Hedgehog Signaling Strength Is Orchestrated by the mir-310 Cluster of MicroRNAs in Response to Diet.

Since the discovery of microRNAs (miRNAs) only two decades ago, they have emerged as an essential component of the gene regulatory machinery. miRNAs have seemingly paradoxical features: a single miRNA is able to simultaneously target hundreds of genes, while its presence is mostly dispensable for animal viability under normal conditions. It is known that miRNAs act as stress response factors; however, it remains challenging to determine their relevant targets and the conditions under which they function. To address this challenge, we propose a new workflow for miRNA function analysis, by which we found that the evolutionarily young miRNA family, the mir-310s (mir-310/mir-311/mir-312/mir-313), are important regulators of Drosophila metabolic status. mir-310s-deficient animals have an abnormal diet-dependent expression profile for numerous diet-sensitive components, accumulate fats, and show various physiological defects. We found that the mir-310s simultaneously repress the production of several regulatory factors (Rab23, DHR96, and Ttk) of the evolutionarily conserved Hedgehog (Hh) pathway to sharpen dietary response. As the mir-310s expression is highly dynamic and nutrition sensitive, this signal relay model helps to explain the molecular mechanism governing quick and robust Hh signaling responses to nutritional changes. Additionally, we discovered a new component of the Hh signaling pathway in Drosophila, Rab23, which cell autonomously regulates Hh ligand trafficking in the germline stem cell niche. How organisms adjust to dietary fluctuations to sustain healthy homeostasis is an intriguing research topic. These data are the first to report that miRNAs can act as executives that transduce nutritional signals to an essential signaling pathway. This suggests miRNAs as plausible therapeutic agents that can be used in combination with low calorie and cholesterol diets to manage quick and precise tissue-specific responses to nutritional changes.

ORGANISMS are constantly subjected to changes in nutrient availability and composition, which depend on quantity and quality of consumed food. Currently, there is a considerable amount of data regarding the cellular metabolic processes and signaling pathways involved in metabolism regulation; however, we know little about the mechanisms that efficiently readjust these pathways in response to ever-changing dietary fluctuations. MicroRNAs (miRNAs) are great candidates for such regulation due to their unique features: miRNA expression is extremely dynamic; one miRNA can regulate hundreds of different targets; and more than one miRNA may coordinately regulate a single target. This presents a great number of combinatorial possibilities, which allows for greater precision in regulation of gene expression.

miRNAs have been shown to be involved in virtually all studied biological processes, including regulation of cellular metabolism and organismal homeostasis (Xu ; Teleman ; Barrio ), development of metabolic disorders, and the highly energy-demanding process of carcinogenesis (Bhattacharyya ; Leung and Sharp 2010; Ross and Davis 2011). However, it remains extremely difficult to decipher specific in vivo requirements for each miRNA due to the facts that their mutant phenotypes are very subtle (Lai 2015), and most miRNA mutants are viable, fertile, and apparently normal in well-controlled lab conditions. Furthermore, correlating causal targets to miRNA phenotypes remains the key challenge. Even though multiple algorithms and databases predicting miRNA–messenger RNA (mRNA) interactions based on sequence and physical-chemistry properties exist, they have large numbers of false positives and currently only very few interactions have been experimentally validated. It has been shown that dietary modulations modify miRNA expression profiles, but to date there is a paucity of in vivo functional studies that aim to decipher the complex networks involving nutrition-dependent miRNAs and their targets. Such studies may offer new concepts for preventive and therapeutic strategies for metabolic disorders, including obesity and diabetes.

Since the dietary requirements for major nutrients (sugars, fats, and amino acids) appear to be universal and the signaling pathways involved in the basic logic of nutrient signaling are conserved, studies in model organisms have proven to be beneficial for the understanding of metabolic stress. In Drosophila, similarly to vertebrates, steroids, insulin, and TOR signaling play a critical role in regulation of nutritional responses, suggesting that Drosophila can be used as a relevant model to study nutritional stress (Drummond-Barbosa and Spradling 2001; Konig ; Wei and Lilly 2014). Particularly, the Drosophila ovarian germline stem cell community is a very attractive model to study how adult stem cell self-renewal and differentiation is coordinated with organismal metabolism. In the Drosophila germarium, there are two stem cell types of extremely different origin: the germline stem cells (GSCs) and the somatic follicle stem cells (FSCs). These stem cells also have very distinctive stem cell niche types: the stationary, cell–cell adhesion-dependent GSC niche and the dynamic, cell–matrix adhesion-dependent FSC niche (Song and Xie 2002; Nystul and Spradling 2007; Morrison and Spradling 2008). Interestingly, the GSC niche not only controls GSC maintenance, but also has a distant influence on FSC division and differentiation. The FSC gives rise to somatic ovarian cells that come in different types: the follicular epithelium, stalk, polar, and border cells, all of which protect and assist the germline, ensuring sufficient egg differentiation. Therefore, for proper oogenesis progression, it is extremely important that GSC and FSC divisions and the differentiation of their progeny are synchronized (Gilboa and Lehmann 2006; Chang ; Konig and Shcherbata 2015). Dependent on nutrient availability, insulin ligands are produced in the brain to activate insulin signaling in the GSCs to cell-autonomously control their division rate; in contrast, the Hh ligand is locally produced by the GSC niche, it travels three to five cell diameters to the posteriorly located FSCs to stimulate their proliferation (Forbes ; Drummond-Barbosa and Spradling 2001; Zhang and Kalderon 2001; O’Reilly ; Rojas-Rios )

Importantly, Hh signaling is highly dependent on the diet, because its multiple components are regulated by cholesterol and lipid levels (Panakova ; Sieber and Thummel 2012; Hartman ). Upon dietary restriction, an organism has to quickly change its cellular metabolism and adapt to unfavorable conditions; however, it is very unlikely that levels of cholesterols and lipids would drop instantly (Efeyan ), resulting in sufficient downregulation of Hh signaling. This highlights the importance of the existence of other levels of regulation to ensure the quick and robust response of Hh to dietary changes. While downstream Hh effectors have been well studied in different systems, the upstream regulators of Hh signaling and their roles in energy homeostasis are yet to be revealed. Our data for the first time demonstrate that Hh signaling strength upon nutritional fluctuations can be modulated by miRNAs.

Here we used a new workflow allowing for effective identification of miRNA-regulated processes and relevant targets. First, we applied quantitative proteomic analysis of miRNA mutants to identify the major biological processes affected by miRNA loss. Second, tissue-specific dissection of miRNA mutants was performed to identify the most prominent phenotypes caused by miRNA insufficiency. Third, based on the vast amount of previously published data, we compared these phenotypes to the key phenotypes associated with major signaling pathways. Fourth, we used several databases (Enright ; Kheradpour ; Betel ) to predict potential miRNA targets, among which several genes relevant to the identified signaling pathway were selected and further confirmed using in vitro and in vivo assays. Finally, genetic analyses and rescue experiments under normal and defined stress conditions were performed to validate miRNA roles in certain biological processes.

