Dysregulation of miRNA biogenesis impacts innate immunity signaling

The ribonuclease Drosha is a core component of the miRNA processing machinery essential for the biogenesis of most miRNAs [60]. To examine whether defects in miRNA biogenesis affects innate immunity signaling, we first silenced Drosha expression in vivo by crossing a shRNA transgenic line targeting Drosha (sh-Drosha) to the daughterless-Gal4 (da-Gal4); tub-Gal80^ts composite line [61]. A control cross was set up between a shRNA transgenic line targeting gfp and da-Gal4; tub-Gal80^ts flies. To minimize lethality due to the requirement for select miRNAs in development, fly crosses were kept at permissive temperature (18°C). Upon eclosure progeny of appropriate genotype was shifted to restrictive temperature (29°C) for 5 days to allow for shRNA transgene expression and target gene silencing. This strategy allowed us to achieve significant knockdown of the Drosha mRNA, as measured by RT-qPCR (Fig 1A). Consistent with the critical requirement for Drosha in miRNA biogenesis, depletion of Drosha leads a marked accumulation of several primary miRNA transcripts compared with a control knockdown (tub-Gal80^ts da>sh-gfp) (Fig 1A). Importantly, we found that tub-Gal80^ts da>sh-Drosha flies display a decrease in AMP expression levels compared with control animals, under both non-infection and E. coli infection conditions (Fig 1B). Interestingly, crosses between da-Gal4 and shRNA-Drosha at room temperature yielded under-represented but viable da>sh-Drosha progeny. As expected, we detected a decrease in levels of the Drosha mRNA and an accumulation of primary miRNA transcripts in these flies (S1 Fig). However, in contrast to the observed decrease in AMP expression in tub-Gal80^ts da>sh-Drosha flies, da>sh-Drosha flies display elevated basal levels of AMP expression compared with control animals (S1B Fig). The seemingly disparate phenotype in AMP expression in da>sh-Drosha progeny between the presence and absence of tub-Gal80^ts could be attributable to the difference in the onset and duration of Drosha depletion: While Drosha depletion occurred in da>sh-Drosha animals throughout development, it did not take place in tub-Gal80^ts da>sh-Drosha flies until fully developed adult animals were shifted to the restricted temperature. For instance, the barrier function of the digestive tract could be compromised in da>sh-Drosha animals due to a requirement for Drosha/miRNAs during development, leading to escape of the microbes and an increase in AMP expression. Nonetheless, these data demonstrate that dysregulation of miRNA biogenesis as a whole impacts the IMD innate immunity signaling pathway.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Over-expression of miR-34 activates innate immunity signaling.

(A) Total RNA was isolated from male progeny from crossing flies carrying the ubiquitously expressed da-Gal4 driver and a temperature-sensitive Gal80 transgene (da>Gal4 tub-Gal80^ts) to UAS-shRNA lines targeting Drosha or the control gfp. Flies crosses were kept at 18°C and progeny were shifted to 29°C for 5 days upon eclosure to induce shRNA expression. Steady-state levels of mRNAs encoding Drosha and several primary miRNA transcripts were measured by qRT-PCR, and normalized to levels of the RpL32 mRNA. RNA isolated from da>gfp shRNA males serves as negative control. (n = 3). (B) Flies were left untreated (- E. coli) or infected by E. coli via septic injury (+ E. coli), total RNAs were extracted 6 hours post-infection and mRNAs encoding the AMP Diptericin was measured and normalized to levels of RpL32 (n = 5; mean + standard deviation (SD)). (C-E) Select miRNAs were over-expressed in flies by crossing UAS-miRNA transgenic lines da>Gal4 tub-Gal80^ts flies. Flies crosses were kept at 18°C and progeny were shifted to 29°C for 5 days upon eclosure to induce miRNA expression. (C) Northern blot shows levels of select miRNAs (right) in control and miRNA over-expression flies. 2S rRNA serves as loading control. In addition, flies were either uninfected (D) or infected with E. coli via septic injury (E). Total RNA was isolated from flies 6 hrs post-infection and levels of Diptericin mRNA were measured by RT-qPCR and normalized to the RpL32 mRNA. RNA samples from da>gfp shRNA flies serves as control. Note that levels of the Diptericin mRNA in non-infected and E. coli-infected da>gfp shRNA flies serve as baseline controls in both D and E (n≥4). (F) A Northern blot shows miR-34 expression levels in naïve S2 cells and miR-34 overexpression cells (both were treated with 20-HE at 1 μM for 24 hrs). (G) S2 cells over-expressing miR-34 and control cells were both treated with 20 hydroxy-ecdysone (20-HE) at 1 μM for 24 hrs. Subsequently cells were either left untreated or treated for 6 hrs with a crude lipopolysaccharide sample at 10 μg/mL, which contains the immune stimulator peptidoglycan (PGN). Total RNA was isolated and levels of Diptericin mRNA were measured by RT-qPCR and normalized to the RpL32 mRNA (n = 3). (H) Canonical components of IMD signaling were depleted in miR-34 over-expressing cells using dsRNAs targeting IMD pathway components (below) or a control dsRNA against the firefly luciferase gene. Cells were first treated with 20-HE for 24 hours, and subsequently were either left untreated or treated with PGN, and levels of Diptericin mRNA were measured by RT-qPCR and normalized to the RpL32 mRNA (n = 3). (I) UAS-miR-34 or the control UAS-sh-gfp flies were crossed to da>Gal4 tub-Gal80^ts flies. Flies crosses were kept at 18°C. Upon eclosure, progeny of appropriate genotypes were collected and shifted to 29°C for 7 days. Flies in groups of 45 were subsequently injected with a concentrated culture of Erwinia carotovora carotovora 15 (Ecc15) or PBS (non-infection control) and kept at 29°C. Fly survival was recorded daily up to day 8 post-infection and plotted (n≥3; p<0.05 between Ecc15-infected control and miR-34^OX (da>miR-34 Gal80^ts) flies). (J-K) Control or miR-34^OX flies were infected by injecting a concentrated culture of Ecc15 (J) or overnight culture of Enterobacter cloacae (K). At various time points post-infection, groups of 3 flies in J (and groups of 4 flies in K) were collected and homogenized in sterile PBS. Fly homogenates were diluted and plated onto Ampicillin- (in J) or Nalidixic acid-containing (in K) LB plates, and the resultant colonies were counted one day later. Shown are colony-forming units (CFUs) per fly (n≥4).

