Dietary lipids modulate Notch signaling and influence adult intestinal development and metabolism in Drosophila

Summary

Tissue homeostasis involves a complex balance of developmental signals and environmental cues that dictate stem cell function. We found that dietary lipids control enteroendocrine cell production from Drosophila posterior midgut stem cells. Dietary cholesterol influences new intestinal cell differentiation in an Hr96-dependent manner by altering the level and duration of Notch signaling. Exogenous lipids modulate Delta ligand and Notch extracellular domain stability and alter their trafficking in endosomal vesicles. Lipid-modulated Notch signaling occurs in other nutrient-dependent tissues, suggesting that Delta trafficking in many cells is sensitive to cellular sterol levels. These diet-mediated alterations in young animals contribute to a metabolic program that persists after the diet changes. A low sterol diet also slows the proliferation of enteroendocrine tumors initiated by Notch pathway disruption. Thus, a specific dietary nutrient can modify a key intercellular signaling pathway to shift stem cell differentiation and cause lasting changes in tissue structure and physiology.

Introduction

Development is a unique partnership between a flexible genetic program and the environment. Major advances have occurred in recent years in understanding the genes and pathways that program embryo and tissue development, many of which are highly conserved among diverse invertebrate and vertebrate animals. In addition, numerous examples have been documented where environmental variables, nutrition in particular, exert or are believed to exert strong modulating effects on developmental outcomes and disease susceptibility (reviewed in Preston et al., 2018). Examples, include folate and neural tube closure (Imbard et al., 2013), cholesterol and heart disease (Orho-Melander, 2015), and caloric intake and longevity (Templeman and Murphy, 2018). Despite the great interest and importance of understanding how the environment modifies development, gaining a mechanistic understanding of these unprogrammed effects on animal form and function has proved much more difficult than deciphering the developmental program itself.

Diet is one of the most important and ubiquitous environmental variables impacting animal life cycles. A nutrient poor diet during embryonic development may establish a scarcity metabolic program in the offspring that can lead to metabolic syndrome (review: Ramakrishnan, 2004). Dietary intake of certain nutrients may also increase the risk of cancer. For example, there is a strong correlation between a high fat diet and colon cancer (Beyaz et al., 2016; O’Neill et al., 2016) raising the possibility that a high fat diet may influence developmental signaling pathways that promote cell proliferation. However, nutrients such as cholesterol can potentially affect cancer incidence in many ways, complicating our understanding of the connections between dietary cholesterol intake and oncogenesis (Silvente-Poirot and Poirot, 2014; Kloudova et al., 2017).

Cholesterol uptake and utilization is controlled by mechanisms that are highly conserved between Drosophila and mammals. For example, intestinal enterocytes in both groups take up dietary cholesterol by endocytosis and vesicle trafficking into lysosomes. Subsequently, under control of Niemann-Pick type C genes NPC1 and NPC2, cholesterol is moved from lysosomes to other organelles such as endoplasmic reticulum (ER) and mitochondria (Huang et al., 2005). Once in the ER a large portion of the cholesterol is esterified by acetylCoA cholesterol acyltransferase (ACAT), packaged into lipoprotein particles and transported to peripheral tissues. In mammals, cholesterol biosynthesis is controlled by the lipid-mediated regulation of protein stability in the endoplasmic reticulum (Goldstein and Brown, 1990). When cholesterol levels in enterocytes are high, the rate limiting biosynthetic enzyme, the membrane protein HMGCoA reductase, is destabilized by ubiquitination, dislocated out of the ER and degraded via the endoplasmic reticulum-associated degradation (ERAD) pathway. In contrast, in lipid-deprived cells, Insig-1, a key protein involved in HMGCoA turnover and in processing sterol regulatory element binding proteins (SREBPs), is degraded via ERAD. Even though Drosophila is a cholesterol auxotroph, mammalian lipid regulatory proteins expressed in Drosophila are still subject to sterol-regulated ERAD showing that lipid mediated control of protein stability in the ER via ERAD has been highly conserved (Faulkner et al., 2013).

The expression of genes mediating cholesterol uptake, as with many other dietary lipids, is mediated largely through nuclear receptors (NR). These ligand-regulated transcription factors recognize small lipophilic molecules and mediate broad changes in gene expression to regulate development, homeostasis, and metabolism (Evans and Mangelsdorf, 2014). Cholesterol efflux in mammals is regulated primarily by NR subfamily 1 members LXRα and LXRβ. In response to high intracellular oxysterols, LXR induces the expression of cholesterol transport proteins thereby stimulating cholesterol trafficking from the peripheral tissues to the liver where it is eliminated in the form of bile (Kalaany and Mangelsdorf, 2006). In Drosophila, a single subfamily 1 ortholog, Hr96, is expressed in the intestine and the fat body, and is required for development in cholesterol-depleted environments (Horner et al., 2009). Consequently, NR1 members and their conserved target genes are strong candidates for mediating cholesterol-induced effects on intestinal development.

The Drosophila midgut is a powerful model system in which to analyze the relationship between diet and intestinal development at the cellular, molecular and functional levels, in part because nutrient processing is highly regionalized (Buchon et al., 2013; Marianes and Spradling, 2013). The activity within each gut sub-region is strongly impacted by distinct self-renewing intestinal stem cells (ISCs) (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006) whose daughter cells, known as enteroblasts (EBs), differentiate in response to a high Notch signal into nutrient-processing enterocytes (ECs). In the presence of low intracellular Notch activity, ISCs produce enteroendocrine mother cells that differentiate into multiple types of hormone-secreting enteroendocrine cells (ee’s) (Ohlstein and Spradling, 2007; Guo and Ohlstein, 2015). Notch signaling is largely regulated by its ligand, the transmembrane protein Delta, whose activity is influenced by factors that impact the trafficking, recycling and turnover within endosomal vesicles of both Delta and Notch itself (Seugnet et al., 1997; Weber et al., 2003; Coumailleau et al., 2009; Conner, 2016). In the mammalian intestine, Notch signaling also specifies enterocytes and enteroendocrine cells (Demitrack and Samuelson, 2016). Abnormalities in Notch signaling are associated with increased cell proliferation in both Drosophila and mammals, including colon cancer (Suman et al., 2014).

We have found that dietary cholesterol influences intercellular Notch signaling and enteroendocrine cell differentiation in the adult midgut. Differences in enteroendocrine cell number established in young animals persist and appear to contribute to a metabolic memory, as flies raised initially under lipid deprivation later in life accumulate 18% more sterols on a standard diet than controls. Moreover, dietary lipid levels affect the progression of enteroendocrine tumors. Thus, our results reveal a specific and potentially widespread mechanism through which a dietary nutrient affects Notch intercellular signaling and cellular differentiation with possible consequences for adult metabolic programming and cancer susceptibility.

Results

Dietary sterols influence adult midgut enteroendocrine cell number

The Drosophila midgut completes development shortly after adult eclosion, (Ohlstein and Spradling, 2006; Takashima et al., 2013; Marianes and Spradling, 2013) like the mammalian intestine, which develops rapidly after birth. Delayed differentiation is proposed to allow intestinal metabolism to adapt to an individual’s nutrient environment (reviewed in Reynolds et al., 2015). We observed a delayed appearance of intestinal SREBP signaling (Figure 1A–B), a key pathway in the regulation of lipogenesis (Brown and Goldstein, 1997; Kunte et al., 2006). Only sparse, low level activity is present in the intestine of newly eclosed flies (Figure 1A), consistent with previous studies (Reiff et al., 2015). In contrast, the pathway is highly activated in the posterior midgut of 5-day old flies (Figure 1B). Consequently, we investigated whether midgut differentiation in young adults is influenced by the lipid content of their initial diet. We collected late stage pupae and allowed these animals to eclose and feed on either a control or lipid-depleted diet for 10 days. We then dissected these flies and stained both the anterior and posterior midgut with markers to assess their cellular composition (Figure 1C).

Figure 1.

