Insight into Notch Signaling Steps That Involve pecanex from Dominant-Modifier Screens in Drosophila

Abstract

Notch signaling plays crucial roles in intercellular communications. In Drosophila, the pecanex (pcx) gene, which encodes an evolutionarily conserved multi-pass transmembrane protein, appears to be required to activate Notch signaling in some contexts, especially during neuroblast segregation in the neuroectoderm. Although Pcx has been suggested to contribute to endoplasmic reticulum homeostasis, its functions remain unknown. Here, to elucidate these roles, we performed genetic modifier screens of pcx. We found that pcx heterozygotes lacking its maternal contribution exhibit cold-sensitive lethality, which is attributed to a reduction in Notch signaling at decreased temperatures. Using sets of deletions that uncover most of the second and third chromosomes, we identified four enhancers and two suppressors of the pcx cold-sensitive lethality. Among these, five genes encode known Notch-signaling components: big brain, Delta (Dl), neuralized (neur), Brother of Bearded A (BobA), a member of the Bearded (Brd) family, and N-ethylmaleimide-sensitive factor 2 (Nsf2). We showed that BobA suppresses Dl endocytosis during neuroblast segregation in the neuroectoderm, as Brd family genes reportedly do in the mesoderm for mesectoderm specification. Analyses of Nsf2, a key regulator of vesicular fusion, suggested a novel role in neuroblast segregation, which is distinct from Nsf2’s previously reported role in imaginal tissues. Finally, jim lovell, which encodes a potential transcription factor, may play a role in Notch signaling during neuroblast segregation. These results reveal new research avenues for Pcx functions and Notch signaling.

CELL signaling via the Notch receptor is an evolutionarily conserved mechanism that mediates local cell–cell communications required for development and homeostasis in metazoans (Artavanis-Tsakonas et al. 1999; Cau and Blader 2009; Kopan and Ilagan 2009). Notch signaling plays essential roles in many biological events, including cell-fate decisions, pattern formation, and the regulation of cell physiology (Kim et al. 1996; Palmeirim et al. 1997; Bessho and Kageyama 2003; Le Bras et al. 2011; Guruharsha et al. 2012). Therefore, defects in Notch signaling are associated with various human diseases, such as leukemia, cancers, cerebral infarction, Alagille syndrome, and multiple sclerosis (Ellisen et al. 1991; Joutel et al. 1997; Li et al. 1997; John et al. 2002; Nicolas et al. 2003). The molecular mechanisms of Notch signaling have been extensively studied, and its main process is well understood (Kopan and Ilagan 2009; Guruharsha et al. 2012). However, various additional Notch-signaling components continue to be identified through a wide range of genetic and biochemical studies (Artavanis-Tsakonas et al. 1995; Bray 2016). In many cases, the specific roles of these components remain elusive, although they are required for Notch signaling activation or regulation at least in some contexts. Understanding the functions of these factors will help elucidate the detailed mechanisms of Notch signaling, and could lead to new therapeutic strategies for Notch pathway-related human diseases.

The Notch receptor family consists of transmembrane proteins containing up to 36 epidermal growth factor-like repeats in their extracellular domain (Wharton et al. 1985). In the endoplasmic reticulum (ER) or Golgi, Notch undergoes post-translational modifications such as N-glycosylation, O-fucosylation, and O-glucosylation (Bruckner et al. 2000). The Notch extracellular domain is then cleaved by Furin (S1 cleavage), and the processed Notch fragments rebind to each other within the Golgi (Logeat et al. 1998; Kidd and Lieber 2002; Lake et al. 2009). At the cell surface, Notch binds to Delta- or Serrate-type ligands that are presented on the surface of neighboring cells (Fehon et al. 1990; Rebay et al. 1991). The extracellular domain of Notch is mechanically pulled by the ligands; this pulling is achieved by ligand endocytosis, which requires ubiquitination of the ligands by Neuralized and Mindbomb (Wang and Struhl 2005; Meloty-Kapella et al. 2012). The pulling force induces a conformational change in the Notch extracellular domain, leading to its cleavage at the juxta-membrane region by Kuzbanian/ADAM10 (S2 cleavage) (Stephenson and Avis 2012). Subsequently, the membrane-tethered Notch intracellular domain is cleaved within the transmembrane domain by γ-secretase (S3 cleavage), which releases the Notch intracellular domain from the membrane (Struhl et al. 1993; Mumm and Kopan 2000). The Notch intracellular domain is then translocated to the nucleus, where it functions as a coactivator of transcription, leading to the promoter activation of various target genes, such as Enhancer of split [E(spl)] in Drosophila (Bailey and Posakony 1995).

In Drosophila, in addition to the core components of Notch signaling, several other genes have also been identified as components of the Notch signaling pathway, although the roles of their products remain to be clarified (Hori et al. 2013). One such gene is pecanex (pcx). Embryos homozygous or hemizygous for pcx lacking its maternal contribution (pcx^mz) exhibit neural hyperplasia, termed the neurogenic phenotype (Perrimon et al. 1984); this phenotype is considered evidence for the depletion of Notch signaling (Simpson 1990). Thus, pcx is a maternal neurogenic gene. In addition, the expression of E(spl) m8 and single-minded, which are target genes of Notch signaling, is reduced in pcx^mz embryos (Yamakawa et al. 2012). Therefore, pcx is essential for Notch signaling, at least in certain contexts, such as in neuroblast segregation, which occurs through lateral inhibition in the neuroectoderm. pcx encodes an ER-resident multi-pass transmembrane protein that consists of 3433 amino acids and is highly conserved from Drosophila to humans (LaBonne et al. 1989). In addition to the neurogenic phenotype, ER enlargement is observed in a subset of cells in pcx^mz embryos (Yamakawa et al. 2012). Both the neurogenic phenotype and the ER defects of pcx^mz embryos are rescued through ectopic induction of the unfolded protein response, which promotes ER functions such as protein folding (Kaufman 1999; Yamakawa et al. 2012). These findings suggest that Pcx is required for normal functions of the ER, which are essential for activating Notch signaling (Yamakawa et al. 2012). However, the biochemical functions of Pcx in Notch signaling remain unclear.

We considered that a genetic modifier screen might be a useful approach for obtaining additional insights into the function of pcx. Genetic modifier screens have been used in various organisms including Drosophila to identify genes functioning in a common biochemical pathway (St Johnston 2002). Genetic modifiers identified by such screens can be informative for deducing the function of the affected gene. Such information might be particularly valuable in the case of pcx, given that Pcx does not carry any protein motifs that suggest its biochemical functions (Gilbert et al. 1992). Thus, in this study, we performed genetic screens to identify dominant modifiers of a cold-sensitive lethality that is associated with the pcx mutation condition. The dominant enhancers and suppressors of pcx identified from the screens provide insight into the steps of the Notch signaling cascade where pcx might be involved.

Materials and Methods

Fly strains and genetics

A standard Drosophila medium was used. The culture was maintained at 25° unless otherwise stated. Canton-S was used as the wild-type strain. The mutants used were: pcx³, a hypomorphic allele (Mohler 1977; Mohler and Carroll 1984); big brain¹ (bib¹), a null allele (Lehmann et al. 1981); Delta^5F102(Dl^5F102), a loss-of-function allele (Vӓssin and Campos-Ortega 1987); jim lovell⁴⁷ (lov⁴⁷), a hypomorphic allele lacking a 1.4-kb sequence upstream of lov (a gift from Dr. Kathleen Beckingham; Bjorum et al. 2013); neuralized¹ (neur¹), a loss-of-function allele (Leviten and Posakony 1996); and N-ethylmaleimide-sensitive factor 2^A15 (Nsf2^A15), a hypomorphic allele (Mahoney et al. 2006). Deficiency (Df) kits used for the initial screens were obtained from the Bloomington Drosophila Stock Center (BDSC) (Cook et al. 2012). Small-Df strains that uncovered smaller parts of the regions deleted in the original Dfs included in the Df kits were also obtained from BDSC. Furthermore, 30 additional small Dfs that uncovered the gap regions of the Df kits used for the initial screens were also provided by BDSC. Df(2R)gsb and Df(2R)GGG^d13 were kindly provided by Markus Noll (He and Noll 2013).

Determination of pupa and eclosed fly numbers under the cold-sensitive lethality of pcx/+^m

To characterize the cold-sensitive lethality, which provides a useful assay for pcx function in our genetic screens and in our characterization of mutants identified from the screens, we scored the numbers of pupae and eclosed flies. Five pcx homozygous or wild-type (control) virgin females were mated with five wild-type males in plastic vials containing a standard Drosophila medium. The cultures were maintained at 18, 25, and 28° for 18 days, and then the pupae and/or adult female progenies were counted. The male progenies obtained from pcx homozygous females (pcx/Y^m) are known to be embryonic lethal (Yamakawa et al. 2012). The number of female wild-type pupae was deduced by assuming that half of the total pupae were female (Figure 1B). The eclosed females were identified by their female genitalia and counted (Figure 1C and Figure 5, A–C). These experiments were performed in triplicate. For the temperature-shift analysis in Figure 1, C and F, the parental flies were maintained at 28° until third-instar larvae appeared, and then the cultures were maintained at 18°. In Figure 5, A–C, to evaluate the dominant modifiers identified in the screens, genotypes of the female adults were determined by the absence of appropriate balancer chromosomes (CyO or TM3); otherwise, the same procedures were used.

