Mastermind Mutations Generate a Unique Constellation of Midline Cells within the Drosophila CNS

Background The Notch pathway functions repeatedly during the development of the central nervous system in metazoan organisms to control cell fate and regulate cell proliferation and asymmetric cell divisions. Within the Drosophila midline cell lineage, which bisects the two symmetrical halves of the central nervous system, Notch is required for initial cell specification and subsequent differentiation of many midline lineages. Methodology/Principal Findings Here, we provide the first description of the role of the Notch co-factor, mastermind, in the central nervous system midline of Drosophila. Overall, zygotic mastermind mutations cause an increase in midline cell number and decrease in midline cell diversity. Compared to mutations in other components of the Notch signaling pathway, such as Notch itself and Delta, zygotic mutations in mastermind cause the production of a unique constellation of midline cell types. The major difference is that midline glia form normally in zygotic mastermind mutants, but not in Notch and Delta mutants. Moreover, during late embryogenesis, extra anterior midline glia survive in zygotic mastermind mutants compared to wild type embryos. Conclusions/Significance This is an example of a mutation in a signaling pathway cofactor producing a distinct central nervous system phenotype compared to mutations in major components of the pathway.


Introduction
The central nervous system (CNS) of metazoan organisms consists of many different types of neurons and glia generated through the combinatorial action of intrinsic transcription factors and extrinsic signaling inputs from neighboring cells [1][2][3]. During CNS development and in a number of developmental contexts, the Notch pathway functions as a prominent signaling system providing positional input between cells in direct contact with one another [4,5]. Previously, several roles for Notch during the development of specific cell lineages within the CNS midline of Drosophila melanogaster embryos have been described [6]. Here, we characterize functions of the co-activator, mastermind (mam) during the development of midline lineages.
One of the most surprising findings from comparative developmental biology is the extensive conservation of signaling pathways both within multiple tissues of a given organism as well as within the same tissue across diverse organisms. The Notch signaling pathway is a salient example and is used repeatedly to construct tissues during development and maintain homeostasis in adults [4,[7][8][9]. Notch signaling occurs between contacting cells when the Notch protein, a transmembrane receptor on the surface of one cell, binds one of its ligands, Delta (Dl) or Serrate/Jagged, on an adjacent cell. After binding one of these ligands, the Notch receptor is cleaved and its intracellular domain (NICD) transported to the nucleus where it interacts with the DNA-binding protein CSL (CBF1 in mammals, Suppressor of hairless (Su(H)) in Drosophila, and LAG-1 in C. elegans; hereafter referred to as Su(H); [10]). In cells devoid of Notch signaling, Su(H) functions as a repressor; whereas, in cells containing activated Notch, the NICD binds to both Su(H) and the co-activator Mam, resulting in a complex that activates transcription of target genes [11][12][13][14]. A striking example of the pleiotropic effects of Notch on a cell lineage can be found during CNS midline cell development in fruit flies [6]. In that study, Dl mutants were used to show that Notch promotes formation of midline glia and several midline neurons, while inhibiting the formation of other midline neurons.
The CNS is located on the ventral side of the Drosophila embryo and consists of a repeated unit found within all thoracic and abdominal segments. Midline cells of Drosophila are located in the center of the embryonic CNS ( Figure 1A) and they signal to and organize axons in a manner analogous to floor plate cells within the spinal cord of vertebrates, using similar signaling molecules [15,16]. Because of its simplicity, the fly midline is used to study axon guidance as well as transcription factors and signaling pathways involved in nervous system development [17][18][19]. Previous studies indicate the initial specification of Drosophila midline cells depends on expression of single-minded (sim), the master regulator of this lineage [20][21][22][23]. Activation of sim in the cells that will give rise to the midline is directly controlled by dorsal/ventral patterning genes such as Dorsal, Twist and Snail, together with Notch signaling [24][25][26]. In subsequent stages (8)(9), segment polarity genes such as engrailed (en), wingless and hedgehog determine midline cell fates by separating the midline progenitor cells into anterior and posterior compartments [18,27]. By the end of embryogenesis, the mature Drosophila midline consists of a small number of glia and neurons per segment ( Figure 1A, C and D): approximately 3 anterior midline glial cells (AMG), 2 midline precursor 1 (MP1) neurons, 2 MP3 interneurons (the H cell and H cell sib), 3 ventral unpaired median interneurons (iVUMs), 3 ventral unpaired median motorneurons (mVUMs), and approximately 5-8 interneurons and motorneurons derived from the median neuroblast (MNB) [17,28,29]. Posterior midline glia arise transiently, but die by the end of embryogenesis [30,31]. In summary, midline cells provide a tractable system for understanding how CNS neurons and glia are generated during embryogenesis.
