Disruption of Drosophila melanogaster Lipid Metabolism Genes Causes Tissue Overgrowth Associated with Altered Developmental Signaling

Developmental patterning requires the precise interplay of numerous intercellular signaling pathways to ensure that cells are properly specified during tissue formation and organogenesis. The spatiotemporal function of many developmental pathways is strongly influenced by the biosynthesis and intracellular trafficking of signaling components. Receptors and ligands must be trafficked to the cell surface where they interact, and their subsequent endocytic internalization and endosomal trafficking is critical for both signal propagation and its down-modulation. In a forward genetic screen for mutations that alter intracellular Notch receptor trafficking in Drosophila melanogaster, we recovered mutants that disrupt genes encoding serine palmitoyltransferase and acetyl-CoA carboxylase. Both mutants cause Notch, Wingless, the Epidermal Growth Factor Receptor (EFGR), and Patched to accumulate abnormally in endosomal compartments. In mosaic animals, mutant tissues exhibit an unusual non-cell-autonomous effect whereby mutant cells are functionally rescued by secreted activities emanating from adjacent wildtype tissue. Strikingly, both mutants display prominent tissue overgrowth phenotypes that are partially attributable to altered Notch and Wnt signaling. Our analysis of the mutants demonstrates genetic links between abnormal lipid metabolism, perturbations in developmental signaling, and aberrant cell proliferation.


Introduction
Developmental patterning in metazoans requires the coordinated activity of several intercellular signaling pathways. In D. melanogaster, Notch and Wnt signaling are critical for the formation of diverse organs and tissues, and both pathways also regulate cell proliferation and apoptosis during development [1][2][3][4]. Notch signaling is activated by binding of DSL ligands to the Notch receptor, which induces proteolytic cleavage of Notch by ADAM/ TACE metalloproteases and subsequent cleavage by gammasecretase to generate the Notch intracellular signaling fragment NICD [1,2]. NICD translocates to the nucleus where it regulates target gene expression by displacing co-repressors from transcriptional complexes and converting them into active complexes [5,6].
Notch signaling is strongly modulated by additional posttranslational mechanisms, including glycosylation, ubiquitylation, and endosomal trafficking [2,7]. Endocytosis of Notch and its ligands in both signal-receiving and signal-sending cells is required for productive signaling [8], generating tensile forces needed to expose the ADAM/TACE cleavage site in Notch and facilitate receptor proteolysis [9]. Several mutants with impaired trafficking of ligand-activated Notch are associated with reduced signaling, while others that perturb trafficking of non-activated Notch exhibit Notch hyperactivation and tissue overgrowth [10][11][12][13][14]. Intracellular Notch trafficking thus regulates both signal activation and the degradation of inactive Notch receptors that might otherwise contribute to inappropriate signaling. Moreover, this endocytic regulation of Notch signaling is strongly influenced by its membrane lipid microenvironment; mutations in D. melanogaster phosphocholine cytidylyltransferase alter Notch endosomal routing and activation [15] and mutations in D. melanogaster alpha-1,4-Nacetylgalactosaminyltransferase-1 affect endocytosis and potency of the Notch ligands Delta and Serrate [16].
Canonical Wnt/Wingless signaling is similarly needed for a diverse array of tissue patterning processes and is also influenced by membrane trafficking [3,4]. Wnt, a secreted ligand, binds to receptors of the Frizzled and LRP/Arrow families, leading to recruitment of Disheveled and stabilization of ß-catenin/Armadillo. Accumulation of ß-catenin/Armadillo in turn triggers downstream gene activation through TCF transcription factors [17,18]. Wnt proteins are modified by lipid attachment [19,20], and endocytosis limits secreted Wnt diffusion and modulates intracellular signaling [4,21].
To identify new genes required for trafficking of developmental signaling molecules, we performed a forward genetic screen for mutations that alter the intracellular accumulation of Notch in developing D. melanogaster wing tissues. Among ,40 new mutants recovered, two mutants displayed a strikingly similar phenotype of abnormal, non-cell-autonomous accumulation of Notch, Wingless, and other membrane proteins in endosomes and lysosomes. These two mutants disrupt essential enzymes of lipid metabolism, serine palmitoyltransferase (SPT) and acetyl-CoA carboxylase (ACC), which are encoded by the lace and ACC genes, respectively. SPT catalyzes the covalent attachment of serine to long chain fatty acyl-CoA to produce 3-ketodihydrosphingosine during sphingolipid biogenesis [22,23]. Sphingolipids are major constituents of lipid rafts, a specialized membrane microdomain involved in endocytosis and signaling [24,25]. In addition, bioactive sphingolipids participate in various signaling events, regulating cell proliferation, differentiation, apoptosis, and other cellular functions [26]. ACC catalyzes the carboxylation of acetyl-CoA to produce malonyl-CoA in the de novo synthesis of fatty acids [27,28], which are needed for various cellular functions including energy storage, membrane biogenesis, and serving as precursors for phospholipid biosynthesis.
Notably, the D. melanogaster SPT and ACC mutants also exhibit tissue overgrowth phenotypes indicative of effects on cell proliferation. Analysis of downstream targets of Notch and Wingless reveals that this overgrowth is likely to involve modulatory effects on these pathways, reflecting both Notch hyperactivation and impaired Wingless signaling. Epistasis studies demonstrate that the overgrowth partially depends upon the Notch effector Su(H) as well as gamma-secretase function, and can also be partially suppressed by activated Armadillo, confirming that both Notch and Wingless dysregulation contributes to the lace and ACC mutant overgrowth phenotypes. Our findings emphasize the importance of lipid metabolism for establishing and maintaining the membrane compartments in which key developmental signaling pathways operate, and illustrate how general metabolic processes can exert complex, pleiotropic effects on multiple pathways needed for tissue growth and patterning.

