Evidence for the involvement of dominant-negative Notch molecules in the normal course of Drosophila development

Authors


  • This work is dedicated to our undergraduate colleague Tristan Thayer's short and serene life.

Abstract

Notch signaling is used to specify cell types during animal development. A high level specifies one cell type, whereas a low level specifies the alternate type. The effector of Notch signaling is the Notch intracellular domain. Upon its release from the plasma membrane in response to Delta binding the Notch extracellular domain, the Notch intracellular domain combines with the transcription factor Suppressor of Hairless and promotes the expression of target genes. Using a panel of antibodies made against different extracellular and intracellular regions of Notch, we show that cell types and tissues with low levels of Notch signaling are enriched for Notch molecules detected only by the extracellular domain antibodies. This enrichment often follows enrichment for Notch molecules detected only by antibodies made against the Suppressor of Hairless binding region. Notch molecules lacking most of the intracellular domain or containing only the Suppressor of Hairless binding region are produced during development. Such molecules are known to suppress Notch signaling, possibly by taking away Delta or Suppressor of Hairless from the full-length Notch. Thus, it is possible that dominant-negative Notch molecules are produced in the normal course of tissue differentiation in Drosophila as part of an auto–down-regulation mechanism. Developmental Dynamics 235:411–426, 2006. © 2005 Wiley-Liss, Inc.

INTRODUCTION

Notch (N) is a cell surface protein that is required for differentiation of almost all tissues in animals from worms to humans. Its actions specify two cell types from a population of equipotent cells or establish boundaries between populations of two different cell types. The mechanism of N signaling is as follows. When a ligand such as Delta (Dl) expressed on one cell binds N expressed on the neighboring cell, N is proteolytically cleaved, first by the Kuzbanian or TACE metalloproteases (called the S2 cleavage) and subsequently by the Presenilin (Psn)/γ-secretase complex (called the S3 cleavage). The Notch intracellular domain (Nintra) is released from the plasma membrane, translocated to the nucleus, and in association with the transcription factor Suppressor of Hairless (SuH) activates transcription of target genes such as the Enhancer of split Complex (E(spl)C) genes. We refer to this signaling as the SuH/Nintra signaling. Cells that initially generate high rates or levels of SuH/Nintra signaling, augment this rate or level and become specified as one cell type; cells that initially generate low rates or levels, suppress SuH/Nintra signaling completely and become specified as the alternate cell type (Heitzler and Simpson, 1991; Artavanis-Tsakonas et al., 1999; Mumm and Kopan, 2000; Brou et al., 2000; Lieber et al., 2002; Schweisguth, 2004; Ahimou et al., 2004). This process, often referred to as the lateral inhibition process, is used repeatedly during development for differentiation of various tissues with variations or changes in target genes.

The structural features of N and other components important for SuH/Nintra signaling are shown in Figure 1A. The N protein is composed of the following, in order from the amino terminus (extracellular) to the carboxyl terminus (intracellular): 36 tandem epidermal growth factor-like repeats (EGF-like repeats), which includes the Dl binding site; three cysteine-rich repeats called the lin12/B repeats; a potential Furin-mediated S1 cleavage site (see below); the S2 cleavage site; the transmembrane domain (TM), within which lies the S3 cleavage site; the Ram 23 region, the ankyrin repeats (anks), and the potential phosporylation domain (PPD), which are involved in binding SuH; a polyubiquitination site (ubi) implicated in endocytosis; a transcription activation domain (TAD); and a PEST sequence implicated in protein turn over (Wharton et al., 1985; Kidd et al., 1986, 1998; Rechsteiner, 1988; Fehon et al., 1990; Rebay et al., 1991; Lieber et al., 1992; Tamura et al., 1995; Matsuno et al., 1997; Logeat et al., 1998; Schroeter et al., 1998; Kurooka et al., 1998; Brou et al., 2000; Struhl and Adachi, 2000; Lieber et al., 2002; Le Gall and Giniger, 2004; Wilkin et al., 2004; Sakata et al., 2004). N receptors at the surfaces of mammalian cells are predominantly the noncovalently linked heterodimeric forms of the extracellular and the intracellular domains generated by Furin cleavage at the S1 site (Logeat et al., 1998). N receptors at the surfaces of Drosophila cells appear to be predominantly the covalently linked (collinear) full-length form (Kidd and Lieber, 2002). The reason for this difference is not understood but might be related to the role of N and Dl binding strength in the regulation of the rate of SuH/Nintra signaling (Ahimou et al., 2004).

Figure 1.

Structure of N and the epitope regions of N antibodies. A: The structure of the full-length N molecule (NFull) and the major components of SuH/Nintra signaling. See text in the Introduction section for the meaning of the abbreviated terms. B: Epitope regions of the various antibodies used in the study. Filled bars represent epitope regions determined, confirmed, or refined by us using in vivo immunocytochemistry and immunofluorescence procedures and ex vivo immunoprecipitation and Western blot procedures, with materials obtained from flies, S2 cells, and bacteria expressing N fragments. Unfilled bars represent epitope regions that are published or determined by others.

One of the better-understood instances of lateral inhibition is the differentiation of the central nervous system (CNS) and the epidermis (cuticle) from clusters of 5–20 proneural cells that form within a monolayer of cells in the periphery of the Drosophila embryo. Most cells in the proneural clusters accumulate a high level of SuH/Nintra signaling, become the epidermal precursor cells (EPCs), remain in the periphery of the embryo, and differentiate the epidermis. One or a few cells in the proneural clusters suppress SuH/Nintra signaling, become the neuronal precursor cells (NPCs), move inside the embryo, and differentiate the CNS (see Artavanis-Tsakonas et al., 1999; Schweisguth, 2004). Production of SuH/Nintra signaling at any time during the differentiation of the NPCs into neurons suppresses the production of neurons (Struhl et al., 1993; Lieber et al., 1993). However, N continues to be expressed and is required, in some other manner, during differentiation of neurons from the NPCs (Kidd et al., 1989; Fehon et al., 1991; Kooh et al., 1993; Giniger et al., 1993; Giniger, 1998; Crowner et al., 2003). This finding raises a significant question for neurogenesis: How is production of SuH/Nintra signaling suppressed or prevented during differentiation of neurons from the NPCs? One mechanism that suppresses SuH/Nintra signaling at an early stage in the process is known. It involves Numb, an endocytic protein, thought to target N for degradation (Guo et al., 1996; Spana and Doe, 1996; Santolini et al., 2000). Here, we present evidence for another mechanism covering both the early and late stages that would involve enrichment for dominant-negative N molecules lacking most of the intracellular domain or containing the SuH binding sites but not the TAD region. These molecules would titrate Dl or SuH away from the full-length N, the receptor capable of producing a high rate or level of SuH/Nintra signaling.

