Development 124, 673-681 (1997) 673 Printed in Great Britain © The Company of Biologists Limited 1997 DEV7540

Mutations in lottchen cause cell fate transformations in both and lineages in the Drosophila embryonic central

Marita Buescher and William Chia* Institute of Molecular and Cell Biology, National University of Singapore, Singapore 119260 *Author for correspondence (e-mail: [email protected])

SUMMARY

The Drosophila embryonic (CNS) distinct GMC4-2a-like cells that do not share the same develops from a stereotyped pattern of neuronal progeni- parental NB, indicating that a second NB has acquired the tor cells called (NB). Each NB has a unique potential to produce a GMC and a which is identity that is defined by the time and position of its normally restricted to the NB4-2 lineage. Moreover, the ltt formation and a characteristic combination of genes it mutations lead to a loss of correctly specified longitudinal expresses. Each NB generates a specific lineage of ; this coincides with severely defective longitudinal con- and/or glia. Here we describe the genetic and phenotypic nectives. Therefore, lottchen plays a role in specifying the analysis of lottchen (ltt), a novel gene whose loss of function identity of both neuroblast and glioblast lineages in the causes a change in the identity of at least one NB as well as Drosophila embryonic CNS. We discuss the possibility that cell fate transformations within the lateral glioblast lineage. ltt may act to differentiate NB identity along the medial In wildtype the parental NB of the motoneuron lateral axis. RP2 is NB4-2. ltt embryos are distinguished by an addi- tional RP2-like neuron which appears later in development. Key words: Drosophila embryonic CNS, neuroblast identity, We show that the two RP2 neurons are derived from two longitudinal glia, RP2 motoneuron

INTRODUCTION differences between medial and lateral NBs are as yet unknown. NBs initiate unique cell lineages; each NB The central nervous system (CNS) is made up of two distinct undergoes several asymmetric cell divisions to generate, in a cell types: neurons and glia. Individual cells of both types show distinct temporal order, a set of unique intermediate precursor great diversity with respect to position, morphology and phys- cells called ganglion mother cells (GMCs); each GMC divides iological and biochemical properties. The relative simplicity of to produce two post-mitotic neuron/glia. These observations its embryonic CNS makes Drosophila a suitable model system (Doe, 1992) suggest that NB identity is dynamic and evolves for the study of developmental mechanisms that result in this in a unidirectional manner. Each GMC has a unique identity diversity. The neurons and glia of the Drosophila embryonic that is defined initially by the identity of the parental NB and CNS are arranged in a stereotypic pattern which is bilaterally ultimately by a specific combination of expressed genes. symmetric and segmentally reiterated. Each hemisegment It has been shown that individual neuroectodermal cells can contains ~300 neurons (Bossing et al., 1996b) and ~30 glia (Ito be labelled to reveal complete embryonic NB lineages et al., 1995) many of which can be identified according to their (Bossing and Technau, 1994). The terminal lineages of 17 NBs position and morphology. Furthermore, a large number of have been delineated (Udolph et al., 1993; Chu-LaGraff et al., genes have been described that are expressed in subsets of 1995; Bossing et al., 1996a,b): NB1-1 was found to generate neurons and glia and antibodies against these gene products the aCC/pCC neurons, several unidentified neurons and the A- provide molecular tools for the identification of specific cells. and B-glia. These findings indicate that neurons and glia are The neurons and glia are derived from ~30 progenitor cells not necessarily produced by distinct progenitors but that the called neuroblasts and (NBs, GBs), respectively decision to adopt a neural or glial fate can occur within one (Goodman and Doe, 1993; Doe and Technau, 1993; Jimenez lineage (Udolph et al., 1993). This type of progenitor cell can and Modolell, 1993). NBs form in a characteristic manner and be described as neuro-glioblast. However, another group of each NB has a unique identity that is defined by the time of its glia, the longitudinal glia (LG) are derived from a progenitor birth and its position. NB identity is manifested in a specific cell whose ‘mode of division’ is different from that of NBs. At combination of expressed genes. Recent experiments have least six of the LG are derived from the lateral glioblast (LGB) shown that segment polarity genes play a role in the specifica- (Jacobs et al., 1989). The LG lineage has not yet been fully tion of NB identity along the anterior-posterior axis (for described. review: Doe and Skeath, 1996). In contrast, genes that specify To understand the developmental mechanisms that lead to 674 M. Buescher and W. Chia cell diversity within the CNS, it is important to define the TM3 male progeny (F1) was collected and crossed to DTS/TM3,Sb genetic and molecular components that act to specify cell fate. virgins. The F2 generation was raised at 28¡C to eliminate progeny Previous studies have led to the identification of a large number that carry the dominant temperature-sensitve (DTS) lethal mutation. of genes and markers that are differentially expressed in The surviving flies, heterozygous for the EMS-treated chromosome and TM3,Sb were inbred to establish the final generation. Only lines subsets of NBs and GMCs (Broadus et al., 1995). However, + with several exceptions, their roles in cell fate determination with no Sb progeny, indicating the presence of a recessive lethal mutation, were screened by anti-Eve staining. are still poorly understood. From this point of view the best- studied lineage is that of NB4-2. The first-born GMC of NB4- Gamma-irradiation-induced mutagenesis 2, designated GMC4-2a (nomenclature according to Doe, Additional alleles of ltt were induced by irradiating st e iso3 males 1992) divides to produce the RP2 motoneuron and a sibling with 4500 rads from a Cobalt source. The mutagenised males were cell. Several genes have been described whose activities are mated en masse with w; TM6,Tb, e/TM3,Sb, e virgins. Single males required for the generation of a correctly specified RP2 neuron. of the F1 generation, heterozygous for the