Here, we provide the first study of mam functions in the various CNS midline lineages of Drosophila. The results indicate that both anterior and posterior midline glia (AMG and PMG) appear to form normally in mam mutant embryos, in contrast to midline glia in Notch and Dl mutants, which are completely absent. The presence of midline glia in mam mutants allows us to follow their development in late embryogenesis, when zygotic mam mutants cause an increase in the number of midline glia that survive in the mature CNS. In addition, mam and Notch mutants differ in the composition of MP1 neurons, whereas the other midline neural phenotypes observed in mam mutants are also observed in Notch and Dl mutants [6]. Further comparisons of Notch and mam mutants indicate that differences in the expression of the midline gene, sim, contribute to the observed difference in midline phenotypes. Taken together, the results demonstrate that zygotic mutations in the mam co-factor result in a midline cellular composition distinct from zygotic Notch mutations.

Results
Mam was identified in a screen for genes that function in midline development To identify genes involved in Drosophila midline development, we used EMS to introduce mutations throughout the genome of the fly and then examined midline cells using a reporter gene combination that drives GFP expression in all midline cells (UAS-GFP sim-GAL4). In this way, GFP could be visualized and followed in live embryos during late embryonic and larval development; stages that are difficult to examine using routine immunostaining techniques. 1037 lines carrying lethal mutations on the second chromosome were established and embryos from each line were collected and examined for midline cell defects (Figure 2A). Of the 1037 lethal lines screened, 21 showed midline defects based on the UAS-GFP sim-GAL4 reporter. These mutations were mapped within the genome using complementation; first with deficiency lines and then with fly lines containing mutations in single genes. In this report, we focus on one of the mutations that disrupted midline development and mapped to the mam locus [32][33][34]. Mam encodes the transcriptional co-activator of canonical Notch signaling [12] and is a glutamine-rich nuclear protein with a predicted 1596 amino acid sequence [35]. The protein contains a highly conserved basic domain within the N-terminus that binds to both the NICD and Su(H); and 3 glycine-valine (GV) runs and 2 acidic clusters in the C-terminal region needed for 1) interactions with p300 and RNA polymerase and 2) stability of the NICD/ Mam/Su(H) complex ( Figure 2B; [32,[36][37][38][39]). Sequence analysis of the mam DC allele isolated in our screen predicts it encodes a truncated protein lacking both the C-terminal acid cluster and the GV runs ( Figure 2C) and our phenotypic analysis indicates it behaves as a strong loss of function mutation (see below). The midline of mam DC mutant embryos was disorganized and less compact than the midline of wild type embryos during late embryonic stages ( Figure 2D and E). Numerous studies have described mam functions in CNS development [40][41][42], yet its role in midline development has not been reported. This, the midline phenotype of mam DC mutant embryos and the previously characterized roles of Notch signaling during midline cell development, led us to investigate how various midline lineages were affected in mam DC mutant embryos.
AMG and PMG are present in mam DC , but not N 55e11 mutant embryos Previous lineage analysis suggested midline glial precursors undergo multiple divisions to give rise to 2 populations of midline glia at late stages [6,30,43]. At stage 13, each segment contains about 6 AMG derived from the anterior compartment of the segment that express runt but not en; and 4 PMG cells, derived from the posterior compartment that express en but not runt ( Figure 1B). Later, at stage 16, only 3 AMG survive to enwrap the axon commissures, while all of the PMG and remaining AMG are depleted by apoptosis [43][44][45]. Both AMG and PMG are missing in Dl 3 mutants, suggesting the Notch pathway is required for development of both glial lineages [6]. To examine midline glial development in mam DC mutant embryos, we monitored Wrapper, an immunoglobulin protein required for midline glial survival, and expressed almost exclusively in the midline glia, at a high level in AMG and a lower level in PMG [43,46]. The development of AMG can be followed using the co-localization of Wrapper and Runt, while the PMG can be identified using co-localization of Wrapper and En ( Figure 3A-P). During mid and late embryogenesis, Wrapper protein was never detected in the midline of N 55e11 homozygous embryos, a null allele of Notch ( Figure 4F and S1A), but present at high levels in the AMG and at lower levels in the PMG of wild type and mam DC mutant embryos ( Figure 3A-P). At stage 13 (mid embryogenesis), both wild type and mam DC mutants contained 6 AMG per segment ( Figure 3A and I; Table 1). Wild type embryos contained 4 PMG, whereas mam DC mutants contained about 3 per segment (P = 0.0001; Figure 3A and I; Table S2). By stage 16 (late embryogenesis), wild type embryos contained just 3 AMG ( Figure 3D and L; Table 1), whereas mam DC mutants contained approximately 5 AMG (P = 0.0001; Table 1; Figure 3H and P). The PMG were not detectable at stage 16 in wild type or mam DC mutant embryos (Table S2). In addition, midline segmental compartments were less clearly defined and in many cases, glial processes extended into the posterior compartment in mam DC mutants ( Figure 3E-H and M-P) compared to wild type embryos ( Figure 3A-D and I-L). These results show that mam DC mutants, in marked contrast to N 55e11 mutants, contained AMG and PMG and that additional AMG survived during late embryogenesis in mam DC mutants compared to wild type embryos.