https://doi.org/10.1371/journal.ppat.1006034.g001

Over-expression of miR-34 leads to hyperactivation of innate immunity signaling both in cultured cells and in vivo

To further reveal the identities of cellular miRNAs that underlie the innate immunity phenotype, we conducted a miRNA over-expression screen in vivo using 101 UAS-miRNA transgenic lines and the da-Gal4; tub-Gal80^ts composite line [62]. A control cross was set up between a shRNA transgenic line, which expresses an artificial sh-gfp RNA embedded in the miR-1 cassette [61], and da-Gal4; tub-Gal80^ts flies. Fly crosses were kept at permissive temperature (18°C) until adult progeny of the appropriate genotype emerged. Subsequently the progeny was shifted to restrictive temperature (29°C) for 5 days to allow for miRNA transgene expression. This strategy allowed us to achieve marked over-expression of individual miRNAs. Fig 1C shows results of Northern blot analyses for select miRNAs. Under these experimental conditions, viable adult progeny of the appropriate genotype that over-expresses individual miRNAs was readily recovered. One exception was let-7, which causes lethality at pupal stage (S1 Table). Subsequently, these flies were either left untreated or infected with E. coli via septic injury. Total RNA was isolated 6 hours post-infection and levels of the Diptericin (Dpt) mRNA were measured and normalized to the RpL32 control mRNA. S1 Table documents the relative Diptericin mRNA levels in uninfected and E. coli-infected flies compared to the respective controls (da>sh-gfp flies that are uninfected and infected by E. coli). The AMP expression phenotype for select miRNAs (miR-34, miR-92a, miR-9a and miR-989) is shown (Fig 1D and 1E). In subsequent studies, we focused on miR-34 because 1) miR-34 over-expression causes a concordant increase in levels the Diptericin mRNA in both uninfected and E. coli-infected flies; and 2) miR-34 expression is regulated by ecdysone signaling, which impacts innate immunity [33, 63].

Besides IMD signaling, which is activated upon Gram-negative bacterial infection, we also asked whether miR-34 additionally impacts other signaling routes, such as the Toll signaling pathway, which mediates host defense against infection by fungi and Gram-positive bacteria. Interestingly, our analysis revealed that Drosomycin expression in response to M. luteus infection was significantly decreased in miR-34 over-expressing flies compared with controls (S2 Fig), suggesting that miR-34 differentially impacts IMD and Toll signaling. Next, we examined the impact of miR-34 over-expression on AMP expression in immuno-competent cultured S2 cells. We detected higher levels of the Diptericin mRNA, as well as a panel of mRNAs encoding additional AMPs, including Cecropin A1 (CecA1), Attacin A (AttA), Metchnikowin (Mtk) and Defensin (Def), in miR-34 overexpressing cells than in control cells, both in the absence and presence of peptidoglycan (PGN) treatment (Fig 1F and 1G, S3 Fig). In addition, levels of the pirk mRNA, which is transcriptionally activated by IMD signaling and encodes a negative regulator of IMD signaling, were also increased both in untreated and PGN-treated cells upon miR-34 over-expression (S4 Fig). Importantly, depletion of core components of the IMD signaling pathway, such as PGRP-LC, subunits of the DmIKK complex (Ird5 and Kenny), or Relish, significantly alleviated the immune-activation phenotype of miR-34 (Fig 1H), indicating that the canonical IMD signaling pathway is required for the immune-stimulating function of miR-34. Interestingly, in ecdysone-treated S2 cells the effect of miR-34 overexpression appears to be PGRP-LC-dependent, even in the absence of PGN stimulation. It is possible that miR-34 overexpression may impact ecdysone-mediated regulation of PGRP-LC expression [33], thereby affecting IMD signaling. Taken together, these data demonstrate that mis-expression of miR-34 affects the IMD innate immunity signaling pathway both in cultured cells and in vivo.

miR-34 impacts innate immunity signaling in vivo

To further define the role of miR-34 in innate immunity signaling in vivo, we injected either sterile PBS or a suspension of a concentrated culture of the fly pathogen Erwinia carotovora carotovora strain 15 (Ecc15) into miR-34 over-expression or control flies, and monitored fly survival at different time points post injection. As expected, flies of both genotypes survive well in response to PBS injection (Fig 1I). Importantly, while flies of both genotypes show a decrease in survival following Ecc15 infection, miR-34 over-expression flies display a significantly higher survival rate than control flies (Fig 1I), consistent with the observed higher levels of AMP expression in miR-34 over-expression flies. To further examine whether enhanced survival of miR-34 over-expression flies upon Ecc15 infection is attributable to improved pathogen clearance or immune tolerance, we monitored pathogen load at various time points post Ecc15 infection. Our analysis revealed that while infected flies of both genotypes carry comparable amount of Ecc15 at the starting time point, miR-34 over-expression flies appear to out-perform control animals in clearing invading pathogens, as indicated by the lower pathogen load in miR-34 over-expression flies than in controls (Fig 1J). Similar observations were made upon infecting flies with, Enterobacter cloacae, a second Gram-negative bacteria (Fig 1K). Furthermore, to examine whether miR-34 over-expression affects hemocyte-mediated bacterial phagocytosis, which is a key mechanism of the cellular immune response, we injected into flies E. coli bio-particles conjugated with a PH-sensitive dye, which becomes fluorescent only after being engulfed by hemocytes and sorted into the acidic endosomal compartment. This analysis reveals a similar degree of phagocytosis between control and miR-34 over-expression flies (S5A and S5B Fig). We conclude that miR-34 over-expression enhances innate immunity and host survival upon Gram-negative bacterial infection, at least in part, by promoting AMP expression and pathogen clearance.