Abnormal Notch trafficking in D. melanogaster lace and ACC mutants
To identify new genes required for proper Notch trafficking, we designed a forward genetic screen in which homozygous mutant tissue is directly examined for aberrant Notch accumulation using antibody immunoanalysis. Because important trafficking genes would likely encode products essential for organismal viability, we created clones of homozygous mutant tissue in developing imaginal wing discs of otherwise heterozygous D. melanogaster using the FLP-FRT mosaic method [29] (see Materials and Methods). This approach also allows mutant tissues to be compared directly to adjacent heterozygous tissue in each sample, eliminating variability in fixation time, antibody penetration, and other parameters. Following screening of 3335 mutagenized second chromosome arms, we identified over 40 genes which, when mutated, alter the pattern of Notch trafficking as visualized using an antibody directed against the Notch intracellular domain.
Two lethal mutants, subsequently identified as lace and acetyl-CoA carboxylase (ACC) mutants, exhibited similar effects on Notch trafficking in homozygous mutant clones. In both cases, mutant cells display large, abnormal intracellular Notch-positive vesicles confined to internal clone regions approximately 3-4 cell diameters away from clone boundaries ( Figures 1A and 1D). In homozygous mutant cells near clone boundaries, mutant effects on Notch trafficking are evidently rescued by non-cell-autonomous activity provided by nearby wildtype cells. The D. melanogaster wing disc is an oriented columnar epithelial monolayer, and the aberrant Notch-containing vesicles are observed throughout mutant cells from basement membrane to apical surface, in contrast to the predominantly apical accumulation of Notch in wildtype cells ( Figures 1A and 1D). This non-autonomous Notch trafficking defect is consistently observed in both newly isolated alleles of ACC (ACC 1 and ACC 2 ), both newly isolated alleles of lace (lace 18 and lace 19 ), and a previously isolated amorphic allele of lace (lace 2 [30]) ( Figures 1A and 1D; Supplemental Figure S1A-E). In the lace and ACC mutants, the Notch ligand Delta is similarly mislocalized in a non-cell-autonomous manner (Figures 1C and  1F).
We examined the localization of additional cell-surface molecules to determine whether the requirement for lace and ACC is specific to Notch signaling or also pertains to other developmental pathways. Wingless, the secreted ligand of the Wnt pathway, the Epidermal Growth Factor Receptor (EGFR), a receptor tyrosine kinase that activates Ras/MAPK signaling, and Patched, the receptor for the Hedgehog signal, also accumulate non-cellautonomously in large intracellular vesicles in lace or ACC homozygous mutant cells (Figure 2A-L). Moreover, expression of an exogenous mCD8::GFP fusion protein also leads to its weak overaccumulation in the enlarged Notch-associated vesicles in lace or ACC mutant cells, although this co-accumulation of Notch and mCD8:GFP is primarily observed in apical but not basal cell regions ( Figure 2M-P). Thus lace and ACC are likely to play a general role in intracellular protein trafficking and may influence the proper routing of numerous membrane proteins.