RESULTS

Preface to N Signal Patterns Described and the Procedures Used

The N antibodies used in the study and their epitope regions are shown in Figure 1B. The N antibody signals we describe can be grouped into four classes: (1) signals observed with all antibodies, (2) signals observed with the extracellular domain antibodies, (3) signals observed with the Ram 23 + Ankyrin repeat region antibodies; and (4) signals observed with the carboxyl terminus antibodies. αN203, αVT19, α466 signals were good representatives of signals 2, 3, and 4, respectively. Therefore, more data with these antibodies are shown. However, data from at least two different antibodies for each region are shown for many patterns. Signals 2 and 3 were exceedingly dynamic. Often, morphologically indistinguishable embryos showed apparently evolving patterns that were highly reproducible from batch to batch. This dynamism rendered immunofluorescence- and confocal microscopy-based procedures exceedingly inefficient and prohibitively wasteful of resources. Therefore, for basic characterization, we relied on the alkaline phosphatase- or horseradish peroxidase-based immunohistochemical procedures, which enabled us to study thousands of identically processed, developmentally timed embryos that could be ordered according to their relative ages. We imaged embryos mounted loosely on plain glass slide using coverslip props (so that the embryos can be rolled), illuminated by high-intensity light reflected off a white base, and captured by a Spot CCD camera attached to a Nikon SMZ 1500 stereo microscope and a computer. The resolution of the images, therefore, is limited. However, it is sufficient to capture the patterns and dynamism, in relation to known aspects of N function. Where possible, we examined immunofluorescent signal patterns and found no discrepancy with signals obtained from the immunohistochemical procedures. All antibody signals described are based on at least 10 repetitions. Each repetition used large numbers of developmentally timed embryos, produced by flies entrained to the circadian cycle that yielded at least 10 embryos of a particular pattern. Thus, although not all embryos of a morphologically defined stage showed a particular pattern belonging to a dynamic series, a similar or identical pattern was represented 100% of the time, at similar frequencies relative to other patterns in the series, in all repetitions. Signals observed with only one antibody have been ignored.

Specificity and Epitope Regions of N Antibodies Used

αN203, αVT19, and α466 signals are N-specific as indicated by the following experiments. We tested the epitope region specificity of antibodies in S2 cells, as many exogenous N molecules expressed in embryos are cleared rapidly (Struhl et al., 1993; Wesley and Mok, 2003). αN203, αVT19, and α466 detected only those N molecules that contained their epitope regions (Supplementary Fig. S1A–D, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat; N1-1789 containing 18 amino acids from the αVT19 epitope region was recognized weakly by this antibody). All other antibodies gave similar results, detecting only N molecules containing their epitope regions (data not shown). The antibodies made against the different intracellular domain regions gave different signal patterns not only in vivo but also ex vivo. Therefore, we tested the specificity of these antibodies on Western blots (ex vivo). All intracellular antibodies detected only those N fragments containing their epitope regions (Supplementary Fig. S1E–G). These observations indicated that the N antibodies we have used are specific to N and detected only N molecules containing their epitope regions. The N antibody signals were also drastically reduced or eliminated in zygotic N null (N/Y) embryos relative to the wild-type embryos (Supplementary Fig. 1H; the dark color of unused yolk in N embryos is not due to signals). Signals from the primary antibody minus control embryos served as the baseline for our assessment. αHM10 and α7477 also showed drastically reduced or no signals in N/Y embryos (data not shown).

N Signals in the CNS

The four extracellular domain antibodies used gave strong signals in the commissures and connectives (neuropile) of the CNS, whereas the seven intracellular domain antibodies gave weak signals, if any (Fig. 2A; αNT and αNPCR data are not shown but can be seen in Kidd et al., 1989, and Wesley and Saez, 2000, respectively). The strength of the signals was assessed relative to the signals in the ventral nerve cord (VNC) and the developing cuticle in the same embryos. The relatively strong horizontal segmental signal pattern observed only with αHM10 (Fig. 2A, embryo 8) was ignored.

Figure 2.

Embryonic central nervous system (CNS) signals obtained with the different N antibodies. A: Embryos immunostained (horseradish peroxidase) with the different N antibodies. BG: Immunofluorescence and confocal microscopy images of the embryonic CNS probed with different combinations of N antibodies or an N antibody and the hunchback antibody. Conc. = ∼10× concentrated antibody preparation. Green color, Alexa Fluor 488 secondary antibody; red color, Alexa Fluor 647 secondary antibody. All images are from stage 16 embryos. The right images in B–G are the merges of the middle and the left images. Here and in other figures, staging is according to Campos-Ortega and Hartenstein (1997). Here and in many other figures, epitope regions of the antibodies and the structure of N molecules used or described are shown at the bottom.

In the immunofluorescence and confocal microscopy procedure, the extracellular αN203 and αB antibodies gave strong signals in the commissures and the connectives of the CNS and weaker signals in the surrounding cells, whereas the intracellular αVT19, αC17.9C6, or α466 antibodies gave uniformly low signals in all cells of the CNS (Fig. 2B–G). Although the αC17.9C6 signals were quite similar to those of αVT19 and α466 at the most commonly used concentration range in the field (1/500–1/800), it gave relatively faint signals in the commissures and the connectives of the CNS at 10× that concentration (Fig. 2E,F). αC17.9C6 signals were also found to be more intracellular (compare Fig. 2E,F with B–D). Single channel images of αVT19, αC17.9C6 (ascites), and α466 signals showed a weak negative image of the commissures and the connectives of the CNS (Fig. 2C–E, images on the left). The Hunchback antibody (αHb) did not show such a negative image in the same area (Fig. 2G), indicating that it is not due to any physical barriers to antibody penetration.

The above antibody signal patterns suggest that the N intracellular domain is relatively inaccessible or deficient in the commissures and the connectives of the embryonic CNS or is present or accessible at similar levels in this tissue as well as the surrounding tissue. On the other hand, the N extracellular domain is relatively more accessible or enriched in the commissures and the connectives of the embryonic CNS.