mutagenised chromosome The secreted gene product of wingless (Chu-LaGraff and Doe, and TM6,Tb, e, were crossed to virgin females of the genotype: lttEMS, 1993) is required for the formation and specification of NB4- e+/TM6, Tb, e. e males from vials that failed to yield Tb+ progeny (failed to complement lttEMS) were crossed to w; TM6,Tb, e/TM3,Sb, 2. In addition the homeobox genes fushi tarazu (ftz) and even- EMS skipped (eve) (Doe et al., 1988a,b) and the POU domain genes e virgins and balanced stocks were established. A ltt like phenotype was established by staining each line with anti-Eve, pdm1 and pdm2 are expressed in GMC4-2a and are necessary mAbBP102 and/or anti-Fasciclin II. for specifying its cellular identity and for the formation of the RP2 neuron (Yeo et al., 1995). The homeobox gene prospero Immunocytochemistry (pros; Doe et al., 1991; Vaessin et al., 1991) is expressed in Embryos were fixed and stained as previously described (Yang et al., many GMCs including GMC4-2a. Total loss of pros function 1993). For immunofluorescence FITC-conjugated anti-mouse affects the fate of many neurons including that of RP2. In antibody (Boehringer Mannheim) and biotinylated anti-rabbit addition, the putative transcription factor Huckebein appears to antibody in combination with avidin-Texas Red (Vector Laboratories) regulate aspects of GMC and neuronal identity required for the were used. ltt alleles were balanced over a ‘blue balancer’ chromo- correct differentiation of motoneurons arising from the NB4-2 some carrying an ubx-lacZ insertion to facilitate the identification of lineage (Chu-LaGraff et al., 1995). homozygous mutant embryos. The following antibodies were used: anti-Eve (rabbit, polyclonal; Despite this progress in understanding the specification of M. Frasch, The Mount Sinai Medical Center, N.Y.); anti-Eve the RP2 neuron, many components involved in this process (mouse, monoclonal; K. Zinn, Caltech), anti-Ftz (rabbit, polyclonal; are still unidentified. Here, we report the isolation of a set of W. Gehring, Biozentrum, Basel); mAbBP102, mAb22C10, anti-Fas- mutations in a novel gene, designated lottchen (name of twins ciclinII (mouse, monoclonal; C.S. Goodman, University of Califor- in the book ‘Das doppelte Lottchen’ by Erich Kaestner). In nia, Berkeley); anti-Pros (mouse, monoclonal, C.Q. Doe, University ltt loss-of-function mutant embryos the RP2 motoneuron of Illinois, anti-β-galactosidase (rabbit, polyclonal; ORGANON, appears to be duplicated and cell fate transformations are seen TEKNIKA CORPORATION). anti-Repo (RK2) (rabbit; G. in the LGB lineage. The extra RP2-like cell is not derived Technau, University of Mainz; rat; A. Tomlinson, Columbia Uni- from GMC4-2a, but from a GMC which appears later in versity). anti-Pdm1 (mouse and rabbit polyclonal, Yang et. al., development. This GMC expresses several molecular markers 1993). Anti-Zfh1 (mouse polyclonal; Z. Lai, University of Pennsyl- vania). indicative of a GMC4-2a identity, suggesting that the ltt mutations cause a duplication of GMC4-2a; however, the two Generation of β-gal-expressing ‘flipout’ clones GMCs don’t share the same parental NB, indicating that the To label individual NB lineages through the expression of TaulacZ ltt mutations cause a second NB to acquire the identity of we used the FRT/FLP system (Golic and Lindquist, 1989; Xu and NB4-2 (at least in part). In addition, the ltt mutation leads to Rubin, 1993). Using a homozygous viable third chromosome a defect in the LGB lineage resulting in the lack of correctly insertion carrying hsFLP and a second chromosome insertion specified LG. Furthermore, ltt embryos have a severe axon carrying a TaulacZ cassette (A. Marti-Subirana and R. Holmgren, defect which predominantly affects the longitudinal connec- personal communication), the stock lttEMS, hsFLP/TM3, Sb, e was tives. established; virgins from this stock were crossed to males of the genotype TaulacZ cassette/+; lttEMS/+. To irreversibly activate the expression of taulacZ, embryos were collected from the above cross in 2-hour intervals and aged for 2 hours at 25¡C. Heatshock was MATERIALS AND METHODS performed by placing embryos into a moist chamber which was submerged in a waterbath for approx. 15 minutes at 34¡C. Subse- Fly strains quently, the embryos were allowed to develop at 18¡C until stage 16, An isogenised third chromosome carrying the markers scarlet and fixed and stained with anti-β-gal (rabbit) and anti-Eve (mouse). Non- ebony was generated (st e iso3) and used for mutagenesis; a multi- heatshocked control embryos, handled in the same manner, had no marked third chromosome (ru h Ki roe p(p) cu sr e ca) Ð used for β-gal-positive clones. recombinational mapping of the ltt mutation Ð was obtained from the Bloomington Stock Center; y,w; 5953/5953 flies, carrying a lacZ Microscopy and staging of embryos insertion in huckebein were obtained from C.Q. Doe and used to Embryos were staged according to Campos-Ortega and Hartenstein generate 5953;lttEMS/TM3,Sb. (1985). During stages 10-11, staging was based on the neuroblast pattern (Doe, 1992). Embryos were observed using DIC optics with EMS mutagenesis a Zeiss Axiophot microscope. Confocal microscopy was performed EMS mutagenesis was performed according to the method of Grigli- using a BioRad MRC600 scanhead equipped with a krypton/argon atti (1986). Mutagenised males were mated en masse with laser and a Zeiss Axiophot microscope. Adobe Photoshop was used DTS/TM3,Sb virgins, and the males were removed after 5 days. Single for image processing. ltt is required for neural/glial cell identity 675