Using this paradigm, we found that in Drosophila during adulthood, the s orchestrate Hh signaling strength in accordance with nutritional status. We identified three new s targets, , , and , all of which are involved in Hh pathway regulation. By simultaneous targeting of multiple regulators of the pathway, the s safeguard a quick and robust response of Hh signaling to dietary fluctuations. Additionally, we discovered a molecular function for the membrane trafficking protein Rab23 in the process of Hh ligand intracellular transport and secretion in the stem cell niche. Plausibly, Hh signaling management by the s is just one example of many diet-dependent processes regulated by these miRNAs. Our proteomic data, generated by SILAC labeling accompanied by mass spectrometry analysis, revealed that multiple critical metabolism-related genes are deregulated due to the s deficiency under normal and dietary restrictive conditions, suggesting that in general, the molecular function of these miRNAs is management of organismal homeostasis upon dietary fluctuations.

Materials and Methods

Fly stocks

All fly stocks were maintained and crosses were set up on standard food with yeast, cornmeal, and agar at 25°, constant humidity, and a 12-hr light–dark cycle. The nutrient restriction experiments were done using 2% agar-agar (Serva), 25% apple juice, and 2.5% sugar medium. The nutrient-starved flies were fed this medium plain, whereas the well-fed flies were given additional fresh yeast paste made of dry yeast and 5% propionic acid (∼50% w/v). Food vials of both conditions were refreshed every 2 days. The following fly stocks were used: Oregon-R-C and as controls; s deletion lines KT40 (Tsurudome ), w*; Df(2R)mir-310-311-312-313 P(neoFRT)42D/CyO, P(GAL4-twi.G)2.2, P(UAS-2xEGFP)AH2.2 (no. 58923 Bloomington Drosophila Stock Center, BDSC), and the deficiency line w[1118]; , P(w[+mC]=XP-U)Exel6070/CyO (no. 7552 BDSC) as mutant alleles; s-Gal4 (P(GawB)NP4255 from Drosophila Genomics and Genetic Resources, Kyoto) (Yatsenko ), UAS-mCD8::GFP, UAS-nLacZ line (gift from Frank Hirth) for expression analyses; tub-Gal80; bab1-Gal4/TM6 and w; +; bab1-Gal4/TM6 (no. 6803 BDSC), UAS-hh (gift from Christian Bökel), UAS-Rab23 RNAi (y[1] v[1]; P(y[+t7.7] v[+t1.8]=TRiP.JF02859)attP2 (no. 28025 BDSC), UAS-hh RNAi (Sahai-Hernandez and Nystul 2013), and a s rescue line (w-; Sco/CyO; attB2 mir-310s res long 2/TM6B) carrying a large genomic region encompassing the s as a transgene in the 3rd chromosome (gift from Eric Lai) for rescue experiments. To generate the UAS-Rab23 line, we cloned cDNA into the UASt vector (gift from Alf Herzig). Cloning was performed by standard cloning techniques, digesting the cDNA vector (RH23273 clone from Drosophila Genomics Resource Center) and the UASt vector with EcoRI and KpnI restriction enzymes. The microinjection and recovery of the transgenic flies was done by Bestgene. The site-specific integration on the 3rd chromosome (76A2 site) was achieved by the att sites in the UASt-Rab23 plasmid and PhiC31 into the PBac[yellow[+]-attP-9A]VK00013 strain. ::YFP::4xmyc line (also referred as ::YFP in the text) was generated by ends-in homologous recombination, and the initial genomic duplication was resolved using the I-Cre system. The size of the homologous sequence 5′ from the YFP start codon is 4045 bp, and the size of the homologous sequence 3′ from Myc tag DNA fragment is 3601 bp. The donor construct was verified by sequencing. Recombination events were verified by PCR. In our analyses, homozygous flies bearing endogenously tagged copies were used.

For SILAC analysis, qRT-PCR (list of primers, Table S11), immunohistochemistry, luciferase assay, coupled colorimetric assay (CCA), and coimmunoprecipitation, refer to File S1.

Results

mir-310s loss of function causes defects in energy metabolism and deregulation of nutritional homeostasis-associated genes

In our previously performed screen for stress-dependent miRNAs (Marrone ), we found that the miRNAs from the newly evolved s family are differentially expressed under stress and disease conditions. Therefore, we aimed to decipher the potential role for these miRNAs in maintenance of a healthy physiological state. To begin with, we studied global changes in protein expression caused by s deficiency. The quantitative SILAC proteomics data of miRNA mutant flies were generated for the first time using previously described (Sury ) mass spectrometry of heavy isotope-labeled Drosophila. This analysis resulted in a sizeable list of proteins with altered expression levels caused by s deficiency. Since miRNAs are generally identified as fine tuners of gene expression, we considered proteins with a moderate (≥30%) relative increase or decrease with a P-value of 0.1 for filtering for the significant data (Supporting Information, Table S1), which resulted in the identification of 264 proteins that were up- or downregulated in s mutants, among which 24 are predicted s direct targets (Table S1, bold boxes).

Next, using the STRING database (Franceschini ), we created functional association networks of deregulated genes; then, we grouped these genes into functional groups according to their gene ontology (GO) terms from the UniProt database (UniProt Consortium 2014). This analysis revealed distinct functional groups: lipid and energy metabolism, protein homeostasis, nucleotide synthesis, mitochondria, muscle and neural development/function, cuticle formation, and others (Figure 1A). Furthermore, 20% of the altered genes were reported to be lipid droplet associated (Kuhnlein 2011). Importantly, the common denominator of these affected gene functions was their involvement in energy metabolism and homeostasis, suggesting that the s are involved in regulation of these processes, which can be achieved directly by the s regulation of their target genes and indirectly by secondary effects of their targets. It is important to stress that due to the current limitations of quantitative mass spectrometry analyses, only 30% of all predicted Drosophila proteins could be identified in this study, which is comparable to previously described SILAC proteomic data (Sury ). Detected proteins mainly represent the most highly expressed, but not regulatory proteins or transcription factors that are known to efficiently operate even in very low quantities. Despite the limitations of this analysis in identifying direct miRNA targets that belong to these functional groups, it allowed for meaningful identification of the processes regulated by the miRNAs.

Figure 1

The mir-310s-affected genes suggest energy metabolism-related defects. (A) Interaction network of globally up- or downregulated genes in mir-310s loss-of-function (KT40/KT40) mutant females compared to control (w) females exhibits eight interconnected gene ontology groups: energy metabolism, lipid metabolism, protein homeostasis, muscle and neural development and function, mitochondrial, nucleotide synthesis, and cuticle related (Table S1). Large node size indicates the availability of protein structure information. Node colors have no particular meaning. Line colors indicate different evidence types used to generate node interactions (see key). (B) Starvation-sensitive genes have altered mRNA expression levels in mir-310s mutant females compared to controls under well-fed and/or nutritionally restricted conditions (10 days), demonstrating the role of the mir-310s in the response to changes in nutritional status (Table S2). Functional groups are color coded as in A. In B the bar graph indicates the arithmetic mean (AVE) ± the standard error of the mean (SEM). Significances were calculated using two-tailed Student’s t-test. *P < 0.05, **P < 0.005, ***P < 0.0005.

