Maternally inherited siRNAs initiate piRNA cluster formation

Maternal

inherited siRNAs initiate piRNA cluster formation

Yicheng Luo

, Peng He

1,2

, Nivedita Kanrar

Katalin Fejes Toth

and

Alexei

Aravin

Division of Biology and Biological Engineering, California Institute of Technology, Pasadena,

CA 91125, USA

present address:

European Molecular Biology Laboratory, European Bioinformatics Institute

(EMBL

EBI),

Wellcome

Genome Campus, Cambridge, UK

To whom correspondence should be addressed:

Alexei Aravin aaa@caltech.edu

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Abstract

PIWI

interacting RNAs (piRNAs)

guide

repression of transposable elements in germline

of animal

. In

Drosophila

, piRNA

are produced

from heterochromatic genomic loci,

called piRNA cluster

that act as a

repositories

of information about genome invaders.

piRNA generation by dual

strand clusters depend on

the

chromatin

bound Rhino

Deadlock

ff (RDC) complex

, which is deposited on clu

sters

guided by piRNA

forming

feed

forward loop

in which

piRNA

promote their own biogenesis. However,

how piRNA clusters are formed initially

, before

cognate piRNA

s are present,

remained

unknown.

Here we report spontaneous

de novo

formation of a piRNA

cluster from

repetitive transgenic sequences.

We show that cluster formation occurs gradually over

several generations and require

continuous trans

generational transmission of small

RNAs from mothers to their progeny. We

discovered

that

maternally

suppl

ied

siRNA

are

responsible for

trigger

ing

de novo

cluster activation

in progeny.

contrast

the

siRNA

pathway is

dispensable for

cluster function after its establishment. These results revealed

unexpected cross

talk between

the

siRNA and piRNA pathways and

suggest

mechanism for

de novo

formation of piRNA clusters triggered by production of siRNA

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Highlights

A t

ranscribed repetitive transgene is

spontaneously

converted into dual

strand piRNA

cluster

Establishment of piRNA cluster occurs over multiple generations and requires cytoplasmic

inheritance of cognate

small

RNA from mothers

ognate siRNAs initiate the act

ivation of piRNA cluster

, but are dispensable after its

establishment

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Introduction

Binary complexes of small non

coding RNAs and Argonaute (Ago) proteins play essential

roles in regulating gene expression and suppressing foreign and selfish nucleic aci

ds. Small

RNAs guide the Argonautes to complementary RNA targets upon which Agos either cleave the

target or recruit additional factors to repress them by other mechanisms. Despite the common

architecture of Ago

small RNA complexes, there are three distinc

t classes of small RNAs

siRNA,

miRNA and piRNA

that differ in their biogenesis, functions as well as the specific members of

the Ago family they partner with.

Both siRNA and piRNA were reported to suppress activity of endogenous (transposable

elements

and other selfish genes) and

exogenous

(viruses) elements in various animal species

(Aravin et al., 2006; Brennecke et al., 2007; Gammon and Mello, 2015; Gitlin et al., 2002; Lindbo

et al., 1993; Vance and Vaucheret, 2001; Voinnet et al., 1999)

, however, t

he two pathways are

believed to work independently of each other. Despite the complete abrogation of the siRNA

pathway in Ago2 deficient flies, these flies are viable and fertile and show only mild activation of

a few transposons in somatic tissues

(Czech et al., 2008; Ghildiyal et al., 2008; Kawamura et al.,

200

8; Pelisson et al., 2007)

. In contrast, flies deficient for piRNA pathway components

demonst

rate strong activation of multiple TE families in the germline associated with DNA damage

and complete sterility

(Brennecke et al., 2007; Sienski et al., 2015; Vagin et al., 2006; Yu et al.,

2015b)

siRNAs are processed from double

stranded or hairpin precursors by the Dicer nuclease and

then loaded into their Ago protein partner

(Bernstein et al., 2001; Matranga et al., 2005; Rand et

al., 2005)

. siRNA/Ago complexes cleave complementary RNA targets in the cytoplasm leading to

their degradation

(Matranga et al., 2005; Rand et

al., 2005)

. In

Drosophila

, siRNAs associate

exclusively with Ago2 and mutation of Ago2 abrogates siRNA

guided repression

(Czech et al.,

2008; Ghildiyal et a

l., 2008; Kawamura et al., 2008; Okamura et al., 2004; Vagin et al., 2006)

The biogenesis and function of piRNAs

, however,

is much more complex than that of siRNAs.