Figure 1.

pcx exhibits cold-sensitive lethality. (A) Mating scheme to analyze the cold-sensitive lethality of pcx/+^m. Five pcx homozygous females were mated with wild-type males. All females of the F1 progeny were pcx heterozygotes lacking maternal pcx (pcx/+^m) and exhibited cold-sensitive lethality. All males of the F1 progeny were pcx hemizygotes lacking maternal pcx (pcx/Y^m) and exhibited embryonic lethality. (B) Numbers of wild-type female (+/+, 50% of pupae were predicted to be female) and pcx/+^m (F1 female) pupae cultured at 18, 25, or 28°. SD are indicated. (C) Numbers of wild-type female (+/+) and pcx/+^m (F1 female) adults obtained from cultures at 18, 25, or 28° or cultures with a temperature shift from 28 to 18°. In the temperature-shift experiment, the cultures were maintained at 28° until third-instar larvae appeared, and then they were maintained at 18° for 18 days in total. SD are indicated. (D) Numbers of wild-type female (+/+) and pcx/+^m (F1 female) eggs at 25 and 28°. Sex of embryos was determined by anti-Sxl antibody staining. SD are indicated. (E) Lateral views of wild-type (+/Y) and pcx^mz (pcx/Y^m) embryos cultured at 28° and immunostained with a neuron-specific antibody (anti-Elav antibody) at stage 13–14. These embryos were identified as males based on the absence of anti-Sxl antibody staining. (F) Phenotypes of wild-type (+/+) and pcx/+^m adults cultured with the temperature shift from 28 to 18°. From top to bottom, adult females, wings, eyes, and thoraxes. (G) Percentage of total embryos cultured at 25 or 28° that were female embryos (+/+ and pcx/Y^m). The number of total embryos examined is shown as (N) at the top. * P > 0.01, ** P < 0.01.

Figure 5.

Dominant modification of the cold-sensitive lethality of pcx/+^m by pcx modifiers identified as single loci. (A) The numbers of female adult flies of wild type (+/+), pcx/+^m, heterozygotes of each enhancer (bib¹, Dl^5F102, or neur¹), and pcx/+^m combined with bib¹/+, Dl^5F102/+, or neur¹/+ at 28° are shown. Genotypes were determined by sex and marker phenotypes associated with balancer chromosomes. (B) Numbers of female adult flies of wild type (+/+), pcx/+^m, heterozygote of lov⁴⁷, and pcx/+^m combined with lov⁴⁷/+ at 28° are shown. Genotypes were determined by sex and marker phenotypes associated with balancer chromosomes. (C) Numbers of female adult flies of wild type (+/+), pcx/+^m, heterozygotes of each suppressor (BobA¹ and Nsf2^A15), or pcx/+^m combined with BobA¹/+ and Nsf2^A15/+ at 25°. Genotypes were determined by sex and marker phenotypes associated with balancer chromosomes. (D) Survival rate of wild-type (+/+) females, pcx/+^m, and heterozygous females of each enhancer (bib¹, Dl^5F102, or neur¹), and pcx/+^m combined with bib¹/+, Dl^5F102/+, or neur¹/+ to the third-instar larval stage at 28°. Percentages of surviving female third-instar larvae with the indicated genotype in the deduced total number of female embryos with that genotype are shown. (E) Survival rate of wild-type (+/+) females, pcx/+^m, heterozygous females of lov⁴⁷, and pcx/+^m combined with lov⁴⁷/+, to the third-instar larval stage at 28°. Percentages of surviving female third-instar larvae with the indicated genotype in the deduced total number of female embryos with that genotype are shown. (F) Survival rate of wild-type (+/+) females, pcx/+^m, heterozygous females of each suppressor (BobA¹ and Nsf2^A15), or pcx/+^m combined with BobA¹/+ and Nsf2^A15/+, to the third-instar larval stage at 25°. Percentages of surviving female third-instar larvae with the indicated genotype in the deduced total number of female embryos with that genotype are shown. In (D–F), genotypes were determined by sex and marker phenotypes associated with balancer chromosomes at the third-instar larval stage. In (D–F), for the calculation, the numbers of embryos with the indicated genotypes were predicted based on Mendel’s laws, and half of the embryos were assumed to be female. In (D–F), the number of female embryos examined is shown as (N) at the top of each graph. We rounded the deduced N down. All experiments were performed in triplicate. In (A–C), SD are indicated. In (D–F), * P > 0.01, ** P < 0.01.

Measurement of egg numbers

The number of eggs laid by wild-type and pcx homozygous females was counted (Figure 1D). Five pcx homozygous or wild-type virgin females were mated with five wild-type males in a 50-ml tube containing a glass plate with grape agar and yeast paste. Eggs were collected for 48 hr, with four replacements of the glass plate, at 25 or 28°.

Determination of female embryo percentage

The percentage of female embryos was analyzed (Figure 1G). Five pcx homozygous or wild-type virgin females were mated with five wild-type males in a 50-ml tube containing a glass plate with grape agar and yeast paste. Eggs were collected for 48 hr with four replacements of the glass plate, at 25 or 28°. The vitellin membrane of the eggs was removed by heptane-methanol emulsion. Remaining vitellin membrane was removed from embryos with a tungsten needle after placing the embryos on a piece of double-sided tape covered with PBT (137 mM NaCl, 10 mM Na₂HPO₄, 2.7 mM KCl, 1.8 mM KH₂PO₄, 10% TritonX100, pH 7.4). The embryos were then stained with a female-specific anti-Sexlethal (Sxl) antibody (Bopp et al. 1991), and the percentage of female embryos was determined. Experiments were performed in triplicate. Statistical significance was determined by the chi-square test.

Measurement of hatching rate

The percentage of female eggs from which first-instar larvae successfully hatched was determined as the hatching rate (Figure 2A). Five pcx homozygous or wild-type virgin females were mated with five wild-type males in a 50-ml tube containing a glass plate with grape agar and yeast paste. Eggs were collected for 48 hr with four replacements of the glass plate, and then incubated for another 3 days, at 25 or 28°. The number of female eggs was deduced by assuming that half of the total number of eggs were female. The number of wild-type female larvae was deduced using the same assumption. All of the first-instar larvae obtained from pcx homozygous females were female (pcx/+^m).

Figure 2.

Survival rate of pcx/+^m is decreased at lower temperature. (A) Hatching rate of wild-type (+/+) and pcx/+^m females by the first-instar larval stage cultured at 25 or 28°. Percentages of total female embryos that became hatched female first-instar larvae are shown. For the calculation, half of the total embryos and hatched first-instar larvae were assumed to be female. The number of female embryos examined is shown as (N) at the top. (B) Survival rate of wild-type (+/+) and pcx/+^m females to the third-instar larval stage, cultured at 25 or 28°. Percentages of total female embryos that became surviving female third-instar larvae are shown. For the calculation, half of the total embryos were assumed to be female. The number of female embryos examined is shown as (N) at the top. (C) Pupation rate of wild-type (+/+) and pcx/+^m females cultured at 25 or 28°. The percentages of female third instar larvae that became pupae are shown. The number of female larvae examined is shown as (N) at the top. (D) Eclosion rate of wild type (+/+) females and pcx/+^m cultured at 25 or 28°. Percentages of pupae (sex was determined at the third-instar larval stage) that successfully eclosed as adults are shown. The number of female pupae examined is shown as (N) at the top. In (A and B), we rounded the deduced N down. All experiments were performed in triplicate. * P > 0.01, ** P < 0.01.

Measurement of the rate of survival to the third-instar larval stage

In Figure 2B and Figure 5, D–F, the rate of survival to the third-instar larval stage was determined. The rate of survival to the third-instar larval stage is the number of female third-instar larvae with the indicated genotype divided by the deduced total number of female embryos with that genotype (×100). Five pcx homozygous or wild-type virgin females were mated with five wild-type males in a 50-ml tube containing a glass plate with grape agar and yeast paste. Eggs were collected as described above and the total eggs were counted. The number of eggs with each genotype was estimated based on Mendel’s laws and the sex ratio. All of the hatched larvae were transferred to the standard medium and cultured at 25 or 28°. Thereafter, the sex of all larvae was determined by observing the gonad size at the third-instar larval stage. In addition, the presence of dominant modifiers in heterozygotes was determined by the absence of appropriate balancer chromosomes, judged by β-galactosidase (LacZ) staining (Kozlova et al. 1998). Experiments were performed in triplicate. The statistical significance of the survival rate to the third instar larval stage was examined by the chi-square test.