Dietary sterols influence adult midgut enteroendocrine cell number. (A-B) Midguts from 6–12 hour old (A) or 5 day old (B) adult females. SREBP reporter (green) DAPI (blue). Scale bar = 400μm. (C) Experimental strategy to test the effects of diet on differentiation. (D-K) Posterior (D-G) or Anterior (H-K) midgut regions from animals fed a control diet (CD) (D,H), lipid-depleted diet (LD) (E, I), or lipid-depleted diet supplemented with 1X cholesterol (1X C, Fig. S1B) or 4X cholesterol (4X C) (F, J). DAPI (blue); Prospero (red). Scale bar = 40μm. (G, K) % enteroendocrine (ee) cells (Prospero⁺) relative to total posterior (G) or anterior (K) cells. Circles show individual values color-coded for diet as indicated (N = 20 for all samples). Bar = Average value; *** = p < 0.005 (t-test). (L-O) Notch signaling reporter GBE-Su(H)-GAL4 driving UAS-mCD8::GFP in posterior midgut. Animals were fed on CD (L), LD (M) or 4X C (N), for 10 days prior to analysis. DAPI (blue); GFP (green). Inset scale bar = 10μm. (O) % of total cells with Notch signaling (GFP⁺) in a 5000 μm² field. GFP⁺ cells were subdivided by ploidy based on nuclear size and DAPI intensity: blue (< 4C), red (4C-8C), green (>8C). See also Figure S1.

When we looked at the distribution of midgut cell types in adult animals fed a lipid-depleted diet, we observed a significant decrease in posterior ee cells (recognized by Prospero staining) relative to controls (Figure 1D–E). This reduction was not observed in the anterior midgut (Figure 1H,I). To confirm that the absence of dietary lipids caused the reduction in ee number, we fed newly eclosed flies a lipid-depleted diet supplemented with a single lipid species in the form of cholesterol, fatty acid, or triglyceride, and found that providing these dietary metabolites prevented the reduction in ee number (Figure1G, Figure S1B,E–H). The effect was specific to the period of initial adult gut development, because flies shifted to a lipid-depleted diet after one week on control media did not show a significant change in ee number 10 days post-shift (Figure S1D).

To determine whether elevated dietary lipid levels could likewise affect ee number, we supplemented a lipid-depleted diet with high levels of specific lipid species. Increasing the concentration of dietary fatty acid resulted in death within a week of feeding, however, animals fed a high cholesterol diet are viable and fertile. In newly eclosed animals fed a diet with four times the amount of cholesterol in a control lab diet the frequency of posterior ee cells was significantly increased (Figure 1F). When cholesterol supplementation commenced one week after eclosion, this increase did not occur, even after 14 days of high cholesterol (Figure S1D). We found that, as expected, growth on a lipid-depleted or 4X cholesterol diet significantly changed the levels of intestinal cholesterol and cholesterol ester (Figure S1C). Since Drosophila are sterol auxotrophs that normally acquire sterols from yeast rather than cholesterol, we fed newly eclosed flies on lipid-depleted media supplemented with 4x ergosterol, the major yeast sterol and observed a similar increase in the percentage of ee cells in the posterior midgut (Figure S1G–I). Cholesterol supplementation or depletion does not act by altering rates of ISC division, because the effects of these treatments on the mitotic index (ISC divisions) did not correlate with ee production (Figure S1K). Furthermore, ee’s did not change in frequency due to altered death rates, because the number of TUNEL-positive cells per midgut was unchanged following cholesterol supplementation and cell death increased rather than decreased following lipid depletion (Figure S1L).

Notch signaling regulates the differentiation of ISCs into enterocytes or enteroendocrine cells. To determine whether Notch pathway activation is affected by dietary lipids, we used a reporter for Notch signaling, GBE-Su(H)-GAL4, and found that animals fed a low cholesterol diet had significantly more midgut Notch signaling activity than animals fed a control diet (Figure 1L,M). Moreover, a lipid-depleted diet caused Notch signaling activity in enteroblasts to remain high for an extended period as they developed into enterocytes as evidenced by poyploidization (Figure 1M,O). In contrast, animals fed a 4X cholesterol-supplemented diet had significantly fewer cells with Notch reporter activity (Figure 1N,O). This suggests that increasing or reducing dietary cholesterol levels decreases or increases the level and kinetics of Notch signaling during differentiation, and hence increases or reduces enteroendocrine cell production, respectively.

Hr96 mediates cholesterol-dependent control of enteroendocrine cell number

Physiological effects of dietary cholesterol on enteroendocrine cell number would require cholesterol uptake and cellular utilization, pathways that depend on the nuclear hormone receptor Hr96 (Horner et al., 2009). Consequently, we analyzed Hr96¹ null mutant animals, fed a control diet. Hr96¹ midguts contained significantly fewer ee’s compared to heterozygous controls in the posterior region (Figure 2A–B), and also throughout the midgut (Figure S2A–H). In the posterior this reduction strongly resembles the effects of a lipid-depleted diet on young wild type adults, since ee differentiation during early adulthood in this region will be affected when cholesterol levels are reduced either via diet or Hr96 mutation. Most ee differentiation in the anterior and middle midgut occurs prior to adult eclosion, which explains the insensitivity of this region to adult dietary cholesterol restriction (Fig. 1K), but its sensitivity to Hr96 mutation, which would effectively reduce cholesterol and influence Notch signaling during ee differentiation both before and after adulthood.

Figure 2.

Hr96, NPC2b and ACAT mediate sterol availability and enteroendocrine cell number. (A-F) Posterior midguts from animals of the indicated genotypes following culture on the indicated diets. (A,D) Hr96¹ / + (as a control); (B,E) Hr96^1/1; (C,F) Hr96^2X. Animals were fed a control diet (A-C) or 4X cholesterol-supplemented diet (D-F) for 10 days prior to analysis. DAPI (blue); Prospero (red); scale bar = 40μm. (G) % ee cells relative to total posterior cells. Data points show individual values shape-coded for genotype and color-coded for diet as indicated (N = 15 for all samples). Bar = Average value; *** = p < 0.005 (t-test), n.s. = not significant. (H-I) % ee cells relative to total midgut posterior cells for animals expressing the indicated RNAi constructs driven in enterocytes by Act-GAL4 (H) or in ISCs and enteroblasts by esg-GAL4 (I) and raised on a control diet. Midgut images for these experiments are in Fig. S2L–O: RNAi control is in Fig. S2P. (J) The relative expression based on triplicate RNAseq in anterior (yellow) or posterior (blue) midguts of the nine known enteroendocrine peptide hormone genes from Hr96^2X animals (elevated ee number) compared to controls. * = Significant increases compared to control (t-test), p-value < 0.05. bars-standard deviation. See also Figure S2.

We investigated whether the normal gene dosage of Hr96 is rate-limiting for sterol utilization by examining midguts from animals carrying two tandem copies of Hr96 on each homolog (Hr96^2X) (Sieber and Thummel, 2009). Hr96^2X flies contained significantly more enteroendocrine cells in the posterior midgut (Figure 2C) and also in middle and anterior regions (Figure S2C,G), as expected if Hr96^2X increases ee differentiation both before and after eclosion. Interestingly, a 4X sterol diet did not further increase ee’s in Hr96^2x animals beyond the high levels that form on a control diet (Figure 2C,F–G). Over-expressing Hr96 only in enterocytes using Myo1A-GAL4 was sufficient to increase ee number relative to controls (Figure S2I–K). This suggests that increased dietary sterols taken up and processed by the over-expressing enterocytes may be transferred to ISCs/EBs where they would depress Notch signaling and enhance ee cell differentiation.

