A somatic piRNA pathway in the Drosophila fat body ensures metabolic homeostasis and normal lifespan

Abstract

In gonadal tissues, the Piwi-interacting (piRNA) pathway preserves genomic integrity by employing 23-29 nucleotide (nt) small RNAs complexed with argonaute proteins to suppress parasitic mobile sequences of DNA called transposable elements (TEs). Although recent evidence suggests that the piRNA pathway may be present in select somatic cells outside the gonads, the role of a non-gonadal somatic piRNA pathway is not well characterized. Here we report a functional somatic piRNA pathway in the adult Drosophila fat body including the presence of the piRNA effector protein Piwi and canonical 23-29 nt long TE-mapping piRNAs. The piwi mutants exhibit depletion of fat body piRNAs, increased TE mobilization, increased levels of DNA damage and reduced lipid stores. These mutants are starvation sensitive, immunologically compromised and short-lived, all phenotypes associated with compromised fat body function. These findings demonstrate the presence of a functional non-gonadal somatic piRNA pathway in the adult fat body that affects normal metabolism and overall organismal health.

ARTICLE

Received 18 Jul 2016 | Accepted 4 Nov 2016 | Published 21 Dec 2016

DOI: 10.1038/ncomms13856 OPEN

A somatic piRNA pathway in the Drosophila fat body ensures metabolic homeostasis and normal lifespan

Brian C. Jones1, Jason G. Wood1, Chengyi Chang1, Austin D. Tam1, Michael J. Franklin1, Emily R. Siegel1

& Stephen L. Helfand1

In gonadal tissues, the Piwi-interacting (piRNA) pathway preserves genomic integrity by employing 2329 nucleotide (nt) small RNAs complexed with argonaute proteins to suppress parasitic mobile sequences of DNA called transposable elements (TEs). Although recent evidence suggests that the piRNA pathway may be present in select somatic cells outside the gonads, the role of a non-gonadal somatic piRNA pathway is not well characterized. Here we report a functional somatic piRNA pathway in the adult Drosophila fat body including the presence of the piRNA effector protein Piwi and canonical 2329 nt long TE-mapping piRNAs. The piwi mutants exhibit depletion of fat body piRNAs, increased TE mobilization, increased levels of DNA damage and reduced lipid stores. These mutants are starvation sensitive, immunologically compromised and short-lived, all phenotypes associated with compromised fat body function. These ndings demonstrate the presence of a functional non-gonadal somatic piRNA pathway in the adult fat body that affects normal metabolism and overall organismal health.

1 Department of Molecular Biology, Cell Biology, and Biochemistry, Brown University, Providence, Rhode Island 02912, USA. Correspondence and requests for materials should be addressed to S.L.H. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856

Transposable elements (TEs) parasitize the DNA of their hosts and account for a large portion of eukaryotic genomes1,2. To combat the invasion and expansion of

TEs, small RNA (smRNA) silencing pathways have evolved to suppress TEs across species from plants to humans3. The short interfering RNA pathway suppresses TEs in all tissues of plants and animals, whereas the activity of the Piwi-interacting RNA (piRNA) pathway is thought to be primarily restricted to the gonads of metazoans4,5. Loss or decline of these pathways results in genomic instability and cellular dysfunction caused by TE reactivation and transposition69.

The piRNA pathway is best known for its role in gonadal tissues where it protects against genomic damage caused by TE reactivation4,5. The pathway silences TEs by employing complementary small RNAs called piRNAs, generated from large TE-rich genomic regions called piRNA clusters. In ies, these clusters transcribe long single-stranded RNA precursors that are then further processed into smaller 2329 nucleotide (nt) piRNAs. These piRNAs partner with argonaute effector proteins (Piwi, Aubergine or AGO3) that are then able to silence TEs via their homology to TE transcripts4,10. This process is accomplished by one of two silencing mechanisms. In the primary piRNA pathway, active in both the germline and ovarian somatic follicle cells, Piwi represses TE transcription by establishing heterochromatin5,11. In the secondary piRNA pathway, active only in the germline, Aubergine and AGO3 silence TEs post transcriptionally in the cytoplasm via messenger RNA cleavage4,5,10. Although the role of the piRNA pathway was previously thought to be restricted to the gonads, recent evidence in a diversity of organisms suggests that this pathway may also be present in somatic cells outside of the gonad12.

Over the past decade, new evidence has begun to reveal non-gonadal examples of the piRNA pathway including a role for the piRNA pathway in stem cell function12. In planaria, piRNA pathway proteins are essential in maintaining stem cell pluripotency as well as the regenerative capacity of these animals13. piRNA pathway components are active in multiple types of cancer12, including specic cancers in mammals and ies12,1418, and in ies piwi has been shown to contribute to malignant tumour growth19. Less information is available for a role of the piRNA pathway in normal differentiated somatic tissues, although evidence for the activity of the secondary piRNA pathway in specic neurons of the adult y brain has been reported20. As more non-gonadal examples of an active RNA interference system are discovered, it appears that the piRNA pathway may have other important roles beyond its known functions in gonadal tissue.

Here we show the presence of a functional somatic piRNA pathway in the adult y fat body. The piRNA pathway in the fat body exhibits all the canonical characteristics of a primary piRNA pathway and actively suppresses TE mobilization in this tissue. We observe that loss of this pathway correlates with compromised fat body function and shortened lifespan. These ndings demonstrate a novel role for the piRNA pathway outside of the gonads in a fully differentiated somatic tissue.

ResultsThe fat body exhibits components of an intact piRNA pathway. Given the recent evidence that the piRNA pathway may also be active in select non-gonadal somatic cells12, we examined the expression of piRNA pathway genes outside of the gonads. We found that examination of RNA sequencing (RNA-seq) libraries from adult y eviscerated abdomen, but not heads or thorax, had signicant enrichment of piRNA pathway genes relative to other somatic body segments (Fig. 1a, Supplementary Fig. 1, and

Supplementary Dataset 1). Immunouorescent microscopy demonstrated specic localization of the Piwi protein to the nuclei of adult abdominal and pericerebral fat body cells, but not to nuclei of cells from other tissues outside of the gonads or in fat bodies of homozygous piwi null mutants (Fig. 1b). Immunoblotting of isolated puried fat body, thorax and head also demonstrated the presence of Piwi protein in the fat body (Fig. 1c and Supplementary Fig. 2). The presence of Piwi protein in the fat body suggests the possibility of an intact piRNA pathway in this tissue.

Activated piRNAs in functional piRNA silencing complexes are 20-O-methylated at their 30 termini and can be selectively enriched and detected using periodate oxidation4,21. We found a broad piRNA-like peak of smRNAs ranging from 23 to 29 nt in oxidized small RNA-seq (smRNA-seq) libraries from pure adult abdominal fat body (Fig. 1d), suggesting the presence of piRNAs with 20-O-methylation at their 30 termini as is typical of gonadal piRNAs actively loaded into a piRNA-argonaute complex22. Of the oxidized fat body piRNAs, 49% (2329 nt smRNAs) mapped to TEs (Fig. 1e) and these reads exhibited a strong antisense bias (Supplementary Fig. 3a) and a canonical rst position nucleotide bias for uracil10 (Fig. 1f), indicating that they likely have the capacity to target TE transcripts for silencing. Oxidized smRNA libraries from puried abdominal fat bodies of two distantly related drosophilids, Drosophila simulans and Drosophila yakuba (Supplementary Fig. 3b), also showed 2329 nt piRNAs mapping to TEs (Supplementary Fig. 3cf) with antisense TE-mapping piRNAs exhibiting a strong rst position uracil bias (Supplementary Fig. 3g,h), demonstrating that adult fat body piRNAs are conserved across diverse drosophilid species. Together, these data suggest the presence of fat body piRNAs that exhibit canonical piRNA characteristics, are associated with an active piRNA-argonaute silencing complex and are evolutionarily conserved.