Measurement of pupation and eclosion rates

To analyze the lethality of pcx/+^m females from the third-instar larval to pupal stages, we calculated the “pupation rate,” which is the number of pupating females divided by the number of collected third-instar larvae (×100). The sex was determined at this stage based on the gonad size (Figure 2C). The pcx/+^m and wild-type larvae were collected as described above.

To determine the lethality of pcx/+^m females during eclosion, we also calculated the “eclosion rate,” which was the number of eclosed females divided by the number of pupae (×100) (Figure 2D). The sex of pupae was determined at the third-instar larval stage, as described above. Experiments were performed in triplicate. The statistical significance of the pupation and eclosion rates was examined by the chi-square test.

Dominant modifier screens based on the cold-sensitive phenotype of pcx/+^m

Df kits of the second and third Drosophila chromosomes were used for the dominant modifier screens (Cook et al. 2012) (Supplemental Material, Table S1). Together, the Df kits cover ∼98% of the genome in each chromosome (Cook et al. 2012). In addition, to fill in the gaps of the Df kits, we picked 30 available small Dfs located in the gaps of the primary Dfs (Table S2) and subjected them to the same screens. For the initial screens, five pcx homozygous virgin females were mated with five wild-type (control) or Df/Balancer males and cultured at 25 or 28°. At 25°, the Dfs were evaluated as positive when the subsequent generations without Balancers (pcx/+; Df/+ embryos lacking maternal pcx) eclosed (>10 adults), while the control progenies (pcx/+; +/+ lacking maternal pcx) were virtually embryonic lethal (no adults) under the same conditions. These deficiencies represented dominant suppressors of the pcx/+^m lethality, and potential genes encoding negative regulators of pcx might exist in the genomic regions uncovered therein. At 28°, the Dfs were evaluated as positive if pcx/+; Df/+ flies lacking maternal pcx failed to eclose (no adults), while the control progenies (pcx/+; +/+ lacking maternal pcx) developed to adulthood (>10 adults). These deficiencies represented dominant enhancers of the lethality, and potential genes positively acting in conjunction with pcx might exist in the respective uncovered genomic regions.

To identify the dominant modifiers as single loci, smaller deficiencies that uncovered smaller parts of the genomic regions deleted in the Dfs identified in the first screens were subjected to the same analyses. After defining the minimal genomic regions that encompassed the loci responsible for the dominant modifications of the pcx lethal phenotypes, we searched for genes known to be involved in Notch signaling within these minimal regions. This mapping analysis using Dfs suggested that the Brother of bearded A (BobA) gene was a dominant suppressor; however, mutants affecting only the BobA locus were not available. Therefore, in this study, we generated a deletion mutant of BobA, BobA¹, as described below. For the lov gene, analyses using small Dfs defined the lov locus as a potential enhancer of pcx, and this possibility was confirmed using a preexisting lov mutant allele (Bjorum et al. 2013).

Mutagenesis of the BobA locus

A deletion mutant of BobA was generated using the FLP-FRT recombination method (Parks et al. 2004). Males carrying a P-element (PBac{RB}e00913) inserted in the 455-bp region upstream of Twin of m4 (Tom) on the third chromosome were crossed with females carrying a FLP recombinase transgene (Thibault et al. 2004). Male progeny carrying both the P-element and FLP recombinase transgene were then mated with females carrying another P-element (PBac{WH}BobA^f03024) inserted in the 3′ untranslated region of the BobA gene (Thibault et al. 2004) (see Figure S7). After 2 days, the parents and progeny were transferred into a 37° water bath and incubated for 1 hr. The cultures were then incubated at 25°, and the parents were removed after 72 hr. Subsequently, the progenies were subjected to heat-shock treatments at 37° for 1 hr once a day for 4 days. The virgin females of the progeny were then collected and mated with males containing a balancer, TM6B, which carries the lacZ gene, to establish stocks for further analysis.

Sequence analysis of the BobA¹ mutant allele

We established 14 potential lines of BobA mutants. Mutation of the BobA locus was confirmed by DNA sequence analysis. All polymerase chain reaction (PCR) assays were performed using purified genomic DNA from five embryos homozygous for each potential BobA mutant line. These homozygous embryos were selected based on the absence of X-gal staining (lacZ-negative) using a standard protocol (Ashburner 1989). Genomic DNA was extracted using a standard protocol (Parks et al. 2004). PCR primers flanking the two P-elements were designed (Forward primer 1, 5′-ACA GGG AGC ACA CAA TAA CGA TC-3′; Reverse primer 1, 5′-CAG AAG GGG TTT GAC CAG TTT C-3′). PCR was carried out using TaKaRa Prime Script GXL and a rapid protocol according to the manufacturer’s instructions: 10 sec at 98°, 15 sec at 55°, and 20 sec at 68° (40 cycles). We isolated three lines that carried a deletion uncovering the BobA genomic region, based on the size of the PCR product. Unincorporated primers, nucleotides, and salts were removed using a Wizard SV Gel and PCR Clean-Up System (Promega). Then, 4 μl of each eluted PCR product was added to 1 μl of Big Dye Terminator v3.1 sequencing cocktail (Applied Biosystems) containing the following sequencing primers (Reverse primer 1 and Forward primer 2, 5′-ACA GAA AAC TGC CGA ACC AA-3′). After 25 cycles of PCR (10 sec at 96°, 5 sec at 50°, and 4 min at 60°), the products were separated using an Applied Biosystems 3130xl sequencer. The results revealed that we obtained a BobA¹ allele in which the entire BobA gene and a noncoding RNA gene, CR45978, were deleted.

Immunohistochemistry

Antibody staining of embryos was performed as previously described (Rhyu et al. 1994). The following primary antibodies were used: rat anti-Elav [7E8A10, 1:500; NIH Developmental Studies Hybridoma Bank (DSHB)], mouse anti-Sxl (M18, 1:10; DSHB), mouse anti-Dl (C594.9B, 1:500; DSHB), guinea pig anti-Dl (gp582, 1:100, a gift from S. Artavanis-Tsakonas), chicken anti-β-galactosidase (ab9361, 1:500; Abcam), rabbit anti-Rab11 (1:5000, a gift from A. Nakamura; Tanaka and Nakamura 2008), and rabbit anti-Rhomboid (1:500, a gift from E. Bier; Sturtevant et al. 1996) antibodies. The following secondary antibodies were used: Cy3-conjugated donkey anti-mouse (1:500; Jackson ImmunoResearch), Cy3-conjugated donkey anti-guinea pig (1:500; Jackson ImmunoResearch), Alexa488-conjugated donkey anti-rabbit (1:500; Jackson ImmunoResearch), Alexa488-conjugated donkey anti-rat (1:500; Jackson ImmunoResearch), Cy5-conjugated donkey anti-chick (1:500; Jackson ImmunoResearch), and Cy5-conjugated goat anti-mouse (1:500; Jackson ImmunoResearch) antibodies. Confocal microscopy images were captured using an LSM 700 (Zeiss) and analyzed on an LSM Image Browser, ZEN2010 (Zeiss).

Data availability

Strains are available upon request. File S1 contains all supplemental figures, tables and reagent table. Tables S1 and S2 contains lists of all D. melanogaster strains used for this screen. Supplemental material available at Figshare: https://doi.org/10.6084/m9.figshare.6395831

Results

pcx exhibits a cold-sensitive lethal phenotype

pcx is a maternal neurogenic gene, given that embryos homozygous or hemizygous for pcx without a maternal contribution of pcx (referred to as pcx^mz in this study), exhibit the neurogenic phenotype, which is a benchmark implicating a deficiency in Notch signaling (Perrimon et al. 1984; LaBonne et al. 1989). Furthermore, the maternal neurogenic phenotype can be paternally rescued, as the introduction of a paternal pcx locus into pcx heterozygotes lacking maternal pcx results in a much weaker neurogenic phenotype than that of pcx^mz (pcx/Y^m in males) at 25° (optimal conditions for Drosophila) (LaBonne et al. 1989; Yamakawa et al. 2012) (Figure 1A). Here, we refer to the heterozygote for pcx lacking maternal pcx (but containing paternal pcx) as “pcx/+^m.” However, in a previous study, we discovered that pcx/+^m is lethal at 25° in a standard Drosophila medium (Yamakawa et al. 2012). To analyze the lethality of pcx/+^m further, in this study, we mated five pcx homozygous or wild-type virgin females with five wild-type males at 25°, and maintained the cultures at 25° (Figure 1A). Wild type showed normal development to pupae and adults at this temperature (Figure 1, B and C). However, only a small fraction of pcx/+^m individuals reached the pupal stage, and none reached adulthood, under the same condition (Figure 1, B and C). Thus, the lethal phenotype associated with pcx^mz was not paternally rescued at 25°.