Drosophila has two NP1 class C genes, and eight NP2 class C genes, and several of these genes are regulated by Hr96 (Horner et al., 2009). These include Npc1b which is required for dietary cholesterol absorption (Voght et al., 2007), and at least two Npc2 class genes involved in intracellular cholesterol trafficking, Npc2a and Npc2b (Huang et al., 2005; 2007). To examine whether Hr96 acts on ee number by controlling sterol transport genes, we used two different RNAi lines to suppress the Hr96 target gene Npc2b in the adult flies (Fig. S2R). The posterior midguts of animals expressing either RNAi construct contained fewer ee’s compared to controls (Figure 2H; Figure S2L–M). This was similar to the behavior of animals on a lipid-depleted diet, consistent with the predicted inability of Npc2b mutant animals to take up and utilize sterols. Interestingly, knocking down Npc2b only in stem cells and undifferentiated cells using esg-GAL4 in the intestine still reduced ee number (Figure 2I; Figure S2O–P). Thus, altering cholesterol transport and availability only in undifferentiated cells is sufficient to influence the differentiation of intestinal stem cell daughters.

Another important and conserved gene required for steroid metabolism in mammals and Drosophila is ACAT, which mediates sterol esterification, transport and storage. We used RNAi to knock down the gene encoding the Drosophila ACAT homolog CG8112 (Fig. S2R; Horner et al., 2009), and found a subsequent increase in the percentage of ee’s in the midgut, similar to the effects of high dietary cholesterol (Figure 2H–I, Figure S2N,Q). This is consistent with the expectation that ACAT knock-down will increase free intracellular sterols. These experiments strongly support the view that the Hr96 and Npc2b-dependent uptake of dietary sterols in the midgut influences the differentiation of enteroendocrine cells in young adults.

To address the autonomy of Hr96 function, we generated Hr96¹ homozygous ISC clones and stained them for a clonal marker and Delta (Figure 2K–L). Clonal mutant cells showed higher Delta immunofluorescence and a greatly increased frequency of Delta persistence in pairs (Fig. 2L), as expected if Hr96 function is required cell autonomously.

Dietary lipids increase the abundance of diverse posterior enteroendocrine cell types

Enteroendocrine cells comprise multiple subtypes expressing different peptide hormones that are organized regionally along the anterior-posterior midgut axis, presumably in response to transcription factor differences (Marianes et al., 2013; Veenstra and Ida, 2014). All these ee subtypes are generated from ISCs and their differentiation is influenced by Notch signaling in a similar manner. If dietary lipids modulate ee differentiation primarily through changes in Notch signaling, then the production of all classes of ee cells in posterior regions should be affected. In contrast, only one or a few subclasses of ee’s might be affected if dietary lipids only stimulate cells that function in concert with lipid-metabolizing enterocytes. To assess the diversity of ee subtypes following dietary or genetic treatments, we performed RNAseq in triplicate on anterior and posterior midguts from wild type animals fed a control, lipid-depleted, or cholesterol-supplemented diet. Parallel studies were done on anterior and posterior midguts from Hr96¹ and Hr96^2X mutants.

The results of these studies strongly support the view that dietary lipids affect all or nearly all subclases of ee cells through a general influence on ee differentiation mediated by altered Notch signaling. Since Hr96^2X animals contain the largest increase in enteroendocrine cells, approximately three-fold in the posterior, we examined the expression levels of known enteroendocrine hormone genes in Hr96^2X midguts to query which ee subtypes were likely to have increased. Six of the seven enteroendocrine hormone genes known to be expressed in posterior ee cells (Marianes et al., 2013; Veenstra and Ida, 2014) were significantly increased 3–8 fold in posterior midguts from Hr96^2X animals compared to controls (Fig. 2J). The bHLH transcription factor Dimmed, which mediates peptide secretion via the large dense core vesicle pathway used by ee cells (Beehler-Evans et al., 2015), was also significantly increased (1.27 +/− 0.18 vs 0.76 +/− 0.26 p<0.05, t-test) in Hr96^2X posterior midguts, consistent with the overall increased number of functional, neuropeptide-secreting ee cells. Expression of RNAs encoding two enteroendocrine hormones, AstC and Orcokinin, also rose in the anterior midgut of Hr96^2X animals (Figure 2J). An enteroendocrine cell subtype expressing these hormones has been described in both the anterior and middle midgut (Veenstra and Ida, 2014).

Dietary sterols modulate Delta levels and trafficking

A lipid-depleted diet only produced modest increases in Notch pathway gene expression (Fig. S2Q). Consequently, to better understand the dramatic changes in Notch signaling induced by such a diet (Fig. 1L–O) we immunostained the Notch ligand Delta to visualize its expression and subcellular localization in the posterior midguts of 10-day old animals fed various diets. Delta was found in cytoplasmic vesicles in control animals within scattered diploid cells that correspond to ISCs (Figure S3A; Ohlstein and Spradling, 2006). ISCs from animals fed a lipid-depleted diet consistently stained more strongly for Delta compared to controls (Figure S3B,D). Moreover, in these animals Delta staining could be readily detected within daughter enteroblasts, generating pairs of Delta-labeled cells (Figure S3B). Since posterior midgut size did not increase as a result of lipid depletion, these cell pairs were likely caused by reduced Delta turnover in enteroblasts under lipid-deprived conditions rather than stem cell amplification divisions (O’Brien et al., 2011). Conversely, feeding young animals a high sterol diet reduced the level of Delta staining within ISCs (Figure S3C–D).

We examined Delta levels in animals mutant for cholesterol uptake genes to verify that these changes in Delta content depended on physiological lipid uptake. Hr96¹ mutants on a normal diet showed elevated levels of Delta in ISCs compared to controls (Figure 3A–B). When these animals were grown on lipid-depleted media, the levels of Delta were further increased and abundant Delta-rich cytoplasmic vesicles were now readily apparent (Figure 3E–F). In contrast, Delta levels were scarcely detectable and foci were rare in Hr96^2x animals, even when fed a lipid-depleted diet (Figure 3C,G). NPC2b mutant animals, which like Hr96¹ mutants are defective in cholesterol incorporation and mimic wild type animals on a lipid-depleted diet, showed higher levels of cytoplasmic Delta staining compared to control (Figure S3E–G). Like Hr96¹, in Npc2b RNAi animals, Delta protein expression persisted in enteroblasts and young enterocytes (Figure S3F–G), like control animals on a lipid-depleted diet (Figure 3E).

Figure 3.

Hr96-mediated sterol metabolism regulates Delta protein levels and trafficking. Posterior midguts are shown from (A,E) Hr96^1/+ (B,F) Hr96^1/1 and (C,G) Hr96^2X animals fed a control diet (A-C) or a lipid depleted diet (E-G) for 10 days prior to analysis. DAPI (blue) and Delta (white). arrows: ISCs (A,C); ISC-enteroblast pairs (B). (D,H) Corrected fluorescence intensity per Delta⁺ cell is plotted in arbitrary units from animals of the indicted genotypes and is comparable between all 6 experiments. Individual values are shape-coded for genotype as indicated. N = 15 (D); N = 20 (H). Bar = Average value; *** = p<0.005; (t-test). Scale bar = 20μm for all images. (I-J) Flies were raised on food containing 30μM vehicle (I) or NMS-873 (J) and after four days posterior midguts were analyzed for Delta expression. (K) Integrated Delta fluorescence (arbitrary units) of cells from I, J showing 4-fold increase in Delta staining by the p97 inhibitor. (L) Survival plot of control flies or Hr96¹ mutant flies treated with the indicated concentrations of NMS-873. See also Figure S3.

Taken together, these observations show that when lipids including cholesterol are low due to diet or a genetic defect, ISCs continue to proliferate but turn over Delta more slowly than normal. This change in Delta persistence matches the increased Notch signaling activity previously observed on lipid-depleted diets (Figure 1M) and is likely to favor EC over ee production from ISCs. Conversely, Hr96^2x animals on a control diet (Figure 3C) or ACAT RNAi animals on a control diet (Figure S3H) showed reduced Delta expression. This is similar to wild type animals fed a high sterol diet (Figure S3C) and is consistent with their reduced Notch signaling activity (Figure 1N), a condition likely to favor ee over EC production. These data strongly suggest that gut lipid uptake and utilization influences endomembrane lipid composition, and that these changes influence Delta abundance through their effects on Delta subcellular localization, trafficking and turnover.