RESULTS eve expression in one RP2 proximal neuron. This additional cell is found in about 70% of the hemisegments. Occasionally, Isolation of the ltt alleles we observed two additional eve-expressing neurons in close The ltt gene was discovered in a 3rd chromosome screen for proximity to the aCC/pCC neurons. Although not changed in ethyl-methanesulfonate (EMS) induced lethal mutations which number, the CQ neurons appeared spatially disorganised. In alter the fate of identified GMCs and neurons in the embryonic contrast, the ltt mutation had no effect on the number and CNS. We used antibodies against the homeobox protein Even- position of the EL neurons. skipped (Eve) to monitor defects in its expression. The lttEMS Mature RP2 neurons are characterised by the ipsilateral allele was discovered by its abnormal eve expression pattern. extension of an axon that can be stained with mAb22C10 ltt embryos have no cuticle defects indicating that the mutation (Fujita et al., 1982). The additional Eve-positive neuron in ltt does not affect segmentation. Four additional alleles of ltt were embryos is labelled by mAb22C10 and in a few cases exhibits induced by gamma-irradition and identified by their failure to an ipsilateral axon extension (Fig. 1C,D), consistent with this complement the lttEMS allele. All alleles are embryonic lethals neuron having a RP2 identity. This conclusion was corrobo- with similar CNS defects (Table 1). rated using an antibody against an additional marker for RP2 To localise the ltt gene we obtained gamma-irradiation- identity: during late stages of the induced alleles. However, we failed to detect any cytological defect associated with any of the four gamma-induced alleles. By recombination mapping we located the ltt gene to the right of the visible marker ebony (cytological position: 093D). However, we were unable to uncover the ltt gene using simple deficiencies from the 093-100F region. To examine deficien- cies which removed haploid insufficient regions, we crossed the lttEMS line with the appropriate deficiency/duplication stocks and analysed the phenotype of the resulting defi- ciency/lttEMS progeny by anti-Eve staining. Df(3R)L127 (99B5-6;099E4-F1)/lttEMS embryos displayed a phenotype similar to that of lttEMS/lttEMS. In contrast, Df(3R)B81 (99C8; 100F5)/lttEMS embryos were wild type (wt) indicating that ltt is located within 99B5-6 to 99C8. The expressivity of the phenotype of lttEMS/Df(3R)L127 is similar to that of lttEMS/lttEMS (Table 1) indicating that lttEMS is close to being an amorphic allele. The ltt mutations cause a duplication of the RP2 motoneuron In late stages of embryonic development (stage 13 and later) eve is expressed in about 20 neurons per hemisegment. The rather small number of eve-expressing cells enabled us to screen for mutations which have effects as subtle as a fate change of a single neuron. In wt embryos eve is expressed in the RP2 neuron, the aCC and the pCC neurons, six CQ neurons and six to ten neurons making up the eve-lateral (EL) cluster (Patel et al., 1989). ltt embryos display several deviations from this pattern (Fig. 1A,B). The most obvious defect is a gain of

Table 1. Expressivity of different ltt alleles Fig. 1. ltt embryos exhibit a frequent duplication of the RP2 neuron. (A-D), Dorsal views of dissected stage 15 wt (A,C) and ltt (B,D) Percentage of hemisegments embryonic CNS stained with anti-Eve (A,B) or double stained with Alleles with duplicated RP2 Axon phenotype anti-Eve (black) and mAb22C10 (brown; C,D). Note that in ltt CNS lttEMS 68% (102/150) longitudinal connectives (B) many hemisegments contain an extra cell near the RP2 position missing (arrows indicate RP2 and duplicated neuron). Like RP2, the ltt2-x 55% (57/103) longitudinal connectives duplicated neuron can also extend an ipsilateral projection (C,D missing arrowheads). Note that in the ltt CNS some CQ and EL neurons can ltt4-D 58% (58/100) longitudinal connectives be seen in the same focal plane as RP2. This reflects a spatial missing disorganization of the ltt CNS along the dorsal/ventral axis late in ltt4-E 61% (50/80) longitudinal connectives embryogenesis. (E,F), Ventral views of stage 15 wt (E) and ltt (F) missing whole-mount embryos double stained with mouse anti-Zfh1 (with 5-A FITC-conjugated secondary antibody; green) and rabbit anti-Eve ltt 51% (51/100) longitudinal connectives (with rhodamine-conjugated secondary antibody; red). Note that RP2 missing expresses Zfh1 (E, yellow nuclei) as do both of the Eve-expressing lttEMS/DefL127 72% (21/29) n.d. cells near the RP2 position in ltt embryos (F, yellow nuclei). 676 M. Buescher and W. Chia putative transcription factor Zfh1 is highly expressed in RP2 results indicate that the two RP2 neurons are derived from two (Lai et al., 1991). Double-label experiments with anti-Eve and distinct GMCs which appear sequentially and that both GMCs anti-Zfh1 revealed that the additional neuron in ltt embryos express Pdm1, Eve and Pros, markers consistent with a GMC4- expresses Zfh1 (Fig. 1E,F). Therefore, we will refer to this cell 2a identity. as the ‘duplicated RP2 neuron’. The two RP2s do not share the same parental NB The RP2 neurons are derived from two GMCs which Since the identity of a GMC is thought to be determined by the appear sequentially parental NB and its birth order (Goodman and Doe, 1993), the The RP2 motoneuron is derived from GMC4-2a which extra GMC can only be generated by a NB that has the capacity expresses eve. Because of its isolated position within the to initiate a GMC4-2a-like fate and therefore by a NB that has pattern of eve-expressing cells, the fate of GMC4-2a can be itself at least partial NB4-2 identity. Several models are con- easily monitored (Fig. 2A-F): GMC4-2a is born dorsal to NB4- ceivable that would explain the presence of an additional 2 (late stage 10) and eve expression is detectable shortly after- GMC4-2a-like cell in ltt embryos: NB4-2 would be duplicated, wards. GMC4-2a divides to generate the RP2 neuron and its sibling cell. The sibling cell gradually extin- guishes eve expression, whereas RP2 remains Eve positive throughout embryonic development. GMC4- 2a is the only eve-expressing GMC within the NB4-2 lineage. To determine the origin of the duplicated RP2 neuron, we stained ltt embryos at different stages of development with anti-Eve (Fig. 2G-L). At early stage 11, ltt embryos are indistinguishable from wt: the first- born GMC of NB4-2 expresses eve and gives rise to two Eve-positive daughter cells. Soon after, one daughter cell gradually loses anti-Eve reactivity. However, at late stage 11 one additional Eve-positive cell appears in the vicinity of NB4-2 (Fig. 2I). At early stage 12 this cell has been replaced by two Eve- positive cells which presumably represent its daughter cells (Fig. 2J). The later fate of these cells appears to replicate the fate of RP2 and its sibling: anti-Eve reac- tivity is gradually lost in one daughter cell, whereas the other cell maintains eve expression. These results indicate that the extra RP2 in ltt embryos is produced in a manner reminiscent of that of RP2. However, it is important to note that this neuron is not formed at the same time as RP2 but appears later in