For some mutants of genes, including those encoding Notch-signaling components, the associated phenotypes are sensitive to a change in temperature (Leonardi et al. 2011; Shimizu et al. 2014; Ishio et al. 2015). The heat- or cold-sensitive properties of such mutations may be useful for performing genetic analyses, including genetic modifier screens, because the gene activity can be controlled by temperature shifts to reveal the phenotype (O’Connell et al. 1998; Dellinger et al. 2000; Lindsay et al. 2008). Thus, in the current study, we examined the lethal phenotypes of pcx/+^m at different temperatures. As performed at 25°, we mated five pcx homozygous or wild-type virgin females with five wild-type males at 18 and 28° (Figure 1A). The cultures were maintained at each temperature, and the average number of female pupae and adults obtained from each culture vial was counted (Figure 1, B and C). We found that pcx/+^m developed to pupae and adults at 28° but not at 18°, while wild type reached the adult stage at both temperatures (Figure 1, B and C). In particular, the numbers of pcx/+^m pupae and adults increased markedly at 28°, compared with 25°. This difference was not attributed to temperature-dependent changes in the number of eggs laid by females of parental crosses at these two temperature, because we did not observe a marked difference in the numbers of eggs laid by wild-type and pcx homozygous females at 25° vs. 28° (Figure 1D). pcx/Y^m is reported to exhibit a neurogenic phenotype and embryonic lethality at 25° (Yamakawa, et al. 2012). However, because we did not determine the sex in Figure 1B it was possible that a portion of the pcx/Y^m embryos became viable at 28° until the pupal stage, to explain the increase in pupae observed. Therefore, to confirm that the pcx/Y^m embryos were embryonic lethal at 28°, we examined whether all of the male embryos cultured at this temperature showed a neurogenic phenotype. Male embryos (pcx/Y^m) were sorted out based on the absence of anti-Sexlethal (Sxl) antibody staining, and their neuronal cells were detected by anti-Elav antibody staining (Figure 1E) (Bopp et al. 1991). The repetitive structure of the ladder-like nervous system was observed in wild-type embryos (Figure 1E, +/Y), while the pcx/Y^m embryos exhibited a severe neurogenic phenotype in all cases examined at 28° (Figure 1E, pcx/Y^m). Thus, we concluded that pcx/Y^m males did not contribute to the observed change in the lethality, and that pcx/+^m exhibited a cold-sensitive lethal phenotype.

Given that pcx/+^m exhibited a clear cold-sensitive lethal phenotype, a change in the sex ratio could also have influenced the number of pupae and adults in our experiments. However, we found that the females of parental crosses laid almost the same percentage of female eggs (+/+ and pcx/+^m, respectively) at 25 and 28° (Figure 1G). In addition, the percentage of pcx/+^m females in the total embryos (pcx/+^m plus pcx/Y^m) laid by pcx homozygote mothers was almost the same at 25 and 28° (Figure 1G). Therefore, the sex ratio of the eggs laid by pcx homozygous females did not show a significant difference at these two temperatures. These results excluded the possibility that a change in the sex ratio contributed to the cold-sensitive lethality.

To determine the developmental stage responsible for the cold-sensitive lethality, we next compared the survival rate of pcx/+^m and wild type between these two temperatures at different developmental stages. First, we analyzed the hatching rates of wild-type and pcx/+^m embryos. To obtain the hatching rate, the number of female first-instar larvae (+/+ or pcx/+^m) was divided by the number of total female embryos, which was deduced by assuming that half of the total embryos were female. For the wild-type control, the number of female first-instar larvae was also deduced from the sex ratio. We found that the hatching rate of pcx/+^m was significantly increased at 28° compared with 25° (Figure 2A). In contrast, the hatching rate of wild type was slightly reduced at 28° compared with 25°, probably due to the nonoptimal temperature (Figure 2A). Therefore, we concluded that pcx/+^m shows cold-sensitive lethality during its embryonic stage. We also determined the rate of survival to the third-instar larval stage (Figure 2B). We calculated the percentage of deduced female wild-type and pcx/+^m embryos based on the sex ratio that reached the third-instar larval stage. The survival rate of pcx/+^m to the third-instar larval stage was significantly greater at 28° than at 25° (Figure 2B). Therefore, the cold-sensitive lethality of pcx/+^m could also be detected as the rate of survival to the third-instar larval stage, which was similar to the cold-sensitive hatching rate of the pcx/+^m embryos. Thus, the survival rate to the third-instar larval stage could be used to analyze the cold-sensitive lethality of pcx/+^m in combination with other mutants in the latter part of this study, which was useful because the sex and genotypes could be determined more easily at this stage, compared with those of first-instar larvae.

We also examined whether other developmental stages, such as pupation and eclosion, were involved in the cold-sensitivity of pcx/+^m lethality. To determine the pupation (percentage of third-instar larvae that became pupae) and eclosion (percentage of pupae that successfully eclosed) rates, we counted the female larvae at the third instar (pcx/+^m and +/+ in each experiment) based on gonad size and cultured them until pupation and eclosion. However, the pupation rates of pcx/+^m and wild-type females were not markedly different at 25° vs. 28° (Figure 2C). Therefore, once pcx/+^m embryos reached the third-instar larval stage, most of them developed to the pupal stage. We found that the eclosion rate of the pcx/+^m pupae was significantly greater at 28° than at 25° (Figure 2D). However, the eclosion rate of pcx/+^m vs. wild-type was not significantly different at 28°, probably because the eclosion rate of wild type was slightly lower at 28° than at 25° (Figure 2D). Thus, although the eclosion rate has some influence on the cold-sensitive lethality of pcx/+^m, it may not have a major effect. Collectively, these results showed that the cold-sensitive lethality of pcx/+^m was mostly associated with a reduction in the hatching rate.

The pcx/+^m that were cultured continuously at 28° reached adulthood, and these adult flies did not exhibit detectable defects in their overall morphology (data not shown). Thus, to detect the potential requirement of pcx during the pupal development, we also performed a stage-specific temperature-shift experiment from 28 to 25° during the larval and pupal stages. Such a temperature shift could cause some developmental defects in adulthood, because pcx/+^m individuals showed a slightly reduced eclosion rate at 25° compared to 28°, which suggested potential defects in their pupal development. We cultured pcx/+^m at 28° until the third-instar larval stage, then shifted the culture temperature to 18°. However, we did not observe a marked reduction in the number of eclosed flies under this temperature-shift condition (Figure 1C). Furthermore, the adult flies that experienced the temperature shift did not exhibit detectable defects, in particular those suggesting a reduction in Notch signaling, such as wing notch, wing-vein thickening, or rough eye (Figure 1F) (Shellenbarger and Mohler 1978; De Celis and García-Bellido 1994; Go et al. 1998). Thus, the cold-sensitive lethality of pcx/+^m did not appear to be directly attributable to a decline in Notch signaling after the third-instar larval stage. This idea is consistent with our observation that the cold-sensitive lethality of pcx/+^m was mostly observed as a reduction in the hatching rate (Figure 2A).

The cold-sensitive phenotypes of pcx is due to a reduction in Notch signaling at lower temperatures

We next examined whether the cold-sensitive lethality of pcx/+^m was caused by a further reduction in Notch signaling at the lower temperature during embryogenesis. To test this possibility, we examined whether the cold-sensitive lethality of pcx/+^m embryos coincided with the extent of the neurogenic phenotype, detected by anti-Elav antibody staining, even though the neurogenic phenotype of pcx/+^m embryos was less prominent owing to paternal rescue (weak neurogenic phenotype), compared with that of pcx^mz embryos (strong neurogenic phenotype) at 25° (Figure 3, A–D) (Yamakawa et al. 2012). Wild-type embryos exhibited a normal central nervous system (CNS) and peripheral nervous system (PNS) at stage 14 (100%, n = 47) (Figure 3A). As previously reported, a strong neurogenic phenotype, characterized by neural hyperplasia of the CNS and PNS, was observed in the pcx^mz (pcx/Y^m) embryos, in which the zygotic and maternal pcx function was absent (100%, n = 19) at 25° (Figure 3B) (Yamakawa et al. 2012). We also observed the paternal rescue of the neurogenic phenotype in pcx/+^m embryos, in which the strong neurogenic phenotype was suppressed in all cases examined at 18° (n = 47), 20° (n = 52), 22° (n = 102), 25° (n = 44), 28° (n = 51), and 30° (n = 34), and the strong neurogenic phenotype was not observed at all (Figure 3, C–E). Among the paternally rescued embryos, we identified two classes of phenotypes in the embryonic nervous system: one exhibited a mostly normal nervous system as determined by its overview structure (quenched neurogenic phenotype) (Figure 3, C and E), whereas the other still showed a weak neurogenic phenotype, especially in the CNS, although the neural hyperplasia of the PNS was effectively suppressed (Figure 3, D and E). We observed that the percentage of embryos classified as weak neurogenic increased, and the percentage of quenched neurogenic embryos decreased, as the culture temperature was decreased (Figure 3E). The percentage of embryos classified as weak neurogenic was 80.9, 69.2, 68.6, 52.3, 31.4, and 14.7% at 18, 20, 22, 25, 28, and 30°, respectively (Figure 3E). These results suggested that the Notch signaling activity in pcx/+^m embryos decreased with the decline in temperature. Therefore, we concluded that the cold-sensitive lethality could be attributed, at least in part, to a reduction in Notch signaling in pcx/+^m embryos.