Since our proteomics data indicated that the s could be associated with maintenance of metabolism and energy homeostasis, we compared this dataset with genes previously reported to be starvation-sensitive by transcriptome analysis (Farhadian ) and found 31 genes in common. Next, we measured mRNA expression levels of these genes by qRT-PCR in wild-type and s mutant animals under well-fed and nutrient-restricted conditions (Figure 1B; Table S2). We used two diets: well-fed (sugars + yeast paste) and starved/protein restricted (just sugars, no yeast paste). As has been reported by pioneering studies and recent efforts, the Drosophila life cycle (development and adult homeostasis) greatly depends on the nutritional input from the yeast source, which can be reconstituted by addition of amino acids, cholesterol, nucleic acids, folic acid, inositol, biotin, riboflavin, nicotinic acid, pyridoxine hydrochloride, calcium pantothenate, thiamine, choline chloride, ergosterol, and metal ions (Piper ). Our starvation conditions supply only simple sugars and lack these essential components needed for optimal homeostasis. In agreement with the proteomic data, most of these genes were aberrantly expressed in s mutants. In addition, starvation induced uncoordinated alterations in the gene expression profiles, consistent with a suggested role for s in dietary response (Figure 1, A and B). For instance, one of the genes, (Lsp1beta), was found to be 10-fold higher in s mutants under well-fed conditions in comparison to controls. Nutrient restriction caused a sharp decrease (>30-fold) of the Lsp1beta transcript levels in control flies, while almost no change was detected in s mutant flies (only a one third-fold decrease). The transcript levels of another gene, () were downregulated close to zero upon nutrient restriction in control flies; however, in s mutants, levels were only slightly decreased. As a result, mRNA levels in nutrient-restricted s mutants were ∼40-fold higher in comparison to those in controls. In general, most of the nutrition-dependent genes that were analyzed showed atypical alterations in their expression levels in s when compared to wild-type flies under both normal and restrictive conditions, showing a role for the s in nutritional homeostasis and response to starvation.

Defects caused by mir-310s deficiency depend on nutrition

In correlation with the abnormal expression of the energy- and lipid metabolism-related genes, the analysis of s-deficient flies revealed several gross morphological and physiological phenotypes that are known to be related to nutrient availability. One of the most prominent phenotypes detected upon dissection of s mutants was the enlarged food storage organ or crop (Figure S1, A and A′), the size of which is highly diet dependent and is capable of expansion after starvation and refeeding (Lemaitre and Miguel-Aliaga 2013). It is also known that the enlarged crop is a persisting signature of poststarvation response, since females switched from nutrient-poor to nutrient-rich food consume more food (Edgecomb ; Al). We found that even under normal feeding conditions, average crop size of s females was 30% larger than crops of wild-type females of comparable size (Figure S1A′′). This suggests that due to their abnormal metabolism, s females exhibit a phenotype consistent with the physiology elicited during poststarvation.

Interestingly, studies on the physiology of starvation-selected flies demonstrate that their entire life history is disturbed; as adults, these animals contain more lipids, but at the cost of reduced fecundity (Masek ). To determine whether s flies exhibit these phenotypes, first we evaluated the fat storage characteristics of mutant females using a colorimetric assay. Under well-fed conditions, the total body fat content of s females was ∼2-fold lower than that of controls (Figure S1B). Consistent with previously reported data (Musselman ), 10 days of protein starvation resulted in a 1.3-fold increase in the total body fat content in controls. However, upon the same restriction, s females accumulated dramatically higher amounts of lipids, 2.5- and 4-fold increases in comparison to the starved and well-fed controls, respectively (Figure S1, B and B′).

It is well known that the nutrient-sensitive and energy-demanding egg production process is stopped due to nutrient deficit (Drummond). In s females, the cessation of egg production is delayed compared to wild type in response to starvation (Figure S1C). However, even on a normal diet, s females laid ∼2.5-fold fewer eggs (Figure S1, D–E). If egg-laying ability is a direct readout of metabolic status, these results imply that s-deficient females in general have deficient energy resources and in addition, they cannot properly respond to dietary restriction. Taken together, the proteomic and qRT-PCR expression assays (Figure 1) in combination with the physiological defects caused by s loss, which include increased crop size and reduced egg production under normal diet and dramatic fat accumulation under starvation (Figure S1, Table S3), confirm that the s are essential factors in regulation of energy metabolism in various physiological and cellular elements of the whole organism.

The mir-310s function in the ovarian soma

Next, we aimed to dissect the s function at the cellular level and identify their direct targets involved in starvation response. Therefore, we focused on oogenesis, which is one of the best-studied nutrition-dependent processes. Drosophila oogenesis takes place in the ovaries, which are paired organs consisting of individual ovarioles—the egg production units made of progressively developing egg chambers. While developing egg chambers move toward the posterior and develop into mature eggs, they stay attached to the neighboring egg chambers by small groups of cells forming stalks (Figure 2A). Each egg chamber is surrounded by a monolayer of epithelium composed of follicle cells, and specialized polar cells are specified at each end of the egg chamber.

Figure 2

The mir-310s are expressed in the ovarian soma. (A–C) The mir-310s are expressed in the somatic cells in the germarium, stalk, and follicular epithelium, as visualized by nuclear β-gal and membrane-bound GFP reporters (mir-310s-Gal4/+; UAS-mCD8::GFP, UAS-nLacZ/+). Some of the stalk (B) and follicular epithelium cells (C) express exclusively β-gal (arrowheads) or GFP (concave arrowheads), whereas others coexpress both reporters (arrows). Different turnover rates of the reporter proteins indicate the dynamic mir-310s locus activity. (D) In mir-310s mutants (KT40/Df6070), ovarioles contain excessive numbers of cells at the stalk region, deformed and multilayered follicular epithelia, and abnormal numbers of nurse cells per egg chamber as a result of defects in egg chamber encapsulation. These phenotypes resemble phenotypes caused by Hh signaling deregulation (Forbes ). (E) Overexpression of hh in TF and CpCs by shifting 2-day-old tub-Gal80; bab1-Gal4/ UAS-hh females to restrictive temperature (29°) for 7 days causes the stalk region and egg chambers to be filled by excessive numbers of epithelial cells in multilayers and egg chambers to bear abnormal numbers of nurse cells. These phenotypes look similar to those of the mir-310s mutant shown in D. (F) Schematic of the Drosophila germarium. Drosophila oogenesis depends on the presence of two to three adult germline stem cells (GSCs) per germarium that continuously divide. The GSCs reside in a specialized microenvironment, the GSC niche, which consists of specialized somatic cells, namely terminal filaments and cap cells (TFs and CpCs). The differentiating GSC progeny is enveloped by the escort cells (ECs) that assemble the differentiation niche. Another type of somatic cells, the follicle stem cells (FSCs) and their progeny, the follicular epithelium (FE) cells divide and surround the 16-cell germline cysts at region 2b. At region 3, the germline cyst encapsulated by the FE pinches off of the germarium as an individual egg chamber. The Hh ligand is expressed in the TF and CpCs and acts long range to FSCs, inducing their division and the differentiation of their progeny. (G) The mir-310s are expressed in the stem cell niche, TF and CpCs (arrowheads), and in the differentiation niche, ECs (concave arrowheads), as visualized by anti-GFP staining (mir-310s-Gal4/+; UAS-mCD8::GFP, UAS-nlacZ/+). (H) The mir-310s (mir-310 and mir-312) are significantly upregulated upon starvation. Whole ovary extracts from 7-day-starved females show an ∼1.5-fold increase in miRNA levels compared to well-fed controls (Table S5). The bar graph indicates AVE ± SEM. Significances were calculated with two-tailed Student’s t-test. *P < 0.05, **P < 0.005, ***P < 0.0005. At least three biological replicates per genotype and condition were analyzed. (I) Upon 7 days of nutritional restriction, the number of mir-310s expressing CpCs significantly increases as visualized by anti-GFP staining (mir-310s-Gal4/+; UAS-mCD8::GFP, UAS-nLacZ/+) AVE ± SEM values are reported from the measurements done from 20 germaria (4.1 ± 0.34 well-fed, 5.6 ± 0.42 starved, statistical significance is calculated using Mann–Whitney U-test and Z-statistic, P = 0.0078). In A–G, anterior is to the left. B and C represent single optical sections. A, D, E, and G represent maximum intensity projections of confocal Z-stacks. Bars, 20 µm in A, D, and E and 5 µm in B, C, and G.