The fat body piRNA pathway suppresses TEs. We next examined whether the fat body piRNA pathway exhibits canonical hallmarks of an active piRNA pathway. Loss of Piwi in gonadal tissues results in a dramatic reduction of piRNAs and a derepression of their corresponding TEs11,23. In the fat bodies of piwi null mutants, we observed a signicant increase in the transcript levels of multiple TEs and a corresponding decrease of their associated piRNAs (Fig. 2a and Supplementary Fig. 4). Total piRNAs also decreased (28.1% decrease) as well as TE-mapping piRNAs (70.5% decrease) (Fig. 2b, side panel). We used a reporter of transposition for the endogenous gypsy retrotransposon (gypsy-TRAP)9,24, a known target of Piwi in the ovary25, to detect transposition in fat body cells and found that piwi mutant fat bodies displayed signicantly higher levels of gypsy transposition relative to controls (Fig. 2c). Together, these data show that a somatic piRNA pathway actively suppresses the expression and mobilization of TEs in adult fat body cells.

The piRNAs that target and suppress TEs in ovarian tissues are known to originate from genomic regions called piRNA clusters10. Mapping fat body piRNAs to previously annotated y piRNA clusters showed that many map to the amenco cluster, which is known to be specic to somatic follicle cells (Fig. 2d,e)10,11. In agreement with studies of ovarian follicle cells10,11, piRNAs mapping to amenco were depleted in the fat bodies of piwi mutants (Fig. 2d,e). These data further support the presence of a functional fat body piRNA pathway where piRNAs produced from somatic piRNA clusters pair with Piwi to suppress TEs as the primary piRNA pathway does in ovarian follicle cells10,11.

Previous studies in ies have shown that primary piRNAs are also derived from the 30 untranslated regions (UTRs) of coding genes26,27. We observed that 17% of fat body piRNAs from the

2 NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856 ARTICLE

a

89.4

45.7 16.7 17.5 35.4 63.1 46.6 30.4 71.9 52.4

Head

Thorax

Eviscerated abdomen Ovary

10

8

RPKM

6

4

2

0

Squash

1 piRNA pathway 2 piRNA pathway

1/2

Head Thorax Fat body Ovary

Piwi

Actin

Piwi Armitage Zucchini Vreteno Traffic jam Hen1 AGO3 Aubergine SpindleE Vasa

b

c

1 piRNA

pathway

2 piRNA

pathway

DAPI Piwi Merge

100

100

100

50

AGO3

Aub

d e f

70

6%

60

Total small RNAs

Counts (thousands)

50

2329 nt smRNAs

1.0

40

TE-mapping antisense smRNAs (2329 nt)

30

TEExon Intergenic snoRNA/tRNA miRNA

Bits

0.5

20

10

0.0

5 10 15 20

0

19 20 21 22 23 24 25 26 27 28 29 30

Position from 5 end

Small RNA length (nt)

Figure 1 | Canonical signatures of a piRNA pathway in the adult y fat body. (a) Expression of primary and secondary piRNA pathway genes generated from total RNA-seq libraries of head, thorax, eviscerated abdomen and ovary. piRNA pathway genes are more highly expressed in the eviscerated abdomen than in the head or thorax. Data values for ovary libraries that exceed the range of the plot are shown above each relevant bar. RPKM, reads per kilobase per million. Error bars represent s.e.m.; n 3 replicate libraries. In comparing the eviscerated abdomen with head and thorax controls, 10 of 11 genes

(excluding tj) are statistically signicant (Po0.0001). See Supplementary Dataset 1 for statistics. (b) Piwi protein localizes to the nuclei of abdominal fat body cells. DAPI labels fat body nuclei. Staining in the membrane is autouorescence typical of fat body cells. Scale bars represent 20 mm. (c) Piwi protein is present in the fat body. All piRNA argonautes are present in the ovary samples. Actin serves as a loading control. (d) Fat body smRNA size prole from oxidized smRNA-seq libraries. Oxidation allows for enrichment of 20-O-methylated smRNAs. Peak at 21 nt likely represents short interfering RNA (siRNA) population. Broader peak from 23 to 29 nt represents putative fat body piRNAs. Reads aligning to rRNA and miRNA were excluded from analysis.

(e) Fat body piRNAs (2329 nt) aligned to the y genome map primarily to TEs. (f) Sequence composition of TE-mapping fat body piRNAs (2329 nt) displays a rst position nucleotide bias for uracil.

oxidized fat body library mapped to coding genes (Fig. 1e) and that 30UTR sense-mapping piRNAs were depleted in piwi mutants (17.5% decrease; Fig. 2b, side panel). These 30UTR-derived piRNAs are between 78% (piwi / ) and 87% (wt)

sense-mapping, as is expected of genic piRNAs (Supplementary Fig. 5a). Fat body piRNAs mapping to the 30UTR of trafc jam (tj), a known source of genic piRNAs in ovarian follicle cells26,27, were depleted in piwi mutants (Fig. 2f,g and Supplementary Dataset 2) along with other 30UTR-derived piRNAs (Supplementary Fig. 5be and Supplementary Dataset 2). Together, these data provide evidence of 30UTR-derived fat body genic piRNAs, another hallmark of the primary piRNA pathway.

In order to rule out the possibility that fat body piRNAs result from contamination by ovarian tissues during dissection and library preparation, we performed smRNA-seq on oxidized fat body libraries isolated from ovoD1 ies. Because of a dominant female-sterile mutation in the ovo gene, these ies exhibit severely degenerated ovaries, thus signicantly decreasing the likelihood that any piRNAs isolated from the fat body of these animals would be due to contamination from ovarian tissues. We observed in both wild-type and ovoD1 fat bodies 2329 nt

smRNAs that mapped uniquely to the amenco locus (Supplementary Fig. 6a,b), suggesting that piRNAs observed in oxidized fat body libraries are in fact originating from this tissue and not a result of ovarian contamination. In addition, immunoblotting demonstrates the presence of Piwi protein in the eviscerated abdomen and isolated fat body of ovoD1 ies (Supplementary Fig. 6c). These data, combined with our observation of the Piwi protein in the nuclei of fat body cells (Fig. 1b and Supplementary Fig. 2), strongly support the presence of a functional canonical piRNA pathway in the fat body.

DNA damage and metabolic dysregulation in piRNA mutants. The replicative mobility of TEs can contribute to mutagenesis via their ability to insert into new genomic loci6. TE reactivation and transposition has been shown to correlate with chromosomal rearrangements, double-strand DNA breaks and apoptosis7,8. Phosphorylation of the histone variant H2A.v (g-H2A.v) during DNA repair serves as a reliable marker of double-strand DNA breaks and has been shown to correlate with increased TE activity in the fat body9,28.

NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856

a b

c

RNA-seq

piRNA

piwi /+ piwi /

1,200

600

1

Total small RNAs

TE-mapping smRNAs

3UTR-mapping smRNAs

1,000

100%

800

400

15

600

* * * * * * *

* * * *

R2_DM

200

400

Normalized read count (thousands)

R1_DM

200

80%

0

0

52

HETA

19 20 21 22 23 24 25 26 27 28 29 30

19 20 21 22 23 24 25 26 27 28 29 30

19 20 21 22 23 24 25 26 27 28 29 30

% of flies per category

350

0

Gypsy6_LTR

200

0

300

High

Medium

Low

MAX_LTR

150

250

60%

Gypsy1I_DM

200

100

150

MAX_I

50

100

Bica_LTR

50

0

40%

84

16

TAHRE

Copia_I

120

80

ROOA_I

100

80

60

20%

I_DM

60

40

32

/ / piwi /+ )

Copia1LTR_DM

40

DIVER_I

20

20

Copia2_I

0%

2329 nt smRNAs

High

Medium

Low

Copia1I_DM

Small RNA length (nt)

0

piwi:

/+

/

Log 2fold change (piwi

MDG1_LTR

d

Fold ratio of uniquely mapping piRNAs piwi /+ /piwi /

e

1,500

wt

Copia_LTR

1,000

ROOA_LTR

500

0 10 20 30 40 50 60 70 80 42AB

Cluster_02 Cluster_05 Cluster_06 flamenco

Cluster_08 Cluster_10 Cluster_11 Cluster_12 Cluster_13

BATUMI_LTR

0

1,000

Counts (RPM)

piwi /+

piwi /

BARI_DM

Gypsy3_I

500

500

DM412

0

1,000

DOC

Chouto_I

Gypsy7_LTR

0

ZAM_LTR

21.54 21.56 21.58 21.60 21.62 21.64 21.66

chrX Mb

flamenco cluster

FW2_DM

f

800

g

200

0

wt

FROGGER_I

100

Bica_I

600

Counts (RPM)

0

G2_DM

wt

BEL_I

400

NINJA_I

piwi /+

piwi /

Counts (RPM)

200 piwi /+

200 piwi /

100

BLOOD_I

200

100

0

19.4650 19.4655 19.4660 19.4665 19.4670

chr2L Mb

0 5UTR CDS 3UTR

4

2

0 2 4

tj gene body

tj

Figure 2 | The piRNA pathway is active in the fat body. (a) Heat map of log2 fold change of TE transcript levels (total RNA-seq) 41.2 fold change and corresponding piRNAs (smRNA-seq) in piwi mutant (piwi / ) compared with heterozygous control (piwi / ) fat bodies; n 3 replicate libraries. *False

discovery rate (FDR) o0.05. (b) Fat body smRNA size prole from oxidized smRNA-seq libraries of piwi mutant fat bodies (red) and heterozygous controls (black). Shown are total smRNAs (top), TE-mapping smRNAs (middle) and 30UTR-mapping smRNAs (bottom). Panels at right show levels for 2329 nt smRNAs for each genotype. (c) A fat body-specic transposition reporter line, gypsy-TRAP/r4-GAL4::UAS-GFP, in a piwi mutant background (see Methods).

GFP-positive cells are cells in which a transposition event has activated reporter function. The 10-day-old piwi mutants show elevated levels of GFP-positive cells compared with heterozygous controls. Images are of representative ies exhibiting high, medium and low levels of GFP-positive cells (left panels). Scale bars represent 150 mm. Panel on right shows the distribution of ies for each group by genotype; numbers within each bar are corresponding percentages for each group. Fishers exact test compared the combined High Medium groups and the Low group of each genotype. piwi / : n 140

ies; piwi / : n 142 ies; Po0.0001. (d) Fold ratio of uniquely mapping cluster piRNAs in piwi mutants compared with heterozygous controls. The

amenco cluster shows the greatest response to loss of piwi. (e) Unique piRNA reads (2329 nt) map to the amenco locus in wild-type (wt) and piwi heterozygotes and are lost in piwi mutants. (f,g) Unique piRNA reads (2329 nt) map to the tj gene body in wt and piwi heterozygotes and are lost in piwi mutants. Thick lines in gene model (g) represent coding sequence, and thin lines represent 50 and 30UTRs. See Supplementary Dataset 2 for raw data.

Using immunouorescent microscopy, we observed an increase in the intensity of g-H2A.v staining in piwi mutants relative to controls (Fig. 3a,b). These data suggest that the fat body piRNA pathway normally protects fat body cells from the accumulation of DNA damage that may be caused by TE reactivation.

The y fat body is a functional analogue of the mammalian liver and adipose tissue, with one of its primary roles being storage of lipids and glycogen29. Our observation that piwi

mutant fat body cells showed increased TE mobilization and elevated levels of DNA damage (Figs 2c and 3a,b) led us to hypothesize that this could result in disrupted fat body function. Nile Red staining of fat body lipid droplets revealed a reduction of lipid droplet size in piwi mutants compared with controls, with larger lipid droplets (4200 mm2) greatly reduced in their abundance (Fig. 3ce and Supplementary Dataset 3). These data correlate with a signicant reduction of two of the major

4 NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856 ARTICLE

a

b

c

d e

50100 m2

150200 m2

-H2A.v/DAPI

3,500

-H2A.v fluorescence intensity (a.u.)

*

Nile Red/DAPI

Nile Red/DAPI

100150 m2 >200 m2

Lipid droplets >200 m2

1.0

3,000

700

piwi/+

piwi/

Fraction of total lipid droplets

piwi /+

piwi /

0.8

Lipid droplet area (m2 )

2,500

600

2,000

0.6

500

-H2A.v/DAPI

1,500

0.4

400

1,000

0.2

300

500

200

0.0

0 /+

piwi: /

/+

piwi: /

/+

piwi: /

f h i

g

120

120

120

120

100

100

100

100

Normalized (TAG) %

60

Normalized (TAG) %

80

60

Normalized (glycogen) %

Normalized (glycogen) %

80

80

80

**

60

60

**

**

40

40

40

40

20

20

20

20

*

0

0

0

0

/+ /

piwi: /+ /

piwi:

/+ /

flamenco: flamenco: /+ /

Figure 3 | Loss of the piRNA pathway disrupts normal metabolic functions of the fat body. (a) Representative images of g-H2A.v staining in piwi mutants and heterozygous controls. piwi mutants exhibit higher levels of g-H2A.v staining compared with heterozygous controls. Scale bars represent 20 mm. (b) Quantication of g-H2A.v staining in (a). piwi / and piwi / : error bars are s.e.m. Students two-tailed t-test compared with heterozygous control; n 104 nuclei per genotype. *Po0.0001. (c) Representative images of Nile red staining of fat body lipid droplets in piwi mutants and heterozygous

controls. piwi mutants exhibit smaller lipid droplets relative to heterozygous controls. Scale bars represent 20 mm. (d) Quantication of lipid droplet staining in (c). piwi / : n 6 ies, piwi / : n 5 ies. See Supplementary Dataset 3 for raw data. (e) Box plot showing distribution of lipid droplets 4200 mm2

from (d) comparing piwi mutants to heterozygous controls. (fi) Measurements of whole-body adult y TAGs (f,g) and glycogen (h,i) of piwi (f,h) or amenco (g,i) mutants compared with heterozygous controls. Data were normalized to total protein concentration of each sample and represented as a percent of the heterozygous control. Error bars are s.e.m. For each assay, n 5 biological replicates per genotype. Students two-tailed t-test compared with

heterozygous control. *Po0.01; **Po0.001. All assays were performed using 10-day-old ies.