Given that lipid-stimulated-ERAD requires the activity of p97 to facilitate the turnover of target proteins we hypothesized that p97 inhibition would mimic elevation of Delta levels seen in the intestine of Hr96 mutants and in animals fed a lipid-depleted diet. Consistent with this prediction, treating flies with the allosteric p97-inhibitor, NMS-873, stimulated Delta protein accumulation, indicating that p97-mediated protein turnover regulates Delta protein levels (Figure 3I–K). Moreover, these data are consistent with the model that altering lipid levels in the intestine induces an ERAD-like degradation of Delta protein similar to what has been observed with proteins like HMG-CoA reductase (DeBose-Boyd, 2008). Consistent with the fundamental nature of this relationship between sterol metabolism and p97-mediated protein turnover we observed that, while control animals showed no defects in survival upon p97 inhibition, Hr96 mutants showed a pronounced sensitivity to decreases p97 activity (Figure 3L). This synthetic interaction between Hr96 and p97 supports the idea that sterol homeostasis and p97 converge to regulate Delta, and potentially many other proteins, in the intestine.

Notch is also modulated by lipid uptake and utilization

We investigated whether the abundance and trafficking of Notch is also affected by dietary lipid utilization. Proper recycling of the Notch extracellular domain (NECD) is crucial for accurate Notch activation or suppression and is thought to couple to Delta turnover (Coumailleau et al., 2009). If lipid depletion impairs Delta endosomal recycling or degradation, we reasoned that the NECD may also display aberrant turnover and trafficking. Using antibodies against the NECD we examined the intestines of flies fed a lipid-depleted diet and the intestines of Hr96 mutants, and found that NECD staining was present in putative ISC-EB cell pairs (Figure 4B,E). In contrast, NECD was only detectable in single cells in control animals (Figure 4A,E). Furthermore, NECD-positive cells had less antibody staining in high sterol or Hr96 overexpression animals (Figure 4C,H). Taken together, these data further support the finding that cellular sterol levels strongly influence the trafficking and turnover of Notch pathway proteins, including Delta and Notch NECD. As with Delta protein, these results with NECD were paralleled by altering cholesterol metabolism with Hr96 mutation (Figure 4F–G,I–J) or duplication (Figure 4H–J).

Figure 4.

Diet and Hr96 regulate Notch extracellular domain (NECD) protein levels and trafficking. (A-C) DAPI (blue) and NECD (white) is shown in posterior midguts of wild type animals raised on a control (A, A’), lipid-depleted (B, B’) or 4X cholesterol (C,C’) diet, or from Hr96^1/+ (F) Hr96^1/1 (G) or Hr96^2X (H) animals raised on a control diet. (D,I) Corrected fluorescence intensity per NECD⁺ cell is plotted in arbitrary units from animals of the indicted diets or genotypes and is comparable between all 6 experiments. Individual values are shape-coded for genotype as indicated. N = 15 (D); N = 20 (I). Bar = Average value; *** = p<0.005; (t-test). (E,J) The % NECD-positive cells per field in the experiments plotted in (D,H) is shown according to ploidy, as colored coded. Scale bar = 10 μm for all images. See also Figure S4.

Sterol regulation of Delta is also observed in the ovary and in human cells

To investigate whether the relationship between cellular sterol levels and Notch signaling described above for the Drosophila midgut also applies to other tissues, we examined the Drosophila ovary. Oogenesis is highly dependent on efficient nutrient processing, since proteins, lipids, and carbohydrates, all initially processed by the midgut, are accumulated in a tightly regulated, stepwise manner to support the growing oocyte (Sieber and Spradling, 2015). Like midgut enterocytes and enteroendocrine cells, the follicle cells of the ovary also rely on Notch signaling to induce follicle cell differentiation at the mitotic-endocycle (ME) transition, and to coordinate somatic and germline development (Sun and Deng, 2005). Diet strongly influences the rate of follicle development and passage through checkpoints (Drummond-Barbosa and Spradling, 2001), so we used genetic manipulation of sterol metabolism to look for connections to Notch signaling.

Delta was present on membranes at only low levels in wild type mid-stage follicles (Figure 5A), consistent with previous reports (Bender et al., 1993). However, significantly more Delta was present on membranes from similarly staged Hr96¹ mutant follicles (Figure 5B). Furthermore, large cytoplasmic Delta foci were observed in Hr96¹ mutant follicles (Figure 5D), but not in controls (Figure 5C), reminiscent of the Delta foci in midgut ISCs and developing enterocytes. Similarly, we found Hr96¹ ovaries displayed increased levels of NECD (Figure 5F) relative to controls (Figure 5E). Large NECD aggregates were present in stage 9–10 follicles of Hr96¹ mutants (Figure 5F), but not controls (Figure 5E). Corresponding increases in membrane Delta and NECD were observed in ovaries from NPC2b-RNAi animals as well (Figure S5A–D). To determine whether these effects on protein levels and aggregation were specific to components of the Notch pathway, or affected other transmembrane proteins, we stained control and Hr96¹ mutant follicles with an antibody against the LDLR homolog LpR2 and found no significant change in LpR2 levels or localization (Figure 5I–J). These observations argue that sterol levels have similar effects on Notch signaling components in ovarian cells as in the midgut.

Figure 5.

Delta and NECD levels and trafficking are specifically modulated by lipids in the ovary and in mammalian cells. Notch (A-D), NECD (F-G), or Lpr2 (I-J) immuofluorescence is shown in Drosophila ovarian follicles from Hr96^1/+ (A,C,F,I) or Hr96^1/1 (B,D,G,J). C and D show Delta expression in a large polyploid nurse cell. DAPI (blue), Delta (red). Scale bar = 50μm. (E) Corrected Delta fluorescence intensity is plotted in arbitrary units (R.F.U) from Control (Hr96^1/+) Hr96^1/1 or NPC2b-RNAi (Fig. S5B) follicles. ** = p<0.01; (t-test). (H) Corrected NECD fluorescence intensity is plotted in arbitrary units (R.F.U) from Control (Hr96^1/+) Hr96^1/1 or NPC2b-RNAi (Fig. S5D) follicles. ** = p<0.01; (t-test). (K-L) show HCT116 cells treated with vehicle (K), or Simvastatin (L) and stained with the fluorescent cholesterol-binding agent filipin, Scale bar = 25μm. (M,N) show HCT116 cells treated with vehicle (M), or Simvastatin (N) and stained for Notch1-ECD. Scale bar = 10μm. O. Integrated fluorescence per cell from M,N. *** = p<0.001. See also Figure S5.

Notch signaling regulates the ME transition in follicle cells during stage 5–6 of oogenesis (Sun and Deng, 2007). Late stage follicles in Hr96¹ mutants had significantly more follicle cells than similarly staged controls (Figure S5E–F,I), indicating that the Notch-induced ME transition had been delayed. Moreover, Hr96¹-mutant stage 14 follicles (Fig. S5H) contained follicle cell nuclei that had been extruded from the follicular layer (arrows), further suggesting that a delayed or incomplete switch to the endocycle resulted in excess follicle cells (Figure S5G–H). Normally, the Notch-regulated gene hindsight (hnt) is expressed in follicle cells after the ME transition (Figure S5K). In contrast, Hr96 mutants expressed Hnt slightly earlier than normal, but Hnt levels remained lower than in control follicles (Figure S5L) suggesting that altered Notch signaling ultimately delayed rather than accelerated normal follicle development. This defect in Notch signaling may stem from the fact other sterol dependent pathways such as ecdysone signaling also function to regulate the M/E transition.