development. The sequence of events as seen with anti-Eve suggests that the duplicated RP2 is derived from a GMC which appears later in development than GMC4- 2a. To confirm this result we stained ltt embryos with antibodies against the proteins Pdm1 (Dick et al. 1991; Lloyd and Sakonju, 1991; Billins et al., 1991) and Pros (Doe et al., 1991; Vaessin et al., 1991) both of which accumulate to high levels in GMC4-2a. After the division of GMC4-2a, Pdm1 and Pros proteins are rapidly lost from its daughter cells and the mature RP2 and the sibling cell stain negative. We performed Fig. 2. Origin of the duplicated RP2 in ltt embryos. Ventral views of whole- double label experiments with anti-Eve and anti-Pros mount wt (A-F) and ltt embryos (G-L) stained with anti-Eve. (A,G) Early or anti-Eve and anti-Pdm1, respectively, to distinguish stage 11, GMC4-2a; (B,H) mid-stage 11, GMC4-2a has divided into RP2 and between Eve-positive GMCs and neurons of the NB4- its sibling; (C,I) late stage 11, RP2 and its sibling (C); in ltt embryos one 2 lineage (Fig. 3). In wt and ltt embryos at early stage additional Eve-positive cell can be seen (I); (D,J) early stage 12, Eve 11, GMC4-2a is strongly positive for Pros and Pdm1 expression in the sibling cell is fading (D) but in ltt embryos two RP2s and proteins. At late stage 11 its post-mitotic progeny, RP2 two ‘fading’ siblings can be seen (J); (E,K) late stage 13; sibling cell is no and the sibling, are only weakly positive for Pros and longer Eve-positive (E) but in ltt, two RP2s and one sibling (presumably the Pdm1, due to perdurance of the proteins. At this stage, later born) are Eve-positive (K); (F,L) stage 15, one RP2 in wt and two RP2s in ltt are Eve-positive. Arrows mark RP2 and its sibling; arrowheads indicate the additional Eve-positive cell which is observed in ltt GMC4-2a and the putative extra GMC4-2a-like cell in ltt embryos. These embryos is strongly positive for Pros and Pdm1 indi- panels represent a temporal series; the age of embryos can be distinguished cating that this cell is a GMC rather than a neuron (Fig. from stages 10-13 based on the neuroblast pattern and the appearance of Eve- 3I,K). At stage 15, RP2 in wt and both RP2s in ltt positive cells. Schematic representations of the course of events in wt and mutants stain negative for Pros and Pdm1. These mutant embryos are shown. ltt is required for neural/glial cell identity 677

Fig. 3. Two GMC4-2a-like cells appear sequentially inltt embryos. (A-F) Double staining of wt (A-C) and ltt (D-F) embryos with mouse anti-Pdm1 (green) and rabbit anti-Eve (red). At early stage 11, (A,D) GMC4-2a (yellow) expresses Eve and Pdm1 (arrowhead). At late stage 11 (B,E), GMC4-2a has divided to produced the post-mitotic RP2 and its sibling both of which express Eve, but are only weakly positive for Pdm1 (arrows, B); in ltt embryos one additional cell which is Eve- and Pdm1-positive (yellow) is seen (arrowhead, E). By stage 15 (C.F), the mature RP2 is Eve-positive (red), but Pdm1- negative (arrow, C); the duplicated RP2s in ltt embryos are also Eve-positive and Pdm1-negative (arrows, F). (G-L) Double staining of wt (G-I) and ltt (J-L) embryos with mouse anti-Pros (green) and rabbit anti-Eve (red). At early stage 11, GMC4-2a (yellow) expresses Eve and Pros (arrowhead, G, J). At late stage 11, GMC4-2a has divided into RP2 and its sibling both of which express Eve but are only weakly positive for Pros (arrows, H, K); in ltt embryos one additional cell which is Eve- and Pdm1-positive (yellow) is seen (arrowhead, K); this cell, which expresses markers consistent with a GMC4-2a identity, is always lateral to the two Eve-expressing post-mitotic neurons derived from the earlier born GMC4-2a. By stage 15, the mature RP2 (red) is Eve-positive but Pros-negative (arrow, I); both RP2-like cells (red) in ltt embryos are also Eve-positive and Pros-negative (arrows, L). Arrowheads mark GMC4-2a (and GMC4-2a-like cell); arrows label RP2 and sibling. a neighbouring NB could have adopted NB4-2 identity or subset of NBs. Examination of ltt embryos at stage 11 reveals NB4-2 itself could have generated GMC4-2a twice. that the β-gal expression pattern is identical to wt (Fig. 4A,B). To distinguish between these possibilities we used the only Since there is no extra β-gal-expressing NB, NB4-2 is not marker that is currently available for a molecular identification simply duplicated in the mutant. In addition, it excludes the of NB4-2: the viable lacZ insertion line 5953. 5953 is located possibility that the neighbouring NB3-2 and NB4-1, as well as in the huckebein gene which has been shown to be necessary other NBs which do not normally express 5953lacZ, have for the proper specification of RP2 identity (Chu-LaGraff et adopted the identity of NB4-2. However, we cannot exclude al., 1995). 5953 directs β-galactosidase (β-gal) expression to a the possibility that NB4-3 has adopted a NB4-2 identity: since

Fig. 4. The duplicated RP2 neurons in ltt mutants are not derived from the same parental neuroblast. (A-B) Ventral views of whole mount stage 11 5953 (A) and ltt; 5953 (B) embryos stained with anti-β- gal. 5953 is expressed in NB 2-1, 2-3, 2-4, 4-2, 4-3, 4-4, 5-4, 5-5 and 7-3 in wt background and this expression pattern is not changed in ltt embryos. Broken lines indicate the midline, triangles mark the tracheal pits and arrows label NB4-2. (C-F), Ventral views of whole-mount embryos stained with rabbit anti-β-gal (membrane staining; brown) and mouse anti-Eve (nuclear staining; black). C is a combined view of the three progressively more dorsal focal planes shown in (D-F). The anti-β-gal-positive cells represent a Tauβ-gal-expressing clone (area delineated by dotted lines) which has been generated by a single ‘flip-out’ event. Only one RP2 neuron, showing nuclear and membrane staining, is part of this clone (arrowhead); the second ‘RP2’ (arrow) does not express Tauβ-gal (nuclear Eve staining only) indicating that it is not part of the clone. 678 M. Buescher and W. Chia

Fig. 5. ltt, klu double mutants exhibit additive affects with respect to gain of RP2 neurons. Ventral views of stage 15 klu, ltt embryos stained with anti- Eve (A) or anti-Eve (red) and anti-Zfh1 (green) (B,C). B and C represent two different focal planes of the same segment, each focal plane contain two of the four RP2s (arrowheads) present. Four RP2s can be seen in A. Asterisks indicate aCC neuron.