Next, we examined whether miR-34 deficiency impacts innate immunity by measuring AMP expression both in miR-34 knockout (miR-34^KO) flies [64] and in control (Res) animals (miR-34^KO flies carrying a miR-34 genomic rescue construct) (Fig 2A). This analysis revealed that miR-34^KO flies express significantly lower levels of the Diptericin mRNA than control animals, under both non-infection and E. coli infection conditions (Fig 2B). To alleviate the concern that endogenous or environmental microbes could potentially impact host innate immunity and cause variations in levels of AMP expression, we also analyzed miR-34^KO and wildtype flies raised in media containing multiple antibiotics. We note that antibiotics treatment alone may not completely eliminate microbes. Nonetheless, under these conditions, lower levels of Diptericin expression was detected in miR-34^KO animals than in control animals (Fig 2C and 2D). Consistent with lower levels of Dpt expression, miR-34^KO flies present poorer survival and pathogen clearance in response to Ecc15 compared to control animals (Fig 2E). Lastly, since miR-34 expression steadily increases with age, we also examined AMP expression, host survival and pathogen clearance upon Ecc15 infection in various groups of age-matched miR-34^KO and control flies. Consistent with previous reports [65–67], we detected higher basal levels of Diptericin transcript in aged control flies than in young flies (Fig 2C). Notably, this age-dependent increase in basal levels of AMP expression persists in miR-34^KO animals, suggesting that age-dependent increase in miR-34 expression does not significantly contribute to the observed difference in AMP expression between young and old flies. Importantly, in both young and aging settings, miR-34^KO flies display lower basal and E. coli infection-induced levels of Diptericin expression, poorer survival and pathogen clearance in response to Ecc15 challenge compared to age-matched control animals (Fig 2C–2H). Lastly, our analysis reveals that miR-34^KO flies display a defective survival and pathogen clearance in response to Enterobacter cloacae infection (Fig 2I and 2J), and that miR-34 deficiency did not significantly impact hemocyte-mediated phagocytosis (S5C and S5D Fig). These data demonstrate that miR-34 deficiency compromises AMP expression and impairs IMD signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. miR-34 deficiency compromises innate immunity.

(A) A Northern blot shows levels of miR-34 and the control 2S rRNA in miR-34 knockout flies (KO) or knockout flies carrying a miR-34 rescue transgene (control, ctr). (B) Flies were either uninfected or infected with E. coli via septic injury. Total RNA was isolated and levels of the Diptericin mRNA were measured by RT-qPCR and normalized to the RpL32 control (mean + SD; n = 3). (C-D) miR-34^KO and control flies were reared in standard food supplemented with antibiotics. Age-matched fly progeny (young– 3d, old– 22d) were either uninfected (C) or infected with E. coli via septic injury (D), and levels of the Diptericin mRNA were measured by RT-qPCR and normalized to the RpL32 control (n≥3). In both panels, the Dpt/RpL32 ratio in non-infected 3d old control flies serves as baseline. (E-F) Groups of 45 age-matched miR-34^KO and control flies (E, young– 3d; F, old– 23d) were injected with a concentrated culture of Ecc15 or PBS. Fly survival was recorded daily and plotted (n≥3; p<0.001 between Ecc15-infected control and miR-34^KO flies). (G-H) A similar group of age-matched control or miR-34^KO flies (as in E and F) were infected by injecting a concentrated culture of Ecc15. At various time points post-infection, groups of 3 flies were collected and homogenized in sterile PBS. Note that due to lethality, 1 fly per group was used for a subset of data points in miR-34^KO flies 2 days post-infection. Fly homogenates were diluted and plated onto Ampicillin-containing LB plates, and the resultant colonies were counted one day later. Shown are colony-forming units (CFUs) per fly (n≥5). (I) Groups of 4–7 d miR-34^KO and control flies were injected with a concentrated culture of Enterobacter cloacae or PBS. Fly survival was recorded daily and plotted (n≥3; p<0.001 between Enterobacter cloacae-infected control and miR-34^KO flies). (J) A similar set of flies (as in I) were infected by injecting an overnight culture of Enterobacter cloacae. At various time points post-infection, groups of 4 flies were collected and homogenized in sterile PBS. Fly homogenates were diluted and plated onto Nalidixic acid-containing LB plates, and the resultant colonies were counted one day later. Shown are colony-forming units (CFUs) per fly (n = 5).

https://doi.org/10.1371/journal.ppat.1006034.g002

Identification of miR-34 target genes relevant to innate immunity signaling

To identify miR-34 targets, we first employed two widely used bioinformatics algorithms (TargetScan and PicTar) [46, 68, 69]. In addition, we performed mRNA sequencing and compared the mRNA expression profiles in cells over-expressing miR-34 with that in control samples. As most miRNAs only mildly reduce the expression of cognate target genes, we undertook an inclusive approach and considered all mRNAs that display a decrease in expression upon miR-34 over-expression. Integration of these datasets identified a list of 27 genes that not only scored positively using both bioinformatics algorithms, but also displayed a decrease in mRNA levels upon miR-34 over-expression (S6 Fig & S2 Table). Among these 27 genes is Eip74EF, which is a well-characterized miR-34 target, thereby validating our approach [64]. Next, to examine whether any of these 27 genes are functionally relevant to innate immunity signaling, we knocked down these genes individually in S2 cells and examined the impact on Diptericin expression prior to and after PGN treatment. This analysis revealed that inactivation of three genes, CG8468, dlg1 and mura, led to an increase both in basal levels and/or PGN-induced Diptericin expression (Fig 3A). While the magnitude of changes in Diptericin expression upon silencing of individual genes appears to be weaker than that elicited by miR-34 overexpression, possibly due to low knockdown efficiency and/or slow turnover of target proteins, these data nonetheless suggest that CG8468, dlg1 and mura are candidate miR-34 target genes that could potentially downregulate IMD signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. dlg1 is a miR-34 target gene relevant to innate immunity signaling.

