Jones, T., Bejsovec, A., Peifer, M., Mortin, M., Clevers, H., 1997. Armadillo coactivates transcription driven by the product of the Drosophila segment polarity gene dTCF. Cell 88, 789–799. Vincent, S., Vonesch, J.L., Giangrande, A., 1996. glide directs glial fate commitment and cell fate switch between neurones and glia. Development 122, 131–139. Wegner, M., Drolet, D.W., Rosenfeld, M.G., 1993. Regulation of JC virus by the POU-domain transcription factor Tst-1, Implications for progressive multifocal leukoencephalopathy. Proc. Natl. Acad. Sci. USA 90, 4743–4747. Willert, K., Nusse, R., 1998. Beta-catenin: a key mediator of Wnt signaling. Curr. Opin. Genet. Dev. 8, 95–102. Xiong, W., Okano, H., Patel, N., Blendy, J., Montell, C., 1994. repo encodes a glial-specific homeo domain protein required in the Drosophila nervous system. Genes Dev. 8, 981–994.