NB4-3 itself expresses 5953lacZ in wt embryos, a switch to a abnormalities indicating that the ltt mutation does not cause NB4-2 identity cannot be detected with this marker. wide spread cell fate changes (data not shown). To determine whether both RP2 neurons in ltt embryos share Pros is present at high levels in GMC nuclei but absent in the same parental NB we used the ‘flipout’ technique (see mature neurons. ltt embryos show no obvious defects in their Methods) which enables individual cell lineages to be marked Pros expression pattern at early stages in development (not through the expression of Tauβ-gal (Anna Marti-Subirana and shown). However, at stage 15 ltt embryos appear dramatically Robert Holmgren, personal communication). We applied con- different from wt embryos (Fig. 6). At this stage most if not ditions which resulted on average in one ‘flipout’ event per all GMCs have divided and substantial Pros expression is hemineuromere (see Methods). Subsequently, the embryos present only in the nuclei of a small number of non-neuronal were allowed to develop until stages 15-16, fixed and double- cells including six LG cells, the belt cells and some midline labeled with anti-β-gal and anti-Eve. ‘Flipout’ events in RP2- cells. ltt embryos exhibit the complete lack of Pros-positive generating NBs were indicated by RP2 neurons double positive LG. In contrast, the number of Pros-positive cells in the for β-gal and Eve (Fig. 4C-F). position of the belt cells is increased. The belt cells are located Double positive RP2 neurons were surrounded by additional at the lateral surface of the nerve chord and extend processes cells that were β-gal positive, but negative for Eve; these cells around the dorsal and ventral surface of the nerve chord. It has presumably represent the later progeny of the NB lineage. We been suggested that the belt cells function in circumferential found that the clones generated in ltt mutant embryos never condensation of the nerve chord (Doe et al., 1991): the ltt included both of the RP2 neurons. Identical results were phenotype is consistent with this hypothesis since the increase observed in three independent clones. It is unlikely that these in Pros-positive cells at the lateral surface coincides with results reflect incomplete clones that are due to ‘flipout’ events defects in the mediolateral condensation resulting in a dented in NB4-2 after the generation of GMC4-2a: we analysed clones appearance of the nerve cord (see Fig. 6B). in a different mutant (klumpfuss, klu) that also shows a dupli- cation of the RP2 neurons. In klu mutants both RP2s share the ltt mutations affect the longitudinal glioblast lineage same parental NB. Using identical experimental conditions we Two models could explain the absence of Pros-positive LG: the observed several clones which included both RP2s. We never ltt mutation may cause a loss of pros expression in otherwise observed clones in which only one RP2 was labelled by β-gal normal LG; alternatively, ltt may cause a complete loss of LG. (X. Yang, personal communication). Therefore we conclude To distinguish between these possibilities we used an antibody that the two RP2s in ltt mutant embryos are not derived from the same parental NB. Both, the ltt and the klu mutation result in the duplication of the RP2 neuron. However, the klu mutation acts within the lineage of NB4-2 (X. Yang, personal communication). If the duplication of RP2 in ltt embryos indeed arises outside of the NB4-2 lineage, as suggested by our clonal analysis, then we would expect the ltt and klu mutant phenotypes to be additive. Anti-Eve staining of ltt/klu double mutant embryos at stage 15 revealed that most hemisegments contained three Eve-positive cells; occasionally, we observed four cells (Fig. 5A). We found that the four Eve-expressing cells were also positive for Zfh1 (Fig. 5B), indicating that these cells express the combination of markers characteristic for RP2 identity. These results lend further support to our conclusion that the duplicated RP2 in ltt arises outside the NB4-2 lineage. ltt loss of function does not cause wide spread cell fate changes To address the question of whether ltt mutations cause cell fate changes in Eve-negative cells we analysed the expression patterns of the pdm1, ftz and pros at different stages in devel- Fig. 6. ltt is required for Pros-expressing longitudinal glia. Ventral opment. These genes are widely expressed within the CNS in views of stage 15 whole-mount embryos stained with mouse anti- a number of different cell types. Staining of ltt embryos with Pros. (A) wt; (B) ltt; Pros-expressing LG (arrowhead); belt cells anti-Pdm1 and anti-Ftz at stages 11-14 did not reveal any gross (arrow). ltt is required for neural/glial cell identity 679

Fig. 7. The first LGB cell division is unaffected in ltt embryos. Ventral views of wt (A-C) and ltt (D-F) embryos double stained with mouse anti-Pros (green) and rabbit anti-Repo (red). (A,D) Early stage 11; the LGB has divided into two daughter cells. In wt, but not in ltt, the more dorsal daughter cell is double-positive for Repo and Pros (yellow). (B,E) stage 12; in wt, but not in ltt six double-positive LG form a rhomboid pattern (yellow). Note that in E a slightly different focal plane was chosen because in ltt embryos no Repo-positive cells are seen in a focal plane corresponding to that of B. (C,F) Stage 15; in wt, the LG form two rows along the anterior- posterior axis (yellow); in ltt essentially no Repo-positive cells are found in the position of the LG. Asterisks indicate the LGB progeny. against the homeodomain protein Repo (Xiong et al., 1994; formation and/or maintenance of the longitudinal connectives Campell et al., 1994; Halter et al., 1995) in combination with (Jacobs et al., 1989; Hidalgo et al., 1995). Moreover, anti-Pros antibody as molecular markers for the LG. Repo expression of