storage metabolites in fat body, triacylglycerides (TAGs) and glycogen, in both piwi and amenco mutants (Fig. 3fi and Supplementary Fig. 7a,b). We next asked whether these phenotypes correlated with an altered fat body transcriptome as transcription of TEs alone can cause RNA toxicity and contribute to cellular dysfunction8. We generated RNA-seq libraries from piwi mutant and control fat bodies and found that many differentially expressed genes in metabolism-associated pathways were signicantly changed (Supplementary Fig. 7c). These data support a model in which a loss of the piRNA pathway function results in a decrease in lipids and stored metabolites, disruption of metabolic homeostasis and a decline in cellular function, possibly because of reactivation of TEs.

piRNA mutants show altered fat body function and lifespan. The y fat body plays a major role in resisting stressors such as fasting and, through its role in innate immunity, in resisting pathogenic infections29. We observed that piwi mutants were highly sensitive to starvation conditions as well as infection by a pathogenic insect bacterium (Fig. 4a,b and Supplementary

Table 1). Because of the central role that the fat body plays in regulating longevity30,31, we next examined the effect of disrupting the piRNA pathway on lifespan. We found that mutations in either piwi or the amenco locus dramatically shorten lifespan (Fig. 4c,d and Supplementary Table 1). Finally, we tested whether the shortened lifespan of piRNA pathway mutants was dependent upon TE activity. Many TEs in Drosophila are retrotransposons, including gypsy, and depend upon reverse transcriptase for their replicative mobility32. Administration of a known reverse transcriptase inhibitor, 3TC, inhibits the normal age-related increase in gypsy mobilization and extends the shortened lifespan of Dcr-2 mutants, another condition in which derepression of TEs occurs9. We administered 3TC to amenco mutants and observed a signicant lifespan extension (Fig. 4e and Supplementary Table 1), suggesting that a shortened lifespan phenotype is at least partially dependent upon TE mobilization. These results suggest that loss of the fat body piRNA pathway and an increase in TE activity and mobilization correlates with compromised fat body function including its ability to otherwise mitigate the detrimental effects of environmental stressors.

NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856

a b

100

piwi /+

100

80

piwi /

/+

/

80

Percent surviving

Percent surviving Percent surviving

60

60

40

40

piwi /+ (+)

20

20

piwi / ()

piwi / (+)

0

0 24 48 72 96 120 144 168Age (hours)

0

0 2 4 6 8 10

Age (days)

c

d

100

100

flam /+

flam /

80

80

Percent surviving

60

60

40

40

20

20

0

0 0 20 40 60 80Age (days)

0 20 40 60 80 Age (days)

e

100

10 mM 3TC

80

Percent surviving

60

40

20

0 0 20 40 60 80

Age (days)

Figure 4 | piRNA pathway mutants are stress sensitive and short-lived. (a) Survivorship curves for starvation of piwi mutants and heterozygous controls. piwi mutants are more sensitive to starvation than heterozygous controls. Log rank test compared with heterozygous controls; nE50; Po0.0005.

(b) Survivorship curves for immune challenge of piwi mutants and heterozygous controls. piwi mutants are more sensitive to infection than heterozygous controls. Flies were either infected with a mock EtOH control ( ) or a culture of E. carotovora ( ). Log rank test compared with heterozygous or mock

EtOH control ( ); nE50; Po0.0005. (c) Survivorship curves of piwi mutants and heterozygous control. piwi mutants are shorter lived compared with

heterozygous controls. Wilcoxon rank sum test compared with heterozygous control; nE250; Po0.0005. (d) Survivorship curves of am mutants and heterozygous control. am mutants are shorter lived compared with heterozygous controls. Wilcoxon rank sum test compared with heterozygous control;

nE250; Po0.0005. (e) Survivorship curves of am mutants fed 10 mM 3TC. am mutant ies fed 3TC live longer than untreated controls. Wilcoxon rank sum test compared with control; nE250; Po0.0005. See Supplementary Table 1 for assay parameters and statistics. All assays were repeated twice with similar results.

DiscussionHere we have shown evidence for a fully functional piRNA pathway in a non-gonadal somatic tissue, the adult y fat body, that is likely to be necessary for proper tissue function and overall organismal health. These results demonstrate that the adult fat body piRNA pathway exhibits canonical characteristics found in gonadal somatic cells, and its activity likely positively affects the function of a tissue important to metabolic homeostasis and physiological health. Although we are not able to entirely rule out a contribution of the gonadal piRNA pathway to fat body function, many of the phenotypes we observe are opposite to those typically seen in animals with compromised gonadal tissue function and therefore likely represent the effect of a loss of the fat body piRNA pathway. For example, the shortened lifespan and reduced lipid stores in piRNA pathway mutants

demonstrates that the piRNA pathway is essential in the health and functioning of non-gonadal somatic tissues, as reduction or ablation of gonadal function in ies often extends lifespan and increases lipid stores rather than decreasing lifespan and fat storage31. Recent studies in wild-type ies have also demonstrated an important link between TE activity and longevity9, and our studies demonstrating partial rescue of the shortened lifespan in amenco mutants upon administration of a reverse transcription inhibitor further support this association.

Interest in a function for the piRNA pathway in the soma has increased recently as new roles for this pathway are being illuminated. The piRNA pathways association with tissues that maintain a degree of immortalization similar to that exhibited in the germline is of particular interest12. For example, the somatic stem cell niches of Hydra maintain an active piRNA pathway that

6 NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856 ARTICLE

represses TEs, possibly contributing to this organisms remarkably long lifespan33. These studies, together with our ndings, suggest that the presence of a piRNA pathway in normal somatic tissues may offer an additional cellular defence against TE reactivation and possible somatic genomic damage. Our nding of a role for the piRNA pathway in preserving metabolic homeostasis and the overall health of the y suggests the potential importance of the piRNA pathway in other somatic tissues. Finally, interventions specically augmenting the piRNA pathway may provide signicant benets to maintaining genomic integrity, tissue function and healthy lifespan.

Methods

Fly stocks and husbandry. Drosophila stocks were all maintained on standard media (30.5 g l 1 autolysed yeast, 121.8 g l 1 sucrose, 52.3 g l 1 cornmeal,8.75 g l 1 agar and 2.62 g l 1 tegosept, all weight by volume) at 25 C with a 12-h light/dark cycle at 60% relative humidity. Unless otherwise noted, ies used in all experiments were collected over a 48-h period, placed in density-controlled, mixed-sex vials and aged for 10 days on food containing 150 g l 1 autolysed yeast,150 g l 1 sucrose and 20 g l 1 agar, all weight by volume. Unless otherwise noted, all experiments were performed using mated female ies grown under these conditions.