Embryos containing mam deletions also contain AMG
Because the mam DC mutation introduces a premature stop codon, the N-terminus of the resultant protein is still present and may be able to interact with the NICD and Su(H) to form an activation complex [47]. If so, the mam DC allele may retain some function and act as either a hypomorph or dominant negative allele. To test this, we examined midline phenotypes of embryos homozygous for a characterized point mutation in mam (mam 8 ; [34,48]) as well as several chromosome deletions that lack all or part of the mam gene: Df(2R)BSC383, Df(2R)50C-38, and Df(2R)BSC18 ( Figure S2A). Midline glia were clearly present in homozygous mam DC ( Figure 4B; see also Figure 3) and mam 8 mutants ( Figure 4C), as well as mam 8 /mam DC transheterozygotes ( Figure 4D and E). The N-terminal region of the mam protein is absent in Df(2R)BSC383 and the entire mam gene is deleted in Df(2R)BSC18 and Df(2R)50C-38 ( Figure S2A). In homozygous mam deficiency mutants, Wrapper protein was also clearly detectable ( Figure S1C-E), indicating the presence of AMG. The CNS midline in homozygous Df(2R)50C-38 and Df(2R)BSC383 embryos appeared more disorganized than in homozygous Df(2R)BSC18 embryos (Figure S1B-E), possibly due to additional genes missing in these larger deletions. The results indicate that the midline glia were present in all homozygous point and deficiency embryos tested, similar to results obtained with mam DC mutants ( Figure 3E-H and M-P), but different from those obtained with N 55e11 mutants which lack midline glia ( Figure 4F and S1). These results suggest that the mam DC mutant behaves as a strong loss of function allele and that midline glia do form in embryos completely lacking zygotic mam activity.
Additional AMG survive in mam mutant embryos As described above, analysis of mam mutants indicated they contained additional AMG during late embryonic stages. To further investigate the AMG in mam DC mutant embryos, we investigated the interaction between mam and the EGFR signaling pathway, which is known to affect AMG survival. For these cells to survive, they must receive Spitz (Spi) from lateral CNS axons that cross the midline [44]. In AMG that die, the Head Involution Defective (HID) protein is active and stimulates apoptosis, whereas in surviving AMG, cell surface EGFR binds to Spi, leading to HID phosphorylation. Phosphorylated HID is inactive, and therefore, Spi-activated glia survive. Because Notch and EGFR signaling act antagonistically in many tissues [49][50][51], we wanted to determine their relationship in AMG. However, this is not possible in Notch mutants because they lack midline glia. Instead, we investigated interactions between mam and EGFR in AMG by overexpressing the secreted form of Spi in the midline of mam DC mutant embryos. As described above, we found approximately 6 AMG per segment in both wild type and mam DC mutant embryos during mid embryogenesis, using the co-localization of Sim and Runt (Table 1; Figure 5A and B). Embryos overexpressing spi had a significant increase in AMG (P = 0.001; Table 1 and Figure 5C) to 8 per segment at stage 13. During this stage, embryos overexpressing spi in a mam DC mutant background could not be distinguished from embryos expressing spi in a wild type background or wild type embryos (Table 1 and Figure 5D). By late embryogenesis, the number of AMG in wild type embryos decreased to approximately 3 per segment as previously reported ( [6]; Figure 5E). Interestingly, all 3 classes: 1) mam DC mutants, 2) embryos overexpressing spi, and 3) embryos overexpressing spi in a mam DC mutant background each had around 5 AMG per segment and each class was significantly different from wild type embryos (Table 1 and Figure 5F-H). This, together with the known neurogenic nature of mam mutations [42], suggested midline glia may be exposed to additional spi provided by the extra neurons generated in mam DC mutants. To investigate this, we compared the interaction of the midline glia with lateral axons in wild type and mam DC mutant embryos using Wrapper and the BP102 monoclonal antibody ( Figure S1D-I). The results indicate that the additional AMG present in these embryos do enwrap lateral axons and have increased glial processes that stain with the wrapper antibody (see also Figure 3M-P). Moreover, the nerve cord does not retract normally in mam DC mutant embryos (data not shown), which may also be a consequence of extra neural tissue present in these embryos. These results suggest the greater number of neurons generated in mam DC mutant embryos may provide excess spi that allows additional AMG to survive.