the transcription factor Pointed in the LG has (=RK2) is expressed in all embryonic CNS glia, with the been shown to be crucial for the induction of 22C10 expression exception of the midline glia and two of three segmental nerve in the MP2 neurons (Klaes et al., 1994). This loss of 22C10 root glia. Six LGs are derived from the LGB which forms at expression in early pointed embryos might explain the disrup- stage 10 in the lateral-most row of NBs at the anterior margin tion of the longitudinal connectives which is observed in late of each segment (Jacobs et al., 1989). Repo expression can be pointed embryos. In ltt embryos no correctly specified LG are detected in the LGB shortly before its first division. At early formed. To examine whether ltt also affects the formation of stage 11 the LGB divides along the apical/basal axis to the axonal scaffold, we stained ltt embryos with mAb22C10 generate two progeny of approximately equal size. The dorsal (stage 12) and mAbBP102 (stage 15). We observed that the cell is positive for pros whereas the ventral cell remains Pros- early expression of 22C10 in the MP2 neurons was reduced negative (Fig. 7A). Both daughter cells migrate medially and (Fig. 8A,B) and immunostaining with mAbBP102 revealed a anteriorly. During stages 11/12 the LGB progeny undergo frequent loss of the longitudinal connectives in ltt embryos further divisions that result in six Pros/Repo double positive (Fig. 8C,D). In contrast, the commissures were formed quite cells that are arrayed in a characteristic rhomboid pattern (Fig. normally. 7B). At stage 15, eight to ten Repo-positive glia form two rows on the dorsal surface of the connectives; six of these cells are also positive for Pros (Fig. 7C). DISCUSSION At early stage11 ltt embryos were indistinguishable from wt: the LGB was formed at its correct position and expressed ltt embryos have one additional Eve-positive neuron. This normal levels of Repo. The division of the LGB occurred but neuron shares many of the characteristics exhibited by the RP2 we failed to detect any pros expression in the dorsal progeny motoneuron: (1) it is found near the RP2 position; (2) it (Fig. 7D). The two LGB progeny migrated medially and ante- expresses Zfh1, 22C10 and Eve markers that are characteristic riorly, but during stage 12 we did not observe any further for mature RP2 neurons; (3) it is produced by a GMC that divisions and the characteristic rhomboid pattern was not expresses the combination of genes indicative of a GMC4-2a formed (Fig. 7E). However, we did observe a normal increase identity; (4) the sibling of the RP2-like cell undergoes the same in other Repo-positive cells. At stage 15 no Repo-positive cells fate as the RP2 sibling with respect to eve-expression; (5) were found to form a regular pattern along the anterior- mature RP2 neurons are characterised by the ipsilateral pro- posterior axis and we were not able to identify the two LGB jection of an axon. We found that the RP2-like cell in ltt progeny among other Repo-positive cells (Fig. 7F). These embryos does, at low frequency, project an ipsilateral axon, results indicate that the ltt mutation does not affect the indicating that this neuron has the same potential as RP2. Our formation and the first division of the LGB. Furthermore, the observation that this potential is realised only occasionally mutant LGB expresses the glial-specific marker Repo. could be explained by the fact that axonal pathfinding is not an However, in ltt mutants pros expression is lost in the entire intrinsic property of neurons but is dependent on external cues LGB lineage. whose presence is tightly regulated during development. The RP2-like cell in ltt embryos forms later than the RP2 neuron ltt mutants exhibit defective longitudinal and therefore axon formation may be prevented by the lack of connectives external cues. It has been postulated that LG play an important role in the We have shown here that in ltt embryos two GMCs are 680 M. Buescher and W. Chia

of a particular row of NBs into that of a different row of NBs. Loss of gooseberry (gsb) function leads to a transformation of row 5 NBs into that of row 3 NBs (Skeath et al., 1995 ). Con- versely, ectopic expression of gsb can lead to a row 3 to row 5 transformation (Zhang et al., 1994). The protein Wingless (Wg) is secreted from row 5 and NBs. Loss of wg function alters the fate of adjacent NBs in rows 4 and 6 (for review see, Doe and Technau, 1993 ). gsb and wg are segment polarity genes and act to specify NB identity along the anterior- Fig. 8. ltt embryos show defects in the longitudinal posterior axis. Genes that differentiate between the identities axon tracts as well as 22C10 of NBs in medial and lateral columns of NBs are as yet expression in the MP2 unknown. ltt is a likely candidate for a gene which functions neurons. Ventral views of wt in medial-lateral specification. (A) and ltt (B) stage 12 The ltt mutation causes a loss of pros expression in the embryos stained with LGB lineage. However, this alone cannot account for the LG mAb22C10. Dorsal views of phenotype: it has been shown previously that in pros loss-of- dissected stage 15 wt (C) and function mutants the six LG are formed (Doe et al., 1991), ltt (D) CNS stained with although these mutant LG appear spatially disorganised and mAbBP102. fail to undergo terminal differentiation (Jacobs, 1993). Nev- ertheless, in pros loss-of-function mutants, the LG do express formed which both have the identity of GMC4-2a with respect repo and can be identified unambigiously (M. Buescher, to position, expression of marker genes and their progeny cell unpublished observations). Since in ltt embryos the six LG fate. These GMCs appear sequentially and do not arise from are either not present or fail to express Repo, we conclude the same parental cell suggesting that the ltt mutations act on that the ltt mutation