(A) Schematic representation of Notch protein, and of derivatives used below. Functional domains are drawn approximately to scale. LNR: Lin12/Notch repeats, TM: transmembrane domain, cdc10: ankyrin/cdc10 repeat domain. Asterisks in the block diagram indicate positions of Su(H) binding sites. Thick lines represent protein coding sequences present in the indicated transgenes.
(B-M) Embryos of the indicated genotypes were raised to embryonic stage 16/17 either at 25° (panels I-M) or by an appropriate temperature-shift regimen (panels B-H), fixed, incubated with anti-Fasciclin 2 to label particular axon tracts (B-H) or with anti-Elav to label neuronal nuclei (I-M) and visualized with peroxidase histochemistry. Experiments to assay axon patterning (C-H) employed the allele Notch^ts1 and used the temperature protocol described in Giniger (1998); experiments assaying lateral inhibition (J-M) employed Notch^55e11. For rescue of axonal phenotypes (D-H) Notch transgene was under control of elav-GAL4; for rescue of neurogenic phenotypes (K-M) Notch transgenes were expressed under control of sca-GAL4. (B, I) wild type embryos; (C, J) Notch^- (Notch^ts1 and Notch^55e11, respectively); (D, K) Notch^-; UAS-Notch^wt; (E, L) Notch^-; UAS-Notch^Δcdc10. While this transgene gave some rescue of longitudinal axons (see quantification in (N)), rescue was incomplete and often restored intersegmental connection of only a single fascicle (arrowhead); (F, M) Notch^-; UAS-Notch^Δ2155; (G) Notch^ts; UAS-Notch^{ΔEGF 10-12}. Expression of this transgene consistently produced more severe axonal defects than those observed in the parent Notch^ts1 stock, suggesting a dominant inhibitory activity. (H) Notch^ts; UAS-Notch^ECN. Experiments published by others have documented previously that Notch^ECN and Notch^Δ10-12 lack all function in lateral inhibition (Jacobsen et al., 1998; Lieber et al., 1993). Arrows in C and H highlight breaks in longitudinal axon tracts. Anterior is to the left in all panels; dorsal is to the top in (I-M).
(N) Histogram quantifying the fraction of missing CNS longitudinal connections observed in the experiments of (C-H). As described in the Experimental Procedures, the number of broken connections between successive thoracic and abdominal hemisegments was counted, and expressed as a percentage of the total. Expression of Notch^Δ10-12 consistently produced more disconnections than the Notch^ts1 control. Average variation in the percentage of disconnected hemisegments for a given genotype in different experiments was ± 5%. N > 400 hemisegments for all genotypes except Notch^ts1; UAS-Notch^Δ10-12, for which approximately 50% of embryos had a disrupted morphology due to the antimorphic effect of the transgene, and were excluded from the quantification.

A) Schematic of the ISNb motor nerve. Wild type ISNb motor axons defasciculate from the main ISN trunk upon contact with Delta-expressing cells at the choice point; inactivation of Notch with a temperature-sensitive mutation prevents defasciculation. Numbered ovals represent muscles along the ISNb path and in its target field.
B) Embryos of the indicated genotype were subjected to a temperature-shift protocol that produces a partial failure of ISNb defasciculation. Embryos were fixed, immunostained with anti-Fasciclin 2 antibody to label motoneurons and visualized by peroxidase histochemistry. The fraction of hemisegments in which some or all ISNb motor axons failed to turn at the choice point was quantified in filet preparations and is indicated by the horizontal bars. Thin lines indicate SEM. In all cases, expression of the indicated transgene was driven with elav-GAL4. The degree of rescue of the mutant phenotype was not significantly different among UAS-Notch^WT, UAS-Notch^Ram* and UAS-Notch ^Ram*^Δ3; rescue by UAS-Notch^RamΔA was significantly impaired (∼40% reduced; p < 0.01 (ANOVA)).

Notch^ts1 embryos were generated that did, or did not, overexpress wildtype Disabled under control of elav-GAL4. Embryos were temperature-shifted, fixed and assayed for ISNb patterning as for the experiment of Fig 3. Thick vertical bars report the percentage of hemisegments in which ISNb axons bypassed the first choice point; thin lines represent SEM.