Molecular lesions in lace and ACC
Genetic mapping and complementation established that the newly recovered mutations are alleles of lace, which encodes serine palmitoyltransferase (SPT), and ACC encoding acetyl-CoA carboxylase. We determined the genomic sequence of the two lace alleles isolated from our screen as well as that of the previously isolated amorphic allele lace 2 [30], all of which bear point

Author Summary
The development of complex, multicellular animal tissues requires the coordinated function of many different cellcell communication pathways, in which secreted or cellsurface-anchored ligands from one cell typically activate a receptor on the surface of other cells, which in turn regulates downstream gene transcription and other cellular processes. We used a genetic approach in the fruit fly Drosophila melanogaster to search directly for mutations that perturb intracellular trafficking of a major signaling receptor, namely the Notch receptor, which controls cell differentiation in various tissue contexts. The Notch signaling pathway, like other key developmental signaling pathways, is evolutionarily conserved and functions in a similar manner in D. melanogaster and mammals, including humans. We recovered and characterized mutations in two genes that encode different enzymes involved in cellular lipid metabolism. Both mutants alter not only Notch signaling but also downstream activity of another highly conserved signaling pathway mediated by the Wingless protein, illustrating that alterations in cellular enzymes of lipid metabolism can exert complex effects on multiple critical signaling pathways. We also found that the new mutants exhibit dramatic cell overproliferation effects, reinforcing findings from mammalian studies suggesting that lipid metabolism might play an important role in oncogenesis and tumor progression.
mutations altering a single amino acid ( Figure 3A). As noted above, we confirmed that the null allele lace 2 shows the same noncell-autonomous effect on Notch accumulation (see Supplemental Figure S1A-C), and thus lace 2 was utilized for all subsequent analyses.
The newly recovered ACC mutants fail to complement Pelement insertion line B131, in which the transposon is inserted into the intron of ACC [31]. ACC encodes D. melanogaster acetyl-CoA carboxylase (ACC), an enzyme that is highly conserved from bacteria to humans [27]. Over three separate domains, .60% of the amino acids are identical between D. melanogaster ACC and human ACC1 ( Figure 3B). We sequenced ACC mutant alleles ACC 1 and ACC 2 , and found that ACC 1 contains a premature stop codon, indicating that ACC 1 likely represents a null allele, while ACC 2 encodes an amino acid substitution within the N-terminal conserved domain ( Figure 3A). For subsequent studies, we used the presumptive null allele ACC 1 .
To confirm that the Notch trafficking defects observed in these mutants are specifically attributable to loss of lace and ACC activity, we expressed wildtype UAS-laceHA [32] (in which Lace contains an HA-epitope tag) and UAS-ACC cDNA constructs in posterior compartment clones of lace and ACC mutant cells, respectively, utilizing a hh-GAL4 driver line with UAS-FLP (see Materials and Methods). Expression of UAS-laceHA almost completely suppresses the Notch accumulation phenotype of lace 2 clones, and expression of UAS-ACC fully suppresses this phenotype in ACC 1 clones (Supplemental Figure S1F and S1G). To test whether non-specific hh-GAL4-mediated expression of either transgene can rescue the Notch accumulation defect, we also expressed UAS-laceHA in ACC 1 clones, and conversely UAS-ACC in lace 2 clones using the same approach. No significant rescue was observed in either case (Supplemental Figure S1H and S1I), indicating that the nonautonomous Notch trafficking defects seen in both mutants reflect specific requirements for lace and ACC gene activities rather than a general reduction in lipid homeostasis.