N Signals During NPC (Neuroblast) Specification

At the onset of lateral inhibition, αVT19 and α7477 gave very strong signals in the predelamination stage NPCs (neuroblasts) at the surface of the embryo when compared to the signals in the surrounding cells (Fig. 3A, embryos 3–5, 8–10). The monoclonal antibody αC17.9C6 also gave stronger signals in these NPCs relative to the surrounding cells, although the overall signals were weaker than those obtained with the polyclonal αVT19 and α7477 antibodies. We attribute this difference to multiple binding of the relatively low-expressed N molecules by the polyclonals. The dynamism of the signal pattern obtained with αVT19 and α7477 antibodies is shown in embryos 3–5 and embryos 8–10 of Figure 3A. The three embryos in each set are separated by not more than a few minutes and are morphologically indistinguishable. Some of the signals in embryos 5 and 8 of Figure 3A could be from proneural cells as αVT19 and α7477 gave strong signals both in the NPCs and the proneural cells (see Fig. 4D,E for αVT19 signals in these cells). The strong αVT19 and α7477 signals were very transient, disappearing even before the NPCs have completed their delamination. For identification of NPCs, we relied on (1) their relatively large size and round morphology (Campos-Ortega and Hartenstein, 1997; see the magnified image in Fig. 3A, 15), (2) partial correspondence with the well-known markers (see below), and (3) the low level of the expression of E(spl)C m5 + m8 RNAs compared with the surrounding EPCs (data not shown). The well-known horizontal, segment-wise arrays of NPCs were vaguely discernible in the αVT19 patterns, to some degree resembling the achaete (ac) RNA pattern a short time later when the achaete expression is restricted to single NPCs within the proneural cluster (Fig. 3A, compare Embryo 5 with Embryos 6–7; note the doublet spots along the vertical midline in all three embryos). This suggests that the strong αVT19 and α7477 signals might precede the restriction of achaete expression to the NPCs and delamination of the NPCs. It also suggests that, although all the NPCs become part of the regular segmental arrays some time after lateral inhibition, their actual specification might not be in unison as both the achaete and αVT19/α7477 signals indicate. Our studies indicate that the strong αVT19 and α7477 signals might be the earliest markers of the NPCs.

Figure 3.

Signals obtained with the different N antibodies in embryos undergoing lateral inhibition. A: Signals obtained with the different N antibodies (embryos 1–5, 8–16) or the probe for achaete (ac) RNA (embryos 6–7) in immunocytochemical or in situ RNA hybridization procedures using alkaline phosphatase-conjugated secondary antibodies. Embryos 1–5, 8–12, 15 are stage 8d; embryos 13–14, 16 are stage 9–9a; embryos 6–7 = stage 9c–10a. Panel 15 is a magnified image of a segment of an embryo comparable to 4, in which the outlines of cells around the incipient neuronal precursor cells (NPC, the large cell in the middle) are marked. Panel 16 is a magnified image of a segment of an embryo comparable to 13; some cells are outlined to show the location of the spots near the surface and inside the cells. BF: Immunofluorescence and confocal microscopy images of the lateral inhibition stage embryos probed with different combinations of N antibodies and NPC marker antibodies αSca and αHb. In D–G, αN203 signals are green; αHb, α466, αC17.9C6, and αVT19 signals are red. Embryos in B and C are stage 8d; in D–G are stage 9–9a. The second image in B is the merged imaged of the last two images. The top image in C is a merged image of the bottom two images. All images in D–G are merged images.

Figure 4.

Signals obtained with the different N antibodies and the probe for achaete RNA in embryos at various stages of development. A: In stage 1 embryos. B: In stage 1–2 embryos. C: In stage 4 embryos. D: In stage 8d embryos. E: In stage 8c embryos. All embryos were probed with antibodies, except the third embryo from the top in D and E, which were probed with the digoxigenin-labeled achaete (ac) DNA. All embryos were immunochemically stained using alkaline phosphatase-conjugated secondary antibodies.

Antibodies made against the extracellular domain also gave stronger signals in the predelamination stage NPCs compared with the signals in the surrounding cells. However, these signals were more transient and much weaker than the signals obtained with the αVT19 or α7477 antibodies (Fig. 3A,11–12; αN203 and αNO gave similar signals, data not shown). The extracellular domain antibodies gave strong signals in localized spots near the cell surfaces and inside the delaminating/delaminated NPCs (Fig. 3A, embryos 13–14; see 16 for a magnified image). Strong αVT19 and α7477 signals were not observed on or in these late stage NPCs. Antibodies made against the carboxyl terminus, α466 and αHM10, gave uniform signals in all cells of the embryo at these stages (Fig. 3A, embryos 1 and 2).

Among the many NPC markers tested, only Scabrous showed some correspondence with the αVT19 and α7477 signals in the early stage NPCs and Hunchback showed correspondence with the αN203 signals in the delaminating or delaminated NPCs (data not shown). Accordingly, immunofluorescence and confocal microscopy images of doubly probed embryos showed transient overlap between Scabrous and αVT19 or α7477 signals and good overlap between Hunchback and αN203 signals (Fig. 3B–G). The strong αVT19 and α7477 signals in the incipient NPCs appeared to derive from these cells becoming filled with signals (see Fig. 3B and C, αVT19 and α7477 single images; note the strong signals inside and at the surfaces of cells showing strong signals). On the other hand, the strong αN203 signals in the delaminating/delaminated NPCs appeared to derive from localized spots near the surface or inside these cells (Fig. 3D–G).

The above-described signal patterns suggest that the Ram 23 + Ankyrin repeat region of N is the most enriched or accessible part of N in the predelamination stage NPCs, i.e., in the NPCs at the periphery or the surface of the embryo. The extracellular domain of N is modestly enriched or accessible in these NPCs. In the later stage NPCs, i.e., the delaminating or the delaminated NPCs, the extracellular domain of N is the most enriched or accessible part of N, in localized spots at or near the cell surface.

N Signals at Other Stages of Embryogenesis

In the same pool of embryos used to study the CNS development, αN203, αVT19, and α466 gave very similar signal patterns at the beginning of embryogenesis, with αN203 giving the strongest signals. Very soon after, αN203 and αVT19 gave a similar pattern of signals that was distinct from the signal pattern of α466 (Fig. 4A,B). αVT19 gave the strongest signals in germ cells, followed by α203, and then α466 (Fig. 4C). On the other hand, αN203 gave the strongest signals in the amnio serosa and αVT19 in the sensory organ precursor cells that are the NPCs (Fig. 4D). An embryo probed for achaete RNA, a marker for proneural cells, is also shown for comparison. At a slightly earlier stage (by just a few minutes), αVT19 gave strong signals overlapping with the proneural cells; αN203 or α gave very weak signals (Fig. 4E). The strong αVT19 or α7477 signal domains appeared to be generally larger than the domains of the proneural cells (marked by achaete expression), suggesting that the former might define the limits within which the proneural clusters can form. All of these observations indicate that the differences in N signals obtained with antibodies specific to the extracellular domain, the Ram 23 + Ankyrin repeats region, and the carboxyl terminus are not limited to the CNS development and are apparent in many types of embryonic cells and tissues, keeping in line with the widespread function for N during embryogenesis. Indeed, the differences shown in this article constitute a minor fraction of the differences observed throughout embryogenesis.