must cause defects in the LGB lineage the NB level. Which cell is the parental NB for the extra in addition to the loss of pros expression. Interestingly, we GMC4-2a-like cell? The extra GMC is always found in the found that in wt embryos many glial precursor cells do vicinity of NB4-2 therefore possible progenitors are NB4-2 express Pros and the ltt mutation does not abolish pros itself and any NB which is located directly adjacent to NB4-2; expression in these cells. In contrast to the LG, the post- the parental NB should express 5953 since huckebein (5953) mitotic progeny of these glial precursors are Pros-negative. function appears to be absolutely required for the generation This suggests that pros expression is regulated differently of an Eve-expressing GMC4-2a. No change in 5953lacZ within the LGB lineage and other glial lineages and that the expression was seen in the ltt background, therefore either ltt gene product is required for pros expression in the LGB NB4-2 or NB4-3 must be the progenitor cell. Our results using lineage but not in other glial lineages. the ‘flipout Tauβ-gal cassette’ indicate that the two RP2-like The lack of correctly specified LG in ltt mutants coincides neurons are not part of the same lineage and therefore the two with a reduction of 22C10 expression in early MP2 neurons ‘GMC4-2a’s must be generated by different NBs. This con- and severe defects of the longitudinal axon tracts. However, the clusion is supported by the observation that in klu/ltt mutants causal relationship between these defects is difficult to assess. 3-4 RP2-like cells are formed frequently. Taking into account The presence of correctly specified pros-expressing LG may all the available data, we postulate that NB4-3 is the most likely be an absolute requirement for axon pathfinding and loss of the candidate for the parental NB of the extra GMC4-2a (and RP2). ltt function within the LGB lineage may be sufficient to cause NB4-3 delaminates significantly later than NB4-2 (NB4-2: S2; a lack of longitudinal connectives. Alternatively, the neurons NB4-3: S5) and at a more lateral position. This is consistent whose axons contribute to the longitudinal connectives may be with both the observed time and position of appearance of the affected by the mutation and may not be able to recognise the later GMC4-2a-like cell. positional cues required for axon pathfinding. These scenarios If NB4-3 is the progenitor of the extra GMC then the loss are not mutually exclusive. of the ltt gene product enables this NB to produce a daughter In this study we have described ltt, a novel gene whose loss cell that is normally produced by NB4-2. This scenario would of function causes a duplication of the RP2 neuron and a lack predict that the identities of NB4-2 and NB4-3 are very similar, of correctly specified LG. Our results suggest that ltt function at least with respect to early events within the lineage, and that is required to restrict the number of RP2 neurons to one per ltt+ normally acts to prevent NB4-3 from producing a cell with hemisegment and to ensure that six Pros-positive LG per a GMC4-2a identity. If the ltt mutation causes NB4-3 to adopt hemisegment are formed. The strongest ltt allele causes a NB4-2 identity, the normal ltt gene product presumably acts duplication of RP2 in approx. 70% of the hemisegments but very early in the hierarchy of cell specification genes. At the LG are affected in all hemisegments. This observation present a molecular characterization of NB4-3 and its lineage suggests that the ltt function may be indispensable for the in wt and ltt embryos is not possible due to the lack of appro- formation of the LG but may be partially redundant with priate markers; furthermore, the progeny of NB4-3 are not respect to RP2 formation. Our results raise the interesting pos- known. sibility that ltt may belong to a class of genes which acts to It has been shown that NB identity is specified by positional differentiate NB identities between medial and lateral columns cues which act prior to NB delamination (for review: Doe and of NBs. Skeath, 1996). So far, two examples have been described in which the loss of a single gene product transforms the identity We thank K. Weigmann for initiating the screen and for her enthu- ltt is required for neural/glial cell identity 681 siasm and stimulating discussions; X. Yang for being an excellent Golic, K. G. and Lindquist, S. (1989). The FLP recombinase of yeast catalyses teacher of , microscopy and image processing; C. Q. Doe, site-specific recombination in the Drosophila genome. Cell 59, 499-509. M. Frasch, W. Gehring, C. S. Goodman, Z. Lai, G. Technau, A. Goodman, C. S. and Doe, C. Q. (1993). Embryonic development of the Tomlinson and K. Zinn for providing antibodies and/or flies; A. Marti- Drosophila central nervous system. In The Development of Drosophila Subirana and R. Holmgren for providing transformants of the taulacZ melanogaster (ed. Bate, M. and Martinez Arias, A.), pp. 1091-1131. Cold Spring Harbour Laboratory Press, Cold Spring Harbour, New York. cassette prior to publication; Kathy Matthews and the Bloomington Grigliatti, T. (1986). Mutagenesis. In Drosophila: a Practical Approach (ed. stock center for efficiently providing stocks. We also thank Nick D. B.. Roberts), pp. 39-58. IRL Press, Oxford, UK. Harden and members of the Chia lab for assistance and helpful dis- Halter, D. A., Urban, J., Rickert, C., Ner, S., Ito, K., Travers, A. and cussions. Technau, G. (1995). The homeobox gene repo is required for the differentiation and maintenance of glia function in the embryonic nervous system of . Development 121, 317-332. Ito, K., Urban, J., and Technau, G. M. (1995). Distribution, classification and REFERENCES development of Drosophila glial cells in the late embryonic and early larval ventral nerve cord. Roux’s Arch. Dev. Biol. 204, 284-307. Billin, A. N., Cockerill, K. A. and Poole, S. J. (1991). Isolation of a family of Jacobs, J. R. (1993). Perturbed glial scaffold formation precedes axon tract Drosophila POU domain genes expressed in early development. Mech. Dev. malformation in Drosophila mutants. J. Neurosci. 