Notch accumulates abnormally in endocytic compartments of mutant cells
To identify the cellular compartment in which Notch accumulates in the lace and ACC mutants, we performed antibody uptake studies on clone-bearing wing discs. When live, unpermeabilized discs are incubated with antibodies that bind to the extracellularly exposed domain of Notch, the antibodies specifically detect the subpool of Notch at the cell surface and in surface-derived endocytic compartments [33]. Incubating lace and ACC mosaic mutant discs with anti-Notch extracellular antibody C458.2H, followed by a brief incubation period to allow antibody uptake by the live cells, subsequent tissue fixation and imaging revealed that for both mutants, the abnormal Notch vesicles reside in the endocytic trafficking pathway ( Figures 1B and 1E).
To determine the specific endocytic compartment in which Notch accumulates, we performed double-labeling studies with Notch and various organelle markers, including Rab5-YFP, Rab7-YFP, Rab11-YFP, and LAMP-HRP, which label early endosomes, late endosomes, recycling endosomes, and late endosomes/ lysosomes, respectively [34,35]. In lace mutant cells, 38% of abnormal Notch vesicles colocalize with Rab5-YFP, 16% with LAMP-HRP, and 11% with Rab7-YFP (Figure 4E-4H; Supplemental Table S1). In ACC mutant cells, 55% of Notch-positive vesicles colocalize with LAMP-HRP, 15% with Rab5-YFP, and 12% with Rab7-YFP (Figure 4J-4M; Supplemental Table S1). In both mutants, the Notch-positive vesicles exhibit a much lower degree of colocalization with other organelle markers, including Rab11-YFP, Clathrin light chain-GFP (Clc-GFP), PDI-GFP (an ER marker), Golgi-YFP, and Sara ( Figure 4G Table S1). Interestingly, LAMP-positive late endosomes/lysosomes are dramatically enlarged in lace and ACC mutant cells ( Figures 4I and 4N, Supplemental Table S2), while other compartments are only slightly affected (Supplemental Table S2). Collectively, these results indicate that Notch accumulates primarily in early endosomes, late endosomes and lysosomes in lace mutant cells, but predominantly in lysosomes, and to a lesser extent in early and late endosomes in ACC mutant cells. The distinct patterns of Notch mislocalization in the two mutants might reflect different alterations in these endocytic compartments, and might also contribute to the differential effects of these mutants on Notch and Wingless signaling (see below). However, it should be noted that these organelle marker studies involve overexpression of endosomal machinery components under UAS control, which although widely used to label different endocytic compartments, might lead to abnormal endosomal compartment morphogenesis and/or function.

Loss of lace and ACC activity causes tissue overgrowth
The lace and ACC mutants were examined for whether they might also show a cell proliferation phenotype, since several D. melanogaster endocytic trafficking mutants cause cell overproliferation, especially in large clones produced using the Minute system [36] or through ectopic expression of the Caspase inhibitor p35 [10][11][12][13][14]. Using the FLP-FRT system with Minute chromosomes to generate large clones of either lace or ACC mutant cells, we found that these clones showed significant tissue overgrowth ( Figure 5A-5C). To control for clone size and location, we next produced lace and ACC mutant clones in specific disc regions by expressing UAS-FLP under the control of hh-GAL4. Testing multiple alleles of lace and ACC using this approach revealed that the overgrowth phenotypes are variable, ranging from ,2% up to ,95% for different lace alleles, and from ,10% to ,28% for the two available ACC alleles (Figure 5G-L; Supplemental Table S3). The lace and ACC overproliferation phenotypes are almost completely suppressed by overexpression of Lace and ACC, respectively ( Figure 5N and 5P; Supplemental Table S3). Consistent with our findings above for the Notch trafficking phenotype, the overproliferation phenotypes of lace and ACC mutant cells could not be rescued by the converse overexpression of ACC and Lace, respectively (Figure 5O and 5Q; Supplemental Table S3).
To confirm overproliferation at the cellular level in lace and ACC mutant cells, we examined phosphohistone H3 (pH 3) signals in clones of both mutants. The percentage of pH 3-positive cells is significantly increased in lace and ACC mutant cells compared to neighboring heterozygous cells ( Figure 5R-T). We also examined apical-basal cell polarity in these overgrown mutant discs, since it is disrupted in other D. melanogaster endocytic mutants [10][11][12][13][14], in the glycosphingolipid metabolism mutants egghead and brainiac that disrupt Notch signaling during D. melanogaster oogenesis [37], and in several C. elegans mutants that likewise affect glycosphingolipid biosynthesis [38]. Unexpectedly, despite using four different markers to examine the apicobasal structure of lace and ACC mutant cells, we found that cell polarity is apparently unaffected in these mutant cells (Supplemental Figure S3A-T).