Lab stocks of Canton S or w1118 were used for wild type experiments. amKG/FM4 (Bloomington 16453) was obtained from the Bloomington Drosophila Stock Center at Indiana University. Stocks of D. yakuba and D. simulans were provided by the Drosophila Stock Center at the University of California, San Diego. The piwi mutant fat body gypsy-TRAP reporter line was generated from the gypsy-TRAP line provided to us by Joshua Dubnau24, r4-GAL4 (Bloomington 33832), UAS-GFP (Bloomington 1522) and the piwi2/CyO line from Haifan Lin. ovoD1 mutants lacking ovaries were generated by crossing males of ovoD1 (ovoD1v24/C(1)DX, y1w1f1) (Bloomington 1309) to virgin Canton S females.

RNA and smRNA-seq library preparations. Flies were ash frozen on dry ice and head and thorax body segments were dissected on a 20 C cold block and stored

at 80 C. Eviscerated abdomens and pure adult fat bodies were dissected from

the abdominal wall in cold phosphate-buffered saline (PBS) and also stored at

80 C.

Total RNA was extracted from relevant y tissues using the mirVana miRNA Isolation Kit (ThermoFisher Scientic AM1560). For RNA-seq library prep, 100 ng of total RNA was used as input for the Ovation Universal RNA-seq System kit (Nugen 0343), with Drosophila ribosomal RNA (rRNA) depletion module, according to the manufacturers instructions. Three biologically independent libraries were made for each tissue. For smRNA-seq libraries, 2 mg of total RNA was oxidized in 25 mM sodium periodate with 30 mM borax and 30 mM boric acid (pH 8.6) for 30 min at room temperature (Phillip Zamore Lab Illumina TruSeq Small RNA Cloning Protocol April, 2014; http://www.umassmed.edu/zamore/resources/protocols/

Web End =http://www.umassmed.edu/zamore/ http://www.umassmed.edu/zamore/resources/protocols/

Web End =resources/protocols/ ). RNA was then recovered with the RNA Clean & Concentrator-5 kit (Zymo Research R1015) and used as input to the NEBNext Multiplex Small RNA Library Prep Set for Illumina (New England Biolabs E7300), with modications adapted from34 (2S rRNA block oligo added before 50 ligation step to decrease rRNA reads in nal library).

Tissue RNA-seq. Reads were mapped to the dm3 genome using Tophat. The Bioconductor R package easyRNASeq was used to count reads (using geneModels summarization parameter) and calculate normalized reads per kilobase per million (RPKM) values. Error bars presented represent s.e.m. with three independent biological replicates for each condition. The false discovery rate values in Supplementary Dataset 1 were calculated using the Bioconductor edgeR package, with a GLM correcting for batch effects as described in the vignette.

smRNA size proles and nucleotide bias. Oxidized small RNA libraries were rst trimmed of adapter sequences using cutadapt and then aligned against Drosophila rRNA sequences using Bowtie to remove rRNA aligning reads. For pie charts, reads were rst aligned to the dm3 genome, keeping only reads that aligned. Reads were next consecutively mapped to the following genomic compartments using Bowtie ( v 1): sno tRNA, microRNA (miRNA), TEs, exons and inter-

genic regions (all sequences downloaded from FlyBase, except TEs that came from Repbase). Reads aligning to each compartment were counted to create a chart.

For total small RNA size proles (see, for example, Fig. 1d), miRNAs were rst removed using Bowtie ( v 1), and remaining reads were subsequently fed into a

Perl script that counted the number of reads for each different length between 18 and 50 nucleotides. For size proles of TE-mapping reads (Supplementary Fig. 2), reads were aligned to the set of Repbase Drosophila consensus sequences using Bowtie ( v 1 k 1--best), and sorted into sense/antisense (with respect to TE)

using samtools. Alignments were then converted back into FASTQ les using bedtools, and size proles calculated as described above. For D. simulans and

D. yakuba pie charts and size proles, analysis was performed as described, using relevant genomic compartments for each species downloaded from Flybase. For transposable elements, the set of all Drosophila species transposons from Repbase was used.

For nucleotide bias calculations, 2329 bp TE-mapping antisense alignments were converted into FASTA format using bedtools and EMBOSS, trimmed to uniform length of 23 bp using FASTX-toolkit and used as input for the WebLogo 3 program35.

TE and coding gene analysis. In order to properly account for multi-mapping RNA-seq reads when analyzing TEs, we used the RepEnrich approach36 to quantify read counts for each TE. This approach combines all instances of each annotated TE in the genome together and counts a read if it aligns to any of them at least once, allowing proper quantitation of reads that would otherwise be discarded because of mapping to multiple locations. Read count tables from RepEnrich were processed with the edgeR package to perform normalization and calculate log2 fold change. For small RNA-seq libraries, reads were aligned using RepEnrich and counts were normalized using unique alignments to cisNATs and structured loci, as described in ref. 11. Log2 fold changes (piwi / / piwi / ) were calculated using normalized read counts. TEs with RNA-seq log2 fold changes 40.263(1.2 fold increase) in piwi mutants compared with heterozygous controls were

plotted on the heat map together with the corresponding piRNA-seq change for each element. Heat maps were generated with the gplots package in R.

To create TE alignment proles (Supplementary Fig. 4), reads weremapped uniquely to the relevant TE Repbase consensus sequence using Bowtie( v 1 m 1). Read depth across TE consensus sequence was determined using

bedtools genomecov command, using total uniquely mapping reads in the library to normalize for library size differences. Plots were created in R.

For normal coding gene analysis, edgeR was used to determine differentially expressed genes whose expression was signicantly changed in total RNA-seq libraries of piwi mutant fat body relative to heterozygous controls. We then performed a KEGG pathway analysis on these genes using Flymine.org37.

piRNA cluster analysis. To determine abundance of small RNA reads mapping to annotated piRNA clusters, 2329 bp size selected reads were aligned uniquely to the genome using Bowtie ( v 1 m 1). piRNA cluster-specic reads were

extracted using bedtools using the 15 most highly expressing piRNA clusters, as dened in refs 10,11. Specic cluster coverage plots were calculated using bedtools genomecov, normalizing to total uniquely aligning reads (excluding small nucleolar RNA, transfer RNA and miRNA), and plotted with the Sushi R package from Bioconductor38.

To assay genic piRNA reads, 2329 bp aligned reads were separated into 50UTR, coding sequence and 30UTR regions for each gene (Flybase annotations) and counted and sorted sense/antisense using bedtools. For selected genes with high numbers of antisense piRNAs in the 30UTR (see, for example, Fig. 2g), coverage was calculated as described above. Coverage and gene models were plotted using Sushi38.