Notch activation expands expression of a Wrapper reporter
Results described above as well as previous studies [6] suggest Notch signaling promotes AMG and PMG development, which are completely absent in N 55e11 zygotic mutants. Because the AMG developed normally in mam DC zygotic mutants, we next compared the effect of overexpressing mam to the overexpression of other Notch signaling components. For these experiments, we examined both the presence of AMG using a Runt antibody, as well as the regulation of gene expression within AMG using a wrapper reporter gene. The reporter contains an 884 bp wrapper enhancer sufficient to drive expression of the GFP reporter gene in midline glia ( Figure 6A and F; [52]). Expressing a constitutively active form of Su(H), UAS-Su(H).VP16 [53], in all midline cells using sim-GAL4, causes a three-fold increase of midline glial cells at the expense of midline neurons [6]. Expression of the wrapper transcriptional reporter was greatly expanded when either the NICD ( Figure 6B) or Su(H).VP16 ( Figure 6C) was overexpressed in the midline using the sim-GAL4 driver. Co-localization with Runt indicated the  [44,45]. By stage 16, the PMG are absent in both (D and L) wild type and (H and P) mam DC mutant embryos, whereas wild type embryos contain 3 AMG and mam DC mutants contain about 5 (Table 1). Images are projections of multiple focal planes and cells were counted using stacks of all focal planes. doi:10.1371/journal.pone.0026197.g003 expansion was due to the formation of additional AMG expressing the reporter compared to wild type embryos (Table 1; Figure 6A). Likewise, significantly more AMG survived until stage 16 in the NICD and Su(H).VP16 overexpression embryos compared to wild type embryos (Table 1; Figure 6F, G and H) as previously reported [6]. Therefore, over activation of the Notch pathway in the midline led to an increase in the number of AMG as well as activation of the wrapper reporter in the additional cells.
In contrast, UAS-mam sim-GAL4 embryos at both embryonic stages 13 and 16 appear normal and showed no increase in AMG at stage 16 ( Figure 6D and I). Finally, embryos in which the NICD was overexpressed in the midline of mam DC mutant embryos also contained extra AMG ( Figure 6J), similar to embryos overexpressing the NICD in a wild type background ( Figure 6G). These results suggest AMG can form in the absence of zygotic mam function.

AMG do not form in mam DC germline clones
Midline glia may form in zygotic mam DC mutant embryos because maternal mam transcripts are stable and produce sufficient Mam protein to function during Notch signaling when glia differentiate. To determine if AMG can form in embryos lacking maternal mam transcripts, we generated mam DC germline clones using the FRT, hsFLP system [54] and examined wrapper expression. Both Notch [55] and mam [32] are maternally deposited and germline clones of either gene exhibit a strong neurogenic phenotype [42]. We observed variable phenotypes in mam DC germline clones and many embryos had gross developmental defects. Most embryos did not express wrapper, although some did express this gene at low and variable levels and often in only limited regions of the embryo ( Figure S1B and C). Embryos containing either one or no copies of mam had the same phenotypes, suggesting that it was the maternal and not zygotic mam activity that caused the reduction in wrapper expression.
Because zygotic mam DC mutants expressed wrapper at high levels, while mam DC germline clones did not, we compared midline development in embryos lacking either maternal or zygotic mam at earlier developmental stages. For these experiments, we examined sim expression, which is first activated at the blastoderm stage in the mesectoderm. Mesectodermal cells are located between the mesoderm and ectoderm on both sides of the embryo ( Figure 7A) and Notch is needed in these cells for initial sim activation [22,26,56]. We determined if mam functions together with Notch to activate sim by examining mam DC germline clones. Wild type embryos express sim in the mesectoderm throughout the length of the embryo ( Figure 7A) at the blastoderm stage. In contrast, most embryos derived from homozygous mam DC mutant mothers contained gaps in sim expression, and many embryos expressed sim in only a few cells ( Figure 7B and C). The observed variation in sim expression is similar to that observed in embryos derived from Notch germline clones [26,56]. As development progresses, the mesoderm invaginates at gastrulation and mesectodermal cells move toward and meet at the ventral midline. After this, sim was expressed at high levels in both midline and muscle precursors of wild type embryos ( Figure 7D), whereas sim expression was low or undetectable in the midline, and expanded in muscle precursor cells of embryos derived from mam DC germline clones ( Figure 7E and F). These results indicate that maternal mam, similar to maternal Notch, is required to activate sim during early Drosophila development.