Notch signaling is altered in lace and ACC mutant cells
To determine which signaling pathways might be responsible for these overproliferation effects, we examined developmental gene expression patterns. Expression of Cut, which marks the presumptive wing margin during late larval development, was strongly reduced in regions where the margin extended deeply into lace or ACC mutant clones ( Figure 6A-C). Cut expression depends on both Notch and Wingless activity [39][40][41], so we independently  assessed Notch signaling in mutant clones using two additional Notch-responsive reporters. Expression of the vestigial boundary enhancer-lacZ (vgBE-lacZ) reporter [42] was diminished in lace and ACC mutant cells ( Figure 6D-F). A second reporter, Gbe+Su(H) m8 [43], also showed reduced expression along the dorsal-ventral (D/ V) boundary in lace and ACC mutant clones; however, ectopic weak signal induction was also observed ( Figure 6G-I). This result indicates that loss of either lace or ACC has complex, differential effects on Notch signal activation depending upon its cellular context.
These findings suggest that loss of lace or ACC has variable effects on Notch signaling in different tissues and even different wing disc regions. The Gbe+Su(H) m8 expression data suggest that Notch signaling might be modestly upregulated in non-margin disc regions, potentially contributing to the overproliferation phenotype. To test this idea, we asked whether the overproliferation requires functional gamma-secretase, the proteolytic enzyme complex that cleaves Notch to produce NICD. Wing disc clones mutant for both lace and aph-1 were generated, in which the aph-1 mutation inactivates an essential subunit of gamma-secretase [44]. In these clones, the aph-1 mutation strongly suppressed the overproliferation normally caused by the lace mutation but did not prevent elevated endosomal accumulation of Notch ( Figure 5D and 5F). Taken together, these findings indicate that gammasecretase-mediated Notch signaling activity is likely to be elevated in proliferating zones of the wing disc, leading directly or indirectly to the observed overgrowth. Confirming this interpretation, double mutant clones of lace and Su(H), which encodes a dedicated effector for Notch signaling, also exhibited substantial suppression of the lace wing disc overgrowth phenotype ( Figure 5E and 5M; Supplemental Figure S3).

Wingless signaling is perturbed in lace and ACC mutants
Wingless signaling is also altered in lace and ACC mutant cells based on expression of two downstream targets, Senseless (Sens) and Distalless (Dll), which respond to strong and weak Wingless signaling, respectively [41,45,46]. In lace mutant cells, expression of both Wingless targets is decreased ( Figure 6L and 6M), while Sens expression is only weakly decreased and Dll expression is apparently unaffected in ACC mutant cells ( Figure 6N and 6O). To determine whether altered Wnt signaling might contribute to the lace and ACC mutant overproliferation phenotypes, an activated form of armadillo (encoding b-Catenin, an effector of Wnt signal), termed armS10, was expressed in lace and ACC mutant clones. The overproliferation phenotype was significantly suppressed in lace mutant clones, in terms of both frequency and severity of the phenotype (Figure 7A and 7C; Supplemental Table S4). For ACC clones, the severity of the overgrowth phenotype was reduced by armS10 expression (Figure 7B and 7D), although the percentage of discs exhibiting detectable overgrowth was not significantly  different between control and armS10-expressing discs (Supplemental Table S4). These results are consistent with the idea that altered Wnt signaling partially contributes to the overproliferation phenotype in lace mutant cells, but seems to have a more modest role in the overproliferation observed for ACC-deficient cells.
In the above studies, we consistently observed that overgrowth phenotypes in lace and ACC clones were confined to the pouch and hinge region of the wing and not seen in the notum region, so we examined expression of a third Wnt downstream reporter, fz3-lacZ, in notum cells of mutant clone-bearing wing discs. Expression of fz3-lacZ was not altered in either lace or ACC mutant clones (data not shown), suggesting that the effects of the mutants on Wnt signaling, as with Notch signaling, are tissue-dependent.

Altered Dpp, Hippo, MAPK, Akt and JAK-STAT signaling do not contribute to the overproliferation observed for lace and ACC mutant cells
We also examined two pathways that mediate cell proliferation in D. melanogaster, namely Dpp and Hippo signaling, to determine if their activity was also perturbed by loss of lace or ACC activity. In lace and ACC mutant clones, we failed to detect elevated or ectopic expression of phosphorylated Mad, which transduces the active Dpp signal [47], or Spalt, a transcriptionally-induced target of Dpp and Mad in the wing imaginal disc (Supplemental Figure  S4C, S4D, S4G and S4H). In some large lace clones, expression of these markers was reduced or absent, which might be an indirect consequence of globally disrupted developmental patterning in large mutant clones (cf. Supplemental Figure S4C).
Two downstream markers of Hippo signaling, Cyclin E and DIAP1-lacZ, are upregulated when Hippo signaliing is inactivated and are also associated with overgrowth phenotypes [48]. We observed a slight increase in DIAP1-lacZ expression in lace and ACC clones (Supplemental Figure S4B and S4F), but no significant increase in Cyclin E expression (Supplemental Figure S4A and S4E), indicating that Hippo signaling makes little if any contribution to the cellular overproliferation seen in the mutants.
In addition, we assessed activation of other growth control pathways, including the EGFR, Insulin receptor, Hedgehog, and JAK/STAT pathways, by monitoring levels of their respective downstream markers dpERK, pAkt, Cubitus Interruptus, and pSTAT in lace and ACC mutant wing disc clones. Loss of lace or ACC function was not associated with any obvious disruptions in the levels or subcellular localizations of these pathway markers (Supplemental Figure S5A-L). These findings support the idea that Notch and Wnt signaling are relatively more sensitive to the loss of Lace and ACC enzyme activities, although due to the limitations of these antibody probes, we cannot exclude the possibility that some activities of these other pathways are also subtly perturbed.