Immunoblotting. Fly body segments and tissues were dissected as in RNA/smRNA-seq Library Preparations. Whole-cell lysate samples were prepared in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, 1% Triton-X-100, 1% Na-deoxycholate, 0.1% SDS and 2.5 mg ml 1 of Pepstatin A, Leupeptin, Antipain, Aproptinin and Chymostatin). Then, 15 mg of total protein was loaded and run on a 12% polyacrylamideSDS gel. Proteins were then transferred to polyvinylidene diuoride membrane and the membrane was blocked overnight in Tris-buffered saline with 0.1% Tween-20 in 5% milk at 4 C. Membranes were rst cut between a 75 and 50 kDa marker and the lower molecular weight half of the membrane incubated with anti-Actin (mouse 1:2,000; EMD Millipore MAB1501). The upper molecular weight half of the membrane was incubated with anti-Piwi (mouse 1:200; Santa Cruz sc-390946), anti-Aubergine (rabbit 1:2,000) or anti-AGO3 (rabbit 1:3,000). Before blotting with either anti-Aubergine or anti-AGO3 and between each reprobing, the blot was stripped with stripping buffer (50 mM Tris-HCl, pH 6.8, 20 g l 1 SDS and 0.8%

b-mercaptoethanol) at 60 C. After treating with each primary antibody, membranes were incubated with either horseradish peroxidase-conjugated anti-mouse (goat 1:5,000; ThermoFisher 31430) or anti-rabbit (goat 1:5,000; ThermoFisher 31460). The anti-AGO3 and anti-Aubergine antibodies were gifts from Phillip Zamore. Uncropped images of the original immunoblots can be found in Supplementary Fig. 2.

Immunouorescence assays and lipid staining. Flies were briey dipped in EtOH and blotted, xed in 4% paraformaldehyde in PBS for 20 min, washed with PBS on ice and embedded in Tris-buffered saline Tissue Freezing Medium (Fisher Scientic 15-183-13). Moulds were frozen on dry ice and stored at 80 C.

Moulds were cryosectioned into 10 mm sections and gently washed with PBS. For anti-Piwi immunouorescence, slides were incubated with anti-Piwi antibody (mouse 1:50; Santa Cruz sc-390946) for 1 h at room temperature, washed with PBS, incubated with anti-mouse Alexa Fluor-568-conjugated antibody (goat 1:2,000; ThermoFisher A-11031) for 1 h at room temperature and then washed again with PBS. Anti-Piwi primary and secondary antibodies were both diluted in 5% normal

NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856

goat serum and 0.1% Triton X-100. For lipid staining, sections were incubated with Nile Red (0.5 mg ml 1; ThermoFisher N-1142) for 1 h at room temperature and washed with PBS. For anti-g-H2A.v immunouorescence, slides were washed in PBS and blocked for 30 min at room temperature in 5% normal goat serum and0.5% Triton X-100. Slides were then briey rinsed in PBS, incubated with anti-g-H2A.v antibody (mouse 1:20; DSHB unc93-5.2.1) for 2 h at room temperature, washed with PBS 0.1% Tween-20 (PBST), incubated with anti-mouse Alexa

Fluor-568-conjugated antibody (goat 1:1,000; ThermoFisher A-11031) for 1 h at room temperature and washed twice with PBST followed by one nal wash with PBS. Anti-g-H2A.v primary and secondary antibodies were both diluted in 5%

normal goat serum and 0.1% Tween-20. All slides were mounted and stained with 4,6-diamidino-2-phenylindole (DAPI; ThermoFisher P36935).

Images of Piwi were acquired with a Zeiss LSM 510 meta-confocal laser-scanning microscope and g-H2A.v and lipid staining with a Zeiss AxioImager.Z1

ApoTome microscope. Images of representative nuclei, proteins and lipids were selected and processed using ImageJ and Adobe Photoshop. g-H2A.v intensity was measured and quantied using ImageJ according to the method published inref. 28. Lipid droplet size was measured and quantied using ImageJ from sections of individual ies. For g-H2A.v uorescence intensity and lipid droplet size quantication, Students two-tailed t-test was performed to determine signicance.

gypsy-TRAP transposition reporter. The piwi mutant fat body gypsy-TRAP reporter line was generated using the lines mentioned in Fly Stocks and Husbandry. Briey, the fat body-specic r4-GAL4 driver line was rst recombined with a UAS-GFP line generating a new stable line that expresses green uorescent protein (GFP) exclusively in the fat body. This line was then crossed with the gypsy-TRAP reporter line24, a line containing a GAL80 driven by a tubulin promoter separated by an ovo binding site that attracts the gypsy TE that suppresses GFP expression until TE transposition activates reporter function. Once stable, this new line was then further crossed into a piwi2/CyO mutant background, thereby generating the nal piwi mutant fat body gypsy-TRAP reporter line. Mixed sex heterozygous and homozygous ies from this line were then aged together for 10 days. Flies were then separated by piwi genotype and female ies of each piwi genotype further separated under a uorescent dissecting scope into three groups according to the approximate number of GFP cells visibly uorescing beneath the ventral abdominal wall. Groups were as follows: low (no visible GFP-positive cells), medium (o50% abdominal area showing GFP-positive cells) and high (450%

abdominal area showing GFP-positive cells). Fishers exact test was performed to determine signicance between the combined medium high groups and the low

group.

As was demonstrated for adult mushroom body neurons24, we found that the GFP-positive cells in the fat body were dependent upon endogenous gypsy insertion as mutant ovo binding sites in the wild-type gypsy-TRAP/r4-GAL4::UASGFP reporter line resulted in no GFP-positive cells as expected9. All representative images were taken from heterozygous piwi mutant fat body gypsy-TRAP ies.

Metabolic assays. Assays were performed as in ref. 39 with the following modications. Biological replicates for each genotype assayed were generated using ve whole adult ies. All data were normalized to total protein concentration and calculated as a percent relative to the control genotype. For each assay, Students two-tailed t-test was performed to determine signicance.

TAG assay. Flies were homogenized in 200 ml of cold PBST buffer. The homogenate was heat-treated at 70 C for 5 min to inactivate endogenous enzymes. Protein concentration was measured using a Bradford assay (Bio-Rad 5000006). Samples were diluted with PBST and 15 ml of heat-treated homogenate or glycogen standard (Standbio 2103-030) were incubated with 200 ml of Trigylceride Reagent (Thermo Scientic TR22421) for 5 min at 37 C. Absorbance was measured at 540 nm and TAG content of samples was calculated based on a standard curve of TAG that was run in parallel with experimental samples.

Glycogen assay. Flies were homogenized in 200 ml of cold PBS. The homogenate was heat-treated at 70 C for 5 min to inactivate endogenous enzymes. Samples were centrifuged for 3 min at 16,100 g and the supernatant collected. Protein

concentration was measured using a Bradford assay (Bio-Rad 5000006). Samples were diluted with PBS and 90 ml of heat-treated homogenate were incubated with either 20 ml of amyloglucosidase (Sigma-Aldrich A7420) or 20 ml of PBS. To create a glycogen standard curve, 50 ml of glycogen standards (Ambion AM9510) were treated with either 50 ml of amyloglucosidase or 50 ml of PBS. All samples were incubated at 37 C for 1 h. Then, 30 ml of each sample was added to a 96-well plate. Next, 100 ml of Innity Glucose Hexokinase reagent (Thermo TR15421) was added to all samples and incubated at room temperature for 15 min. The absorbance of samples was then measured at 340 nm and normalized by subtracting the absorbance of the free glucose of untreated samples from the absorbance of the total amount of glucose present in samples treated with amyloglucosidase. Glycogen content was then calculated based on the normalized glycogen standard curve.