Since all the components linking intracellular sterol levels to Notch signaling are highly conserved between Drosophila and mammals, we investigated whether human cells show similar responses. We identified a human colon cancer cell line, HCT116, that responds to the statin Simvastatin by losing membrane cholesterol and accumulating it in internal vesicles (Fig. 5K–L), without altering the amount of ER (Fig. S5M–N). Simvastatin treated HCT-116 cells accumulated Notch1-ECD internally (Fig. 5M–O). Levels of the Delta-like Notch ligand 4 (DLL4) also increased following Simvastatin treatment (Fig. S5P–Q).

We identified another human cell line, HEK293, that expresses the cholesterol regulator LXR, as well as DLL4 and asked whether promoting intracellular reverse cholesterol transport with the LXR agonist GW3965 would affect DLL4 protein levels. DLL4 protein levels were consistent in cells treated with vehicle alone, but after treatment with the LXR agonist, DLL4 protein decreased significantly in 5 independent experiments (Figure S5Q–R). The results with both cell lines suggest that intracellular sterol levels act in a conserved manner on Notch ligand stability and Delta-Notch signaling in at least some mammalian cells.

Early lipid depletion reduces posterior ee numbers and establishes a scarcity metabolic state

Our observation that dietary lipids and cholesterol in particular enhance Notch signaling in the midgut and ovary raises the question of how such a connection affects the animal phenotype. For example, do changes in ee number mediated by dietary lipids during early adult life persist and influence the animal’s response to a different dietary regime? To determine whether nutrients in the early diet influence metabolism after the diet has changed, we raised young flies on lipid-depleted food for one week to increase Notch signaling and decrease posterior ee number. We then transferred the flies to normal yeast food and observed that the reduction in posterior ee number persisted for at least two additional weeks (Figure 6A). Interestingly, flies raised initially on lipid-depleted media, during one week back on normal yeast food, accumulated 18% more whole body cholesterol per fly than animals that had eaten a normal diet beginning at eclosion (Figure 6B). This suggests that low lipid levels early in development generate a scarcity metabolic phenotype that parallels the reductions in ee number, and that persists and programs greater lipid storage in the subsequent week under normal dietary conditions. When tested two weeks after switching from a lipid-depleted to a normal diet, the animals still had fewer posterior ee cells and still stored 14% more whole body cholesterol than controls that were never lipid-deprived (Figure 6A–B). Thus, the cellular and metabolic effects of nutrient starvation early in adult life can persist for a significant period in the absence of the original stimulus.

Figure 6.

A lipid-depleted diet generates a persistent scarcity metabolic state and slows enteroendocrine cell tumor progression. (A) A lower percentage of ee’s is present in the posterior midgut of flies raised on a lipid-depleted diet for 7d and then shifted to control diet for 1 or 2 additional weeks (LD-CD: purple), than in flies raised continuously on a control diet (CD-CD, black). Flies raised and maintained on an LD diet (LD-LD, red) also differed. (N = 10); (B) Total body cholesterol per fly (relative to CD-CD values as 1.0) is greater in LD-CD treated animals. bars-standard deviation. ** = p<0.01, *** = p<0.005 (t-test). (C-H) Posterior midguts from (C) control (Esg::GFP) or (D-G) Notch-RNAi (Esg::N-RNAi) fed (D-E) a control (CD) or (F-G) a lipid-depleted (LD) diet for 9 days. DAPI (blue); GFP (green); Prospero (red, nuclear) and Delta (red, cytoplasmic). Scale bar = 40μm. (H) Plot quantitating reduced tumor burden of flies on LD compared to CD. (I-M) Posterior midguts from animals fed the indicated diets and showing individual MARCM clones marked with GFP for controls (I-J) or N^−/− cells (K-L) 9 days after induction. Scale bar = 10μm. (M) Mean clone size from experiment in (I-L) showing significant reduction in N^−/− tumor clone size on LD food. ***p<0.001 (t-test). (N) Model of dietary sterol-mediated regulation of Notch signaling via changes in Delta and Notch trafficking and stability in the ER.

Dietary lipids influence enteroendocrine tumor development

When Notch signaling is strongly reduced in ISCs they give rise to a rapidly proliferating neoplasia composed of ISC-like and enteroendocrine-like progenitors (Micchelli and Perrimon, 2006; Ohlstein and Spradling, 2006; Patel et al., 2015). Strikingly, these “enteroendocrine tumors” preferentially form in lipid-rich regions of the intestine, which are mostly located in the posterior midgut (Marianes and Spradling, 2013). Our finding that elevated dietary sterol downregulates Notch signaling suggests an explanation for the preferential development of enteroendocrine tumors in lipid-rich midgut regions. ISCs in lipid-rich zones would likely contain elevated levels of membrane cholesterol and other lipids that would further depress Notch signaling by the mechanism described here, enhancing the effect of genetic lesions in the Notch pathway. Thus, tumor formation might represent another adult characteristic affected by diet.

If increased dietary lipids suppress Notch signaling in stem cells and thereby promote enteroendocrine tumors, we reasoned that a lipid-depleted diet, which increases Notch signaling, might slow tumor development. We induced tumor formation by expressing Notch RNAi in progenitor cells throughout the intestine using esg-GAL4, and tracked tumor development quantitatively by counting the number of clustered tumor cells that were labeled using UAS-GFP and Prospero staining. By 9 days after tumor induction, animals on control diets had large to massive tumors (Fig. 6D–E), while animals fed lipid-depleted food had tumor burdens ranging from scarcely detectable to moderate (Figure 6F–H). To further analyze the effects of diet we induced ISC clones lacking Notch and counted clone size as a function of diet (Figure 6I–M). Nine days after induction, Notch mutant clones were significantly smaller in animals on a lipid-depleted diet than in animals on a control diet (Figure 6M).

Discussion

Nutrient regulation of development

Both unicellular organisms and metazoans maximize their utilization of scarce environmental resources such as sterols. Mechanisms for optimizing cholesterol uptake, biosynthesis, transport and excretion evolved early and are widely shared throughout unicellular and metazoan animals and plants. Perhaps because lipid uptake inherently involves endocytosis and vesicle trafficking, and affects membrane lipid composition, regulatory mechanisms controlling lipid utilization in animals are centered on the stability of key regulatory endosomal membrane proteins (Brown and Goldstein, 1990).

Our experiments imply that this ancient system extends beyond nutrient homeostasis and metabolism to encompass tissue development and maintenance. As the locus of nutrient uptake, the intestine is the tissue most directly responsible for bringing environmental nutrients into balance with organismal requirements, both in the short term and by anticipating future needs. The Notch signaling pathway controls the differentiation of enterocytes and enteroendocrine cells downstream from intestinal stem cells that are among the most active in the body. Thus, linking sterols to Notch signaling provides a pathway for these nutrients to rapidly modulate the cellular machinery controlling both their immediate and long-term utilization. The ovary makes great metabolic demands on females by abundantly producing nutrient-laden oocytes surrounded by follicle cells. In Drosophila, ovarian germline and somatic cells, like intestinal cells, arise from adult stem cells and differentiate under the influence of Notch signaling. Thus, sterol-mediated regulation of ovarian cell differentiation potentially provides a control on the tissue making the largest demand for sterol production.

Not all tissues that require Notch are affected by alterations is steroid availability. Unless a significant number of new cells are being generated, altered Notch signaling may have little effect on a tissue’s cellular composition. We observed no defects in wing vein patterning or bristle development in the Hr96 mutant. Even sensitive tissues have limits to the degree of modulation. When Hr96 is overexpressed, the accumulation of additional ee cells could not be increased further by a high sterol diet (Fig. 2G). In the absence of all Hr96 function in null mutant animals a low level of ee cells was still produced regardless of diet. This suggests that the regulation of Notch signaling by environmental lipids has natural limits that maintain at least a basal level of tissue function even under extreme dietary circumstances. This ensures at a minimum that the tissue will still be able to respond if the nutrient environment becomes more favorable.