(A) S2 cells treated with various dsRNAs (below) were first treated with 20-HE for 24 hrs, and were subsequently either left untreated or treated with PGN. Total RNA was isolated and levels of Diptericin were measured and normalized to RpL32 (n = 3). Imd dsRNA serves as a positive control. (B) Levels of Dlg1 protein in dlg1 knockdown cells or control cells were measured by immunoblot. Tubulin serves as a loading control. The remaining Dlg1 protein reflects incomplete depletion. (C-D) Dlg1 was knocked down in flies using a UAS-shRNA and the da-Gal4 tub-Gal80^ts transgenes in a similar strategy as described in Fig 1. Levels of dlg1 (C) and Diptericin mRNA (D) in E. coli-infected flies were measured by qRT-PCR, and normalized to RpL32. RNA isolated from progeny of a parallel cross using the sh-gfp transgene serves as negative control (n = 3). (E-F) Levels of dlg1 mRNA (E) or protein (F) in S2 cells over-expressing miR-34 or in control cells were measured. The RpL32 mRNA and Tubulin protein serve as controls, respectively (n = 3). The remaining Dlg1 protein in F reflects incomplete depletion. (G-H) Levels of Dlg1 protein in miR-34 knockout or control flies were measured by immunoblot (G). Note that multiple Dlg1 protein isoforms were detected in fly extracts. Their collective abundance was quantified and shown in H (n = 3; mean + SD). (I) Renilla luciferase reporter constructs that carry the dlg1 3’ UTR containing either a wildtype (WT) or mutant (mut) miR-34 binding site were generated. Seed region of miR-34 was highlighted in green. (J) The Renilla luciferase reporter constructs (in J) together with a control firefly luciferase construct were transfected into S2 cells with or without a miR-34 expression vector. Reporter activities were measured and relative Renilla/Firefly ratio is shown (n = 3). (K) Various combinations of miR-34 and Flag-Dlg1 expression constructs were transfected into S2 cells. Total RNA was isolated and levels of the Diptericin mRNA were measured and normalized to RpL32 (upper panel; n = 3). Levels of Dlg1 protein were measured by immunoblot (lower panel).

https://doi.org/10.1371/journal.ppat.1006034.g003

Validating dlg1 and Eip75B as bona fide miR-34 target genes that modulate innate immunity signaling

Dlg1 is a member of membrane-associated guanylate kinase (MAGUK) family proteins (albeit it lacks catalytic activity), and is localized to septate junctions [70]. Depletion of Dlg1 in cultured S2 cells led to an increase in both basal and PGN-induced Diptericin expression (Fig 3A and 3B). In addition, we assessed the impact of Dlg1 depletion on innate immunity signaling in flies using UAS-shRNA transgenic lines (sh-dlg1 and the sh-gfp control) and the da-Gal4; tub-Gal80^ts composite line. As expected, levels of the dlg1 mRNA are reduced in da>sh-dlg1 flies compared with that detected in the control da>sh-gfp flies (Fig 3C). Importantly, such a decrease in dlg1 mRNA levels correlated with an increase in levels of the Diptericin mRNA (Fig 3D). These data demonstrate that a reduction in levels of Dlg1 correlates with enhanced IMD signaling both in cultured cells and in vivo. Consistent with the notion that dlg1 is a miR-34 target gene, we observed a reduction in both mRNA and protein levels of dlg1 upon over-expression of miR-34 (Fig 3E and 3F). In addition, higher levels of Dlg1 protein were detected in miR-34 knockout flies than in control animals (Fig 3G and 3H). Furthermore, our bioinformatics analysis predicted a miR-34 binding site in the dlg1 3’ untranslated region (3’ UTR) (Fig 3I). To show definitively that the predicted miR-34 binding site is sufficient to confer gene silencing, we performed reporter assays using luciferase reporter constructs carrying either the wildtype or mutant dlg1 3’ UTR, which abolishes miR-34 binding (Fig 3I). This analysis reveals that miR-34 is capable of silencing the reporter construct carrying a wildtype dlg1 3’ UTR, and this repression is relieved by introducing mutations in the miR-34 binding site (Fig 3J). Lastly, we found that while miR-34 over-expression led to a significant reduction in levels of Dlg1 protein, expression of a dlg1 cDNA construct containing only the open reading frame efficiently restored Dlg1 protein levels in S2 cells even in the presence of exogenous miR-34 (Fig 3K, lower panel). This is consistent with the notion that a functional miR-34 target site is present only in the 3’ UTR of the Dlg1 mRNA. Importantly, expression of dlg1 efficiently blunted the immune-stimulating effect of miR-34 (Fig 3K, upper panel). Taken together, these data demonstrate that dlg1 is a bona fide (perhaps a major) miR-34 target gene relevant to IMD signaling.

Among our list of miR-34 targets is Eip74EF, which encodes a key component of the ecdysone signaling cascade and has been previously identified as a miR-34 target gene [64]. We confirmed these findings by showing that over-expression of miR-34 can lead to a significant reduction in the levels of Eip74EF protein (Fig 4A and 4B). Inspired by the notion that a single miRNA can coordinately regulate the expression of multiple components of a given pathway [71], we performed immunoblot assay and surveyed additional ecdysone response genes to search for additional miR-34 targets. This analysis revealed that protein levels of Eip75B, a transcriptional factor and component of ecdysone signaling, are significantly reduced by miR-34 over-expression (Fig 4A and 4B). Further inspection of the Eip75B mRNA sequence allowed us to identify five potential miR-34 sites in the coding region. In order to test whether any of these sites can confer miR-34-mediated gene silencing, we generated reporter constructs by placing tandem triple repeats of individual miR-34 binding sites derived from the Eip75B mRNA in the 3’ UTR of the firefly luciferase gene, and examined whether these reporter genes can be repressed by miR-34. In addition, we generated a corresponding set of reporter constructs carrying mutations in candidate miR-34 binding sites that abolish miR-34 recognition. This analysis reveals that reporter genes containing wild type sites #1 or #5 were more efficiently repressed by miR-34 than the reporter with mutant sites (Fig 4C and 4D & S7 Fig). In addition, depletion of Eip75B in S2 cells resembles the miR-34 over-expression phenotype, i.e. activation of innate immunity signaling both in the presence and absence of PGN treatment (Fig 4E and 4F). These observations are consistent with previous studies that identified Eip75B as a negative regulator of IMD signaling [33, 72]. Importantly, over-expression of the Eip75B-RC isoform, which lacks site #1, significantly suppressed the immune-activation function of miR-34 (Fig 4G). We conclude that besides dlg1, Eip75B is another miR-34 target gene relevant to innate immunity signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Eip75B is another miR-34 target gene that modulates innate immunity signaling.