Figure 3.

pcx/+^m exhibits a cold-sensitive neurogenic phenotype. (A–D) Lateral views of embryos cultured at 25° and immunostained with a neuron-specific antibody (anti-Elav antibody) at stage 13–14. (A) Wild type (+/+). (B) pcx^mz (pcx/Y^m) embryo showing a strong neurogenic phenotype (100%, n = 19) (“Strong”). (C) pcx/+^m embryo showing a mostly normal nervous system structure (quenched neurogenic phenotype) (47.7%, n = 44) (“Quenched”). (D) pcx/+^m embryo showing a weak neurogenic phenotype (52.3%, n = 44) (“Weak”). In (A–D), Bar, 50 μm. (E) Percentage of pcx/+^m embryos showing a mostly normal nervous system (quenched, green) and a weak neurogenic phenotype (weak, yellow) at the indicated temperatures. Number of embryos examined in each experiment (N) is indicated at the top.

Genetic screens to identify mutations that dominantly modify the cold-sensitive phenotype of pcx

Using the cold-sensitive phenotype of pcx/+^m, we next conducted dominant modifier screens. First, we screened for dominant enhancers of the cold-sensitive lethality (Figure 4A). At 28°, pcx/+^m animals survived until adulthood, whereas this genotype was virtually lethal before adulthood at 25° (Figure 1, B and C). Therefore, we searched for second-site mutations that ameliorated the viability of pcx/+^m adults, thereby resulting in lethality prior to adulthood in a dominant manner at 28°. Second, we screened for dominant suppressors of the lethality that was observed in pcx/+^m at 25° (Figure 1B). In this screen, we looked for second-site mutations that resulted in the emergence of pcx/+^m adult flies at this temperature. In both screens, we used Df kits that carried chromosomal deletions of the second and third chromosomes on a uniform genetic background (see Materials and Methods and Figure 4A). We screened 189 and 179 deficiency lines on the second and third chromosomes, respectively (Table S1). These screens covered ∼98% of the Drosophila genome (Cook et al. 2012).

Figure 4.

Genetic screen procedure and summary of the dominant modifiers of pcx. (A) Five pcx homozygous females were mated with five males of each Deficiency (Df) mutant on the second or third chromosome at 25 or 28°. The genotype of F1 females without balancers was identified as pcx/+^m; Df/+. After 18 days, the numbers of F1 progeny were counted. pcx/+^m individuals survived until adulthood at 28°. In the enhancer screen, the Df lines that showed reduced viability at 28° (i.e., those in which F1 did not survive until adulthood) were identified. pcx/+^m individuals were lethal by the pupal stage at 25°. In the suppressor screen, Df lines that suppressed the lethality at 25° (i.e., those in which F1 survived to adulthood) were identified. (B) Tables summarizing the dominant enhancers [28° lethal (Enhancer)] and suppressors [25° viable (Suppressor)] of pcx/+^m. Df lines (Symbol No.) identified as dominant modifiers of pcx/+^m in the first screens and the deleted genomic region (Deleted region) in each Df line are listed in the left columns (Df). The small Df lines (Symbol No.) uncovering smaller parts of the original Dfs and containing the potential loci of the modifiers are listed in the middle columns (Small Df) with the deleted genomic regions (Deleted region). Dominant modifiers identified as single genes are indicated in the right columns (Tested allele) with information about the alleles (Allele name), molecular functions of the gene products (Molecular function), and potential associated cellular processes (Cellular process). “–” indicates inapplicable.

Identification of dominant enhancers of pcx/+^m

In our dominant enhancer screen, we identified seven deficiencies that enhanced the cold-sensitive lethality of pcx/+^m in a dominant manner at 28° (Figure 4B). To identify the genetic loci responsible for the dominant enhancement, we narrowed down the genomic regions using smaller Dfs that uncovered smaller parts of the deleted regions in the seven positive Dfs identified in our first screen (Figure 4B and Figures S1–S6 and S8). Df(2L)BSC50 and Df(2L)BSC689 were positive Dfs carrying deletions that partially overlapped (Figure 4B and Figure S1). Among the smaller Dfs uncovering the genomic regions deleted in Df(2L)BSC50 and Df(2L)BSC689, we found that Df(2L)BSC205 and Df(2L)BSC251, but not other Dfs, enhanced the pcx/+^m lethality (Figure 4B and Figure S1). The genomic deletions in Df(2L)BSC205 and Df(2L)BSC251 partially overlapped, and the bib locus exists in the overlapping region (Figure S1). bib is a classic neurogenic gene that encodes a component of the Notch signaling pathway (Lehmann et al. 1983; Doherty et al. 1997). The deduced product of bib has high homology to mammalian aquaporin-4, although a water channel activity of Bib was not detected in previous studies (Kaufmann et al. 2005; Park et al. 2008). In the current study, we found that bib¹, an existing allele of bib, dominantly enhanced the cold-sensitive lethality of pcx/+^m (Figure 5A, pcx/+^m; bib¹/+). However, the heterozygote for bib¹ showed normal development to adulthood at 28°, demonstrating a synergistic interaction between pcx and bib (Figure 5A, +/+; bib¹/+). From these results, bib was identified as a dominant enhancer of pcx.

Using the same approach, we identified the genetic loci that were uncovered in each positive deficiency and responsible for dominant enhancement of the cold-sensitive lethality of pcx/+^m (Figures S2–S6 and S8). Among the small deficiencies overlapping Df(3R)ED5938, we found that Df(3R)BSC850 and Df(3R)ED5942, which partially overlapped each other, enhanced the lethality of pcx/+^m (Figure 4B and Figure S2). Notably, the overlapping region contained the Dl locus (Figure S2). Dl encodes a ligand for the Notch receptor (Vӓssin et al. 1987; Kopczynski et al. 1988). We found that Dl^5F102, a preexisting allele of Dl, dominantly enhanced the lethality of pcx/+^m (Figure 4B and Figure 5A, pcx/+^m; Dl^5F102/+). In contrast, heterozygotes for Dl^5F102 developed normally to adulthood at 28° (Figure 5A, +/+; Dl^5F102/+). Thus, we identified Dl as another dominant enhancer of pcx. Another of the seven deficiency lines identified by our screens was Df(3R)ED5330, which contained the neur locus within its deleted genomic region (Figure 4B and Figure S3). We found that neur¹, a preexisting allele of neur, dominantly enhanced the lethality of pcx/+^m (pcx/+^m; neur¹/+), although neur¹/+ flies developed normally to adulthood at 28° (Figure 4B and Figure 5A). Thus, neur was also identified as a dominant enhancer of pcx. Thus, our enhancer screen of the cold-sensitive lethality of pcx/+^m at 28° identified three genes, bib, Dl, and neur, all of which encode known components of the Notch signaling pathway, as dominant enhancers of pcx.

On the other hand, in the genomic regions deleted in Df(2R)Kr10, which was also identified as dominant enhancer of pcx, we failed to find a gene that was known to be involved in Notch signaling (Figure 4B and Figure S4). However, it was possible that gene(s) encoding novel component(s) of the Notch signaling pathway exist in these regions, given that three genes involved in Notch signaling were successfully identified by our enhancer screen as described above. Therefore, using small Dfs, we attempted to narrow down the genomic region containing the dominant enhancer to single loci. The genomic deletion in Df(2R)Kr10 partially overlapped with two small Dfs, Df(2R)GGG^d13 and Df(2R)gsb (Figure 4B and Figure S4). We found that Df(2R)gsb, which overlapped with the lov locus, enhanced the cold-sensitive lethality of pcx/+^m at 28°, but Df(2R)GGG^d13, which did not overlap with the lov locus, did not (Figure S4). These results collectively suggested that lov may be responsible for the dominant enhancement of lethality (Figure S4). We found that an available lov allele, lov⁴⁷, dominantly enhanced the lethality of pcx/+^m at 28° (Figure 5B, pcx/+^m; lov⁴⁷/+). However, lov⁴⁷/+ flies developed normally to adulthood at 28° (Figure 5B, +/+; lov⁴⁷/+). Thus, we concluded that lov is also a dominant enhancer of pcx, and demonstrated for the first time that lov may be involved in the Notch signaling pathway. Finally, we also identified two partially overlapping Dfs, Df(2R)X58-12 and Df(2R)X58-8, from the dominant enhancer screen (Figure 4B and Figure S5). In this case, we failed to identify a preexisting mutation(s) that enhanced the lethality of pcx/+^m in the overlapping region of these two Dfs, although we excluded CG3927, Vacuolar protein sorting 20 (Vps20), bonsai, and Ribosomal protein 624 (Rp624) from the candidates for the modifier gene (Figure S5). Therefore, these Dfs were not analyzed further in this study.