Mechanism of dietary influence on Notch signaling

We have found that endosomal trafficking of the Notch pathway membrane proteins Notch and Delta is lipid sensitive, like other key membrane regulators of nutrient utilization and metabolism. Delta (and NECD) localization changed dramatically on a lipid-depleted diet or in the Hr96 mutant, appearing in more vesicles and larger vesicles spread throughout the cytoplasm, and at higher levels associated with the plasma membrane. These changes, amounting to 3–6 fold increased integrated Delta immunofluorescence per cell (Figure 3), may result partly from small (~2X) increases in Delta mRNA, but mostly from reduced Delta turnover. The greater stability of Delta was the likely explanation for its increased perdurance downstream from the stem cell in the enterocyte lineage. Normally only ISCs have high Delta levels, but in flies starved for cholesterol or bearing an Hr96 mutation, Delta remained at high levels in enteroblasts and in differentiating enterocytes up to and including cells with 8c genomes. This implies that Delta persistence under lipid-starved conditions increases from perhaps an hour in normal enteroblasts to more than the 48 hours needed for an enteroblast to develop into an 8c enterocyte. Conversely, under conditions of excess lipids or Hr96 duplication, Delta levels were reduced 2–3 fold based on fluorescence measurements.

Since the effects of dietary lipids on Notch signaling and ee production depended on the Drosophila LXR homolog Hr96, changes in cholesterol homeostasis in the membrane compartments of the affected cells are likely involved. Many studies have shown that a high-fat diet induces ER stress and stimulates ER associated protein degradation (ERAD). Moreover, the changes in Delta cytoplasmic distribution are exactly those expected for an alteration in stability within endosomes. However, the substantial changes in Delta and NECD stability we observed did not affect all membrane proteins and mostly likely result from regulatory mechanisms that target these proteins specifically. Classical studies have shown that cholesterol levels can control the turnover of specific cholesterol biosynthetic enzymes like HMG-CoA reductase (Goldstein and Brown, 1990; DeBose-Boyd, 2008; Nakanishi et al., 1988). Low cholesterol conditions stimulate ER protein trafficking of SREBP proteins (Wang et al., 1994) and can result in different proteins, such as the ER regulatory protein Insig-1, being targeted for ERAD (Gong et al., 2006). Strong evidence that high sterol levels trigger Delta turnover via an ERAD-like process was provided by the effects of inhibiting the key ERAD enzyme p97, and by the synergistic effects of Hr96 mutation and p97 inhibition, which caused lethality.

Dietary sensitivity of Notch signaling causes midgut adaptation

Our experiments provide new insight into how specific nutrients early in life influence intestinal structure and metabolic programming. Dietary cholesterol early in adult life influences ee production in a direction likely to better adapt the organism to its initial nutrient environment. In mammals, after birth crypts begin to develop, new cell types emerge, and certain regional characteristics are formed (Carulli et al., 2014; Hudry et al., 2016). Many studies (Bates et al., 2006; Rakoff-Nahoum et al., 2015) have suggested that these postnatal effects on development may stem from seeding the intestinal flora. However, given that these later events in epithelial development occur when the animal begins to feed it is also likely that dietary nutrients play a major role in the maturation of the adult intestine in mammals.

Although altered Notch signaling may affect metabolism in multiple ways, changes in the abundance of at least some types of ee’s have the potential to significantly influence nutrient utilization. The enteroendocrine hormone AstA, which increased threefold under conditions of high cholesterol utilization, regulates feeding (Chen et al., 2016). In the brain AstA controls production of adipokinin and insulin, whose levels influence the amount of lipid stored in the fat body (Hentze et al., 2015). We also observed an increase in Tk, an enteroendocrine hormone that negatively regulates lipid biosynthesis in the intestine and transport throughout the body (Song et al., 2014).

Lipid-deprivation-mediated reductions in ee number persisted for at least two weeks after the animals were shifted to a rich diet. Lower ee numbers may explain in part our observation that 18% more whole body cholesterol was stored under these conditions, than in flies that had been raised continuously on rich medium. Persistent changes in tissue cellular composition represent an attractive model for understanding metabolic adaptations in mammals. Mammalian development is strongly dependent on the mother’s nutrition, and children born to malnourished mothers often begin life with a low birth weight (Lechtig et al., 1975). These individuals are more likely to become obese later in life if nutrients are not restricted (Ramakrishnan, 2004). Low nutrient exposure early in life appears to trigger metabolic changes that enhance nutrient scavenging by the developing fetus. After birth, these resets become a permanent state that puts individuals at risk for metabolic syndromes in a normal nutritional environment (Reusens et al., 2011).

Dietary sterols, Notch signaling and cancer

The connection between sterol metabolism and Notch signaling is not only important for metabolic adaptation but has potentially strong implications for cancer etiology and therapeutics. Mutations in the Notch signaling pathway have been associated with cancer deriving from many tissues in the digestive tract including the liver, pancreas and the colon. Furthermore, Notch signaling has been a common therapeutic target for these types of cancer. However, drugs and therapies targeting Notch signaling commonly have severe side effects throughout the body complicating their use with patients. We found that stimulating reverse cholesterol transport using LXR agonists in human embryonic kidney cells (HEK293) caused a dramatic and highly reproducible decrease in DLL4 protein levels. This suggests that sterol metabolism may have a highly conserved role in the regulation of Notch signaling that could be exploited for therapeutic purposes. For example, sterol regulating drugs such as LXR agonists may provide a way to alter Notch signaling in a tissue-directed manner for adjuvant therapy and as a means to reduce cancer risk in genetically susceptible patients. Inhibiting DLL4 is sufficient to reduce tumor growth and initiation in mice and humans (Hoey et al., 2009).

Numerous studies have shown that consuming a high-fat diet increases susceptibility to several specific types of cancer (Huang et al., 2012; Tang et al., 2012; Beyaz et al., 2016). High-fat diets have been proposed to increase chronic inflammation in tissues including liver and intestine (Ding et al., 2010; Miyagi et al., 2010; van der Heijden et al., 2015). Our study provides a specific mechanism to explain these connections. Large quantities of dietary lipids, particularly sterols, may directly inhibit tumor suppressor pathways, such as Notch signaling, and increase cancer susceptibility in a manner that synergizes with inflammation. In contrast, consuming a low-fat diet has been suggested to protect against multiple types of cancer. Consistent with this idea we found that feeding flies a diet low in sterols suppresses enteroendocrine tumor growth and tumor size. Taken together our work suggests that dietary lipid levels play a central role in dictating cancer risk for digestive track tumors and that dietary intervention in conjunction with the use of cholesterol modulating drugs may provide an effective strategy for reducing cancer risk in genetically predisposed patients. In sum, our work reveals new connections between environmental nutrients and developmental signaling. Further characterizing these connections will help reveal how environmental nutrients can positively impact tissues that have become stressed due to age, disease or the presence of cancer.

Star Methods

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Allan Spradling spradling@carnegiescience.edu).

Experimental Model and subject details

Drosophila melanogaster.

Normal and modified dietary treatments

To enhance uniformity, all diets were prepared from lipid-extracted ingredients. The lipid-depleted diet was prepared by extracting yeast extract and agar with chloroform at a 1:5 ratio by weight. Each component was extracted for 4 hours, separated with vacuum filtration, and then extracted for 1 hour with fresh chloroform. Both components were then allowed to dry for five days in a fume hood. Extracted yeast extract and agar were used to make a 1.0 SY diet according the following formula: 30g sucrose, 30g lipid-extracted yeast extract, 3g agar, 300 mL H₂O (Sieber and Thummel, 2009). The control diet consisted of a lipid-depleted 1.0 SY diet supplemented with fresh yeast paste and vehicle alone (1–4% ethanol) to control for the ethanol in the lipid-supplemented diets. Lipid supplemented diets (1X) consisted of the lipid-depleted 1.0 SY diet supplemented with free fatty acid (5 mg/ml stearic acid), triglyceride (5 mg/ml glycerol tristearate), cholesterol (5 mg/ml) or ergosterol (5 mg/ml) to a final concentration of 0.05% supplemental lipid. 4x cholesterol (high sterol diet) and 4x ergosterol were prepared to a final concentration of 0.2% supplemental sterol. Either pupae or 5-day old adult flies were transferred to vials with the diet of interest. Flies moved to a new dietary treatment were tested after an indicated period during which they were transferred to fresh vials containing the media every other day.