N Signals During the Formation of Cephalic and Ventral Furrows

The cephalic furrow and the ventral furrow are formed when a band of cells in the outer layer of the embryo invaginate and move inside to form the mesodermal and endodermal primordia (Campos-Ortega and Hartenstein, 1997). There is some, if only superficial, resemblance between the processes involved in migration of cells inside by the way of cephalic or ventral furrows and NPC delamination. Whereas N function in ventral furrow formation is known (e.g., Morel and Schweisguth, 2000), its function in cephalic furrow formation is unknown. Nevertheless, similar observations in these two similar processes provide compelling evidence for the relationship of the signals from the different N antibodies to SuH/Nintra signaling.

The N extracellular domain antibodies gave strong signals in the cephalic furrow as well as in other furrows forming elsewhere at the same time (Fig. 5A,B). The developmental time from embryo 1 to 5 is just 10–12 min. The strong signals in these embryos were not due to the extracellular domain antibodies nonspecifically accumulating in the crevices or folds as these signals preceded the furrow formation, marking the first row of cells that would later initiate formation of the cephalic furrow (Fig. 5A,B, rows 1–2). Furthermore, the strong signals disappeared when the furrow was fully formed and much deeper (data not shown). A similar “evolution” of signals was observed at a much later stage where the crevice or fold is more extreme (Supplementary Fig. 2). The Ram 23 + Ankyrin repeat region antibodies gave strong signals in the cephalic furrow as well as in the adjacent cells (Fig. 5C,D). A similar signal pattern was observed with the Dl antibody (Fig. 5G). The carboxyl terminus antibodies gave almost a negative image of the extracellular domain antibody signals in the cephalic furrow and uniformly low signals elsewhere (Fig. 5E,F). Comparable signals were observed with the E(spl)C m5 + m8 RNA probe, SuH antibody, and the Psn antibody (Fig. 5H–J).

Figure 5.

Signals obtained with the different N antibodies and the probes against some important components of SuH/Nintra signaling in embryos forming the cephalic furrow. AJ: Embryos forming the cephalic furrow. In A and B, embryo 1 is stage 6, embryo 2 is stage 6a, embryo 3 is stage 6b, embryo 4 is stage 7, embryo 5 is stage 7a. C to J are stage 7a. Insets in A and C are magnified images of embryo 2 and C, respectively, with some cells outlined to show the distribution of signals. All embryos were immunochemically stained using alkaline phosphatase-conjugated secondary antibodies. Embryo in H shows RNA signals generated using digoxigenin-labeled DNA probes; all others show signals generated using antibodies.

Cells invaginating into and forming the ventral furrow are bounded by the rows of mesectodermal cells expressing the E(spl)C genes (Fig. 6F; these are the same cells that express single-minded, Morel and Schweisguth, 2000). The developmental time from embryos 1 to 7 in Figure 5F is estimated to be just 10–12 min. The extracellular domain antibodies gave strong signals in cells at the center of the field of cells bounded by E(spl)C m5+m8 RNA expression, those very likely to invaginate first (Fig. 6A,B, row 1; only embryos stained with αN203 and αB are shown). Soon after, these antibodies gave strong signals within the field of cells bounded by the E(spl)C m5+m8 RNA expression and in cells within the ventral furrow (Fig. 6A,B,F, rows 2–4). Antibodies made against the Ram 23 + Ankyrin repeat region gave strong signals in a complex pattern in the initial stages of the invagination process (Fig. 6D, rows 1–3; only embryos probed with αVT19 are shown). Near the end of the process, these antibodies gave strong signals in the single rows of cells on either side of the ventral furrow (Fig. 6D, rows 4–7 and 6E). These strong signals were coincident with the loss of E(spl)C m5+m8 expression (Fig. 5F, row 4–7). A closer examination of embryos at the very early stages in the process showed that not only the Ram 23 + Ankyrin repeat antibodies but also the extracellular domain antibodies gave a negative image of the E(spl)C m5+m8 expression (Fig. 6K; only antibody staining with α466, αVT19, and αN203 are shown). Our attempts at protein/RNA double labeling have failed so far. Even the minimal protease treatment required for RNA hybridization destroyed N proteins and the substitute acetone treatment gave very poor results, possibly due to the generally low expression of N proteins and the E(spl)C RNAs (data not shown).

Figure 6.

Signals obtained with the different N antibodies and probes against some important components of SuH/Nintra signaling in embryos forming the ventral furrow. AF: Embryos in A–D and F row 1 are stage 6; row 2 are stage 6a, row 3 are stage 6b, row 4 are stage 7, rows 5–7 and E are stage 7a. GJ: Embryo in G is stage 7a, in H is stage 7, in I is stage 6b, and in J is stage 7. Embryos in K are stage 6. Magnified images of important regions of embryos (comparable to the embryo whose number is indicated) are shown at the bottom of columns A and D. All embryos were immunochemically stained using alkaline phosphatase-conjugated secondary antibodies. Embryos in F and embryo 2 in K show RNA signals generated using digoxigenin-labeled DNA probes; all others show signals generated using antibodies.

The carboxyl terminus antibodies gave a negative image of the extracellular domain antibody signals early in the invagination process; near the end of the process, they gave strong signals in the rows of cells on either side of the ventral furrow (Fig. 6C,J). SuH and Psn antibodies gave signals comparable to those of the N carboxyl terminus antibodies (Fig. 6H,I). On the other hand, Dl antibody signals gave signals comparable to those of the N Ram 23 + Ankyrin repeat region antibodies (Fig. 6G, compare with 6D, rows 5–7 and 6E).

The signal patterns described above suggest that the extracellular domain of N is strongly accessible or enriched in the cells invaginating into and forming the cephalic and the ventral furrows. The carboxyl terminus of N, as well as other components of SuH/Nintra signaling, is relatively inaccessible or deficient in these cells. There appears to be an inverse relationship between the expression of E(spl)C m5+m8 RNA and the accessibility or enrichment for the extracellular domain and the Ram 23 + Ankyrin repeat regions of N.

Hypothesis Testing: N Signals in the Neurogenic Embryos and Mutant Flies

If the accessibility or the level of the extracellular domain and the Ram 23 + Ankyrin repeat regions of N was increased in association with the loss of SuH/Nintra signaling, signals from antibodies against these regions were expected to increase in neurogenic embryos, which are null for SuH/Nintra signaling. We tested this hypothesis.