To identify the possible involvement of the s in oogenesis, we analyzed their expression pattern, which was visualized by nuclear β-gal and membrane-bound GFP driven by s–Gal4. Expression of reporters was detected in subsets of different somatic cell types, and their expression levels were fluctuating (Figure 2A). For example, some of the stalk and follicular epithelium cells were expressing nuclear β-gal and/or membrane GFP, but some were not (Figure 2, B and C). Since β-gal and GFP proteins have different turnover rates (Timmons ), we conclude that expression of the s in the ovarian soma is dynamic. Similarly, a dynamic s expression pattern was observed in the brain, where the precision of these miRNAs’ expression is achieved via the perceptive–executive mechanism orchestrated by their target (Yatsenko ).

Upon in-depth examination of s mutant ovaries in well-fed conditions, we identified several phenotypes in the ovarian soma, which could be categorized into three distinct groups. First, supernumerary stalk cells accumulated between egg chambers: in control ovarioles, up to eight stalk cells properly line up between adjacent egg chambers, while in s ovarioles, excessive numbers of disorganized cells at the stalk region formed a multilayered epithelium (Figure 2D). Second, the follicular epithelium cells surrounding egg chambers of different stages were distorted in shape and had irregular cellular polarity, assembling random multilayered patches (Figure 2D). Third, abnormally encapsulated egg chambers, easily identifiable by the aberrant numbers of polyploid nurse cells, appeared (Figure 2D). These defects were very similar to the previously described ovarian phenotypes caused by defective Hh signaling (Figure 2E). Hh signaling is important for cell fate establishment of all ovarian somatic cell types (Forbes ,b; Tworoger ; Besse ; Chang ) and its ligand is produced by the terminal filament and cap cells (TFs and CpCs) forming the stem cell niche, and also escort cells (ECs) forming the germline differentiation niche (Rojas-Rios ). The dynamic s expression was detected in all of these niche cell types (Figure 2, F and G). Importantly, this expression appeared to be diet-sensitive; upon starvation, s expression levels were upregulated (Figure 2H), and the number of s-expressing GSC niche cells (CpCs and TFs) was significantly increased (Figure 2I). These results suggest that the s have a cell-autonomous role in the stem cell niche during dietary changes.

The analysis of the s expression pattern revealed that the s are dynamically expressed in the Hh signal-sending cells (TFs, CpCs, and ECs in the germarium) as well as in the Hh signal-receiving cells (the stalk and follicular epithelium cells in the developing egg chambers) (Figure 2, A–C and G). These results, combined with the similarities of the observed s loss- and Hh gain-of-function mutant phenotypes (Figure 2, D and E), led us to hypothesize that the s regulate Hh signaling via targeting one or multiple components of this pathway.

mir-310s target three genes associated with the Hedgehog signaling pathway

To confirm this hypothesis and define the molecular mechanism responsible for s ovarian phenotypes, we acquired a list of in silico-predicted s targets using several miRNA target search databases (Enright ; Kheradpour ; Betel ) and selected among the putative s targets all known or predicted Hh pathway elements and their interaction partners. The s are recently evolved miRNAs, which have highly evolutionarily conserved seed sequences (Figure 3A). As predicted by different algorithms, the s have 350– 450 putative targets, among which only three [, (), and ()] have been associated with Hh signaling.

Figure 3

The mir-310s target three genes associated with the Hh pathway. (A) The mir-310s share a highly conserved seed sequence (red) with their ancestral miRNAs mir-92a and mir-92b, and their orthologous miRNAs from zebrafish, mouse, and human. (B) Overexpression of mir-310s downregulates Rab23, , and ttk 3′ UTR luciferase reporters in Drosophila S2 cells. The long 3′ UTR of a confirmed target gene (Dg) with mir-310s binding site serves as positive, the short Dg 3′ UTR without a mir-310s binding site serves as negative control (Table S4). (C) mir-310s mutant (KT40/KT40) females have significantly higher Rab23 and mRNA levels (whole RNA extracts from adult females were used). This elevation is even more pronounced under starvation conditions (10 days). The change in ttk mRNA levels, however, illustrates mir-310s-dependent regulation exclusive to the starvation condition (Table S5). (D) Model shows major conserved components of Hh signaling, including the Hh receptor Patched (Ptc), the transmembrane protein Smoothened (Smo), and the transcriptional effector Cubitus interruptus (Ci), which act in the signal-receiving cells (Forbes ; Zhang and Kalderon 2001). Additional Hh receptors and close homologs, Ihog and Boi, promote intrinsic Hh signaling and extrinsic Hh ligand release (Hartman ). Hh signaling governs adult stem cell division and differentiation depending on the cholesterol modification of the ligand, which is required for long-range signaling and is sensitive to changes in the nutritional status of the animal (Panakova ). Particularly, ovarian FSCs rely on this signal to initiate the division and differentiation process, which can be slowed down, stopped, and reinitiated upon changing dietary conditions (Rojas-Rios ; Hartman ). Upon starvation, Hh is sequestered by Boi, while upon feeding, cholesterol binds to DHR96 and promotes Boi phosphorylation and Hh release, which positively affects FSC proliferation (Hartman ). The mir-310s are present in the niche and follicle cells that also express and ttk, respectively (Sun and Deng 2007; Hartman ), which suggests that the mir-310s could intrinsically regulate these targets in both the Hh signal-sending and Hh signal-receiving cells of the ovarian soma in response to nutrient availability. For B and C, bar graphs indicate AVE ± SEM. Significances were calculated with two-tailed Student’s t-test. *P < 0.05, **P < 0.005, ***P < 0.0005.

To verify that the s indeed target these three genes, we performed a Drosophila S2 cell-based luciferase reporter assay, which depends on the readout from a reporter plasmid with a luciferase gene containing the 3′UTR of the gene of interest with the predicted miRNA target site. The luciferase assay showed that in vitro, the s could target the , , and transcripts via their 3′UTRs (Figure 3B; Table S4).

Next, we tested in vivo the responsiveness of these three putative target genes to the s as well as to nutrient restriction. We found that the expression of all three genes is nutrition dependent, showing significant reduction under starvation conditions. In s mutants, and levels were significantly upregulated (>1.5-fold; Figure 3C), and their expression levels were not as efficiently reduced under starvation. In contrast, mRNA expression levels were similar to controls in s-deficient flies under well-fed conditions. expression was controlled by the s only under nutritional stress, where s mutants had 1.5-fold higher mRNA levels when compared to controls (Figure 3C and Table S5). These data demonstrate that the s act to fine tune the expression of the nutrition-dependent genes , , and ; furthermore, the s regulate only upon dietary restriction.