We found that the cold-sensitive lethality of pcx/+^m could be easily assessed as a reduction in the rate of survival to the third-instar larval stage (Figure 2B). Thus, we evaluated the effect of these dominant enhancers on the viability of pcx/+^m based on the survival rate at this stage. We found that all four enhancers, bib, Dl, neur, and lov, significantly reduced the survival rates in a dominant manner, when each was combined with pcx/+^m at 28° (Figure 5, D and E). However, bib¹/+ and neur¹/+ did not show significantly different survival rates from wild type at 28° (Figure 5D). In contrast, the survival rates of Dl^5F102/+ and lov⁴⁷/+ were reduced to about half that of wild type (Figure 5, D and E). However, the survival rates observed in these heterozygous embryos were not as low as those of embryos with these enhancers combined with pcx/+^m (Figure 5, D and E). Therefore, we concluded that bib, Dl, neur, and lov dominantly enhanced the lethality of pcx/+^m from the third-instar larval stage to the embryonic stage.

Identification of dominant suppressors of pcx/+^m

At 25°, pcx/+^m failed to develop to adulthood. We next screened for Dfs that suppressed the cold-sensitive lethality of pcx/+^m in a dominant manner at 25° (Figure 4A and Table S1), and identified Df(3L)ED217. The genomic region deleted in Df(3L)ED217 was partially uncovered in six small deficiencies, Df(3L)BSC441, Df(3L)Exe16125, Df(3L)Exe16126, Df(3L)BSC837, Df(3L)ED218, and Df(3L)Exe16262 (Figure 4B and Figure S6). We found that among these six small Dfs, only Df(3L)BSC441 dominantly suppressed the lethality of pcx/+^m at 25° (Figure 4B and Figure S6). Five genes, CG17839, CG34245, DCX-EMAP, CG13465, and BobA existed in the genomic region encompassed by Df(3L)BSC441, which contained the locus responsible for the dominant suppression. Among these, DCX-EMAP was the only gene with an available mutant (Bechstedt et al. 2010). However, DCX-EMAP did not suppress the lethality, which excluded this gene as the dominant suppressor. To identify the dominant suppressor as a single locus, we next generated a BobA mutant allele, because BobA is a member of the Bearded (Brd) family, which is known to negatively regulate Notch signaling (Lai et al. 2000). Brd family members bind to and inhibit Neur, an E3 ubiquitin ligase that ubiquitinates Dl (Bardin and Schweisguth 2006). The ubiquitination of Dl is required for endocytosis of Dl, which facilitates the signaling activity of Dl (Lai et al. 2001; Le Borgne et al. 2005). However, the specific functions of BobA have remained unknown, because a mutation affecting only the BobA locus has not been available. In addition, the previous study was based on deletion mutants uncovering multiple Brd family genes, including BobA (Bardin and Schweisguth 2006). Thus, in the current study, we generated a mutant allele of BobA, designated as BobA¹, using the FLP-FRT recombination method (Figure S7). Thus, BobA¹ is the first mutant in which a single Brd family gene was disrupted. We found that BobA¹ dominantly suppressed the lethality of pcx/+^m at 25° (Figure 4B and Figure 5C, pcx/+^m; BobA¹/+). In addition, a gene encoding a noncoding RNA, CR45978, was also deleted in BobA¹ (see Figure S7) (Young et al. 2012). However, it is unlikely that CR45978 is the dominant suppressor of pcx/+^m lethality, because CR45978 expression has only been detected in adult males (Young et al. 2012). Thus, our results suggested that BobA is a dominant suppressor of pcx.

Notably, the Df kits used in our screens have small gaps in the coverage of deleted genomic regions, corresponding to ∼2% of the chromosomes (Cook et al. 2012). Thus, the genes located within these gaps had escaped from our screens. To minimize this occurrence, we examined an additional 30 preexisting small Dfs that uncovered the gap regions on the second and third chromosomes (Table S2). Among these, Df(3R)ED5662 dominantly suppressed the cold-sensitive lethality of pcx/+^m at 25° (Figure 4B and Table S2). We found that this Df uncovered the N-ethylmaleimide-sensitive factor 2 (Nsf2) gene, which was previously identified as a modifier of Notch signaling (Figure S8) (Stewart et al. 2001). We here showed that a preexisting allele of Nsf2, Nsf2^A15 dominantly suppressed the cold-sensitive lethality of pcx/+^m at 25° (Figure 4B and Figure 5C, pcx/+^m; Nsf2^A15/+). This result identified Nsf2 as a dominant suppressor of pcx.

Based on the above findings, we speculated that the dominant suppression of the cold-sensitive lethality of pcx/+^m might be due to an increase in the survival rate to the third-instar larval stage. We found that BobA¹ and Nsf2^A15 significantly increased this survival rate in a dominant manner when combined with pcx/+^m at 25° (Figure 5F, pcx/+^m; BobA¹/+ and pcx/+^m; Nsf2^A15/+, respectively). BobA¹/+ by itself showed a slightly decreased survival rate (Figure 5F, +/+; BobA¹/+). However, we might be able to ignore this effect, because the number of eclosed BobA¹/+ flies was comparable to that of wild type at 25° (Figure 5C). These results suggested that BobA¹ and Nsf2^A15 dominantly suppressed the lethality of pcx/+^m from the embryonic to the third-instar larval stage.

pcx modifiers influence Notch signaling activity

Our genetic screens identified four dominant enhancers and two dominant suppressors of the cold-sensitive lethality of pcx/+^m. As described above, in pcx/+^m, we observed a correlation between the extent of cold-sensitive lethality and the neurogenic phenotype, which is an indicator of reduced Notch signaling activity (Simpson 1990). In addition, we found that the dominant enhancers and suppressors reduced and increased the survival rate by the third-instar larval stage, respectively, suggesting that these modifiers affect embryonic development, such as the neuroblast segregation controlled by Notch signaling. Therefore, we next examined whether these enhancers and suppressors also modified the neurogenic phenotype of pcx/+^m. At 28°, 31.4% of the pcx/+^m embryos showed a weak neurogenic phenotype, whereas the rest of the embryos exhibited a largely normal nervous system (quenched neurogenic phenotype), based on the criteria described in Figure 3. However, in combination with pcx/+^m, all of the dominant enhancers identified by our screen; i.e., bib, neur, Dl, and lov, increased the percentage of embryos showing a weak neurogenic phenotype to 75.9, 78.5, 81.8, and 76.3%, respectively, in a dominant manner at 28° (Figure 6A). We also found that bib¹/+, Dl^5F102/+, neur¹/+, and lov⁴⁷/+ exhibited a normal nervous system (normal) upon anti-Elav antibody staining at 28°, suggesting that a synergistic genetic interaction, rather than a simple additive effect, occurred between pcx and each mutant (Figure 6A). Under the same conditions, an arbitrarily selected deficiency, Df(2R)DE2457, did not markedly affect the frequency of the pcx/+^m phenotypes (Figure 6A). These results suggested that the identified dominant enhancers further reduced the Notch signaling in pcx/+^m embryos.

Figure 6.

Dominant modification of the neurogenic phenotype of pcx/+^m by pcx modifiers. (A) Percentages of embryos showing a normal nervous system (normal, blue bars), quenched neurogenic phenotype (quenched, green bars), and weak neurogenic phenotype (weak, yellow bars) at 28°. Embryos of pcx/+^m, pcx/+^m combined with Df(2R)ED2475/+ (an arbitrarily selected deletion as a control), bib¹/+, Dl^5F102/+, lov⁴⁷/+, or neur¹/+, and heterozygotes of each enhancer (bib¹/+, Dl^5F102/+, lov⁴⁷/+, or neur¹/+) were analyzed. Embryos heterozygous for all the enhancers were scored as normal; none of them showed a dominant neurogenic phenotype at 28°. (B) Percentages of embryos showing normal nervous system (normal, blue bars), quenched neurogenic phenotype (quenched, green bars), and weak neurogenic phenotype (weak, yellow bars) at 25°. Embryos of pcx/+^m, pcx/+^m combined with Df(2R)ED2475/+ (an arbitrarily selected deletion as a control), BobA¹/+, or Nsf2^A15/+, and heterozygotes of each suppressor (BobA¹/+ or Nsf2^A15/+) were analyzed. Embryos heterozygous for each suppressor were scored as normal; none of them showed a dominant neurogenic or anti-neurogenic phenotype at 25°. The number of embryos examined in each experiment (N) is indicated at the top of the graph.

We next examined whether the dominant suppressors identified by our screen suppressed the neurogenic phenotype of pcx/+^m. At 25°, 46.7% of the pcx/+^m embryos showed a weak neurogenic phenotype, while the rest showed a quenched neurogenic phenotype (n = 15) (Figure 6B). However, the two dominant suppressors, BobA¹ (6.1%, n = 33) and Nsf2^A15 (0%, n = 21), almost completely suppressed the neurogenic phenotype of pcx/+^m in a dominant manner at 25°, at which almost all of these embryos exhibited a quenched neurogenic phenotype (Figure 6B). Conversely, under the same conditions, an arbitrarily selected deficiency, Df(2R)DE2457, did not markedly affect the frequency of the weak neurogenic phenotype (64.7%, n = 34) (Figure 6B). These results suggested that mutations of the BobA and Nsf2 genes augmented the Notch signaling activity in pcx/+^m embryos. We also found that BobA¹/+ and Nsf2^A15/+ exhibited a normal nervous system (normal) upon anti-Elav antibody staining at 25°, demonstrating that BobA¹ and Nsf2^A15 showed a synergistic genetic interaction with pcx (Figure 6B).