(A-B) S2 cells over-expressing miR-34 or control cells were either left untreated or treated with 20-HE (labeled on top). Cell lysates were subject to immunoblot using various antibodies against Eip74EF, Eip75B, BrC or the control Tubulin (A). Levels of the indicated proteins in 20-HE-treated cell samples were quantified in B (n≥3). (C) Reporter constructs were generated that carry either a wildtype (WT) or mutant (mut) miR-34 binding site derived from the Eip75B ORF. Seed region of miR-34 was highlighted in green. (D) The reporter constructs were transfected into S2 cells together with or without a miR-34 expression construct, and reporter activities were measured (n = 3). (E-F) S2 cells treated with dsRNA against Eip75B or a control dsRNA were first treated with 20-HE, and subsequently were either left untreated or treated with PGN. Total RNA was isolated and levels of the Diptericin (E) and Eip75B mRNA (F) were measured and normalized to RpL32 (n≥3). (G) Various combinations of miR-34 and Eip75B expression constructs were transfected into S2 cells. Cells were treated with 20-HE for 24 hrs, and total RNA was isolated and levels of the Diptericin mRNA were measured and normalized to the RpL32 control mRNA (n = 3).

https://doi.org/10.1371/journal.ppat.1006034.g004

miR-34 is transcriptionally repressed by ecdysone signaling

It has been reported that 20-hydroxyecdysone (20-HE) treatment in S2 cells can profoundly alter the expression levels of a number of miRNAs. For example, expression of let-7, miR-100 and miR-125, which cluster at the same genomic locus, is markedly induced upon 20-HE treatment [63]. In contrast, levels of miR-34 display a significant decrease in response to 20-HE treatment. We confirmed the inhibitory effect of 20-HE on miR-34 expression by performing Northern blot to quantify levels of mature miR-34 in S2 cells before and after 20-HE treatment (Fig 5A and 5B). Such ecdysone-mediated modulation of miR-34 gene expression is most likely at the level of transcription, as changes in levels of the primary miRNA transcripts resemble that of mature miRNAs (Fig 5A–5C) [63]. Further supporting this notion, a miR-34 transgene driven by the metallothionein promoter was robustly expressed in the presence of 20-HE, while its endogenous counterpart under the control of cognate regulatory elements was strongly repressed under the same conditions (Fig 1F).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Identification of trans-acting transcription factors required for ecdysone-mediated repression of miR-34.

(A-F) S2 cells transfected with various dsRNAs were left untreated or treated with ecdysone (20-HE) at 1 μM for 48 hrs. Total RNA was isolated and levels of mature miR-34 were measured by Northern blot (A) and normalized to the 2S rRNA (B; n = 3). In addition, levels of the primary miR-34 transcript (C), or the mRNAs for BrC (D) or EcR (E) were measured and normalized to the control RpL32 mRNA (n = 3; mean + SD). In addition, levels of the BrC protein were measured by immunoblot (F). Note that multiple isoforms of the BrC protein are expressed in S2 cells upon ecdysone treatment and are responsive to dsRNA-mediated knockdown. (G) RNA was extracted from third instar BrC^npr6 mutant or wildtype (WT) larvae and levels of pri-miR-34 were measured and normalized to the control RpL32 mRNA (n = 4). (H) BrC^npr6 mutant and wildtype larvae was either left untreated or infected by a concentrated culture of E. coli via septic injury. Total RNA was isolated 6 hrs post-infection and levels of Dpt mRNA were measured and normalized to the control RpL32 mRNA (n = 4). (I) S2 cells transfected with dsRNAs targeting various transcription factors (below). These cells were left untreated or treated with ecdysone, and levels of pri-miR-34 were measured and normalized to the control RpL32 mRNA (n≥3; mean + SD).

https://doi.org/10.1371/journal.ppat.1006034.g005

Ecdysone receptor (EcR) and the Broad complex (BrC) are essential for ecdysone signaling. EcR heterodimerizes with its co-factor Ultraspiracle, and functions as a stereotypical steroid hormone receptor to modulate the expression of ecdysone target genes, whereas the transcription factor BrC is an ecdysone-induced early gene product that regulates the expression of downstream ecdysone-responsive genes [29, 73, 74]. Consistent with previous reports showing that let-7 expression is activated by ecdysone signaling, depletion of ecdysone EcR or BrC in S2 impaired the increase in levels of the pri-let-7 transcript after 20-HE treatment (Fig 5D and 5E & S8 Fig) [59, 63]. In addition, while BrC has been implicated in ecdysone-mediated repression of miR-34, the requirement for BrC in this process could not be definitively assessed [63]. We therefore measured changes in levels of both the primary miR-34 transcript and mature miR-34 elicited by BrC or EcR knockdown. This analysis revealed that the repressive effect of ecdysone on miR-34 expression was partially relieved in EcR or BrC knockdown cells compared with control RNAi cells (Fig 5A–5F). Furthermore, consistent with the findings of Silverman, Ambros and colleagues [33, 63], we detected higher levels of pri-miR-34 in BrC mutant larvae than in control animals, and found that the BrC mutant larvae expressed lower levels of Dpt in response to E. coli infection (Fig 5G and 5H). Taken together, these data demonstrate that both EcR and BrC are required for optimal ecdysone-mediated transcriptional repression of miR-34.