Mammalian cell culture

HEK293 cultures

Cultures of HEK293 cells grown petri dishes in Dulbecco’s modified Eagle’s medium (+10% FBS, 1x antibiotic) at 37°C and 5% CO₂. Cultures were split at 80% confluency.

HCT116 cultures

HCT116 cells were cultured in petri dishes in McCoy’s medium (+10% FBS, 1x antibiotics) at 37°C and 5% CO₂. Cultures were split at 80% confluency.

Methods details

Drosophila melanogaster

Immunofluorescent staining

Midguts or ovaries were dissected in Grace’s insect medium and fixed for 30 minutes at room temperature (or 15–18 hours at 4°C for midgut Delta staining) in antibody wash with 4% EM grade paraformaldehyde). Tissues were then washed three times for at least 2 hours and then blocked at room temperature in antibody wash +5% BSA for 4 hours. Primary antibodies were added to incubate overnight at 4°C at the following concentrations: anti-Prospero (Developmental Studies Hybridoma Bank MR1A;), RRID:AB_528440) (1:100), 1:100, anti-PH3, Ser10 (Millipore Sigma) 1:1000, (1:100), anti-Delta (DSHB C594.9B, RRID:AB_528194) (1:50), anti-NECD (DSHB C458.2H, RRID:AB_528408), anti-LpR2 (Parra-Peralbo E; PLoS Genet. 2011 Cat# LpR2, RRID:AB_2569135) (Parra-Peralbo and Culi, 2011), anti-Hindsight (DSHB 1G9, RRID:AB_528278) (1:100). Samples were washed for at least 9 hours. Secondary antibodies: Alexa-488, Alexa-568, and Alexa-596 incubated overnight at 4°C in antibody wash. Samples were then washed and counterstained with DAPI at 0.5 μg/ml. Tissue was then mounted with Vectashield and imaged. Antibody wash used for guts contained 1xPBS + 0.3% Triton X-100 + 0.5% BSA. Antibody wash for ovaries contained 1xPBS + 0.1% Triton X-100 + 0.5% BSA. Tunel labeling was performed before incubation with secondary antibodies per manufacturer’s instructions using The DeadEnd™ Fluorometric TUNEL System (Promega, G3250).

Analysis of enteroendocrine cell frequency

Physical characteristics of the Drosophila midgut described in Buchon et al., (2013) and Marianes and Spradling (2013) were used to identify anterior (A2, specifically) and posterior (either P1 or P3) midgut regions. Percent ee’s were scored by counting the number of Prospero-labeled nuclei divided by the total number of DAPI positive cells. For each quantified intestine plotted in Figures 1, 2 and 6, three 2500μm² regions were averaged.

Fluorescence quantification

Quantified Delta and NECD fluorescence reported in Figure 3, 4 and 5 was measured using the integrated density measurement function in ImageJ. Corrected total cell fluorescence was calculated by subtracting mean fluorescent background from the same image. Each graphed point is the average of 3 cells within a 2500μm² region. At least 10 midguts were analyzed for each experiment.

Notch tumor induction

Notch knock-down was performed using UAS-N-RNAi; esg-GAL4; GAL80^ts, UAS-GFP flies raised at 18°C and then moved to 29°C upon eclosion. Knock-down flies with a supplemented diet were introduced to this diet at eclosion, and transferred to fresh media every day. Notch^−/− and FRT19A control clones were induced in 5 day old flies with one 30-minute heat shock at 37°C.

RNA sequencing

Female y¹ w¹ flies aged 10–12 days were dissected in cold Grace’s insect medium. Anterior (a1-a3) and posterior regions(p1-p4) were identified morphologically and isolated using microdissection scissors. 30–50 cut regions per sample were transferred to 400 μl Tripure on ice. Each experimental condition was collected in triplicate. Collected midguts were homogenized and 600μL fresh Tripure added. After 10 min at room temperature 180μl chloroform was added, shaken briefly by hand and allowed to stand at room temperature for 10 minutes. Following centrifugation for 15 min at 12,000 rpm at 4°C, the aqueous layer was removed and RNA precipitated with 400 μl isopropanol. The pellet was recovered by centrifugation, re-suspended in 50uL nuclease free water and stored at −80°C. cDNA libraries were constructed from poly(A)-selected RNA using the llumina TruSeq RNA Library Prep Kit v2 and sequenced on an Illumina NexSeq 500.

Sequenced reads were analyzed using bcl2fastq v2.17.1.14 for base calling, Bowtie 2.2.9 for alignment to the dm6 Drosophila genome, TopHat 2.1.1 for alignment to transcripts defined by BDGP6 and Ensembl.85.gtf. Transcript and gene fpkm was calculated using Cufflinks 2.2.1 and fold change quantified with cuffdiff (v7). We analyzed genes whose expression was >2 FPKM and with a q-value <0.05. Genes reported as significantly changing had at least 2-fold increased or decreased expression levels.

qRT-PCR

RNAi knock-down efficiency reported in Figure S2P was measured in dissected midguts. RNA was collected as described for RNA sequencing and then 1μg of total RNA was reverse transcribed using Bio-Rad iScript Reverse Transcription Supermix. Quantitative PCR was performed using iTaq™ Universal SYBR^® Green Supermix. Biological and technical triplicates were used for each genotype. Results were normalized to rp49.

Cholesterol Measurements

Cholesterol and cholesterol ester (cholesteryl) in single flies or in isolated intestines were measured exactly as described in Sieber and Thummel (2012). Single female flies or 25 midguts from female flies were dissected and rinsed in 1X PBS + 0.5% Triton X-100 (PBST). Tissue was homogenized in 250μl in PBST and the volume raised to 900μl. Samples were sonicated 3 × 30 seconds and split into two samples. 10μl cholesteryl esterase was added to one sample and both samples were incubated at 37°C for 18 hours. Samples were extracted with a 2:1 chloroform:methanol solution. Lipids were solubilized in 500μl PBST and then sonicated 3× 30 seconds. Cholesterol and cholesterol ester were measured using the Amplex Red Cholesterol Assay Kit and samples read per kit instructions.

NMS-873 Treatment

Ten 5-day-old female flies were put with an equal number of male flies in a vial containing 1.0 SY media. Flies were fed fresh yeast paste with either 15μM or 30μM NMS-873, or 30μM DMSO every other day. Lethality of the treatment was assessed every day for 7 days. Delta protein levels were observed by immunofluorescence after 4 days.

Mammalian cell culture

Treatment of HEK293 cells with GW3965

Samples were collected from parallel cultures of HEK293 cells and treated with 2 μM GW3965 or DMSO vehicle for 16hrs.

Treatment of HCT116 cells with Simvastatin

HCT116 cells were cultured with 5μM activated Simvastatin or 5μM ethanol in McCoy’s medium (+10% FBS, 1x antibiotic) for 48 hours, with fresh media and Simvastatin/ethanol added after 24 hours. Simvastatin was activated according to manufacturer’s instruction by dissolving 1.4mg Simvastatin in 100μl of ethanol and incubating the resulting solution with 150μl of 0.1 N NaOH at 50 °C for 2 hours. The final stock concentration was adjusted to a pH of 7.0 with HCl and brought to a final concentration of 4 mg/ml.

HCT116 Immunofluorescent Staining

Cells were cultured as described on round coverglass. Prior to fixation, cells were rinsed 3x with 1X PBS and then incubated with freshly diluted 4% paraformaldehyde for 10 minutes at room temperature. Cells were rinsed again 3x with PBS and then incubated with 1 ml of 1.5 mg glycine/ml PBS for 10 minutes at room temperature. Samples were either immediately stained with 1ml of filipin (0.05 mg/ml in PBS) for 2 hours at room temperature or blocked with 5% NGS in 0.1% PBS + Triton X-100 for 1 hour. Following block, samples were incubated overnight at 4 °C with anti-Notch1 (1:500) and anti-sec61 (1:2000) in blocking solution. Samples were washed the next day 3×5 minutes with PBS and then incubated with secondary antibodies at 1:1000 for 1 hour at room temperature. Samples were washed again for 3×5 minutes with PBS, incubated with 1μg/ml DAPI for 10 minutes, and mounted with Prolong Gold Antifade Reagent.