sim maintenance is disrupted in N 55e11 , but not mam DC zygotic mutants Because germline clones of either mam or Notch lack sim expression early in development, midline cells do not develop [22] and the various midline lineages cannot be examined in these embryos. Therefore, to examine zygotic roles for mam and Notch on sim expression, we used our mam DC allele and the N 55e11 allele. We first determined if early sim activation was affected in mam DC zygotic mutants produced by wild type mothers (mam DC heterozygotes) and compared the results to zygotic N 55e11 mutants. Sim The number of AMG found in a single CNS segment of wild type, mam DC and expression is normal until stage 10 in zygotic mam DC mutants ( Figure 7H) and persists in subsequent stages, although at a reduced level ( Figure 7K). Sim expression in the midline of N 55e11 mutant embryos also appeared normal at stage 10 ( Figure 7I), but completely disappeared by stage 13 ( Figure 7L). These results indicate that, unlike maternal mutations in mam DC and N 55e11 , zygotic mutations in these genes do not affect early sim expression prior to stage 10 and can therefore, be used to study their functions  during subsequent stages of midline development. Moreover, the results indicate maternal mam DC and N 55e11 mutations have similar effects on sim expression during early development, whereas sim expression is maintained in zygotic mam DC , but not N 55e11 mutants during mid and late embryogenesis.

The formation of certain midline neurons requires both Notch and mam
Next, we examined the effects of the mam DC mutation on the development of midline neurons. During embryonic stage 11, midline precursors (MPs) delaminate and divide to produce 6 neuronal subtypes [6,28]. The MPs (1-6) are named based on their anteroposterior position within the segments of the CNS and each midline neural cell type ( Figure 1B-D) expresses a unique gene combination that can be used to follow them during development [31]. We selected tractable markers for the various midline lineages to examine their fate in mam DC mutants.
We first examined the MP1 neural lineage, located within the anterior most region of each midline segment, using an Oddskipped (Odd) antibody [31]. Odd labels 2 MP1 cells and 2 nearby MP2 cells in each CNS segment of wild type embryos ( Figure 8A). To distinguish the MP1 and MP2 neurons, we utilized the UAS-GFP sim-GAL4 reporter that labels MP1, but not MP2 neurons. Mam DC mutant embryos also contained 2 MP1 neurons per  (Table S1). Previous studies demonstrated that Notch mutants contain 2 additional Odd-positive MP2 cells per segment and this was also true in mam DC mutants ( Figure 8E). These results indicate mam DC mutants resemble wild type embryos and differ  Table S1. doi:10.1371/journal.pone.0026197.g008 from Notch mutants in the number of MP1 neurons that form during embryogenesis.
Next, we examined the MP3 lineage which is located just posterior to MP1s within each segment and normally divides asymmetrically to produce 1 H cell ( Figure 8G) and 1 H cell sib neuron ( Figure 8J) in wild type embryos [6]. In mam DC mutant embryos, the H cell sib was not detected as assessed by CG13565 expression (Figure 8K), while approximately 6 H cells that expressed tyrosine hydroxylase (TH) were found in each segment (Table S1; Figure 8H). This was similar to N 55e11 mutant embryos in which the H cell sib was absent ( Figure 3L) and the number of H cells in each segment increased to 10 (Table S1; Figure 8I). These results indicate both mam DC and N 55e11 have similar functions in the asymmetric cell division of the MP3 midline lineage and are needed for the formation of the H cell sib. Moreover, the zygotic N 55e11 mutation had a significantly larger effect on the number of H cells that formed compared to the mam DC mutation (P = 0.0001; Table S1).