The above results confirm that the s are important regulators of at least three components associated with Hh signaling (Figure 3D). encodes a cholesterol receptor responsible for sensing the nutritional status of the cell environment (Horner ; Bujold ; Sieber and Thummel 2012) and promoting Hh ligand release upon dietary cholesterol intake (Hartman ). encodes a transcription factor that acts as a controller of the cell cycle switch during midoogenesis through regulation of Hh target gene expression (Sun and Deng 2007). encodes a membrane organization and trafficking Rab GTPase (Zerial and McBride 2001; Zhang ; Chan ). Mouse Rab23 was shown to act as a negative regulator of the Sonic Hh signaling pathway in signal-receiving cells during embryonic neural patterning (Eggenschwiler ). However, Drosophila Rab23 is known not to function in Hh signaling through the same mechanism [at least in the process of wing development (Pataki )], and its role in Hh signaling has not been confirmed.

The mir-310s and Rab23 regulate Hh ligand release

As the involvement of Rab23 in the Hh pathway remains an open question in Drosophila (Zhang ; Pataki ), we decided to further focus on the s–Rab23 interaction. First, we analyzed the spatial distribution of Rab23 protein using a ::YFP line generated via homologous recombination (see Materials and Methods). Similarly to the s, the endogenous Rab23 protein was detected in the germline stem cell niche (TFs and CpCs) and in the differentiation niche (ECs), and this expression was dynamic: some of the niche cells were Rab23 negative and the others could be classified as high or low Rab23-expressing cells (Figure 2G, Figure 4, A–D). Upon close examination, we found significantly more Rab23-positive CpCs in s mutant germaria in comparison to controls (Figure 4E). Moreover, upon starvation, the number of cells with high Rab23 levels was significantly increased in s mutants (Figure 4E; Table S6).

Figure 4

Rab23 is targeted by the mir-310s, controlling Hh ligand availability. (A–D) The mir-310s negatively regulate Rab23 expression. Rab23 has a stronger and more widespread expression pattern in mir-310s mutant (C, KT40/KT40; Rab23::YFP::4xmyc) compared to control germaria (A, w; Rab23::YFP::4xmyc). As a result of 7-day starvation, in controls, Rab23 has more widespread staining (B), which is more obvious in mir-310s mutants (D). The Hh ligand is produced by CpCs in the niche (outlined in white) and is visualized by anti-Hh antibody (red). Hh protein is detected along the length of the germarium under normal conditions (A). Upon starvation, Hh speckles are confined at the anterior half of the germarium (B, red line), while in the mir-310s mutant this restriction does not happen under the same conditions (D, red line). (E) The mir-310s affect the frequency and intensity of Rab23 expression in CpCs. In well-fed conditions, mir-310s mutants have lower numbers of Rab23-negative CpCs compared to controls (green bars). Upon 7 days of starvation, mir-310s mutant germaria contain a higher number of CpCs expressing high levels of Rab23 compared to controls (red bars), visualized by the intensity of Rab23 fluorescence (Table S6). (F–H) Germaria overexpressing Rab23 in the stem cell niche (bab1-Gal4/UAS-Rab23) similarly to mir-310s mutants, have increased Hh staining (increased number of Hh speckles) compared to controls under well-fed and starved conditions (F and G red lines, Table S7). (I and J) The expression activity of the hh gene locus (as visualized using hh-LacZ) does not change significantly upon starvation and is comparable in the stem cell niche (arrows) and in the escort cells (arrowheads) in well-fed (I) and starved conditions (J). (K) Both Rab23 and Hh proteins are expressed in the CpCs and their expression patterns are dynamic; some of the stem cell niche cells express Rab23 (visualized by Rab23::YFP::4xmyc) and/or Hh (green and red arrows, respectively). In addition, these proteins colocalize in subcellular foci (yellow arrows). (L) Interaction network and related their GO term analysis of Rab23 coimmunoprecipitated multiple coatomer-associated proteins (COPI) that act in intracellular vesicular trafficking are shown. The edges connecting the protein nodes indicate database-derived (Franceschini ) interactions based on coexpression (black edges), experiments (pink edges), and homology (purple edges). The complete list of Rab23 coimmunoprecipitated proteins is given in Table S8. Significances were calculated with two-tailed Student’s t-test. *P < 0.05, **P < 0.005, ***P < 0.0005. A–G and K represent single optical sections; CpCs are outlined; and I and J represent maximum intensity projections of confocal Z-stacks. Anterior is to the left. Bars, 5 µm in A–G, I ,and J and 2 µm in K.

In wild type, Hh is produced in the stem cell niche and travels into the posterior compartment to activate FSC division. We observed that the elevated levels of Rab23 in s mutants in different conditions coincided with higher levels and a broader expression pattern of the Hh ligand, as detected by an anti-Hh antibody (Figure 4, A–D). To confirm the roles of Rab23 and the s in the dispersion of the Hh ligand, we calculated the number of Hh protein speckles in the germarium (Figure 4, A–D, F, and G). Indeed, s loss and Rab23 overexpression in the stem cell niche both resulted in significantly higher numbers of Hh speckles distributed throughout the whole germarium (Figure 4, C and F, red line). Moreover, although starvation results in the restriction of Hh ligand to the anterior part of the germarium (Hartman ) (Figure 4B, red line), in the starved s loss-of-function and Rab23 overexpressing germaria, this spatial restriction was less pronounced (Figure 4, D–H, red lines; Table S7). These results confirm a role for Rab23 in the cell-autonomous positive regulation of Hh release and suggest that the effect of s deficiency on Hh ligand distribution occurs via Rab23.

Next, we tested if starvation-mediated regulation of expression occurs at the transcriptional level. However, expression of a hh-lacZ reporter transgene in the stem cell and differentiation niches did not change upon starvation (compare Figure 4I and 4J, arrows and arrowheads, respectively). These data indicate that upon dietary restriction, Hh is not regulated at the transcriptional level; on the contrary, the s and Rab23 play a role in Hh spatial distribution.