The neurogenic embryos were staged using the shape of the head region and the extent of the shortening of the germ band, which is quite accurate. Stages of Dl null embryos that were beginning to show the effect of loss of SuH/Nintra signaling showed dramatically high levels and numbers of the signals given by the N extracellular domain and the Ram 23 + Ankyrin repeat (Fig. 7A, embryos 3–6). Increased signals were not observed with the carboxyl terminus antibodies (Fig. 7A, embryos 1–2; only α466 signals are shown). Signals by all of the N antibodies used in the study were eventually lost in zygotic N null (N/Y) neurogenic embryos (for example, see Supplementary Fig. 1H). However, at stages that were beginning to show the effect of loss of N, the signals given by the N extracellular domain and the Ram 23 + Ankyrin repeat antibodies also dramatically increased in level and number, in a pattern comparable to the wild-type pattern (compare Fig. 7B rows 1–3 with Fig. 3A embryos 3–5, 8–14, and with Fig. 4E; only αN203 and αVT19 signals are shown). Increased signals were not observed with the carboxyl terminus antibodies (Fig. 7B, row 4; only α466 signals are shown). An interesting pattern could be discerned with the Dl null embryos. Signals by the Ram 23 + Ankyrin repeat region antibodies initially increased in all the NPCs (Fig. 7C, Embryo 1; compare with embryos 3–5, 8–10 in Figure 3A to note the expected increased numbers of NPCs). Subsequently, these signals almost disappeared (Fig. 7C, embryo 2; similar evolution of signals was observed with the αVT19 antibody as well; data not shown). Signals with the extracellular domain antibodies also increased initially, coinciding with the Ram 23 + Ankyrin repeat region antibody pattern but with additional signals in localized spots (Fig. 7C, embryos 3, 5). At later stages, while the signals coincident with the Ram 23 + Ankyrin repeat region antibody signals disappeared, the strong signals in localized spots persisted (Fig. 7C, embryos 4, 6). Similar evolution of signals was observed with αN203 (data not shown). We interpret the extracellular domain and the Ram 23 + Ankyrin repeat region antibody signals in the Dl and N null embryos as an increase over the level of signals observed in control wild-type embryos, because the intensity of signals in the null embryos appeared to be greater than that in the wild-type embryos in the same pool, even with allowance for increased numbers of NPCs.

Figure 7.

Signals obtained with the different N antibodies in wild-type and neurogenic embryos of comparable stages. A: Signals obtained with different N antibodies in wild-type (yw) and zygotic Dl null (Dlx/Dlx) neurogenic embryos of stages 11–12. B: Signals obtained with different N antibodies in wild-type (yw) and zygotic N null (N55e11/Y) neurogenic embryos of comparable stages. Embryos in rows 1, 2, 4 are stage 11; in row 3 are stage 10. C: Signals obtained with different N antibodies in zygotic Dl null (Dlx/Dlx) neurogenic embryos of increasing age and severity of the neurogenic phenotype, from stage 10 (embryos 1, 3, 5) to stage 12 (embryos 2, 4, 6). Different embryos for each antibody were picked from the same pool of stained embryos. D: Signals obtained with the Hunchback antibody in wild-type (yw) and zygotic N null (N55e11/Y) neurogenic embryos. Embryos 1–2 are stage 11; embryos 3–5 are stage 13. All embryos were immunochemically stained using alkaline phosphatase-conjugated secondary antibodies.

We also examined whether or not the N extracellular domain and the Ram 23 + Ankyrin repeat antibody signals increase in embryos that were manipulated to reduce SuH/Nintra signaling. We expressed the dominant-negative Dl transgene Dl-DN (Huppert et al., 1997) or the N RNAi construct 14E (Presente et al., 2002) in a general manner using the da-Gal4 driver. Although these experiments are complicated by many factors, they clearly showed that removal of N or Dl activity results in increased signals from the extracellular domain and the Ram 23 + Ankyrin repeat region antibodies but not from the carboxyl terminus antibodies (Supplementary Fig. 3). Thus, in both the classic and transgenic N or Dl null/hypoactive embryos the N extracellular domain and the Ram 23 + Ankyrin repeat region antibody signal increased but not the carboxyl terminus antibody signals.

If increased signals from the N extracellular domain and the Ram 23 + Ankyrin repeat antibodies was sufficient for the production of neurons, the CNS was expected to be more or less developed in neurogenic embryos. We tested this hypothesis. Immunostaining with the neuronal marker Hunchback antibody showed signal patterns comparable to the patterns obtained with the Ram 23 + Ankyrin repeat antibodies in the neurogenic embryos: the signals increased initially (possibly due to increase in the numbers of NPCs/neuroblasts) but eventually were lost (Fig. 7D). The Elav antibody, another neuronal marker, gave somewhat similar results (Supplementary Fig. 4). It appeared that either the neurons failed to form fully or they failed to persist. Thus, the processes that are associated with increased signals from the N extracellular domain and/or the Ram 23 + Ankyrin repeat region antibodies appear to be insufficient for either producing fully formed neurons or their stable existence. While both N null and Dl null embryos showed similar patterns, we show data for only the more rigorous N null test material. Dl null embryos are less rigorous for this hypothesis testing, as Dl has N-independent activity that might be required for neurogenesis (Mok et al., 2005).

The results described above support the hypothesis that the N extracellular domain and the Ram 23 + Ankyrin repeat regions become more accessible or enriched in association with loss of SuH/Nintra signaling. They also show that this accessibility or enrichment is not sufficient for the formation of stable neurons.

N Signals on Western Blots

Western blotting of αN203 immunoprecipitates from embryonic extracts showed a faster migrating form of N that was recognized αNT, αB, αVT19, but not by αNI, αC17.9C6, α7477, and α466 (Fig. 8A). We will refer to this form as NΔI. Detection by αVT19 indicates that the carboxyl terminus of NΔI lies definitely after the amino acid 1771 (the end of the transmembrane domain), possibly a few amino acids after 1789, as this antibody detects NΔI better than N1-1789 (Fig. 8B, lanes 5–6). Note that, in sodium dodecyl sulfate-polyacrylamide gel electrophoresis with β-mercaptoethanol, NΔI migrates alongside N1-1789 truncated just after the end of the transmembrane domain at 1771 and faster than NΔCterm truncated just after the end of Ankyrin repeats at 2145 (see even numbered lanes Fig. 8A). As we observed quite dramatic differences in the in vivo signals obtained with αVT19 and the extracellular domain antibodies, it appears that αVT19 does not detect NΔI in vivo (if it did, the differences would be an underestimate of the actual differences). This inference is supported by the absence of obvious differences between the in vivo signals of αVT19 and α7477 that does not recognize NΔI (see Fig. 8A, lanes 11–12). In any case, detection by αVT19 distinguishes NΔI from the putative S2/S3 cleaved extracellular domain of the heterodimeric receptor (unavailability αVT19 being the prime reason for not detecting NΔI in our previous study, Wesley and Saez, 2000).