BobA may have a similar function as other Brd family genes in neuroblast segregation as well as in mesoderm specification

The endocytosis of Dl shows dorsal-ventral polarity in the cellular blastoderm at stage 5 (Bardin and Schweisguth 2006). Dl endocytosis is active in the ventral part (mesoderm) of embryos, depends on Neur, and is required for the spatially controlled activation of Notch signaling in the mesectoderm (Bardin and Schweisguth 2006). It was previously shown that neur activity is suppressed in the dorsal region of embryos by Brd family genes, so Dl trafficking is upregulated in this region in mutants of Brd family genes (Bardin and Schweisguth 2006; De Renzis et al. 2006). Thus, the Brd family proteins suppress Neur activity, leading to reduced Dl trafficking in the early embryo. In these experiments, however, multiple Brd family genes were simultaneously removed by a deletion mutation. In the current study, we found that the BobA¹ mutant, in which only the BobA gene was deleted among the Brd family genes, suppressed the neurogenic phenotype (6.1%, n = 33) associated with pcx/+^m, compared with the wild-type BobA gene (46.7%, n = 15) (Figure 6B). In addition, although the role of the Dl-Neur-Brd pathway in specification of the mesectoderm has been extensively analyzed (Bardin and Schweisguth 2006), and it was proposed that the inhibition of Neur by Brd regulates the epithelial polarity (Chanet and Schweisguth 2012), its role in neuroblast segregation in the neuroectoderm has not been studied. Therefore, in this study, it was important to determine whether Dl trafficking was also upregulated in the BobA¹ mutant embryos, especially in the neuroectoderm.

The distribution of Dl was analyzed by anti-Dl antibody staining. As previously shown, Dl was detected in recycling endosomes, which were specifically labeled by anti-Rab11 antibody staining, in the ventral but not the dorsal side of wild-type embryos at stage 5 in all cases examined (n = 5) (Figure 7) (Wang and Struhl 2004; Emery et al. 2005). However, Dl was ectopically detected in Rab11-positive vesicles in the dorsal side of BobA¹ homozygous embryos at stage 5 in all cases examined (n = 8) (Figure 7). This observation was consistent with a previous finding that Brd family genes suppress the neur-dependent endocytosis of Dl in the dorsal but not the ventral side (Bardin and Schweisguth 2006). Importantly, we detected the Dl-containing ectopic vesicles in the neuroectoderm of BobA¹ mutant embryos (n = 7), which was visualized by specific staining with an anti-Rhomboid antibody (left side of each panel in Figure 8; boundaries of Rhomboid-expressing and nonexpressing regions are indicated by a white broken line) (Sturtevant et al. 1996). In addition, notably, we found that BobA¹ homozygous embryos showed an anti-neurogenic phenotype at 41.2% (n = 17) frequency at 28°, as revealed by a reduction in neuronal cells detected by anti-Elav antibody staining, whereas wild-type embryos did not show this phenotype at this temperature in any case examined (n = 27). Thus, our results suggest that the Dl-Neur-Brd pathway also regulates Dl endocytosis in the neuroectoderm during neuroblast segregation. On the other hand, we did not detect a notable difference in the number of Dl-containing vesicles in the ventral region of these embryos (Figure 7 and right sided of each panel in Figure 8), probably because the Brd genes are not expressed in the ventral region of wild-type embryos (Chanet and Schweisguth 2012).

Figure 7.

BobA suppresses Dl endocytosis in the dorsal region of embryos. Subcellular localization of Dl (magenta, left, and right panels) and Rab11 (green, middle, and right panels), a marker of recycling endosomes, in the dorsal (upper panels) and ventral (lower panels) epithelium of wild-type (wt), BobA¹/BobA¹ (BobA), and pcx^mz embryos at stage 5. Right panels are merged images of the left and middle panels. White arrowheads indicate vesicles where Dl and Rab11 are colocalized. Percentage of embryos showing the represented phenotypes is indicated at the upper right. The number of embryos examined is in parentheses. Bar, 5 μm.

Figure 8.

Dl recycling and Notch signaling are hyperactivated in the neuroectoderm of the BobA mutant. (A) Subcellular localization of Dl (magenta, left, and right panels) and Rhomboid (Rho) (green, middle, and right panels), a neuroectoderm marker, in wild-type (wt) or BobA¹/BobA¹ (BobA) embryos at stage 5. Boundary between the neuroectoderm and mesoderm is indicated by a white broken line, and the regions left (Rho-positive) and right (Rho-negative) of the broken lines are neuroectoderm and mesoderm, respectively. The percentage of embryos showing the represented phenotype is indicated at the upper right. The number of embryos examined is in parentheses. Bar, 5 μm. (B) Lateral views of wild-type (wt) and BobA¹/BobA¹ (BobA) embryos cultured at 28° and immunostained with a neuron-specific (anti-Elav) antibody at stage 13–14. Percentage of embryos showing the represented phenotype in each panel is shown at the bottom right, and the number of embryos examined is in parentheses. Bar, 50 μm.

Based on these results, BobA appears to have a similar activity to that of other Brd family proteins. This idea is consistent with the previous finding that the majority of Brd family proteins, including BobA, bind to Neur (Bardin and Schweisguth 2006). Conversely, our results also showed that the function of Brd family genes is not completely redundant regarding the suppression of Dl endocytosis and the ectopic activation of Notch signaling in the neuroectoderm. We predicted that if pcx is also directly involved in the Neur-mediated endocytosis of Dl, the number of Dl-positive vesicles should decrease in the mesoderm of pcx^mz embryos, as previously found in embryos homozygous for neur or overexpressing a Brd family gene (Bardin and Schweisguth 2006). However, we did not observe any abnormality in Dl-positive vesicles in the pcx^mz embryos in any case examined (n = 5) (Figure 7). We also examined whether there was a synergistic effect between pcx and BobA on the Dl trafficking by combining pcx/+^m and BobA/+, in each of which the Dl-containing vesicles did not show any detectable abnormality. However, we did not observe any alteration in the Dl-positive vesicles in pcx/+^m; BobA/+ (Figure 9). Therefore, at this point, the molecular mechanisms underlying the genetic connection between the Dl-Neur-Brd pathway and Pcx remain unknown.

Figure 9.

No synergistic interaction between pcx/+^m and BobA/+ in the Dl-trafficking phenotype. Subcellular localization of Dl (magenta, left, and right panels) and Rab11 (green, middle and right panels), a marker of recycling endosomes, in the dorsal (upper panels) and ventral (lower panels) epithelium of pcx/+^m, BobA/+, and pcx/+^m; BobA/+ embryos at stage 5. Right panels are merged images of the left and middle panels. White arrowheads indicate vesicles where Dl and Rab11 are colocalized. Percentage of embryos showing the represented phenotype is indicated at the upper right. The number of embryos examined is in parentheses. Bar, 5 μm.

Discussion

Genes encoding known and novel components of Notch signaling were identified as dominant modifiers of pcx

Our dominant modifier screens of pcx identified four enhancers and two suppressors (Figure 4B). Among these, four genes encode well-characterized components of the Notch signaling pathway, suggesting that our procedure efficiently screened for Notch-signaling components. Although our primary screens were based on the enhancement and suppression of the cold-sensitive lethality demonstrated in pcx/+^m embryos, such modifications of this lethality coincided well with the enhancement and suppression of the neurogenic phenotype, implying that the lethality is related to a depletion of Notch signaling. Thus, the primary cause of the cold-sensitive lethality was probably a further reduction of Notch signaling with the decrease in temperature, and the modifications of the lethal phenotype reflected up- or down-regulations of Notch signaling in the heterozygous combination with each modifier.

We predicted that the phenotypes of pcx/+^m could have resulted from an alteration in the initial timing of the zygotic expression of pcx. However, in our screen, mutations that might have generally changed the onset of zygotic gene expression in the early embryo, such as chromatin remodeling factors, were not identified. Conversely, most of the identified genes were found to encode factors that directly contribute to the Notch signaling cascade. The highly specific genetic interactions between pcx and these genes support the idea that Pcx is a critical component of Notch signaling, at least during the period of neuroblast segregation in the neuroectoderm. This idea is also consistent with our previous finding that the expression of Notch signaling target genes, including Enhancer of split and single-minded, is severely reduced in pcx^mz embryos (Yamakawa et al. 2012).