Discussion
The importance of lipid metabolism for the formation and maintenance of cell membranes is well established [24,25,49]. Both serine palmitoyltransferase (SPT) and acetyl-CoA carboxylase (ACC) are critical enzymes that control different steps of lipid metabolism, and are highly conserved in diverse animal species. Genetic elimination of ACC1 or the SPT subunits Sptlc1 or Sptlc2 cause early embryonic lethality in mice [50,51], although the cellular basis for this lethality is unknown. In D. melanogaster, RNAinterfering disruption of ACC activity in the fat body results in reduced triglyceride storage and increased glycogen accumulation, and in oenocytes leads to loss of watertightness of the tracheal spiracles causing fluid entry into the respiratory system [52]. Here we demonstrate that D. melanogaster mutants lacking functional SPT or ACC exhibit endosomal trafficking defects, causing Notch, Wingless, EGFR, and Patched to accumulate abnormally in endosomes and lysosomes. These effects are accompanied by significant alterations in Notch and Wingless signaling, as revealed by changes in downstream target gene activation for both pathways. However, the mutants do not fully inactivate these developmental signaling pathways, and instead display phenotypes consistent with more complex, pleiotropic effects on Notch, Wingless, and potentially additional pathways in different tissues. Our findings reinforce the importance of lipid metabolism for the maintenance of proper developmental signaling, a concept that has also emerged from studies demonstrating that D. melanogaster mutants for phosphocholine cytidylyltransferase alter endosomal trafficking and signaling of Notch and EGFR [15], mutants for alpha-1,4-N-acetylgalactosaminyltransferase-1 affect endocytosis and activity of the Notch ligands Delta and Serrate [16], mutants for the ceramide synthase gene shlank disrupt Wingless endocytic trafficking and signaling [21], and mutants for the glycosphingolipid metabolism genes egghead and brainiac modify the extracellular gradient of the EGFR ligand Gurken [53].
Most strikingly, our newly characterized lace and ACC mutants also display prominent tissue overgrowth phenotypes. These tissue overgrowth effects are linked to changes in Notch and Wingless signaling outputs, and they involve gamma-secretase, Su(H), and Armadillo activities, suggesting that the overgrowth reflects an interplay of Wingless inactivation and Notch hyperactivation. Consistent with our findings, both Notch and Wingless regulate cell proliferation and imaginal disc size in D. melanogaster [54]. Moreover, several observations indicate that Notch and Wingless are jointly regulated by endocytosis, with opposing effects on their respective downstream pathway activities, a dynamic process that might be especially sensitive to perturbations in membrane lipid constituents [55]. Wingless itself exerts opposing effects on disc size that might depend on the particular developmental stage or disc territory. For example, hyperactivation of Wingless or inactivation of its negative regulators cause overproliferation [56][57][58], but Wingless activity can also constrain wing disc growth [59]. Similar spatiotemporal effects might underlie the variability we detected in our studies with lace and ACC mutant clones, in which both tissue overgrowth and developmentally arrested discs were observed. Although we did not detect obvious changes in downstream signaling for several other cell growth pathways that were examined, the trafficking abnormalities seen for other membrane proteins aside from Notch, Delta, and Wingless, as well as the incomplete suppression of the overgrowth phenotypes by blockage of Notch and Wingless signaling, suggest that other pathways might also be dysregulated in lace and ACC mutants, possibly contributing to the observed tissue overgrowth.
Wingless is modified by lipid addition [19,20], and lipoprotein vesicles have been suggested to control Wingless diffusion [60]. In D. melanogaster embryos, endocytosis of Wingless limits its diffusion and ability to act as a long-range morphogen [61]. Endocytosis can also affect Wingless signaling in receiving cells, where endocytosis both promotes signal downregulation [61,62] and positively facilitates signaling [63]. The apparently normal diffusion ranges for overaccumulated Wingless in lace and ACC mutant clones, yet reduced downstream target gene expression, is consistent with the idea that SPT and ACC act by promoting endocytic trafficking of Wingless in receiving cells rather than influencing the secretion and/or diffusion of Wingless from signalsending cells.