24, 611-626. 34, 75-84 Jacobs, J. R., Hiromi, Y., Patel, N. H. and Goodman, C. S. (1989). Lineage, Bossing, T. and Technau, G. (1994). The fate of midline progenitors in migration and morphogenesis of longitudinal glia in the Drosophila CNS as Drosophila as revealed by a new method for single cell labelling. revealed by a molecular lineage marker. Neuron 2, 1625-1631. Development 120, 1895-1906. Jimenez, F. and Modolell, J. (1993). Neural fate specification in Drosophila. Bossing, T., Technau, G. and Doe, C. Q. (1996a). huckebein is required for Curr. Opin. Genet. Dev. 3, 626-632. glial development and axon pathfinding in the neuroblasts 1-1 and 2-2 Klaes, A., Menne, T., Stollewerk, A., Scholz, H. and Klaembt, C. (1994). lineages in the Drosophila central nervous system. Mech. Dev. 55, 53-64. The Ets transcription factor encoded by the Drosophila gene pointed direct Bossing, T., Udolph, G., Doe, C. Q. and Technau, G. (1996b). The embryonic glial cell differentiation in the embryonic CNS. Cell 78, 149-160. CNS lineages of Drosophila melanogaster: I. Neuroblast lineages derived Lai, Z., Fortini, M. E. and Rubin, G. M. (1991). The embryonic expression from the ventral half of the neuroectoderm. Dev. Biol. 179, 41-64. pattern of zfh1 and zfh2, two Drosophila genes encoding novel zinc-finger Broadus, J., Skeath, J. B., Spana, E. P., Bossing, T., Technau, G. and Doe, C. homeodomain proteins. Mech. Dev. 34, 123-134. Q. (1995). New neuroblast markers and the origin of the aCC and pCC Lloyd, A. and Sakonju, S. (1991). Characterization of two Drosophila POU neurons in the Drosophila central nervous system. Mech. Dev. 53, 393-402. domain genes, related to oct1 and oct2, and the regulation of their expression Chu-LaGraff, Q. and Doe, C. Q. (1993). Neuroblast specification and pattern. Mech. Dev. 36, 87-102. formation regulated by wingless in the Drosophila CNS. Science 261, 1594- Patel, N. H., Schafer, B. Goodman, C. S. and Holmgren, R. (1989). The role 1597. of segment polarity genes during Drosophila . Genes Dev. 3, Chu-LaGraff, Q., Schmid, A., Leidel, J., Broenner, G., Jaeckle, H. and Doe, 890-904. C. Q. (1995). huckebein specifies aspects of CNS precursor identity required Skeath, J. B., Zhang, Y., Holmgren, R., Carrol, S. B. and Doe, C. Q. (1995). for motoneuron axon pathfinding. Neuron 15, 1041-1051. Specification of neuroblast identity in the Drosophila embryonic central Campell, G., Goering, H., Lin, T., Spana, E., Andersson, S., Doe, C. Q. and nervous system by gooseberry-distal. Nature 376, 427-430. Tomlinson, A. (1994). RK2, a glial-specific homeodomain protein required Udolph, G., Prokop, A., Bossing, T. and Technau, G. M. (1993). A common for embryonic nerve chord condensation and viability in Drosophila. precursor for glia and neurons in the embryonic CNS of Drosophila gives rise Development 120, 2957-2966. to segment specific lineage variants. Development 118, 765-775. Campos-Ortega, J. A. and Hartenstein, V. (1985). The embryonic Vaessin, H., Greel, E., Wolff, E., Bier, E., Jan, L. Y. and Jan, Y. N. (1991). development of Drosophila melanogaster. Springer Verlag, Berlin, Germany. prospero is expressed in neuronal precursors and encodes a nuclear protein Dick, T., Yang, X., Yeo, S. and Chia, W. (1991). Two closely linked that is involved in the control of axonal outgrowth in Drosophila. Cell 67, Drosophila POU domain genes are expressed in neuroblasts and sensory 941-953. elements. Proc. Natl. Acad. Sci. USA 88, 7645-7649. Xiong, W. C., Okano, H., Patel, N. H., Blendy, J. A. and Montell, C. (1994). Doe, C. Q., Hiromi, Y., Gehring, W. J. and Goodman, C. S. (1988a). repo encodes a glial-specific homeodomain protein required in the Expression and function of the segemtation gene fushi tarazu during Drosophila nervous system. Genes Dev. 8, 981-994. Drosophila neurogenesis. Science 239, 170-175. Yang, X., Yeo, S., Dick, T. and Chia, W. (1993). The role of a Drosophila POU Doe, C. Q., Smouse, D. and Goodman, C. S. (1988b). Control of neuronal fate homeodomain gene in the specification of neural precursor cell identity in the by the Drosophila segmentation gene even-skipped. Nature 333, 376-378. developing embryonic central nervous system. Genes Dev. 7, 504-516. Doe, C. Q., Chu-LaGraff, Q., Wright, D. M. and Scott, M. P. (1991). The Yeo, S., Lloyd, A., Kozak, K., Dinh, A., Dick, T., Yang, X., Sakonju, S. and prospero gene specifies cell fate in the Drosophila central nervous system. Chia, W. (1995). On the functional overlap between two Drosophila POU Cell 65, 451-465. homeodomain genes and the cell fate specification od a CNS neural Doe, C. Q. (1992). Molecular markers for identified neuroblasts and ganglion precursor. Genes Dev. 9, 1223-1236. mother cell in the Drosophila embryonic central nervous system. Xu, T. and Rubin, G. (1993) Analysis of genetic mosaics in developing and Development 116, 855-863. adult Drosophila tissues. Development 117, 1223-1237. Doe, C. Q. and Technau, G. (1993). Identification and cell lineage of Zhang,Y., Ungar, A., Fresquez, C. and Holmgren, R. (1994). Ectopic individual neuroblasts in the Drosophila CNS. Trends Neurosci 16, 510-514. expression of either gooseberry-distal or proximal gene causes alterations of Doe, C. Q. and Skeath, J. B. (1996). Neurogenesis in the insect nervous cell fate in the epidermis and central nervous system. Development 120, system. Curr. Opin. Neurobiol. 6, 18-24. 1151-1161. Fujita, S. C., Zipursky, S. L., Benzer, S., Ferrus, A. and Shotwell, S. L. (1982). Monoclonal antibodies against the Drosophila nervous system. Proc. Natl. Acad. Sci. USA 79, 7929-7933. (Accepted 31 October 1996)