Embryos of the indicated genotypes were prepared as for the experiment of Fig 1, except that the temperature shift protocol was refined to allow more precise quantification (see Methods). A schematic of the Notch derivatives used in this experiment is included at the top of the figure; brown asterisks (*) designate Su(H) binding sites; ‘X’ designates a missense mutation that prevents binding of Su(H) to its amino-terminal binding site; purple (#) indicates the Disabled binding site. (A-D) dorsal views of the CNS in embryos that are Notch^ts1 and carry elav-GAL4; (E-H) lateral views of the dorsal and lateral clusters of sensory neurons in four successive segments of embryos that are Notch^55e11 and carry sca-GAL4. For comparison, E’is a lower magnification view of the embryo in panel E, with a box indicating the portion included in the high-magnification panel. The UAS-Notch transgene is indicated to the left of each pair of images (A, E) Notch^-; UAS-Notch^WT; (B, F) Notch^-; UAS-Notch^RamΔA; (C, G) Notch^-; UAS-Notch^Ram*; (D, H) Notch^-; UAS-Notch ^Ram*^Δ3. Notch ^Ram*^Δ3 is a full-length Notch transgene specifically deleted for all three Su(H) binding sites, Notch^Ram* bears a missense mutation inactivating just the Su(H) binding site in the Ram domain, Notch^ΔRamA bears a deletion of the first 35 codons of the Ram domain, thus removing both the Su(H) site and the in vitro Disabled binding site in this domain. Notch ^Ram*^Δ3 and Notch^Ram* are described in Le Gall and Giniger (Le Gall and Giniger, 2004). Arrows in (B) highlight interruptions of the longitudinal axon tracts.
(I) Histogram quantifying the percent rescue of defective longitudinal connections between hemisegments for the experiment of (A - D). Rescue by full-length, wild type Notch was defined as 100%. Average variation in the percentage of disconnected hemisegments for a given genotype in different experiments was ±2%. P-values (χ²) for selected key comparisons are indicated: ** P < 0.01; *** P ≪ 0.001. N > 500 hemisegments for all genotypes.
(J) Histogram quantifying the number of dorsal sensory neurons per hemisegment in Notch^55e11 embryos rescued with the indicated Notch transgene for the experiment of (E - H). For all genotypes, SEM ≤ 1.0 neuron (thin horizontal lines), and N > 35 hemisegments.
(K) Immunoblot of total lysate of wild type embryos (lane 1) or embryos expressing the indicated Notch derivative under control of elav-GAL4 (lanes 2-5). Top: anti-Notch immunostaining. Bottom: anti-β-tubulin loading control.

Wild type embryos bearing a Notch reporter transgene, E(spl)mδ-lacZ, were double-labelled with mAb22C10 and anti-β-gal to examine the CNS pioneer neurons dMP2 and vMP2 (A-C, stage 12.0), or with anti-Fasciclin 2 and anti-β-gal to examine the ISNb pioneer, RP3 (D-F, stage 15). Merged signals are shown in A, D; separated channels in B, E (anti-neuron signal, red) and C, F (anti-β-gal, green), respectively. Growth cones are indicated with orange arrows, cell body position with yellow arrows.

Total extracts were made from wild type embryos (B) or adult heads (A, C) and subjected to immunoprecipitation with the indicated antibodies. Precipitates were separated by SDS-PAGE, immunoblotted with the indicated antibodies and visualized by ECL. Small circles indicate bands attributable to the primary antibodies. Molecular weight markers (in kilodaltons) are indicated on the right in each panel.
(A) Immunoprecipitations of wild type head extract with either anti-Notch (lane 3) or with polyclonal anti-Disabled (lane 4). Probing with anti-Trio (top panel) detects the ca. 250 Kd form of Trio in both Notch and Disabled immunoprecipitates (arrow; compare anti-Trio IP and input, lanes 1 and 2, respectively). A smaller Trio variant (ca 120 Kd, arrowhead) comigrates with a prominent antibody band, so we can not determine whether it is also present in the immunoprecipitates. A very small non-specific background of Trio was often detected in control anti-Islet immunoprecipitates (lane 5), but quantification (Image J) demonstrated that the Trio signal observed in anti-Disabled and anti-Notch IPs was more than 4x above this background level. Probing the same samples with anti-Notch (middle panel) detects full length Notch in the input and in the anti-Notch IP (lanes 2, 3) and also co-precipitated with polyclonal anti-Disabled (lane 4). Probing these samples with anti-Disabled (bottom panel) detects Disabled protein only in the input and in the anti-Disabled IP (lanes 2, 4). mAB P6E11 is not sufficiently sensitive to test for Disabled protein in anti-Notch IPs.
(B) Immunoprecipitation of wild type embryo extract with anti-Notch (lane 2) or monoclonal anti-Disabled (lane 4) followed by Western blotting with anti-Notch antibody detects full-length Notch (arrow), whereas immunoprecipitation with a control IgG (anti-Islet, lane 3) does not. Input lysate is shown for comparison (lane 1).
(C) Immunoprecipitations from adult head extract were assayed by Western blotting with anti-Notch. Notch protein (arrow) is readily detectable in anti-Trio immunoprecipitates of wild type extract (lane 4) but not in immunoprecipitates with a control antibody (anti-Islet, lane 3). Input extracts (lane 1) and anti-Notch IP (lane 2) are shown for comparison.