Next, we examined lineages derived from MP4-6 found within the posterior of each segment. Each of these divide asymmetrically once to produce an iVUM and a mVUM, resulting in 3 of each per segment (see [6] and Table S1). The number of mVUMs increased from 3 cells per segment in wild type embryos ( Figure 8M) to 11 in mam DC mutants as assessed with Tyramine b hydroxylase (Tbh), a specific marker for these midline cells (Table S1; Figure 8N). In N 55e11 mutant embryos, the number of mVUMs also increased to 11 per segment (Table S1; Figure 8O). To follow the iVUMs, which are also derived from MP4-6, as well as the MNB and its progeny, we assayed midline cells for the presence of En which is normally expressed in these midline neural lineages, as well as the PMG (see below). En was undetectable in the midline of N 55e11 mutants after stage 10 (data not shown), suggesting the iVUMs and the MNB and its progeny were absent. En protein levels appear relatively normal in mam DC mutants ( Figure 9E and F) compared to wild type embryos ( Figure 9A and B) until mid embryogenesis. During later developmental stages, each midline segment of wild type embryos contains 3 iVUMs and the progeny of the MNB, which divides multiple times after stage 11 to generate approximately 5-8 GABAergic neurons during embryogenesis [6]. However, only PMG express en in stage 13 mam DC mutants ( Figure 9G) and eventually, these cells also disappear ( Figure 9H; also see Figure 3), as they do in wild type embryos ( Figure 9D). Moreover, all midline cells within mam DC mutant embryos remain at the dorsal side of the nerve cord ( Figure 9G and H), which was also previously observed in Notch ts mutants [56]. The results suggest that mam, like Notch, is needed for the production of iVUMs during the asymmetrical cell divisions of MPs 4, 5 and 6 as well as for the development of the MNB and its progeny. In summary, midline neural phenotypes in mam DC mutant embryos are, in some cases, less severe, but consistent with midline phenotypes previously observed in Dl 3 mutants [6] and N 55e11 mutants (Table S1), with the exception of the MP1 neurons. The MP1 neurons appear unaffected in mam DC mutants, while N 55e11 mutants contain additional MP1s. Taken together, these studies of mam DC and N 55e11 mutants, together with previous experiments with Dl 3 mutants [6], indicate zygotic mutations in all 3 genes produce similar midline phenotypes of most neural subtypes. However, midline glia are eliminated and MP1 neurons expanded in Notch and Dl mutants, but not in mam DC mutants.

Discussion
Notch has been shown to play multiple developmental roles in the CNS of several organisms [4,[7][8][9]. The Drosophila midline, with its easy to identify neural and glial lineages, has provided examples of multiple and reiterative roles of the Notch pathway within a single CNS lineage [6]. Here, the characterization of mam DC mutants indicates how a co-factor within a signaling pathway contributes to the development of different midline cell types and adds to our understanding of Notch signaling complexity.
Initial activation of sim in the mesectoderm depends on maternal Notch expression [26,56,57], as N 55e11 germline clones lack most sim expression and therefore, contain few midline cells. Likewise, mam DC germline clones also show a reduction in sim expression. Thus, maternal contributions of both mam and Notch appear to act in the same pathway to activate sim early in development. Similarly, many midline neural phenotypes in zygotic mam DC mutant embryos are largely consistent with those of N 55e11 and Dl 3 [6], suggesting mam and Notch act together during the development of these neurons. Notch is required for formation of neurons expressing en [6] and may be needed to maintain en expression in midline cells that develop in the posterior compartment of each CNS segment, as first suggested by Bossing and Brand [27]. The results described here suggest mam is also required for the formation of the midline neurons that express en and develop into the iVUMs, the MNB and its progeny ( Figure 10). While these cells of the posterior compartment were absent, the H cell and mVUM midline neurons were expanded in mam DC mutants (Figure 10), similar to N 55e11 and Dl 3 mutants, suggesting that mam function is needed within the Notch signaling pathway to obtain the variety of midline neurons found in wild type embryos [6].
The major difference we observed between zygotic mam DC and N 55e11 mutants was the presence of midline glia in mam DC , but not N 55e11 mutant embryos during mid to late embryogenesis ( Figure 10). Not only were AMG present, but additional AMG survived in the mature CNS midline in mam DC mutants compared to wild type embryos (and N 55e11 mutants). The presence of AMG in mam DC mutants suggests either 1) the mam DC mutation is hypomorphic, 2) mam is not required within the Notch pathway for midline glial differentiation or 3) maternally deposited mam transcripts are stable and functional during the Notch signaling event needed for midline glial formation. Results with mam deficiency embryos indicated that midline glia formed and persisted in the complete absence of zygotic mam activity, suggesting it is not the hypomorphic nature of the mam DC allele that allows the midline glia to form. Currently, we cannot distinguish between the other two possibilities, although we favor the last hypothesis due to the timing of midline cell divisions. At gastrulation, each segment contains 8 mesectodermal cells, which each divide, resulting in 16 MPs per segment at stage 10. Cells that give rise to AMG and PMG do not divide again, whereas MPs that develop into neurons each divide once at stage 11. Because MPs that give rise to glia undergo their last division earlier than MPs that give rise to neurons, the Notch signaling event needed for midline glial differentiation may occur prior to Notch events that dictate midline neural fates at stage 11. Maternal Mam protein may linger just long enough to allow midline glia to form, but not long enough to function when MPs divide to give rise to midline neurons slightly later. We think this is the reason N 55e11 mutants contain more midline cells per segment than wild type (and mam DC ; Table 2). In N 55e11 mutants, MPs that would normally form glia and not divide, instead take on neural fates and do divide. Our data are consistent with this hypothesis, but future, additional experiments are required to properly test it.