We noticed that while BrC dsRNA treatment led to a significant reduction in both mRNA and protein levels of BrC, 20-HE-mediated repression of miR-34 was only partially relieved (Fig 5A–5F). These observations suggest that there are yet-to-be-identified factors that are required for 20-HE-mediated miR-34 repression. A recent study reported that the consensus binding motif for a collection of transcriptional factors, including Pnr, Aef1, Trl, CrebA, Srp, Twi and Ap, are enriched in candidate regulatory regions that confer ecdysone-mediated transcriptional repression [75]. We therefore examined whether these transcription factors contribute to ecdysone-mediated repression of miR-34. In addition, we also included in our test additional components of the ecdysone signaling pathway, such as Hr46, E78C and E93F. Our analysis revealed that three additional transcription factors (Srp, Twi and Ap) are required for optimal repression of miR-34 by ecdysone (Fig 5I, S9 Fig).

Identification of cis-regulatory elements responsible for ecdysone-mediated repression of miR-34

The miR-34 genomic locus contains two additional miRNA genes, miR-277 and miR-317, as well as a protein-coding gene Fmr1 (Fig 6A). While all the three miRNA genes are repressed by 20-HE treatment, little changes in levels of the Fmr1 mRNA were detected (Fig 6B and 6C). As a control, 20-HE treatment led to a marked increase in levels of the BrC and Eip75B mRNA (Fig 6B). These observations suggest that local regulatory sequences may confer 20-HE-mediated repression of these miRNA genes. Recently the Stark group has employed STARR-seq (self-transcribing active regulatory region sequencing) to identify enhancers at a genome-scale [76]. STARR-seq exploits the ability of enhancers to function independently of their relative positions. If candidate enhancers are placed downstream of a minimal promoter, such that active enhancers transcribe themselves, each enhancer's strength will be reflected by its abundance among cellular RNAs. This approach has identified several genomic regions near the miR-34 locus (referred to as ecdysone-responsive peaks P1 through P5) as putative ecdysone-responsive elements (Fig 6A) [75]. To examine whether these DNA elements can mediate ecdysone-dependent repression, we generated reporter constructs by placing individual genomic fragments upstream of a Drosophila Synthetic Core Promoter (DSCP) that drives the firefly luciferase reporter gene. Reporter assays reveal that each of the five regions can confer ecdysone-mediated repression with varying strengths. Notably, among the five genomic fragments tested, peak P2 displays the most robust repressive effect in response to ecdysone treatment (Fig 6D). Of note, all three ecdysone-repressed miRNA genes are transcribed in the same direction, whereas the ecdysone-insensitive Fmr1 gene is transcribed in the opposite direction to the miRNA genes. Consistent with the notion that peak P2 is the major regulatory element that confers sensitivity to ecdysone signaling, it is physically located within a closer distance to the miRNA genes than to the Fmr1 gene (Fig 6A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Mapping cis-regulatory elements required for ecdysone-mediated repression of miR-34.

(A) The miR-34 locus contains five candidate regulatory regions (P1 through P5) that are repressed in response to ecdysone treatment. (B-C) S2 cells were treated with ecdysone for various times (below) and levels of various mRNAs (B) or pri-miRNAs (C) were measured and normalized to the control RpL32 mRNA (n≥3). (D) DNA fragments derived from various regulatory regions were placed upstream of a luciferase reporter gene. S2 cells transfected with these reporter constructs were left untreated or treated with ecdysone (20-HE) and reporter activities were measured. Fold change in ecdysone-mediated repression of reporter activity is shown (n≥3; mean + SD). Cells transfected with a firefly luciferase reporter gene driven by the regulatory region derived from the traffic jam gene, which is not responsive to ecdysone treatment, serve as control. (E) A schematic of the P2 region. Predicted BrC-binding sites are shown as shaded boxes, whereas open and filled boxes represent one of the three predicted Twi- and Srp-binding sites, respectively. F1-F5 represent various fragments tested for BrC occupancy in D. In addition, various truncated fragments tested in reporter assays in G are shown on the left. (F) Chromatin immunoprecipitation assay was employed in cultured S2 cells to measure BrC occupancy in various regions of P2 (schematic shown in E, n = 3; mean + SD). (G) Reporter constructs containing either full length or various truncated P2 fragments (in E) were transfected into S2 cells. Cells were left untreated or treated with ecdysone (20-HE) and reporter activities were measured. Ecdysone-mediated repression of reporter activity is shown (n≥3; mean + SD).

https://doi.org/10.1371/journal.ppat.1006034.g006

Since BrC is required for optimal ecdysone-mediated repression of miR-34, we inspected the nucleotide sequence of peak P2 and identified five putative BrC-binding sites (shown as shaded boxes in Fig 6E). To validate and assess BrC occupancy on these regions in vivo, we performed chromatin immunoprecipitation (ChIP) assay both in naïve S2 cells and cells treated with 20-HE, using an antibody against BrC. As expected, in the absence of ecdysone treatment, BrC is expressed at very low levels and thus no enrichment can be detected. However, upon 20-HE treatment, we detected various degree of BrC enrichment in these regions, with F3 displaying the most significant BrC occupancy compared with a control ChIP using a non-immune (n.i.) serum (Fig 6F).

To further define the minimal cis-regulatory element within peak P2 that are required for miR-34 repression, we placed various truncated P2 genomic fragment upstream of the reporter gene and assessed their capability of conferring ecdysone-mediated silencing. This analysis revealed that any deletions that perturb the region composed of nucleotides 573–587 compromised the repressive effect of ecdysone (Fig 6E and 6G). Notably, the genomic fragment F3, which displays the highest degree of BrC occupancy, centers around nucleotides 573–587 (Fig 6E and 6F), suggesting that BrC binding at this region is required, at least in part, for the repressive effect of ecdysone on miR-34 expression. Taken together, these analyses identified key cis-regulatory genomic elements and trans-acting transcription factors required for ecdysone-mediated repression of miR-34.