Our finding that lace and ACC mutant overgrowth phenotypes are also partially Notch-dependent is reminiscent of similar overproliferation phenotypes seen in certain D. melanogaster endocytic mutants, such as vps25, and tsg101 [11][12][13][14]. The overproliferation of disc tissue in these mutants is attributable to Notch hyperactivation, reflecting the fact that non-ligand-bound Notch receptors that are normally targeted for recycling or degradation are instead retained and signal from endosomes. Analogous effects are likely to contribute to the lace and ACC mutant overgrowth, where we observe significant Notch overaccumulation throughout the endosomallysosomal routing pathway. Some ectopic Notch signaling might emanate from the lysosomal compartment, which is enlarged and accumulates particularly high levels of Notch in lace and ACC mutant clones. Analysis of D. melanogaster HOPS and AP-3 mutants, which affect protein delivery to lysosomes, has identified a lysosomal pool of Notch that is able to signal in a ligand-independent, gamma-secretase-dependent manner [64].
How do SPT and ACC contribute to endosomal trafficking of Notch and other proteins? In the yeast SPT mutant lcb1, an early step of endocytosis is impaired due to defective actin attachment to endosomes, a phenotype that is suppressed by addition of sphingoid base [65]. However, the trafficking abnormalities seen in lace and ACC mutants do not resemble those in the yeast lcb1 mutant, perhaps because endocytic vesicle fission is primarily dependent upon dynamin in D. melanogaster and mammals, instead of actin as in yeast [66]. Nevertheless, the requirement for SPT and ACC in D. melanogaster endosomal compartments might reflect possible functions in endosome-cytoskeleton interactions. Another possibility is that the defective endosomal trafficking seen in lace and ACC mutants is caused by the inability to synthesize specific phospholipids needed for normal membrane homeostasis. Finally, lace and ACC might be important for the formation and/or function of lipid rafts, specialized membrane microdomains that have been implicated in both signaling and protein trafficking [67,68].
A remarkable feature of the lace and ACC mutant phenotypes that suggests an underlying defect in lipid biogenesis is the nonautonomous effect in mutant tissue clones, wherein nearby wildtype cells generate a secreted activity that diffuses several cell diameters into the mutant tissue and rescues the trafficking and signaling defects. One possibility is that these secreted activities are diffusible lipid biosynthetic products of SPT and ACC, which enter the mutant cells and serve as precursors for further biosynthetic steps that do not require SPT or ACC. An intriguing alternative is that the SPT and ACC enzymes are themselves secreted and taken up by the mutant cells. A precedent for this mechanism has recently been demonstrated for D. melanogaster ceramidase, a sphingolipid metabolic enzyme that is secreted extracellularly, delivered to photoreceptors, and internalized by endocytosis to regulate photoreceptor cell membrane turnover [69].
Recent work has highlighted the importance of lipid metabolism for oncogenic transformation, and ACC has been advanced as a promising target for cancer drug development [70]. ACC is upregulated in some cancers, possibly as a result of high demands for lipid biosynthesis during rapid cell divisions. Sphingolipids and their derivatives are also thought to influence the balance of apoptosis and cell proliferation during tissue growth, and thus have also garnered attention as potential cancer therapy targets [71]. Our findings regarding the requirements of SPT and ACC for proper trafficking and signaling of key developmental cellsurface signaling molecules, including Notch and Wingless, provide insights into how lipid metabolic enzymes might influence cell proliferation and tissue patterning in multicellular animals. Complex lipid biosynthesis is essential for the creation of the elaborate, interconnected, and highly specialized membrane compartments in which developmental pathways operate, and perturbations in lipid biosynthesis that are tolerated by the cell might nevertheless exert significant pleiotropic effects on developmental patterning, cell proliferation, and other cellular processes. Exploration of lipid metabolic enzymes as pharmacological targets must therefore take into account potentially unfavorable effects on critical signaling pathways controlling development and organogenesis.