In addition to this temporal sensitivity, mam may also be sensitive to spatially restricted events within the midline. Existing evidence suggests the 16 MPs fall into 3 equivalence groups at stage 10: the MP1s, MP3s and MP4s [6]. MP1s are in the anterior, MP3s in the middle and MP4s in the posterior of each CNS segment and effects of mam DC vary according to these positions. The results indicate that neurons derived from the anterior MP1s are sensitive to N 55e11 , but not mam DC ; the middle MP3s are more sensitive to N 55e11 than mam DC ; while the posterior MP4s are equally sensitive to N 55e11 and mam DC . In other words, mam DC mutants 1) differ with N 55e11 mutants in neurons derived from MP1s (MP1 neurons), 2) have similar, less severe effects compared to N 55e11 mutants in cells derived from the MP3s (the H cell and H cell sib) and 3) the same effects as N 55e11 mutants in cells derived from the posterior MP4s (mVUMs, iVUMS and MNB). These differences may be due to region specific differences in expression of other midline regulators that combine with Notch and/or Mam to control cell fate specification during embryogenesis [58]. Possible candidates include hedgehog and wingless, which are expressed in the midline, affect cell fate [27] and both interact with mam in a Notch-independent manner in other tissues [48,59,60]. In any case, clear differences in zygotic mam and Notch mutations within the midline exist and demonstrate that variations in different Notch signaling components can alter the cellular composition of the CNS in unique ways.
Close examination of mam DC and N 55e11 mutants during mid embryogenesis indicates they also differ in sim expression. After stage 10, sim diminishes in N 55e11 mutants, but persists in mam DC mutants. Likewise, midline glia, which are known to require sim expression to differentiate, do not develop in N 55e11 mutants, but do develop in mam DC mutants. Our data indicate that all midline lineages that normally express sim are absent in N 55e11 mutants, while midline lineages that do not normally express sim are present and expanded in zygotic mutants of N 55e11 (Table 2). Therefore, similar to the initiation of sim expression early, the maintenance of sim expression at this later time also appears to require zygotic Notch activity. In contrast, the results suggest sim expression persists in zygotic mam DC mutants.
In the canonical Notch pathway, Mam normally functions as a co-factor and collaborates with both the NICD and Su(H) to activate target genes. Consistent with this role, overexpression of mam alone does not affect the number of AMG generated at mid embryogenesis, whereas the overexpression of the NICD in wild type embryos increases AMG cell number [6]. Overexpression of the NICD in a mam DC mutant background still increased the number of AMG during this stage, further supporting the idea that zygotic mam is not needed at this time. During late embryogenesis, mam DC mutants contained extra AMG. Mutations in mam are known to promote neural tissue at the expense of ectoderm and this may result in the production of additional Spi, which inhibits apoptosis and allows extra midline glia to survive.
Altogether, the data suggest a high level of complexity in the regulation of CNS target genes of Notch. Notch likely interacts with additional cell-lineage specific co-activators other than, or in addition to, Mam in certain cells. In this way, combinatorial interactions between components of Notch signaling and other signaling pathways can lead to different outputs in various cell types, increasing cell diversity and function. The results described here indicate mam DC mutants contain AMG and PMG, whereas N 55e11 mutants do not. While this report describes major disruptions in mam, less severe mutations, such as small deletions, insertions or polymorphisms could also affect the midline and modify its cellular composition. Because mam mutations have more subtle effects on the midline compared to mutations in Notch or Delta, they may be tolerated more than mutations in major components of the pathway and actually contribute to CNS cellular variation in natural populations. Future experiments are needed to fully explore these functional differences between mam and Notch in the midline, as well as other tissues. Such differences can then be exploited to develop progressively specific research and clinical tools to regulate Notch signaling and the cellular composition of tissues [61,62].