Next, we aimed to understand how Rab23 is involved in regulation of the Hh ligand. Analysis of Rab23 and Hh protein expression in CpCs revealed dynamic expression patterns such that some cells coexpressed both proteins and others were positive for either Hh or Rab23 (Figure 4, A–D). At the subcellular level, the proteins formed puncta, some of which contained Hh or Rab23 only (Figure 4K, red and green arrows, respectively) and some had both proteins colocalized (Figure 4K, yellow arrows). In general, Rab proteins are vesicle-tethering proteins, regulating intracellular trafficking (Zerial and McBride 2001). Therefore, we hypothesized that Rab23 is involved in transport and trafficking of Hh-loaded vesicles in the GSC niche cells. To test this idea, we performed a coimmunoprecipitation using ::YFP::4xmyc flies to identify Rab23 interaction partners. Subsequently, mass spectrometry analysis followed by GO term analysis of identified proteins (UniProt Consortium 2014) and evaluation of functional association networks (Franceschini ) revealed a group of 12 proteins as components of COPI-coated vesicle machinery among a larger number of other identified proteins (Figure 4L; Table S8). Importantly, this implicates Rab23 as a novel regulator of precisely controlled Hh ligand secretion in Drosophila. In summary, our results suggest a model in which s act in the stem cell niche to repress expression of Rab23, which is involved in intracellular vesicle trafficking and release of the Hh ligand in COPI vesicles.

mir-310s moderate ovarian Hh signaling via downregulation of the positive regulator Rab23

If our model is correct, then the diet-dependent s–Rab23–Hh trafficking cascade should have a direct effect on Hh signaling strength, and manipulation of the levels of these components may allow the rescue of phenotypes associated with abnormal Hh signaling. Therefore, we performed such an epistasis analysis, quantifying several ovarian phenotypes previously described as Hh signaling defects (Forbes ,b). First, we analyzed the posterior germarium architecture at the intersection of regions 2a and 2b (Figure 2F). In controls, ∼75% of germaria had germline cysts fully encapsulated by the follicular epithelial cell precursors (Figure 5A, marked by FasIII, arrow), while ectopic Hh expression in the GSC niche results in the accumulation of germ cells in germarial region 2b (Figure 5, A, A′, and B). Next we analyzed s loss-of-function and niche-specific gain-of-function phenotypes and compared their frequencies to those caused by Hh overexpression. Only 20% of s mutant germaria had this region properly structured (Figure 5, A and B). This phenotype is s specific, since it was observed in different s mutant allelic combinations and could be fully rescued by the introduction of a s genomic fragment (Figure 5, A′′′ and B; Table S9). Similar frequencies of disorganization were observed due to Rab23 overexpression (Figure 5, A′′′′ and B). If this defect is caused by the increased levels of the Hh ligand, trafficking and release of which depend on Rab23, which is in turn negatively regulated by the s, then downregulation of Hh or Rab23 should alleviate s-deficient phenotypes in the germarium. Indeed, this germarial defect was significantly rescued by reducing Rab23 or Hh levels via RNAi in a s mutant background (Figure 5B; Table S9).

Figure 5

The mir-310s and Rab23 regulate Hh signaling in the ovary. (A and B) Prior to the pinching off of the egg chamber from the germarium, the germline cysts are encapsulated by follicle cell precursors marked with FasIII, which move toward the interior of the germarium and envelop the cyst (A, arrowhead) as shown in a control (w) germarium. Hh overexpressing (A′, tub-Gal80; bab1-Gal4/ UAS-hh at 29°), mir-310s mutant (A′′, KT40/Df6070), and Rab23 overexpressing (A′′′′, bab1-Gal4/UAS-Rab23) germaria have disorganized architecture at the posterior end, with a significantly lower frequency of properly encapsulated cysts than in controls. This phenotype can be rescued by introducing a mir-310s genomic rescue construct in the mir-310s mutant background (A′′′, KT40/KT40; attB2 mir-310s rescue long2/+) (Table S9). (C–F) mir-310s deficiency causes the appearance of egg chambers with abnormal sizes and an abnormal number of nurse cells (C). In addition, the follicular epithelium becomes multilayered in irregular patches (arrowhead in C). Similarly, Hh or Rab23 overexpression results in the occurrence of egg chambers with similar defects (D and E). The frequency of this phenotype is comparable for mir-310s mutation and Rab23 overexpression. This phenotype can be rescued by downregulating the Rab23 or Hh levels in a mir-310s mutant background (KT40/KT40; bab1-Gal4/UAS-Rab23-RNAi and KT40/KT40; bab1-Gal4/UAS-hh-RNAi) (Table S9). (G–I) Loss of the mir-310s results in an excess number of cells accumulating between the egg chambers (arrowhead), forming an overcrowded, multilayered stalk. This phenotype can be rescued by introducing the mir-310s genomic rescue construct in a mir-310s mutant background (H, KT40/KT40; attB2 mir-310s rescue long2/+) (Table S9). (J and K) mir-310s mutant stalks connecting stages 6–10 egg chambers have disorganized shapes and continue to express the precursor marker FasIII (J′, arrowhead), reproducing the cell specification phenotype caused by very mild hh overexpression (tub-Gal80; bab1-Gal4/UAS-hh at 18°) (J, arrowhead). Higher levels of Hh overexpression result in a severe phenotype (depicted in Figure 2E). Rab23 overexpression causes the same stalk defects (J′′′). This phenotype can be rescued by introducing the mir-310s genomic rescue construct in the mir-310s mutant background (J′′, KT40/KT40; attB2 mir-310s rescue long2/+) (Table S9). In A, C–E, G, H, and J, anterior is to the left. A, C–E, G, and H represent single optical sections. J′–J′′′ represent maximum intensity projections of confocal Z-stacks. Bars, 10 µm. Significances were calculated using Pearson’s chi-square test. *P < 0.05, **P < 0.005, ***P < 0.0005 (Table S9).

Second, we analyzed the germline pinching-off defects. Abnormal cyst encapsulation in the germarium coupled with defective epithelial cell fate determination results in the appearance of egg chambers containing atypical numbers of germline cells (Figure 2, D and E; Figure 5, C–E). This phenotype was detected in ∼40% of s-deficient and Rab23-overexpressing ovarioles and was even more pronounced (∼90%) in the ovarioles with Hh overexpression (Figure 5, C–F). Importantly, the introduction of a s genomic fragment or reduction of Hh or Rab23 in s mutants fully rescued this phenotype, demonstrating that Rab23 and Hh act downstream of the s in this process (Figure 5F; Table S9).

Third, we analyzed the state of stalk cell specification. Increased Hh levels cause abnormal differentiation of the stalk cells, resulting in the accumulation of excessive precursor stage-like cells (Tworoger ). A total of 70% of all ovarioles contained multilayered stalks in s loss of function (Figure 5G) and Rab23 gain of function (Figure 5E), and again this phenotype was even stronger in the Hh gain of function (Figure 2E). This phenotype was fully rescued upon the introduction of a s genomic fragment (Figure 5H) or downregulation of Rab23 or Hh in s-deficient animals (Figure 5I; Table S9).

Normally, the stalk cells are terminally differentiated epithelial cells that never divide; thus, a multilayered stalk phenotype can result if stalk cell differentiation is delayed and proliferation continues. In this case, these cells would express undifferentiated epithelial precursor cell markers and undergo additional divisions. Therefore we analyzed the expression of a precursor cell marker, FasIII, in stalks between late-stage egg chambers (later than stage 6) that normally no longer express FasIII (Figure 5, J–J′′′). Overexpression of the Hh ligand in the GSC niche results in a very dramatic phenotype, in which all the stalk cells were abnormally differentiated (Figure 2E and Figure 5D). Therefore, to obtain stalks with a less severe phenotype more amenable to quantification, we overexpressed Hh in a short pulse during adulthood using the Gal4/Gal80ts system. Hh overexpression in the stem cell niche led to the appearance of stalk cells with persistent FasIII expression between late-stage egg chambers (Figure 5J). Similarly, we found FasIII-positive stalk cells in >50% of the analyzed stalks in s loss-of-function and Rab23 overexpressing ovarioles (Figure 5, J and J′′; Table S9). Importantly, downregulation of Hh signaling, either by s genomic rescue or by downregulation of Rab23 or Hh ligand expression in the GSC niche, rescued the stalk cell specification phenotype (Figure 5K; Table S9).