Trehalose assay. Flies were homogenized in 200 ml of cold trehalose buffer(5 mM Tris-HCl, pH 6.6, 137 mM NaCl and 2.7 mM KCl). The homogenate was heat-treated at 70 C for 5 min to inactivate endogenous enzymes. Sampleswere centrifuged for 3 min at 16,100 g and the supernatant collected. Protein

concentration was measured using a Bradford assay (Bio-Rad 5000006). Samples were diluted with trehalose buffer and 90 ml of heat-treated homogenate were incubated with either 20 ml of trehalase (Sigma-Aldrich T8778) or 20 ml of trehalose buffer. To create a trehalose (Sigma-Aldrich T9531) standard curve, 50 ml of standard were incubated with either 30 ml of trehalase or 30 ml of trehalose buffer. A free glucose (Fisher Scientic 50-99-7) standard curve was also generated. All samples were incubated at 37 C overnight. Then, 30 ml of each sample was added to a 96-well plate. To all samples, 100 ml of Innity Glucose Hexokinase reagent (Thermo TR15421) was added and incubated at room temperature for 15 min. The absorbance of samples was then measured at 340 nm and normalized by subtracting the absorbance of free glucose present in untreated samples from the total amount of glucose present in samples treated with trehalase. Trehalose and free glucose content were calculated based on standard curves of trehalose and glucose respectively.

Glucose assay. Flies were homogenized in 200 ml of cold PBS. The homogenate was heat-treated at 70 C for 5 min to inactivate endogenous enzymes. Samples were centrifuged for 3 min at 16,100 g and the supernatant collected. Protein

concentration was measured using a Bradford assay (Bio-Rad 5000006). Samples were diluted with PBS and 30 ml of supernatant was then added to a 96-well plate.

To all samples, 100 ml of Innity Glucose Hexokinase reagent (Fisher Scientic TR15421) was added and incubated at room temperature for 15 min. The absorbance of samples was then measured at 340 nm and glucose content was calculated based on a standard curve of glucose (Fisher Scientic 50-99-7).

Starvation assay. Before permanent starvation, ies were synchronized in their feeding by fasting on 2% agar for 4 h followed by a nal period of feeding for 2 h. Flies were then sorted under CO2 anaesthesia into separate sex vials containing 2%

agar at a density of 10 males or 10 females per vial, with a total of 5 vials (nE50) for each genotype. Dead ies were scored and counted every 6 h. Starvation analyses were performed and log rank statistics calculated using the online application OASIS40. Starvation assays were repeated at least twice.

Immune challenge assay. A bacterial culture of Erwinia carotovora (Gram-negative insect/plant pathogen), a gift from Neal Silverman, was grown overnight in Luria broth, shaking at 220 r.p.m. on a platform shaker at 37 C. The culture was allowed to reach OD600nmE2.0 before removing 1 ml for centrifugation at

2,000 g for 2 min. The supernatant was removed and the bacterial pellet gently

washed with 1 ml of 10 mM MgSO4 to remove traces of culture media. The wash was removed and the pellet resuspended in 10 ml of fresh 10 mM MgSO4. A0.15 mm needle was then dipped into 80% EtOH and ame sterilized. Flies were sorted under CO2 anaesthesia and inoculated by dipping the needle into either the bacterial suspension or EtOH for mock infection controls, and inserting the needle midway into one side of the thorax. Flies were then sorted into separate sex vials at a density of 10 males or 10 females per vial, with a total of 5 vials (nE50) for each condition/genotype, and passed to fresh food at least every other day. Dead ies were scored and counted every 24 h. Immune challenge analyses were performed and log rank statistics calculated using the online application OASIS40. Immune challenge assays were repeated at least twice.

Lifespan assay. Flies used in lifespan experiments were collected upon eclosion over a 48 h period and were sorted under CO2 anaesthesia and placed in food vials at a density of 25 males and 25 females per vial, with a total of 10 vials (nE250) for each genotype. Lifespan food consists of 50 g l 1 autolysed yeast, 50 g l 1 sucrose and 20 g l 1 agar (all w/v). For the amenco 3TC lifespans, amenco homozygous mutant ies were collected and aged on either food containing 150 g l 1 autolysed yeast, 150 g l 1 sucrose and 20 g l 1 agar, (all w/v) or identical food with 10 mM lamivudine (3TC). Flies were passed to fresh food every other day, and dead ies scored and counted. Lifespan analyses were performed and Wilcoxon rank sum test statistics calculated using the online application OASIS40. Lifespan assays were repeated twice.

Data availability. The data sets generated and analysed in this study have been deposited and are available in the GEO database, under the series accession number GSE89260. Additional relevant data sets and computer code are available either in this published article (and its Supplementary Information Files) or are available from the corresponding author on reasonable request.

References

1. Cordaux, R. & Batzer, M. A. The impact of retrotransposons on human genome evolution. Nat. Rev. Genet. 10, 691703 (2009).

2. Levin, H. L. & Moran, J. V. Dynamic interactions between transposable elements and their hosts. Nat. Rev. Genet. 12, 615627 (2011).

8 NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13856 ARTICLE

3. Shabalina, S. & Koonin, E. Origins and evolution of eukaryotic RNA interference. Trends Ecol. Evol. 23, 578587 (2008).

4. Ghildiyal, M. & Zamore, P. D. Small silencing RNAs: an expanding universe. Nat. Rev. Genet. 10, 94108 (2009).

5. Siomi, M. C., Sato, K., Pezic, D. & Aravin, A. A. PIWI-interacting small RNAs: the vanguard of genome defence. Nat. Rev. Mol. Cell Biol. 12, 246258 (2011).

6. Kaminker, J. S. et al. The transposable elements of the Drosophila melanogaster euchromatin: a genomics perspective. Genome Biol. 3, research0084.10084.20 (2002).

7. Slotkin, R. K. & Martienssen, R. Transposable elements and the epigenetic regulation of the genome. Nat. Rev. Genet. 8, 272285 (2007).

8. Wallace, N. a., Belancio, V. P. & Deininger, P. L. L1 mobile element expression causes multiple types of toxicity. Gene 419, 7581 (2008).

9. Wood, J. G. et al. Chromatin-modifying genetic interventions suppress age-associated transposable element activation and extend life span in Drosophila. Proc. Natl Acad. Sci. USA 113, 1127711282 (2016).

10. Brennecke, J. et al. Discrete small RNA-generating loci as master regulators of transposon activity in Drosophila. Cell 128, 10891103 (2007).

11. Malone, C. D. et al. Specialized piRNA pathways act in germline and somatic tissues of the Drosophila ovary. Cell 137, 522535 (2009).

12. Ross, R. J., Weiner, M. M. & Lin, H. PIWI proteins and PIWI-interacting RNAs in the soma. Nature 505, 353359 (2014).

13. Reddien, P. W., Oviedo, N. J., Jennings, J. R., Jenkin, J. C. &Sanchez Alvarado, A. SMEDWI-2 is a PIWI-like protein that regulates planarian stem cells. Science 310, 13271330 (2005).

14. Liu, X. et al. Expression of hiwi gene in human gastric cancer was associated with proliferation of cancer cells. Int. J. Cancer 118, 19221929 (2006).15. Lee, J. H. et al. Stem-cell protein Piwil2 is widely expressed in tumors and inhibits apoptosis through activation of Stat3/Bcl-XL pathway. Hum. Mol. Genet. 15, 201211 (2006).