Figure 8.

Forms of N in wild-type yw embryos identified by immunoprecipitation and Western blotting procedures. A: Western blots showing N molecules immunoprecipitated by an amino terminus antibody and probed with antibodies made against different regions along the length of the NFull protein. IP Ab, antibody used in immunoprecipitation; WB Ab, antibodies used on Western blots. B: Inference of the structures of the different N intracellular domain fragments based on a systematic study of N fragments obtained with all possible immunoprecipitation/Western blotting combinations of the N intracellular domain antibodies used in the study (except αNPCR). Positions of possible proteolytic cleavage sites are shown at the bottom. S1, previously described Furin cleavage site; S4-6, newly proposed sites. C: A sample of two Western blots showing the different N intracellular fragments immunoprecipitated from embryonic extracts that are described in B. IP Ab, immunoprecipitating antibody; WB Ab, Western blotting antibody; P, immunoprecipitate; F, flow through; pre-is, pre-immune serum; [IP Ab], antibody cleared by precipitation. Lanes 1 and 3 and lanes 5 and 7 represent P and F fractions from the same sample.

Similar analyses with the intracellular domain antibodies showed several N molecules that were recovered and/or detected by at least two different N antibodies and expressed at relatively significant levels (assessed in relation to the level of NFull or the housekeeping protein Hsp 70). One such molecule, called Ni45-50, migrated sometimes at 45 kDa and sometimes at 50 kDa, possibly due to modification. See Figure 8B for the structure and 8C (lanes 1 and 5) for the Western blot identity (lanes 3 and 7 show the extract after the immunocomplexes were cleared, i.e., the flow through). To obtain nicely resolved bands, a reasonable statistical sampling of the different N fragments, and to minimize the IgG-related background possible with the procedure, the immunoprecipitations were performed with limited quantities of the N antibodies. Thus, N molecules are expected in the flow through (Fig. 8C, lanes 3, 7). Ni45-50 was detected by αB and biotinylated in cell surface biotinylation experiments with disassociated embryonic cells (data not shown). Thus, Ni45-50 appears to have the transmembrane domain. Although Ni45-50 appears to be NΔCtermintra (Wesley and Saez, 2000), we use a different name, because it was identified by a different approach. Ni32 appears to be Ni45-50 without the amino terminus transmembrane/juxtamembrane region (see Fig. 8B for the structure and 8C lanes 1 and 5 for the Western blot identity). The other molecules shown in Figure 8B, namely Ni60, Ni52, and Ni35, were expressed at lower or variable levels than Ni45-45, Ni32, or Nintra (see Fig. 8C lanes 1 and 5 for their Western blot identities). Note that the levels can only be assessed in relation to the level of NFull in the lanes as the different fragments transfer to the blots at increasing efficiency from the ∼400-kDa NFull to the ∼30-kDa Ni32.

The above-described immunoprecipitation and Western blotting analyses showed that the wild-type embryos contain high levels of an N molecule composed of the epitope regions of all the N antibodies (NFull), an N molecule mostly composed of the epitope regions of the extracellular domain antibodies (NΔI), N molecules mostly composed of the epitope regions of all the intracellular domain antibodies (Nintra), and N molecules mostly composed of the epitope regions of the Ram 23 region plus the ankyrin repeats region antibodies (Ni45-50 and Ni32). They also contain low levels of N molecules lacking the carboxyl terminus (NΔCterm), N molecules mostly composed of the epitopes regions of the carboxyl terminus antibodies (Ni52, Ni35), or N molecules composed of portions of the epitope regions of the carboxyl terminus and the Ram 23 + Ankyrin repeats region antibodies (Ni60).

DISCUSSION

Interpretation of In Situ and Ex Situ Signal Data

All our controls and comparisons to published reference patterns show that the antibody signals we have described derive specifically from the antigens of the antibodies used. With N antibodies, our controls show that they are specific to the epitope regions of the antibodies. In addition to these controls, the extremely predictable dynamism of N signals, not only within a process but between different processes, manifest with at least three different antibodies for each region that was made in different labs or animals, also indicates signal specificity. Dynamic antibody signals derive from the enrichment or loss in the level or accessibility of the epitopes compared with a baseline level. The strong signals by the extracellular domain, the Ram23 + Ankyrin repeat region, and the carboxyl terminus antibodies appear to be due generally to enrichment rather than loss in the levels or accessibility of their epitopes. The uniform and low level of the carboxyl terminus antibody signals at most stages of embryogenesis that appears to be the baseline level with all antibodies supports this inference. The rare occasions showing enrichment or loss of the carboxyl terminus antibody signals (Figs. 4D, 5E,F, 6C row 5, 6J) indicates that our procedures would have detected if such enrichment or loss were widespread. In the instances where the signal pattern of the intracellular domain antibodies included a weak “negative image” of the signal pattern of the extracellular domain antibodies, the loss in the levels or accessibility of the intracellular epitopes is lower than the enrichment in the levels or accessibility of the extracellular epitopes, as the depth of the “negative” and the “positive” images do not seem to match. In the processes where we can place the signal patterns in a developmental sequence, such as the differentiation of the CNS from the proneural cells, the enrichment in the levels or accessibility of the Ram 23 + Ankyrin repeat region antibody epitopes preceded the enrichment in the levels or accessibility of the extracellular domain antibody epitopes. In general, however, it appears that the enrichment in the levels or accessibility of the Ram 23 + Ankyrin repeat region antibody epitopes is complex and very dynamic, whereas that of the extracellular antibody epitopes is well-defined and relatively stable.

Our ex vivo immunoprecipitation and Western blotting data show smaller N molecules that contain the epitope regions of some antibodies but not of others, paralleling the signal patterns observed in vivo. This correspondence suggests that the weak in vivo signals with antibodies against one N region when there were strong signals with antibodies against other N regions is due to the difference in the level rather than the accessibility of the epitopes. Thus, the enrichment for the extracellular domain signals could be due to the enrichment for NΔI. The enrichment for the Ram 23 + Ankyrin repeat region signals could be due to the enrichment for Ni45-50 and/or Ni32 (with αVT19 better at detecting the former at the cell surface and αC17.9C6 the latter inside the cell). The enrichment for both the extracellular domain and the Ram 23 + Ankyrin repeat region signals could be due to the enrichment for NΔCterm or the simultaneous enrichment for NΔI, Ni45-50, Ni32, and NΔCterm. The enrichment for the carboxyl terminus signals is more likely to be due to the enrichment for Ni52 and Ni35 rather than Nintra, because we did not observe it in association with E(spl)C RNA signals (Fig. 6K) or during lateral inhibition (Fig. 3A). However, Nintra could be the basis in some instances (Fig. 6C,J). In other instances, the low and uniform level of NFull represented by the carboxyl terminus antibody signals (and shared by all antibodies) appears to be permissive for the usual levels of the SuH/Nintra signaling. Due to our ignoring signals (1) given by single antibodies, (2) that could not be related to N functions, and (3) that could not be accurately described due to the extreme dynamism, the differences between the different antibody signals we describe are an underestimate of the actual differences in N epitope patterns during Drosophila embryogenesis.