Together, these phenotypic analyses indeed show that higher levels of Rab23 in s mutants phenocopy overactive Hh signaling and confirm that the effect the s have on Hh signaling is accomplished via their regulation of Rab23. Thus, Hh signaling can be intensified via Rab23-mediated enhancement of Hh ligand trafficking/release and the s moderate this signaling cascade via the upstream targeting of .

Hh signaling is regulated by the mir-310s in response to diet

Since our data show that s mutants show defective metabolic status (Figure S1), and that the s act upstream of Hh signaling via repression of (Figure 3C; Figure 4, C–E), we decided to test whether these miRNAs would aid in adjusting Hh signaling efficiency in response to nutritional stress. Remarkably, we observed that the dramatic ovarian phenotypes associated with excessive Hh signaling were radically improved upon dietary restriction (compare Figure 6A and 6B). Similarly, the appearance of mutant phenotypes in s ovaries was significantly rescued upon starvation (compare Figure 6C and 6D). To quantify the effect of starvation, we focused on the overproliferated stalk phenotype, since it is a hallmark of hyperactive Hh signaling. A total of 100% of Hh-overexpressing ovarioles contained patches of multilayered stalk cells, while upon starvation the frequency of this phenotype was reduced by half; in s, this phenotype was fully rescued by dietary restriction (Figure 6E; Table S9).

Figure 6

The phenotypes caused by mir-310s loss or hh overexpression can be alleviated by dietary restriction. (A–E) hh gain of function causes epithelial defects resulting from somatic cell overproliferation (A, arrows). Upon nutritional restriction for 3 days, the dramatic hh gain-of-function (tub-Gal80; bab1-Gal4/UAS-hh at 29°) phenotypes become significantly less penetrant (B and E). Similarly, the appearance of the atypical multilayered epithelium in mir-310s mutant (C, arrows, KT40/Df6070) ovaries is dramatically reduced upon nutritional restriction (C–E) (Table S9). (F–H) Under nutritional stress, on average less than one mitotically active follicle cell (marked by PH3) per stage 2 egg chamber is found in controls (F and H). In the mir-310s mutant, this number is increased (G and H). After nutritional restriction for 7 days, egg production is slowed down, which results in a reduction of follicular epithelial cell proliferation. However, the number of PH3-positive cells is fourfold higher due to mir-310s loss (H). Similarly, upon starvation, overexpression of Rab23 (tub-Gal80; bab1-Gal4/UAS-Rab23) and Hh (tub-Gal80; bab1-Gal4/UAS-hh) (4 days at 29°) results in an approximately fivefold higher PH3-positive cell number compared to control (H). The high mitotic activity in mir-310s mutant egg chambers is rescued by an independent genomic mir-310s rescue construct (KT40/KT40; attB2 mir-310s rescue long2/+), or by downregulating the Rab23 (KT40/KT40; bab1-Gal4/UAS-Rab23-RNAi) or Hh levels (KT40/KT40; bab1-Gal4/UAS-hh-RNAi) (Table S10). A–D, F, and G represent single optical sections and anterior is to the left. Bars, 20 µm in A–D and 5 µm in F and G. In H, the bar graph indicates AVE ± SEM. Significances were calculated for E using Pearson’s chi-square test (Table S9) and for H, using Mann–Whitney U-test and Z-statistic. *P < 0.05, **P < 0.005, ***P < 0.0005 (Table S10).

It is known (Forbes ,b) that ectopic expression results in excessive somatic cell proliferation and that stimulated Hh release can induce ligand accumulation on the follicle cells, hence promoting their division even under nutrient-restricted conditions (Hartman ). Therefore, we analyzed the number of mitotically active cells among follicular epithelium cells wrapping stage 2 egg chambers using phosphohistone 3 (PH3) as a mitotic marker. Under normal conditions, s mutant, Hh and Rab23 overexpressing egg chambers also have a mild increase (1.2- to 2-fold) in the number of follicle cells in mitosis (Table S10). This tendency became even more pronounced under starvation, as s mutants had almost 4-fold higher numbers of follicle cells in mitosis when compared to controls (Figure 6, F–H). To determine whether this proliferation phenotype is caused by excessive Rab23 levels and, thus, overactive Hh signaling, we overexpressed Rab23 or Hh ligand in Hh-sending cells. As expected, even upon starvation, both of the overexpression experiments resulted in ∼5-fold higher numbers of dividing follicle cells in the follicular epithelium, suggesting that Rab23 cell-autonomous involvement in the Hh signaling pathway is diet dependent. Notably, the excessive s follicle cell division phenotype under starvation was significantly rescued by the introduction of the s genomic locus or downregulation of or in a s mutant background (Figure 6H; Table S10). These results show that the abnormal follicle cell proliferation upon dietary restriction is caused by the higher levels of , and, consequently, overactive Hh signaling as a result of s loss of function. Furthermore, this confirms our hypothesis that the s–Rab23–Hh ligand signaling cascade regulates Hh signaling activity, and this regulation becomes even more prominent in response to dietary fluctuations.

Together, our data show that the miRNAs pathway plays an important role in adjusting the metabolic status of an organism to nutritional signals. In particular, we found that the s are diet-sensitive and that s-deficient flies exhibit severe abnormalities in metabolic homeostasis, including altered gene and protein expression profiles. In addition, multiple diet-sensitive physiological processes, such as crop size, lipid storage, and fecundity are perturbed. Furthermore, we found that the s are capable of targeting at least three genes associated with the Hh signaling pathway, ensuring a robust, fast, and precise response to diet alterations via modulation of this vital signaling pathway. Particularly, in the Hh signal-sending cells, the s represses expression of factors regulating Hh ligand production: DHR96, which senses systemic cholesterol levels and promotes Hh release; and Rab23 which, as we propose here, functions in vesicles required for Hh trafficking. In the Hh signal-receiving cells, mRNA encoding the negative Hh signaling transducer and transcription factor, is targeted by the s. Possibly, targeting of several components of the same signaling pathway is a critical principle of miRNA regulation of stress signaling pathways that should be specifically considered in our understanding of the roles of miRNAs in physiologic and pathophysiologic stress.

Discussion

Here we propose a model for a prompt dietary stress response in which nutritional signals are transduced via miRNAs that act upstream of vital cellular signaling pathways to fine tune their activity and efficiently adapt organismal metabolism to ensure healthy homeostasis (Figure 7). Here we found that the s, via targeting of multiple Hh pathway components, ensure rapid and robust adjustment of Hh signaling in response to dietary signals. Normally, the capacity of organisms to adapt quickly to changeable food conditions is crucial for their survival since dietary components and food availability can vary rapidly. It is known that adult miRNA mutants rarely show extreme phenotypes in well-controlled laboratory conditions; however, upon stress, a miRNA deficiency frequently has a profound effect on organism survival and adaptability (Bhattacharyya ; Leung and Sharp 2010; Mendell and Olson 2012; Edeleva and Shcherbata 2013). Therefore, our interpretations that miRNAs act only upon stress may not be entirely reasonable; their functions may be broader and more basic to control organismal homeostasis. Unique challenges and opportunities for miRNA studies, and in particular for miRNA research focused on the stress response, are to identify the biologically relevant downstream targets regulated by miRNAs, which will allow not only to better understand the mechanisms of stress responses, but also to provide the understanding of how organisms constantly fine tune gene expression to maintain healthy homeostasis in the ever-changing external and internal environments. Here we propose a new workflow that facilitates the identification of miRNA targets and conditions under which studied miRNAs might function.