Drosophila strains
The Drosophila fly strain used in the genetic screen was homozygous for both the UAS-GFP and sim-GAL4 transgenes which were recombined onto the same second chromosome. This combination labels all Drosophila midline cells beginning at developmental stage 10, through the remainder of embryogenesis and during larval stages. Prior to the mutagenesis screen, this line was isogenized using the yw 67 strain. The deficiency kit DK2, the 3 small deficiencies of mam: Df(2R)BSC383, Df(2R)50C-38, and Df(2R)BSC18, the mam 8 mutant line [34] and the UAS-GFP line were obtained from the Bloomington Stock Center. Additional fly lines used were: N 55e11 (described in [63]), Dl 3 [64], mam DC (this study), sim-GAL4 [65], UAS-NICD and UAS-Su(H).VP16 [53], and UAS-mam [66]. The FLP-DFS technique was used to generate mam DC germline clones [54]. For this, the mam DC mutation was first recombined onto the FRT42B chromosome and then w; P [48]42B 42B mam DC /CyO virgins were crossed to yw 67 P{hs-FLP}; P{w + , FRT}42B, P{Ovo D1 }55D/CyO males. Next, 2-3 days old larvae with the genotype y w P{hs-FLP}/w; P{w + , FRT}42B, P{Ovo D1 }55D/P{w + , FRT}42B mam DC generated from the cross were incubated at 37uC for 2 hours to induce recombination. Eclosed virgins were then crossed to w; mam DC /CyO males. Embryos collected from this cross were fixed and subjected to fluorescent in situ hybridization and immunohistochemistry. To test the effect of overexpressing the secreted form of Spi in mam DC mutants, the mam DC mutation was recombined onto both the UAS-sspi4a chromosome [67] and the sim-GAL4 chromosome.

Isolation of EMS generated mam mutants
To screen for genes on the second chromosome that affect midline development, yw 67 ; sim-GAL4 UAS-GFP males were mutagenized with ethyl methylsulfonate (EMS) and then mated en mass to yellow (y) white (w) 67 ; Lobe (L) 2 /CyO Kruppel (Kr)-GFP females. Single F1 male progeny were then backcrossed to 3 yw 67 ; L 2 /CyO Kr-GFP virgin females in a single vial. Next, F2 siblings of the genotype yw 67 ; UAS-GFP sim-GAL4/CyO, Kr-GFP were mated, and the absence of F3 progeny with straight wings indicated a line bearing a lethal second chromosome mutation (Figure 2A). To visually screen the lines bearing a lethal mutation on the second chromosome, embryos were collected every 12 hours, aged for 8 hours at room temperature and then examined for midline defects, first with a Leica MZ FLIII fluorescent stereomicroscope and then positives were more closely examined with a Zeiss Axioskop II fluorescent microscope and either a Zeiss Pascal or 710 confocal microscope. Homozygous mutant embryos were identified based on the absence of Kr-GFP fluorescence.

DNA sequence analysis of the mam DC mutant
Genomic DNA was extracted from homozygous mam DC mutant embryos and used as a template to amplify all mam coding exons. After PCR amplification, each coding exon was cloned into the pSTblue-1 vector (Novagen) and then plasmids were sent to Alpha BioLab, Inc. for sequencing. Sequence analysis was performed using the FinchTV program (Geospiza, Inc.) and indicates the mam DC allele contains a point mutation that creates a premature stop codon. The resulting truncated protein ends at Mam residue 959, eliminating the C-terminal acid cluster and all 3 glycinevaline (GV) runs ( Figure 2C). Based on comparison with mam deficiencies, the mam DC mutation behaves as a strong loss of function allele ( Figure S2).

Immunohistochemistry and in situ hybridization of embryos
Immunohistochemistry and in situ hybridization of whole mount embryos were performed as previously described [29,68]. The following primary antibodies were used: mouse anti-b-galactosidase (1:1000 Promega); rabbit anti-b-galactosidase (1:2000 Cappel); rabbit anti-En (1:100 Santa Cruz Biotech, Inc.); rat anti-Oddskipped (1:100), guinea pig anti-Odd-skipped (1:100) and guinea pig anti-Runt (1:100 or 1:200 East Asian Distribution Center; EADC); rabbit anti-GFP (1:500 Molecular Probes, Invitrogen); rat anti-Single-minded (1:100 [69]; and rabbit anti-tyrosine hydroxylase (1:500 [70]) and mouse anti-Wrapper (1:5 Developmental Studies Hybridoma Bank). The anti-guinea pig Alexa 633 was used at 1:100 and all other secondary antibodies were used at 1:200: anti-rabbit Alexa 488, anti-guinea pig Alexa 488, anti-mouse Alexa 488, antirabbit Alexa 568, anti-rat Alexa 568, anti-mouse Alexa 568 (Molecular Probes, Invitrogen). Embryos were imaged with a Zeiss Pascal in the Forestry Department and Zeiss 710 laser scanning microscope in the Cellular and Molecular Imaging Facility at NCSU. To determine the number of cells belonging to each lineage, midline cells were labeled with specific markers and at least 8 thoracic segments within several embryos were counted and presented as the mean 6 standard error of the mean (SEM) using stacked confocal images. The images shown are projections of multiple focal planes.