The smaller N fragments do not appear to be products of transcriptional or RNA-based posttranscriptional processes (e.g., alternate splicing, etc.), as the N gene lacks appropriate regulatory regions to produce them. They are likely to be produced from NFull by highly regulated proteolytic mechanisms that rapidly produce and destroy them. Otherwise, we would not have detected such dramatic differences in the signals given by antibodies against the different N regions. Our NRNAi data (Supplementary Fig. 3B) also supports a proteolytic mechanism. NΔCterm could be produced by the removal of Ni52 from NFull; NΔI by the removal of Ni35 and Ni60 from NFull and/or Ni32 from NΔCterm. These potential cleavage sites (S4–S6) are shown in Figure 8B. It is possible that NFull, NΔCterm, and NΔI are all substrates for S1 cleavage by Furin to make the heterodimeric forms. In particular, Ni45-50 could be part of a heterodimeric receptor, as our size estimate indicates that this molecule's amino terminus is very close to the S1 cleavage site. Thus, it is possible that, whereas NFull functions as a collinear molecule, NΔCterm and NΔI function as heterodimeric molecules. The N carboxyl terminus has poly-ubiquitination and PEST sites important for endocytosis and turnover (see Fig. 1A; Rechsteiner, 1988; Sakata et al., 2004; Wilkin et al., 2004). We have shown that N molecules lacking the carboxyl terminus are deficient in both Dl-independent and -dependent internalization (Bardot et al., 2005). Thus, the enrichment for molecules lacking the carboxyl terminus (NΔCterm, NΔI, Ni45-50, Ni32) could be facilitated by the loss of endocytosis and turnover signals. On the other hand, the enrichment for molecules containing the carboxyl terminus might be suppressed by the presence of these signals, thereby explaining the uniformly low level of expression of these molecules at most stages of embryogenesis.

The N extracellular domain fragment cleaved at the S2 and S3 sites is thought to be pulled by Dl endocytosis into the NPCs, in association with increased SuH/Nintra signaling in the EPCs (Klueg et al., 1999; Parks et al., 2000; Struhl and Adachi, 2000; Pavlopoulos et al., 2001). The ex vivo NΔI molecule is not such a transendocytosed N extracellular domain fragment, as it contains a small part of the intracellular domain and the transmembrane domain, i.e., it is not cleaved at the S2 or S3 sites (see Fig. 1A for the location of these sites). The in vivo N recognized by all of the extracellular antibodies and none of the intracellular antibodies is also unlikely to be such a fragment, because it is produced in Dl null or N null embryos that are deficient in SuH/Nintra signaling, Dl, or NFull. In fact, we observed increased extracellular domain signals in the neurogenic Dl null and N null embryos (Fig. 7A–C). However, it is possible that the in vivo N or the ex vivo NΔI is a molecule transendocytosed by a novel mechanism that is not directly dependent on Dl/S2 or S3 cleavage/SuH/Nintra signaling but in response to these.

Significance of the Signal Data to Tissue Differentiation in Drosophila

Our study shows that the signals from the N extracellular or the Ram 23 + Ankyrin repeats region antibodies change dramatically in the course of embryogenesis, correlating with the regulation of SuH/Nintra signaling. These changes are possibly due to the production of N molecules composed mostly of the extracellular domain (NΔI) or the Ram 23 + the Ankyrin repeats (Ni45-50 or Ni32). Such molecules are known to behave as dominant-negative molecules with respect to the SuH/Nintra signaling by NFull: NΔI-like molecules by titrating away Dl and NΔCterm, Ni45-50- or Ni32-like molecules by titrating away SuH (Lindsley and Zimm, 1992; Lieber et al., 1993; Lyman and Young, 1993; Sun and Artavanis-Tsakonas, 1997; Jacobsen et al., 1998; Brennan et al., 1999; Wesley and Saez, 2000; Wesley and Mok, 2003). Accordingly, the signals from the N extracellular or the Ram 23 + Ankyrin repeats region antibodies are enriched in cells/tissues with reduced SuH/Nintra signaling. We will briefly describe below the possible significance of our data to the regulation of Drosophila tissue differentiation.

Activation and suppression of SuH/Nintra signaling is used to specify two different cell types from a stem cell population. These cell types go on to produce two different tissues. Neurogenesis and epidermogenesis from proneural stem cells in Drosophila embryos exemplify the use of SuH/Nintra signaling during development. Proneural cells that increase SuH/Nintra signaling become the EPCs and differentiate the epidermis. Proneural cells that suppress SuH/Nintra signaling become the NPCs and differentiate the nervous system. Even a low level of SuH/Nintra signaling during differentiation of the NPCs, even at late stages, will suppress the production of the nervous system (Struhl et al., 1993; Lieber et al., 1993). This finding indicates that the differentiating neuronal cells retain the capacity to transduce the SuH/Nintra signaling but do not produce this signaling even though N and Dl are expressed in these cells and are required for completing the neuronal differentiation program (Shellenbarger and Mohler, 1978; Kidd et al., 1989; Fehon et al., 1991; Kooh et al., 1993; Giniger et al., 1993; Giniger, 1998; Crowner et al., 2003). NΔCterm, Ni45-50 and/or Ni32 molecules might initiate the suppression of SuH/Nintra signaling in a dominant-negative manner by titrating SuH away from NFull. However, this suppression would be only partial as NΔCterm, Ni45-50, or Ni32 molecules are capable of producing some SuH/Nintra signaling (Struhl and Adachi, 1998; Wesley and Saez, 2000; Wesley and Mok, 2003). In contrast, NΔI is completely null for SuH/Nintra signaling and would also dominant-negatively suppress SuH/Nintra signaling by titrating Dl away from NFull. Thus, the cells or tissues requiring suppression or blockage of the SuH/Nintra signaling might enrich for NΔCterm/Ni45-50/Ni32 or NΔI molecules, respectively. If so, Drosophila would have adopted the simple and effective means for inhibiting biochemical reactions: producing a defective substrate that binds its ligands. This mechanism that works with the NPCs might also work during the formation of cephalic furrow, ventral furrow, germ cells, proneural clusters, etc. Of interest, in all these processes, cells within a defined area separate and move away from their neighbors that stick together. We have shown that NFull binds Dl very strongly compared with NΔCterm or NΔI and that the SuH/Nintra signaling is positively correlated with binding strength (Ahimou et al., 2004). Thus, the enrichment for truncated N molecules might serve both the biochemical and biophysical processes regulating tissue differentiation.