Statistical analyses

All statistical analyses in this manuscript were performed using biological replicates and the sample number (n) is shown for each dataset in the corresponding legend. Most analyses were performed using student t-test, except for survival experiments, which involved the log rank test. Unless noted otherwise, data is shown in this manuscript as mean + standard errors of the mean (SEM), * p<0.05; ** p<0.01; *** p<0.001; **** p<0.0001.

DNA constructs and antibodies

Dlg1 cDNA expression construct: a DNA fragment encoding Flag-Dlg1 was amplified by PCR and cloned into pRmHa-3 using SacI and BamHI restriction sites. Reporter construct carrying the UTR sequences from the dlg1 cDNA: a dlg1 3’ UTR fragment was amplified by PCR and cloned into pRmHa-3-Renilla [84] using BamHI and SalI sites. To generate mutations in the candidate miR-34-binding site, the afore-mentioned reporter construct was subject to site-directed mutagenesis using Phusion Site-Directed Mutagenesis Kit (Thermo Fisher) according to the manufacturer’s instructions. Reporter constructs containing candidate miR-34-binding sites identified in the Eip75B ORF: pairs of oligos containing triple copies of individual sites (wildtype or mutant) were annealed and cloned into pRmHa-3-Renilla using BamHI and SalI sites. To generate various miR-34 enhancer-firefly luciferase plasmid, Selected regions were PCR amplified from genomic DNA, cloned into pCR8-TOPO-GW (Invitrogen), and shuttled to pGL3_GW_luc+ [76] by LR clonase II recombination (Invitrogen). Expression construct for miR-34: a ~500 bp DNA fragment encoding pri-miR-34 was amplified by PCR and cloned into pRmHa-3 using EcoRI and BamHI restriction sites. All constructs were verified by Sanger sequencing. The pMT-Eip75B-PC expression construct was a kind gift from Dr. Edward Dubrovsky [85]. Antibodies employed in this study include Anti-Eip74EF and anti-Eip75B antibodies (gift from Dr. Carl Thummel), anti-Dlg1 and anti-BrC antibodies (Hybridoma Bank).

dsRNA synthesis

Templates for generating dsRNAs were either requested from the DRSC (flyRNAi.org) or amplified by PCR in house. dsRNAs were synthesized using Megascript in vitro transcription kit (Ambion) and purified using RNeasy column (Qiagen).

Drosophila genetics and infection

Fly stocks are maintained on a standard fly food. To generate flies over-expressing select miRNAs, we crossed UAS-miRNA transgenic lines with the da-Gal4; tub-Gal80^ts composite line. A control cross was set up between a shRNA transgenic line, which expresses an artificial sh-gfp RNA embedded in the miR-1 cassette. To minimize lethality at early developmental stages due to the requirement for select miRNAs in development, fly crosses were kept at permissive temperature (18°C) until adult progeny of the appropriate genotype emerge. Subsequently the progeny was shifted to restrictive temperature (29°C) for 5 days to allow for miRNA transgene expression. For in vivo knockdown experiments, a similar crossing scheme was employed except that the UAS-shRNA lines were used instead of the UAS-miRNA lines. The npr6 allele of BrC (BDSC stock #36562) was employed in experiments in Fig 5. For fly infection experiments, male progeny of the appropriate genotype were either left untreated (as control) or pricked with a sharp needle previously dipped in a concentrated suspension of either the Gram-negative bacteria E. coli or the Gram-positive bacteria M. Luteus, and flies were harvested 6 and 24 hours post-infection, respectively. Levels of mRNAs encoding the antimicrobial peptide genes Diptericin and Drosomycin, and the control RpL32 mRNA were analyzed by RT-qPCR. Additional infection experiments were carried out by injecting 9.2 nl of bacterial suspension or PBS (control) into flies using Nanoject II (Drummond). For fly survival experiment, a concentrated culture of Ecc15 (OD₆₀₀ ~5); or a concentrated culture of Enterobacter cloacae (OD₆₀₀ ~5) were injected. Fly survival was monitored daily. To determine pathogen load, a concentrated culture of Ecc15 (OD₆₀₀ ~5) or an overnight culture of Enterobacter cloacae (OD₆₀₀ ~0.5) were injected into flies. Subsequently groups of 4 flies were harvested at various time points post Enterobacter cloacae infection (groups of 1 to 3 flies for Ecc15 infection) and homogenized in 200 μl of sterile PBS. Diluted fly homogenates were plated onto LB plates containing nalidixic acid (Sigma) (for Enterobacter cloacae) or Ampicillin (for Ecc15), colony forming units (CFU) were recorded after 24 hours and CFU per fly is calculated. In experiments involving flies reared in antibiotics-containing food, flies were reared in standard fly food supplemented with antibiotics (100 μg/mL each of ampicillin, kanamycin and doxycycline).

Bacterial phagocytosis assay

The assays were performed as previously described [86]. Briefly, 138 nl of pHrodo Red E. coli BioParticles Conjugate (Life Technologies) suspension in PBS was injected into individual flies of appropriate genotype using Nanoject II. Flies were kept at 25°C for 2 hours before imaging using an Axioskop Zeiss microscope.

Cell culture, transfection, and RNAi

Drosophila S2 cells are maintained in Schneider’s medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (Invitrogen). S2 cells stably expressing miR-34 were generated by transfection with pRmHa-3-miR-34 and the selection marker plasmid pHS-neo using the calcium phosphate method, followed by selection in medium containing 400 μg/mL G418 (Calbiochem). dsRNA treatment was performed as described previously [87–89]. Briefly, ~2 × 10⁶ S2 cells were seeded in 6-well plates for 24 h and then transfected with 3 μg of the appropriate dsRNA using the calcium phosphate protocol. Two days later, the cells were harvested, replated in 6-cm plates for 24 h, and then transfected again with another 9 μg of dsRNA. Three days later, the cells were harvested and used in assays. DNA transfection was preformed in the same manner as dsRNA transfections, except that only a single round of transfection was performed.