Western Blot for Tubulin and DLL4

HEK293 and HCT116 cells were collected in cold PBS and combined with an equal volume of 2x Laemmli Sample Buffer. Cells were sonicated for 10 seconds and incubated at 70 °C for 10 minutes. 15μl sample was loaded into 4–15% Mini-PROTEAN® TGX™ Precast Protein Gels and proteins separated at 100V for 90 minutes. Proteins were then transferred onto nitrocellulose membrane at 10mA constant current for 18 hours. The membrane was blocked with Odyssey Blocking Buffer for 1 hour and then incubated overnight at 4°C with 0.1% Tween-20 and primary antibodies anti-DLL4 (1:500) and anti-alpha tubulin (1:1000). Appropriate IRDye secondary antibodies were incubated for 1 hour at room temperature at 1:20,000 in Odyssey Blocking Buffer with 0.1% Tween-20. Membranes were washed 3×5 minutes with 1x TBS after primary and secondary antibody incubation.

Quantification and statistical analysis

The significance of experimental treatments relative to matched controls were determined using Students t-tests (single-tailed) using GraphPad, as described in Figure legends. Significance levels are shown on the Figures using the following code: * = p<0.05; ** = p<0.01; *** = p<0.001.

Data and Software availability

“The RNAseq data described have been deposited in the NIH GEO archive under ID codes GSE111057.”

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies (Drosophila)
anti-Prospero	Developmental Studies Hybridoma Bank (DSHB)	Cat# MR1A RRID:AB_528440
anti-PHC3	Sigma-Aldrich	Cat# HPA039162, RRID:AB_10671038
anti-Delta	DSHB	Cat# C594.9B RRID:AB_528194
anti-NECD	DSHB	Cat# C458.2H RRID:AB_528408
anti-LpR2	Parra-Peralbo E; PLoS Genet. 2011	Cat# LpR2 RRID:AB_2569135
anti-Hindsight	DSHB	Cat# 1G9 RRID:AB_528278
anti-Rab7	DSHB	Cat# Rab7, RRID:AB_2722471
anti-GFP	ThermoFisher	Cat# A-11122 RRID:AB_221569
Antibodies (mouse or human)
anti-Notch1 (A-8)	Santa Cruz Biotechnology, Inc.	Cat# sc-376403 RRID:AB_11149738
anti-DLL4	Abcam	Cat# ab7280 RRID:AB_449562
anti-Sec61alpha	Milipore	Cat# 07–204 RRID:AB_310425
anti-alpha Tubulin	Abcam	Cat# ab28439 RRID:AB_727039
Bacterial and Virus Strains
Biological Samples
Chemicals, Peptides, and Recombinant Proteins
GW3965	Sigma-Aldrich	Cat# G6295
DAPI	Sigma-Aldrich	Cat# D9542
Simvastatin	Sigma-Aldrich	Cat# S6196
filipin	Sigma-Aldrich	Cat# F4767
TriPure Isolation Reagent	Sigma-Aldrich	Cat# 11 667 157 001
NMS-873	Sigma-Aldrich	Cat# SML1128
Vectashield, Mounting medium	Vector Labs Inc.	Cat#H-1000
ProLong Gold Antifade Mountant	TermoFisher	Cat#P36930
TruSeq RNA Library Prep Kit v2	illumina	RS-122–2001
iScript™ Reverse Transcription Supermix for RT-qPCR	BioRad Labs	Cat#170–8840
iTaq™ Universal SYBR® Green Supermix	BioRad Labs	Cat#172–5121
Critical Commercial Assays
DeadEnd™ Fluorometric TUNEL System	Promega	Cat. G3250
Amplex Red Cholesterol Assay Kit	Life Technologies	Cat# A12216
Deposited Data
Dietary lipids modulate Notch signaling and influence adult intestinal development and metabolism in Drosophila	NIH GEO	GSE111057
Experimental Models: Cell Lines
Human: HEK293 transformed cell line	ATCC	Cat# CRL-1573, RRID:CVCL_0045
Human: HCT116 colon cancer cell line	ATCC	Cat# CCL-247, RRID:CVCL_0291
Experimental Models: Organisms/Strains
Oregon-R	Bloomington Drosophila Stock Center (BDSC)	Cat# 25211, RRID:BDSC_25211
y1 w1	BDSC	Cat# 1495, RRID:BDSC_1495
Hr961	BDSC	BDSC Cat# 76592, RRID:BDSC_76592
UAS-Hr96/CyO	BDSC	Cat# 76593, RRID:BDSC_76593
y1 sc* v1; P{TRiP.HMS01682}attP40 (NPC2b-RNAi 1)	BDSC	Cat# 38238, RRID:BDSC_38238
y1 sc* v1; P{TRiP.HMS02607}attP40 (NPC2b-RNAi 2)	BDSC	Cat# 42914, RRID:BDSC_42914
y1 sc* v1; P{TRiP.HMS05309}attP40 (ACAT-RNAi)	BDSC	Cat# 63035, RRID:BDSC_63035
N55e11 FRT19A/FM7	BDSC	Cat# 28813, RRID:BDSC_28813
y1 w1118 FRT19A	BDSC	Cat# 1744, RRID:BDSC_1744
hsFLP, tub-GAL80, FRT19A; UAS-mCD8::GFP	BDSC	Cat# 5134, RRID:BDSC_5134
esg-GAL4; tub-GAL80ts, UAS-GFP	Michelli and Perrimon, 2006
myo1A-GAL4; tub-GAL80ts, UAS-GFP	Jiang et al. 2009
mex-GAL4	Phillips and Thomas, 2006
GBE-Su(H)-GAL4, UAS-GFP	Zeng et al. 2010
UAS-Notch-RNAi	S. Bray
SREBP reporter	BDSC	Cat# 39612 RRID:BDSC_38394
w-;; Hr962x	C. Thummel
w-; Hr962x	C. Thummel
UAS-Hr96/TM6b	BDSC	Cat# 76593 RRID:BDSC_76593
Act-GAL4/CyO; GAL80ts	A. Spradling
FRT82B, tub-GAL80	BDSC	Cat#5135, RRID:BDSC_5135
hsFLP, UAS-mCD8::GFP	BDSC	Cat#5131, RRID:BDSC_5131
Oligonucleotides
RP49-F: 5’-CCCAAGGGTATCGACAACAGA-3’	This paper	N/A
RP49-R: 5’-CGATCTCGCCGCAGTAAAC-3’	This paper	N/A
NPC2B-F: 5’-TGCCGTTCCCAGGATATTACG-3’	This paper	N/A
NPC2B-R: 5’-GCAGGATCTTGAAGCTGTTCTT-3’	This paper	N/A
ACAT-F: 5’-CACGAGTTCCCCAACTTCGA-3’	This paper	N/A
ACAT-R: 5’-CTTACGAGCGGATGTTTCGC-3’	This paper	N/A
Recombinant DNA
Software and Algorithms
bcl2fastq v2.17.1.14	Illumina Corp.	https://support.illumina.com/downloads/bcl2fastq-conversion-software-v217.html
Bowtie 2.2.9	Langmead and Salzberg (2012)	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
TopHat2.1.1	Kim et al. (2013)	http://ccb.jhu.edu/software/tophat
Cufflinks2.2.1 and Cuffdiff		http://cole-trapnell-lab.github.io/cufflinks/install/
ImageJ 1.51w	Schneider et al. (2012)	https://imagej.nih.gov/ij/docs/install/osx.html
GraphPad Prism 7	GraphPad Software	sales@graphpad.com
Other