Our data show that loss of functional N genes or SuH/Nintra signaling might lead to production of incompletely formed or unstable neurons, although there are increased levels of the extracellular domain and the Ram 23 + Ankyrin repeat region epitopes as observed during normal neurogenesis (Fig. 7). These observations indicate that the normal differentiation of the NPCs into the nervous system in the embryos might require suppression of SuH/Nintra signaling and the epidermis or alternate N and Dl functions that should not produce SuH/Nintra signaling. Indeed, our studies show that NΔCterm, Ni45-50, and/or Ni32 up-regulate the expression of neurogenesis genes such as daughterless (Wesley and Saez, 2000), and Dl has neurogenesis-promoting activity independent of its activity as a ligand of N (Mok et al., 2005).

The observation that the extracellular domain and the Ram 23 + Ankyrin repeat region antibody signals increase in N and Dl null embryos (Fig. 7A–C) raises the possibility for an interesting basis for the dominance of N null mutations. Specification of two cell types during lateral inhibition is based on the relative levels of SuH/Nintra signaling (Heitzler and Simpson, 1991). Cells that produce SuH/Nintra signaling at a higher rate or level increase this signaling by a positive feedback mechanism to become one cell type (e.g., the EPCs). Cells that produce SuH/Nintra signaling at a lower rate or level suppress this signaling to become the other cell type (e.g., the NPCs). It is possible that cells activate mechanisms that increase or suppress SuH/Nintra signaling, depending on whether or not they have attained a certain set level of this signaling relative to their neighbors. If that level is not reached, the cells might automatically activate the mechanism that suppresses SuH/Nintra signaling. This kind of an auto–down-regulation mechanism might explain the choice of cells for effecting lateral inhibition (in the classic sense) and the worsening symptoms with age in diseases involving N (Kalimo et al., 1999; Gridley, 2003).

EXPERIMENTAL PROCEDURES

The N antibodies used were the following: αNT made in rabbits against the first two EGF-like repeats (Kidd et al., 1989); αN203 in rats against the first three EGF-like repeats (Wesley and Saez, 2000); αNO in rabbits against EGF-like repeats 17–21 (Kidd and Lieber, 2002); αB in rabbits against the lin12/B repeats (a remake of the DPA antibody, Kidd et al., 1989); αVT19 in chicken and α7477 in rabbits against a bacterially made GST fusion protein containing N amino acids from 1771 to 2155 (numbers according to Kidd et al., 1986); αNI in rabbits against the 1795 to 2157 amino acid region (Lieber et al., 1993; used only in Western blots, as supply is limited); the mouse monoclonal αC17.9C6 (Fehon et al., 1990) from DHSB (University of Iowa), whose epitope we have determined to lie between amino acids 1893 and 2115; α466 in guinea pigs against a bacterially made GST fusion protein containing N amino acids 2148 to 2536; αHM10 in hamsters against a bacterially made GST fusion protein containing N amino acids 2341 to 2536 (the same one described as same α2341 in Wesley and Mok, 2003); and αNPCR in mouse against the 2115 to 2536 amino acid region (Lieber et al., 1993; used only to confirm patterns as its supply is nearly exhausted). Antibodies against Scabrous were generated in guinea pigs against the bacterially made GST fusion protein containing the whole Scabrous protein; SuH antibodies were made in rats (Wesley and Mok, 2003); Psn antibodies were made in rabbits (a remake of the antibody described in Ye and Fortini, 1998); Hunchback antibodies were obtained from Drs. Nipam Patel (Patel et al., 2001) and Paul Macdonald; and Dl (C594.9B), Elav (9F8A9), Prospero (MR1A), and 22C10, were obtained DSHB (University of Iowa). Procedures described in Sambrook and Russell (2001) and Harlow and Lane (1999) were followed for making or using the antibodies.

S2-NFull, S2-N1-2155, and S2-N1-1789 cells have been described previously (e.g., see Bardot et al., 2005). Embryos were collected from cages of yw, N55e11/FM7 actGFP, Dlx/TM3 actGF, P{neoFRT}82B P{Ubi-GFP}/TM3 Sb1P{UAS-Dl-DN}TJ1 X da-Gal4 (sorted using the GFP expression), and UAS-Ni 14E X da-Gal4 flies. Immunohistochemical staining using alkaline phosphate or horseradish peroxidase (HRP) and immunofluorescent procedures, and in situ RNA hybridization, were performed according to Lieber et al. (1993), Corbin et al. (1991), and Sullivan et al. (2000). Species-specific secondary (highly cross-adsorbed) antibodies purchased from Jackson Laboratories and Molecular Probes (Alexa Fluors) were used. The green color is from Alexa Fluor 488 secondary antibody, and the red color from Alexa Fluor 647 secondary antibody. Western blotting and immunoprecipitation procedures described in Wesley and Saez (2000) were followed. Immunohistochemical images were captured by using a Nikon microscope SMZ 1500 fitted with a Spot RT Slider camera. Confocal immunofluorescent images were captured by using the Bio-Rad MRC 1024ES Laser scanning Imaging System. HRP and alkaline phosphates stained embryos were imaged using a SMZ 1500 stereomicroscope fitted with a Spot CCD camera from glycerol loose mounts on a plain glass slide with coverslip props (so that the embryos can be rolled) and regular light reflected off a white base. All images were processed using Photoshop and Canvas programs. Any brightness/contrast adjustment was applied to the whole image or to the same level to all compared images.

Acknowledgements

We thank Drs. Toby Lieber, Simon Kidd, and Nipam Patel for antibodies; Dr. Andrew Andres for the UAS-Ni lines; Drs. Elizabeth Knust and Claire Cronmiller for the da-Gal4 lines; and the Bloomington Drosophila Stock Center for the P{neoFRT}82B P{Ubi-GFP} and P{UAS-Dl-DN}TJ1 fly stocks; Dr. Doug Taatjes and the Vermont Cancer Center Imaging Facility for assistance with confocal microscopy; Dr. Ken Irvine for comments on the manuscript; and the reviewers for very constructive comments and excellent suggestions for improvement. C.W. was supported by the National Institutes of Health (NINDS).

Ancillary