piRNA processing is independent of Dicer but involves multiple other proteins

(Andersen et al.,

2017; Chen et al., 2016; ElMaghraby et al., 2019; Mohn et al., 2014; Vagin et al., 2006; Zhang et

al., 2014)

. piRNAs are expressed

in the germline and closely associated somatic cells of the ovary

and testis and are loaded into a distinct clade of Argonautes called Piwi proteins, which in flies

consist of Piwi, Aub and Ago3

(Brennecke et al., 2007; Saito et al., 2006; Vagin et al., 2006)

Similar to siRNAs, cytoplasmic Piwi/piRNA complexes cleave complementary RNA targets,

however, in addition

to target degradation, this process also amplifies piRNAs through the so

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

called ping

pong cycle

(Aravin et al., 2008; Brennecke et al., 2007; Gunawardane et al., 2007)

thereby connecting their function to their own biogenesis. Furthermore, both flies and mice harbor

a nuclear Piwi/piRNA complex capable of transcriptional repression through recognition of

nascent transcrip

ts followed by recruitment of a chromatin modifying machinery

(Aravin et al.,

2008; Le Thomas et al., 2013; Ninova et al., 2020; Sienski et al., 2015; Sienski et al., 2012; Yu et

al., 2015b)

siRNA precursors are recognized by their

double

stranded nature. Similar to siRNAs, piRNAs

are also processed from longer RNA precursors, however these transcripts are single

stranded

and lack distinct sequence and structure motifs, raising the question of how they are recognized

and channeled in

to the processing machinery. In

Drosophila

, the chromatin

bound Rhino

Deadlock

Cutoff (RDC) protein complex marks dual

strand piRNA clusters, genomic regions that

generate the majority of piRNAs in the germline

(

Klattenhoff et al., 2009; Zhang et al., 2014;

Thomas et al., 2014b;

Mohn et al., 2014;

Chen et al., 2016

)

. RDC is required for transcription of

piRNA precursors by promoting initiation

(Andersen et al., 2017; Mohn et al., 2014)

and

suppressing premature termination

(Chen et al., 2016

)

. RDC also promotes loading of the RNA

binding TREX

(Hur et al., 2016)

and Nxf3

Nxt1

(ElMaghraby et al., 2019)

RNA export complexes

on pre

piRNA. Loading of Nxf3

Nxt1 was proposed

to channel precursors to the cytoplasmic

processing machinery

(ElMaghraby et al., 2019)

. Thus, RDC binding to chromatin of piRNA

clusters might be both necessary and sufficient to sustain piRNA biogenesis.

Though the process of RDC deposition on chromatin is not completely understood, it seems

to be guided by p

iRNAs, as the nuclear Piwi/piRNA complex directs deposition of the H3K9me2/3

mark, which in turn provides an anchor for RDC

binding

(Akkouche et al., 2017; Mohn et al., 2014;

Sienski et al., 2015; Yu et al., 2015b)

. Several studies demonstrate the critical role of cytoplasmic

piRNA inheritance from the mother to the progeny in initiating piRNA production

(Casier et al.,

2019; de Vanssay et al., 2012; Le Thomas et al., 2014b)

. Together these findings suggest that

piRNA biogenesis is governed by a trans

generatio

nal feed

forward loop in which

piRNA

biogenesis is promoted by RDC

complex

, which in turn is deposited on chromatin guided by

cytoplasmically inherited piRNAs. This feed

forward loop explains how piRNA profiles are

maintained through generations. However,

in order to adapt to new transposon invasions, the

pathway must be able to generate novel piRNAs. How novel piRNA clusters arise is poorly

understood.

Here we describe the

de novo

formation of a piRNA

cluster over several generations. This

process is accompanied by increasing piRNA levels and accumulation of the H3K9me3 mark and

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Rhi on cluster chromatin and requires continuous, maternal trans

generational cytoplasmic

transmission of small RNAs. We foun

d that cognate siRNAs trigger initial cluster activation,

however, siRNA are dispensable after the cluster is established. Our results point to a tight

cooperation between the siRNA and piRNA pathways in the fight against genome invaders and

suggest that t

ransposons are first detected by the siRNA pathway, which activates a robust piRNA

response.

Results

Reporters inserted in the dual

strand 42AB cluster are repressed by piRNA, while reporters

in the uni

strand 20A cluster disrupt cluster

expression

To understand how new insertions into piRNA clusters are regulated

we integrated reporters

into

the

two types of piRNA clusters

–

uni

strand and dual

strand clusters. We employed

recombinase

mediated cassette exchange (RMCE) to integrate a repo

rter into specific genomic

sites

using

a collection of

Minos

mediated integration cassette (MiMIC) containing

melanogaster

stocks. The reporter encodes nuclear EGFP expressed under control of the

biquitin

(

ubi

p63E

)

gene promoter, which drives expressi

on in both somatic and germ cells (Fig.