16. Grochola, L. F. et al. The stem cell-associated Hiwi gene in human adenocarcinoma of the pancreas: expression and risk of tumour-related death. Br. J. Cancer 99, 10831088 (2008).

17. Sumiyoshi, T. et al. Loss of l(3)mbt leads to acquisition of the ping-pong cycle in Drosophila ovarian somatic cells. Genes Dev. 30, 16171622 (2016).18. Fagegaltier, D. et al. Oncogenic transformation of Drosophila somatic cells induces a functional piRNA pathway. Genes Dev. 30, 16231635 (2016).

19. Janic, A., Mendizabal, L., Llamazares, S., Rossell, D. & Gonzalez, C. Ectopic expression of germline genes drives malignant brain tumor growth in Drosophila. Science 330, 18241827 (2010).

20. Perrat, P. N. et al. Transposition-driven genomic heterogeneity in the Drosophila brain. Science 340, 9195 (2013).

21. Saito, K. et al. Pimet, the Drosophila homolog of HEN1, mediates 20-O-methylation of Piwi- interacting RNAs at their 30 ends. Genes Dev. 21, 16031608 (2007).

22. Horwich, M. D. et al. The Drosophila RNA methyltransferase, DmHen1, modies germline piRNAs and single-stranded siRNAs in RISC. Curr. Biol. 17, 12651272 (2007).

23. Li, C. et al. Collapse of germline piRNAs in the absence of Argonaute3 reveals somatic piRNAs in ies. Cell 137, 509521 (2009).

24. Li, W. et al. Activation of transposable elements during aging and neuronal decline in Drosophila. Nat. Neurosci. 16, 529531 (2013).

25. Sarot, E., Bucheton, A. & Pe, A. Evidence for a piwi-dependent RNA silencing of the gypsy endogenous retrovirus by the Drosophila melanogaster amenco gene. Genetics 1321, 13131321 (2004).

26. Robine, N. et al. A broadly conserved pathway generates 30UTR-directed primary piRNAs. Curr. Biol. 19, 20662076 (2009).

27. Saito, K. et al. A regulatory circuit for piwi by the large Maf gene trafc jam in Drosophila. Nature 461, 12961299 (2009).

28. Chen, H., Zheng, X., Xiao, D. & Zheng, Y. Age-associated de-repression of retrotransposons in the Drosophila fat body, its potential cause and consequence. Aging Cell 15, 542552 (2016).

29. Arrese, E. L. & Soulages, J. L. Insect fat body: energy, metabolism, and regulation. Annu. Rev. Entomol. 55, 207225 (2010).

30. Partridge, L., Alic, N., Bjedov, I. & Piper, M. D. W. Ageing in Drosophila: the role of the insulin/Igf and TOR signalling network. Exp. Gerontol. 46, 376381 (2011).

31. Hansen, M., Flatt, T. & Aguilaniu, H. Perspective reproduction, fat metabolism, and life span: what is the connection? Cell Metab. 17, 1019 (2012).

32. Touret, F., Guiguen, F., Greenland, T. & Terzian, C. In between: Gypsy in Drosophila melanogaster reveals new insights into endogenous retrovirus evolution. Viruses 6, 49144925 (2014).

33. Juliano, C. E. et al. PIWI proteins and PIWI-interacting RNAs function in Hydra somatic stem cells. Proc. Natl Acad. Sci. USA 111, 337342 (2014).

34. Wickersheim, M. L. & Blumenstiel, J. P. Terminator oligo blocking efciently eliminates rRNA from Drosophila small RNA sequencing libraries. Biotechniques 55, 269272 (2013).

35. Crooks, G. E., Hon, G., Chandonia, J.-M. & Brenner, S. E. WebLogo: a sequence logo generator. Genome Res. 14, 11881190 (2004).

36. Criscione, S. W., Zhang, Y., Thompson, W., Sedivy, J. M. & Neretti, N. Transcriptional landscape of repetitive elements in normal and cancer human cells. BMC Genomics 15, 583 (2014).

37. Lyne, R. et al. FlyMine: an integrated database for Drosophila and Anopheles genomics. Genome Biol. 8, R129 (2007).

38. Phanstiel, D. H., Boyle, A. P., Araya, C. L. & Snyder, M. P. Sushi.R: Flexible, quantitative, and integrative genomic visualizations for publication-quality multi-panel gures. Bioinformatics 19, 28082810 (2014).

39. Tennessen, J. M., Barry, W. E., Cox, J. & Thummel, C. S. Methods for studying metabolism in Drosophila. Methods 68, 105115 (2014).

40. Yang, J. et al. OASIS: online application for the survival analysis of lifespan assays performed in aging research. PLoS ONE 6, e23525 (2011).

Acknowledgements

We thank Gillian Horwitz, Mark Henriksen, Davis Hartnett, Jackson Taylor and Suzanne Hosier for technical assistance and Will Lightfoot for y food preparation. We also thank Haifan Lin, Joshua Dubnau, Robert Reenan, Marc Tatar and the Bloomington Drosophila Stock Center for y stocks; John Sedivy and Neal Silverman for reagents; Chengjian Li, Bo Han and Phillip Zamore for reagents and advice on bioinformatic analysis; Marissa Holmbeck, Leila Rieder, Yiannis Savva, Erica Larschan, Nicola Neretti and Robert Reenan for discussions and comments on the manuscript. This work was supported by a NIA T32 Training Grant AG041688 and a F31 Predoctoral Research Fellowship Award AG047736 to B.C.J. and NIA Grants AG16667 and AG24353, a Glenn/American Federation for Aging Research (AFAR) Breakthroughs in Gerontology award, NIH Program Project Grant AG51449 and Grant P30AI042853 from the Providence/Boston Center for AIDS Research to S.L.H.

Author contributions

B.C.J. conceived and directed the project with contributions from J.G.W. and S.L.H. C.C. isolated all body segment and pure fat body samples. J.G.W. and C.C. performed all sequencing library preparations. J.G.W. performed all bioinformatic analyses with support from B.C.J. B.C.J. performed all immunoblotting. B.C.J. generated and assayed the piwi mutant fat body gypsy-TRAP ies. M.J.F., E.R.S. and B.C.J. performed all sectioning, immunouorescence, lipid staining and microscopy. B.C.J. quantied immunouorescence and lipid droplet staining. B.C.J. processed all immunoblot and microscopy images for publication. A.D.T. and B.C.J. performed all metabolic assays. M.J.F., E.R.S., A.D.T. and B.C.J. performed all starvation assays. E.R.S. and B.C.J. performed all immune challenge assays. C.C., E.R.S. and B.C.J. performed all lifespan assays. B.C.J. wrote the manuscript with assistance from J.G.W. and S.L.H.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Jones, B. C. et al. A somatic piRNA pathway in the Drosophila fat body ensures metabolic homeostasis and normal lifespan. Nat. Commun. 7, 13856 doi: 10.1038/ncomms13856 (2016).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2016

NATURE COMMUNICATIONS | 7:13856 | DOI: 10.1038/ncomms13856 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9