Figure 7

Model of a miRNA-based nutritional stress response signaling relay. In the ovary, upon dietary fluctuations, the mir-310s target multiple components associated with Hh signaling, which ensures fast dietary response and adapts oogenesis. miRNAs can also act on other targets that belong to other critical pathways (for example, Wg), further coordinating the efficiency of the process. See also Discussion for details.

During embryonic development, miRNAs often act only as fine tuners of gene expression, differentiation guardians, and canalization factors, as embryonic development is extremely well programmed and protected from environmental stimuli and, therefore, it should just be stabilized to succeed (Siegal and Bergman 2002; Hornstein and Shomron 2006; Yatsenko and Shcherbata 2014). However, during adulthood, miRNAs often greatly influence the responses of adult tissues to stressful conditions or hormonal fluctuations (Leung and Sharp 2010; Fagegaltier ). To have a profound effect on gene expression, several mechanisms assuring the effectiveness of miRNA-based gene expression regulation have been developed, such as high expression of the miRNA, positive feedback loops, or targeting of multiple components of the critical pathway. For the newly emerged s family, misexpression would be damaging since the s have hundreds of putative and several already confirmed critical targets, such as , (), and , deregulation of which could be fatal (Tsurudome ; Pancratov ; Yatsenko ). Positive feedback signaling is also somewhat unlikely because then miRNAs would be expressed in all Hh signal-receiving cells, which, as we have shown, is not the case for the s. Instead, the s are expressed dynamically only in some of the Hh signal-sending and signal-receiving cells. Previously it was shown that the s gene expression is sensitive to nitric oxide levels (Yatsenko ), which via nitrosylation of histone deacetylases regulates the cellular epigenetic profile. Epigenetic modifications that play a key role in the regulation of gene expression can also be influenced by both the quality and quantity of the diet (Daniel and Tollefsbol 2015). Based on the previous data, it is logical to hypothesize that the dynamic s expression in ovaries could also be dependent on specific histone modifications. Currently, it is unknown which signaling induces s expression in response to deficit of nutrients; however, s ability to target both the factors required for Hh ligand release in the signal-sending cells (Rab23 and DHR96) and the Hh signal transducer (the transcription factor Ttk) in the signal-receiving cells, ensures that the diet-dependent Hh pathway is securely downregulated upon restrictive diet. While previous data propose that modulation of Hh signaling is a primary dietary stress-response mechanism controlling stem cell proliferation (Horner ; Hartman ), we show that the s act upstream of this signaling, demonstrating that miRNAs fine tune a major cell signaling pathway to adjust its strength in the stem cell niche to changing dietary conditions. Even though the miRNAs are generally not well conserved between Drosophila and humans, the processes they regulate are. Therefore, it would be interesting to study whether Hh signaling is also regulated via miRNAs in vertebrates upon diet.

Hh is one of the canonical developmental pathways crucial for the development of a variety of tissues in all bilaterians; thus, finding new components of this pathway is of great importance. We identified the s as a novel upstream regulatory element of this pathway in Drosophila. Namely, the post-transcriptional control of the expression levels of at least three genes from the Hh pathway (, , and ttk) depends on these miRNAs to sustain tissue homeostasis, which has to assume new equilibrium under changing environmental/nutritional conditions. Interestingly, the highly evolutionarily conserved Hh signaling pathway has been shown to play a role in obesity-like fat accumulation in Drosophila and mouse adult stem cells (Pospisilik ).

Importantly, we identified a new regulator of Hh signaling in Drosophila, Rab23. Rab proteins are a family of small GTPases that play key roles in vesicle cargo transport, docking, and fusion and are important for fine tuning of various canonical pathways, safeguarding proper development, tissue morphogenesis, and homeostasis (Zhang ). It is known that Rab proteins can have redundant functions (Chan ); therefore, deficiency or downregulation of only one of them might not have a dramatic effect on animal viability. Indeed, loss of function did not result in any of the analyzed ovarian phonotypes, demonstrating that Rab23 is dispensable for Hh signaling function in the ovary, while its upregulation had an important effect on Hh signaling strength. Based on the proposed Rab23 vertebrate homolog function, Drosophila Rab23 was expected to regulate the trafficking of vesicle-associated components in the Hh signal receiving cells (Evans ; Pataki ). However, our data demonstrate Rab23-based regulation in the Hh signal sending cells. We propose a new mechanism in which Rab23 has a cell-autonomous role in Hh signal-sending cells in the ovary and that diet-sensitive s are potent regulators of and its downstream trafficking events. Interestingly, Drosophila and human Rab23 are highly evolutionarily conserved (with 59% protein sequence homology) and human is a putative target for the human s orthologs mir-25, mir-32, mir-92a/b/c, mir-363, and mir-367 (Enright ; Kheradpour ; Betel ). miRNA-based control of conserved pathways is also generally conserved between species, implying that their regulatory role could have ancient origins. Therefore it will be important to test whether human is regulated via miRNAs as well.

In addition, Rab23 (Wang ) and the COPI complex (Beller ) have been shown to play a role in lipid homeostasis by affecting lipid droplet size, and the COPI complex also takes part in cholesterol-modified Hh ligand release (Lum ; Nybakken ; Aikin ). Together, our data show that miRNAs can fine tune cell signaling to tailor adult oogenesis to changing dietary conditions. Since miRNAs usually are capable of targeting multiple genes, components of the Hh signaling are not the only s targets. At least in the male germline stem cell niche, a Drosophila homolog of vertebrate beta-catenin, is a bona fide s target, and s deficiency has an effect on male fertility (Pancratov ). Arm is not only a major cell adhesion protein that is involved in homophilic cell adhesion via its binding to cadherins, but also it is the major transcription factor involved in Wingless (Wg) signaling (Wodarz ; Somorjai and Martinez-Arias 2008). Recently we found that Wg signaling acts in the germline to regulate the efficiency of germline stem cell progeny differentiation (Konig and Shcherbata 2015). At the same time, the strength of the homophilic cell adhesion between the germline and the soma regulates Wg signaling. Thus, somatic cells communicate to the germline via cell adhesion (Cadherin–Arm complexes), adjusting the speed of germline differentiation (Konig and Shcherbata 2015). Therefore, amounts of Arm available either for cell adhesion or Wg signaling have a profound effect on oogenesis progression. Since the s expression in the GSC niche and differentiation niche cells is very dynamic and depends on dietary fluctuations, it would be interesting to study whether the diet-dependent s could also be involved in regulation of GSC progeny differentiation via targeting of levels (Figure 7). This s-mediated soma–germline communication mechanism (the s-regulating ) could additionally be used to coordinate the speed of oogenesis with the nutritional status of the whole organism. Theoretically, the simultaneous management of different signaling pathways via the same miRNAs may aid in coordinating the stress response. In particular, modification of vital cell signaling via miRNAs in response to dietary changes might be commonly implicated in the process of adapting egg production to dietary conditions to ensure sufficient progeny survival.