1A). Using RMCE we inserted reporters into the major dual

strand cluster 42AB and the uni

strand

cluster 20A (Fig. 1B).

the

20A cluster

the

reporter was integrated 2.5

downstream of

the

cluster promoter and two reporter orientations were obtained.

Reporters inserted in both clusters

were expressed in somatic follicular cells of the fly ovary

(Fig.1C)

, as well as in other somatic

tissues

(data not show

)

, indicating that the transgenes are

functional and that integration into

the repeat

rich cluster environments is compatible with reporter expression in somatic cells. Flies

with reporters in the uni

strand 20A cluster

inserted in either orientation

also expressed GFP in

the germline. Unli

ke 20A cluster transcripts, which localize

to the nucleus, GFP mRNA

was

predominantly in

the

cytoplasm. In contrast,

though fluorescent in situ hybridization revealed

that both strands of

the

42AB reporter sequence

were

transcribed, GFP protein

was not

expressed

from the 42AB cluster insertion

in the germline.

imilar to native 42AB cluster

transcripts, RNA transcribed from both strands of

the

42AB reporter

was

concentrated in the

nucleus.

The exclusive repression

the

42AB reporter in

the

germline

here

the

42AB cluster is

active

suggest that

repression

might occur in a piRNA

dependent fashion. To explore if reporter

sequences generate piRNAs, we cloned small RNA libraries from ovaries of transgenic animals.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Analysis of the libraries revealed

that

reporter

derived

piRNAs

were

abundant in flies with 42AB

cluster insertions

but not

in flies with reporters in

cluster

20A and

in a

control

non

cluster

66A6

region (Fig. 1D). The 42AB reporter

derived piRNAs have the expected bias (69.95%) for U in the

fir

st position. Though piRNA were derived from both strands of the 42AB reporter sequence, 2.45

fold more piRNA are in antisense orientation relative to the GFP mRNA (Fig. 1E), indicating that

they are not processed from reporter mRNA. In contrast,

the

few RN

A reads derived from

reporters inserted in 20A and

the

non

cluster region were predominantly in sense orientation and

did not have

bias, indicating that they likely represent

mRNA

degradation products.

To explore whether

the

repression

GFP reporter

inserted in

to the

42AB cluster depends

on piRNAs

we knock

down Zuc

chini (Zuc)

, a critical piRNA bio

genesis factor

(Ipsaro et al., 2012;

Nishimasu et al., 2012)

, in the germline

epletion of Zuc led to strong (26.4

fold) reduction in the

level of piRNAs mapping to the

42AB

reporter (Fig. 1E) and

its dere

pression (Fig.1F). GFP

protein

expression

upon Zuc GLKD correlated with

the

appearance of sense reporter transcripts

in the cytoplasm of nurse cells, while antisense RNA remained in the nucleus. Thus, insertion of

a gene into the 42AB cluster leads to gene

ration of abundant piRNAs that trigger its repression.

To understand why

20A reporters

do not

produce

piRNA

, we analyzed expression of

20A

cluster transcripts by RT

qPCR using primer sets designed to detect cluster transcripts upstream

and downstream of

the reporter insertion site. Surprisingly, we found that

the abundance of

20A

cluster transcripts was strongly (

fold

) reduced in flies with reporter insertions compared to

both wild

type flies and the original MIMIC

flies

used for RMCE (Fig. 1G). In agreement with

the

decreased cluster transcript level, 20A piRNA level

dropped 84

and 321

fold in 20A[

] and 20A[+]

flies, respectively

throughout the whole cluster as far as 38 kb from

the

insertion site (Fig. 1B, G).

In co

ntrast

the

insertion in

the

42AB cluster

did

not affect

piRNA

level from this cluster

(data not

shown).

As the original MIMIC line contains a promoterless insertion in the same site as the

reporter lines, this result suggests that cluster expression is di

srupted by reporter transcription

rather than insertion of a heterologous sequence

per se

. Thus,

insertion of an

actively transcribed

gene in the 20A cluster close to its promoter disrupts

cluster

expression.