Constructs
A full-length ACC cDNA (GH12002; obtained from the Drosophila Genomics Resource Center, Indiana University) was subcloned into the XbaI and SmaI sites of pBluescript II SK 2 , introducing a new NotI site between KpnI and HindIII. The resulting NotI fragment was excised and inserted into the NotI site of the pUAST vector. Transformants of UAS-ACC were obtained according to standard protocols.

D. melanogaster genetics
Mutagenesis was performed using standard protocols by administering 35 mM ethylmethanesulfonate to isogenic male flies of genotype y w; P{ry[+t7.2] = neoFRT}40A P{w[+mW.hs] = FRT(w hs )}G13, which were used to establish candidate mutant stocks. For screening of 3335 mutagenized second chromosome arms, these stocks were mated to marked 2L and 2R FRT stocks to yield progeny bearing homozygous candidate mutant wing clones using the FLP/FRT method [29]. 10 wing discs of each candidate mutant line were harvested and analyzed for abnormal Notch accumulation by direct immunofluorescence using Notch antibody C17.9C6 as described below.
For live tissue labeling, dissected wing discs were incubated with antibody for 40 min in S2 cell culture medium (Gibco), washed three times for 10 min with S2 medium, fixed and processed further as above. Hoechst-33342 trihydrochloride trihydrate (Invitrogen; 1:1000 dilution) and Phalloidin-Alexa546 (Molecular Probes; 1:20 dilution) stainings were performed for 1 hr at room temperature following immunostaining. Figure S1 Notch accumulates abnormally in homozygous tissue clones mutant for different lace and ACC alleles, and this phenotype is rescued by corresponding wildtype lace or ACC transgene expression. (A-E) Confocal sections through D. melanogaster wing imaginal discs bearing homozygous lace 2 (A-C), lace 19 (D), and ACC 2 (E) mutant clones, depicting apical (A) and basal (B, D, E) horizontal sections and a vertical z-series image compilation (C). For each image pair, panel i shows mutant clone locations (areas devoid of green signal) and Notch protein distribution (red); the Notch signal alone is presented in panel ii. (F-I) Mutant clones encompassing the wing posterior compartment were induced as in Figure 5G-S using the hh-GAL4; UAS-FLP system for lace 2 (F, H) or ACC 1 (G, I), where clones also expressed either UAS-laceHA (F, I) or UAS-ACC (G, H) wildtype cDNA transgenes as indicated. Note that expression of UAS-laceHA rescues the lace 2 Notch trafficking defect, and conversely, expression of UAS-ACC rescues the ACC 1 Notch trafficking defect, but neither transgene rescues the Notch trafficking defects seen in mutant clones for the non-matching gene. In F-I, mutant cells are identified by their lack of green marker signal, and Notch expression is shown in red. Scale bars, 10 mm. Table S1 Quantitative analysis of Notch vesicle colocalization with specific organelle markers in lace and ACC mutant cells. Homozygous lace 2 (top) or ACC 1 (bottom) mutant clones were generated in wing imaginal discs, which were analyzed using Notch antibodies together with several specific organelle markers (listed at left). Confocal z-series optical sections encompassing the entire apicobasal extent of each clone were scored for the total number of enlarged Notch-positive vesicles detected (right column) and the percentage of these Notch-positive vesicles that were colabeled by a given organelle marker (middle column). (DOC)

Table S2
Intensity difference of specific organelle markers between control and lace or ACC mutant cells. Confocal horizontal optical sections of lace 2 (top) or ACC 1 (bottom) mutant clones were compared to non-mutant control cells by measuring the average fluorescence intensity difference of several organelle markers (listed at left) between equivalently sized sectors of mutant and control tissue. Sector sizes were 403.58 mm 2 for PDI-GFP and Golgi-YFP, 408.64 mm 2 for lamp-HRP, or 100.89 mm 2 for Sara, Clc-GFP, Rab11-YFP, Rab5-YFP, and Rab7-YFP. 10 confocal optical sections were measured for each genotype/organelle marker combination; numerical data are mean intensity +/2 standard deviation. (DOC)

Table S3
Quantitative analysis of tissue overgrowth phenotypes in lace and ACC mutant clone-bearing wing discs. Summary of overgrowth phenotypes exhibited by different genotypes used in this study. Leftmost column lists the primary mutant and/or transgenic genotypes, followed by columns showing the percentages of clone-bearing wing discs exhibiting overgrowth, the total numbers of wing discs examined, and the full genotypes used to generate the mutant clones and/or transgene expression for each sample. (DOC) Table S4 Quantitative analysis of cellular overproliferation phenotypes in lace and ACC mutant clones expressing activated Armadillo. Summary of overgrowth phenotypes exhibited by lace 2 and ACC 1 mutant clones in the absence or presence of constitutively activated Armadillo, produced under control of sd-GAL4; UAS-FLP. Leftmost column specifies the mutant clone genotypes (lace 2 sd.FLP or ACC 1 sd.FLP) and whether or not activated Armadillo was also expressed in the mutant clones (armS10), followed by columns showing the percentages of clonebearing wing discs exhibiting overgrowth, the total numbers of wing discs examined, and the full genotypes used to generate the mutant clones and/or transgene expression for each sample. (DOC)