Peptidoglycan and ecdysone treatment in S2 cells

Drosophila S2 cells were treated with 1 μM 20-hydroxyecdysone (Sigma) for 24 hours. Subsequently cells were treated with 10 μg/mL crude LPS prep (Sigma), which contains peptidoglycan (PGN) as a potent inducer of IMD signaling [6], for another 6 hours. Total RNA was extracted and levels of the antimicrobial peptide genes Diptericin, Cecropin, Attacin A, Defensin, Metchnikowin and Drosomycin, and the control RpL32 mRNA were analyzed by RT-qPCR. For miR-34 overexpression experiments, cells were first transfected with a miR-34 expression construct controlled by the metallothionein promoter. One day post-transfection cells were treated with CuSO₄ (250 μM) for 2 days prior to 20-hydroxyecdysone and PGN treatment.

Reporter assay

For luciferase reporter assays, transfections were performed in a 96-well format and the amount of cells and DNA constructs were scaled down accordingly. Briefly, ~5 × 10⁵ S2 cells were seeded in 24-well plates the day before transfection. For miR-34 target site reporter assays, 150 ng of the Renilla luciferase reporter gene containing wild type or mutant miR-34-binding site, 25 ng of a firefly luciferase reporter construct [84], and 600 ng of either the miR-34 expression construct or the empty vector were transfected into these cells. Two days later, cells were treated with 250 μM copper sulfate for 24 hours and the reporter activity was measured using the Dual-Glo system (Promega). For data processing, Renilla/firefly ratio was calculated and normalized against the control sample (wildtype reporter in the absence of miR-34).

To identify the regulatory elements required for ecdysone-mediated repression of miR-34, 100 ng of miR-34 enhancer-firefly luciferase plasmid and 100 ng of pol III-Renilla luciferase or 20 ng of actin-Renilla luciferase were transfected into cells. Two days later, cells were treated with 1 μM 20-hydroxyecdysone (Sigma) for 24 hours and luciferase activity was measured. For data processing, first, the firefly/Renilla ratio was calculated; next changes in the F/R ratio in cognate set of samples upon 20-HE treatment [F/R ratio _{post 20-HE} / F/R ratio _{no 20-HE}] was calculated and normalized to the control sample (transfected with a firefly luciferase reporter gene driven by the regulatory region derived from the traffic jam gene, which is not responsive to ecdysone treatment).

Co-immunoprecipitation

Various combinations of expression constructs for Flag-Dlg1, together with T7-IMD or T7-Kenny were transfected into S2 cells. Two days post-transfection, cells were treated with 250 μM CuSO4 to induce transgene expression and harvested another 24 hrs later. Cells were lysed in lysis buffer (20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 2 mM EDTA, 10% glycerol, 1% Triton X-100, 1 mM DTT, 1 mM orthovanadate) supplemented with protease inhibitor cocktail (Roche). Cleared total lysates were immunoprecipitated with anti-Flag antibodies (Sigma). Both input and immunoprecipitated samples were analyzed by SDS-PAGE followed by immunoblotting with antibodies against the Flag (Sigma) or T7 (Novagen) epitopes.

Chromatin immunoprecipitation (ChIP)

ChIP was performed as described previously with minor modifications [90]. Briefly, Cells were fixed by cross-linking with a final concentration of 1.42% formaldehyde for 15min at room temperature and then quenched with 125 mM glycine. After washing with cold PBS twice, the cells were lysed with IP buffer (150 mM NaCl, 5 mM EDTA, 0.5% NP-40, 1.0% Triton X-100, 50 mM Tris-HCl pH 7.5). The clarified lysate was subject to sonication. The chromatin was then sheared to fragments of 200–1000 bp and cleared of debris by centrifugation. The chromatin is used for ChIP using anti-BrC antibody (Hybridoma Bank). After washing in IP buffer, the precipitated ChIP DNA was eluted, the cross-links were reversed by incubation at 68°C for 2h (or overnight). DNA fragments then were purified from the isolated protein-DNA complex and analyzed by qPCR.

RT-qPCR

Total cellular RNA was isolated with TRIzol (Invitrogen) and reverse transcription was carried out using QuantiTect Reverse Transcription kit (Qiagen). Quantitative PCR was performed using the iQ SYBR-green reagents on a CFX96 Real-Time PCR Detection System (Bio-Rad). The efficiency of various primer pairs was determined using serial dilutions of a standard template (RT samples from naïve or 20-HE treated S2 cells). Fold changes in RNA levels were calculated using the ΔΔCt method for primer pairs with efficiencies of 90–111%. Note that fold changes in the dlg1 mRNA were determined by calculating absolute levels of dlg1 and RpL32 using the detected primer efficiency.

Northern blotting

Northern blotting was performed as previously described [91, 92]. In brief, total cellular RNA was isolated with TRIzol (Invitrogen). Samples of 15 μg RNA were separated on 15% denaturing polyacrylamide gels and transferred to Hybond-N+ membranes (Amersham Biosciences) in 1X TBE buffer. Small RNAs were UV crosslinked to the membranes, and the membranes were prehybridized in hybridization buffer for 2 h. DNA probes complementary to the appropriate strands were 5′ radiolabeled and incubated with membranes overnight at 37°C. Membranes were washed twice in 1X SSC with 0.1% SDS at 42°C, and then exposed to Phosphorimager screens for 12–48 h. Membranes were stripped by the addition of boiling 0.1% SDS solution and incubated for 30 min.

RNA-Seq data analysis

RNA-Seq datasets containing ~25 million 2 x 90 bp reads were generated from control S2 cells or cells over-expressing miR-34 using an Illumina HiSeq 2000 system. The average insert size for the RNA-seq library was ~200 bp. Pair-end reads were mapped to the D. melanogaster genome (UCSC dm3) and to the Flock House virus genome using Bowtie2 (v2.1.0) with TopHat (v2.0.8b) [93]. Differentially expressed transcripts and genes were identified using Cufflinks (v2.1.1) [94].

Oligonucleotides

See S3 Table.