The Dl-Neur-BobA pathway functions not only in mesectoderm specification but also in neuroblast segregation

Our screens identified the Dl, neur, and BobA genes, which have been shown to function in a common biochemical process that regulates the dorsal-ventral pattern of Dl endocytosis, as dominant modifiers of pcx (Bardin and Schweisguth 2006; De Renzis et al. 2006). The activity of Dl as a ligand for Notch requires the endocytosis of Dl, which depends on the ubiquitination of the cytoplasmic domain of Dl by Neur, a RING domain-containing E3 ubiquitin ligase (Lai et al. 2001; Pavlopoulos et al. 2001). During early embryogenesis around stage 5, the endocytosis of Dl is particularly active in the mesoderm located in the most ventral part of the embryo, although Dl endocytosis is much less active in the neuroectoderm that is dorsally adjacent to the mesoderm. Therefore, the high Dl activity in the mesoderm induces the Notch signaling-dependent transcription of sim in single rows of cells next to the mesoderm (mesectoderm) (Morel and Schweisguth 2000).

In the absence of the Brd complex, in which Oche, Tom, BobA, BobB, and BobC are clustered, the active endocytosis of Dl is observed in the entire embryo, indicating that the Brd family genes normally inhibit Neur activity in the dorsal region of the embryo (Bardin and Schweisguth 2006). Under the Brd-deleted condition, the sim expression is expanded (Bardin and Schweisguth 2006). To regulate sim expression, Brd proteins bind to Neur and interfere with the physical interaction between Neur and Dl (Lai et al. 2001; Bardin and Schweisguth 2006). However, unlike this mechanism of mesectoderm specification, the contribution of the Dl-Neur-BobA pathway to the regulation of Notch signaling during neuroblast segregation in the neuroectoderm has not been well studied. In the current study, we found that the neurogenic phenotype associated with the pcx mutation was dominantly enhanced by neur and Dl and suppressed by BobA. Consistent with these observations in pcx/+^m; BobA¹/+ embryos, Dl-containing vesicles were increased in the neuroectoderm of BobA¹ homozygous embryos, as well as in their dorsal side (Figure 8A). Moreover, we found that BobA homozygotes exhibited an anti-neurogenic phenotype at 28°, implying that Notch signaling was augmented during neuroblast segregation in these embryos (Figure 8B). Thus, our results suggest that the Dl-Neur-BobA pathway also functions in the regulation of Notch signaling during neuroblast segregation in the neuroectoderm.

Although the Dl endocytosis was hyperactivated in the neuroectoderm of BobA¹ mutant embryos, we did not observe an attenuation of Dl endocytosis even in the complete absence of pcx function (pcx^mz embryo) (Figure 7). In addition, we did not observe synergistic effects between pcx and BobA on the Dl endocytosis (Figure 8). Therefore, at this point, it is unclear whether Pcx has a direct role in the Dl-Neur-BobA pathway. We are currently investigating the potential connections between Pcx and Dl endocytosis.

lov transcribed in the early embryo may contribute to the activation of Notch signaling during neuroblast segregation

In this study, we identified the lov gene as a potential component of the Notch signaling pathway for the first time. lov encodes a Bric-a-Brac/Tramtrack/Broad/Pox virus and zinc finger (BTB/POZ) domain-containing transcription factor and is expressed in various neurons of the CNS and PNS and in the trachea during Drosophila development (Albagli et al. 1995; Armstrong et al. 2006). A loss-of-function mutant of lov demonstrates multiple behavioral defects in the larval and adult stages, including gravity sensing (Bjorum et al. 2013). These behavioral defects were previously attributed to the abnormal specifications of neurons, given that an absence of lov expression in a subset of neurons in mutants of a particular lov allele was correlated with the behavioral defects (Bjorum et al. 2013). The expression of lov was also observed in specific cells of the CNS midline in embryos (Wheeler et al. 2008). Notably, Notch signaling regulates gene expression in these CNS midline cells, which defines the fates of these cells (Wheeler et al. 2008). However, a detailed analysis of lov expression revealed that the lov gene is not a target of Notch signaling, at least in the ventral unpaired median neurons of the CNS midline (Wheeler et al. 2008). Although it is still possible that lov is involved in Notch signaling activation as a component of the pathway, the contribution of lov to Notch signaling appears to be highly context-dependent, redundant, or subtle, because the defects in the cell-fate specification of neurons found in lov mutants are much milder than those of classic Notch pathway genes (Bjorum et al. 2013).

It is reported that the expression of lov begins in the cellular blastoderm stage and continues during the development of the CNS and PNS (Kearney et al. 2004; Weiszmann et al. 2009). Furthermore, the Lov protein is detected in the neuroectoderm during the stage of neuroblast segregation, as well as in other parts of embryos (Bjorum et al. 2013). Therefore, lov may control the transcription of genes involved in Notch signaling during the periods of neuroblast segregation through lateral inhibition, although this possibility remains to be investigated. In any case, we speculate that the function of lov is not essential for Notch signaling during neuroblast segregation, given that lov homozygous embryos did not exhibit a neurogenic phenotype (data not shown). In addition, a maternal expression of lov was not observed, suggesting that lov does not have a maternal function (Coté et al. 1987). Therefore, it is unlikely that lov plays an indispensable role in lateral inhibition in the embryo, although our results suggest that it makes some contribution to it.

A novel role of Nsf2 supports the contribution of Pcx to ER homeostasis

In this study, we identified Nsf2 as a dominant suppressor of pcx. Drosophila has two Nsf genes, comatose (Nsf1) and Nsf2 (Pallanck et al. 1995a). The Nsf genes encode homohexameric AAA ATPases that are involved in the membrane fusion required for the transfer of membrane vesicles from one membrane compartment to another (Wilson et al. 1989; Söllner et al. 1993a). Nsf1 is mainly expressed in neurons (Ordway et al. 1994), and its loss causes a disruption in synaptic transmission (Pallanck et al. 1995b; Gokhale et al. 2015). To ensure the specificity of membrane fusion, SNARE proteins on vesicles and target membranes form a tight complex (Söllner et al. 1993a). In this process, Nsf promotes the dissociation of SNARE complexes once membrane fusion has been completed, using ATP hydrolysis as the energy source (Söllner et al. 1993a). The dissociation of SNAREs is required for their recycling for further rounds of membrane fusion events (Söllner et al. 1993b). Thus, the absence of Nsf results in an attenuation of membrane trafficking (Beckers et al. 1989). These studies suggested that Nsf1 functions to disassemble SNARE complexes after vesicle docking, which is required to maintain the readily releasable pool of synaptic vesicles (Muller et al. 2002; Gokhale et al. 2015).

On the other hand, Nsf2 is broadly expressed in embryonic and imaginal tissues, as well as in the nervous system (Boulianne and Trimble 1995). Therefore, Nsf2 is likely to encode an Nsf isoform that contributes to vesicular trafficking in non-neuronal cells, including the neuroectoderm of the embryo (Boulianne and Trimble 1995). It was previously shown that the expression of a dominant-negative Nsf2 (Nsf2^E/Q) suppresses Notch signaling in wing imaginal discs (Stewart et al. 2001). Genetic interaction between Nsf2^E/Q and mutants of other genes encoding Notch-signaling components, including Dl, Serrate, fringe, and bib, has also been reported (Stewart et al. 2001). Notably, bib was also identified as a dominant enhancer of the pcx/+^m lethality in the present study (Figure 4B and Figure S1). In addition, it was shown that Notch protein accumulates in the intracellular vesicles in epithelial cells misexpressing Nsf2^E/Q (Stewart et al. 2001). The contribution of Nsf2 to Notch signaling was further supported by the identification of Nsf2 as a dominant enhancer of Presenilin hypomorphic mutants (Mahoney et al. 2006). These results all suggested that Nsf2 positively contributes to Notch signaling (Mahoney et al. 2006). In contrast, in our screen, Nsf2 was identified as a suppressor of pcx, demonstrating that the wild-type Nsf2 gene suppresses Notch signaling activity under the conditions of our experiment. Thus, we speculate that Nsf2 may have different roles in Notch signaling between imaginal tissues and embryonic neuroectoderm, which has not been noted before. This idea is consistent with the finding that various sets of SNAREs control the membrane fusion of different intracellular compartments under the control of an Nsf protein (Söllner et al. 1993b).

Furthermore, we previously reported that the ER is enlarged in a subset of cells in pcx^mz embryos (Yamakawa et al. 2012). In mammalian and yeast cells, Nsf is involved in ER membrane fusion and required for ER reassembly (Uchiyama et al. 2002; Kano et al. 2005). Thus, given that the ER enlargement in the absence of pcx function may be due to excessive assembly of the ER, a loss-of-function mutation of Nsf2 may reduce the excess ER reassembly occurring in the absence of pcx. This scenario is consistent with the behavior of Nsf2 as a dominant suppressor of pcx/+^m lethality. This mechanism is currently under active investigation in our laboratory.

The genetic screens performed in the current study identified dominant modifiers of the cold-sensitive lethality associated with a pcx mutation condition. In particular, although the genetic interactions of pcx with neur or BobA may not reflect a direct relationship with respect to the regulation of Dl endocytosis, the results obtained from our genetic screens revealed intimate relationships between Pcx and the Notch signaling pathway and implied a close relationship between pcx function and vesicular transport during neuroblast segregation in the neuroectoderm. Furthermore, our analysis uncovered a potential role of lov and a previously unknown function of Nsf2 in Notch signaling. There results contribute to our understanding of the complex regulation of Notch signaling.