An u

nusual reporter

behaves like

bona fide

dual

strand piRNA cluster that is enriched in

Rhi and depends on it to generate piRNA

Replacement of the MIMIC cassette with the reporter through recombinase

mediated

cassette exchange (RMCE) leads to random orientation of the inserted sequence. Each

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

replacement experiment generates several independent

Drosophila

lines that we maintained an

in which we determined cassette orientation. Unexpectedly, we found that one of the lines with

insertion into the 20A cluster has distinct properties. Unlike other lines that maintained reporter

expression in both

the

soma and

the

germline, this line, wh

ich we dubbed 20A

X, lost germline

expression after several months of propagation (Fig. 2A). The somatic GFP expression,

which we

also confirmed

by detecting

abundant cytoplasmic GFP

RNA in follicular cells by in situ

hybridization, argues against genetic

damage of the reporter cassette. GFP RNA was also

detected in germline nurse cells, however, unlike 20A[+] and 20A[

] reporters, transcripts

from

20A

X localized exclusively

the nuclei (Fig. 2A). Thus, unlike other

20A insertion

lines and

similar to in

sertions in 42AB, 20A

X shows normal GFP expression in follicular cells and strong

GFP

repression in the germline.

To test involvement of piRNAs

in repression of the 20A

X reporter

in the germline

, we cloned

and analyzed small RNA libraries from ovaries o

f 20A

X flies. This analysis revealed abundant

piRNAs and siRNAs corresponding to the reporter sequence indicating that 20A

X is active as a

piRNA producing locus. In fact, 20A

X generates 20.3

fold more piRNA

than

the 42AB reporter.

Germline knockdown o

f the piRNA biogenesis factor Zuc led to loss of 20A

X piRNAs (Fig. 2B).

Zuc GLKD

also

led to release of

the germline reporter

repression as demonstrated by GFP protein

expression and detection of sense RNA in the cytoplasm (Fig

2C) indicating that, simil

ar to 42AB

insertions, repression of 20A

X is piRNA

dependent.

Several studies revealed essential differences between uni

strand and dual

strand piRNA

clusters

(Chen et al., 2016; Goriaux et al., 2014; Mohn et al., 2014)

. Dual

strand clusters, such as

42AB, are active exclusively in the germline and their transcription, nuclear processing and export

require the Rhin

Deadlock

Cutoff (RDC) complex, which is anchored to these regions by the

H3K9me3 histone mark

(Le Thomas et al., 2014b; Mohn et al., 2014; Zhang et al., 2014)

. In

contrast, uni

strand clusters, such as flamenco and 20A, do not depend on RDC and the

H3K9me3 mark and can be active in the

soma. To explore if 20A

X functions as a uni

or dual

strand piRNA cluster

we analyzed

Rhino

binding and

H3K9me3

enrichment.

ChIP

qPCR and

ChIP

seq

analyses revealed

that, in contrast to the native 20A cluster and 20A[+] and 20A[

]

reporters, 20A

X is

rongly

enriched in both Rhi and the H3K9me3 mark (Fig. 2D)

suggesting

that it acts as a dual

strand piRNA cluster.

To test

whether

20A

activity

depends on RDC, we analyzed GFP repression and small

RNA profile upon germline knockdown of Rhi.

As expected,

Rhi GLKD reduces the level of

piRNA

generated from

the

42AB reporter (Fig. 2E). Rhi GLKD also

caused 4.2

fold reduction in

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

20A

X piRNA levels and

released repression of GFP in germline of 20A

X flies (Fig. 2E). Taken

together, our results indicate

that 20

X acts as a genuine dual

strand piRNA cluster that

generates piRNAs in Rhi

dependent manner.

The repetitive

organization of

the

20A

X locus co

rrelates with its function as dual

strand

piRNA cluster

We employed several approaches to understand how 20A

X differs from 20A[+] and 20A[

]

reporters

. Genomic PCR of flanking regions suggested that 20A

X harbors

the

reporter cassette

in the correct site

in the 20A cluster

(Fig. 3A). In fact, similar to other 20A reporters, expression

of 20A cluster transcripts is decreased in 20A

X flies (F

ig.

). To confirm the insertion site, we

employed

in situ

hybridization on salivary gland polytene chromosomes using

the

Ubi

GFP

reporter sequence as a probe.

In situ

hybridization revealed two signals: the expected signal in

the 20A region of the X chromosome and additional signal on chromosome 3L, which harbor

s the

endogenous

ubi

p63E

gene (Fig. 3B). The absence of signals in other sites indicates that 20A

line does not harbor additional transgene insertions

other genomic regions. To further validate

these findings, we performed whole

genome sequencing and

searched for reads corresponding

to junctions between

the

reporter and genomic sequences. We identified multiple reads

corresponding to

the two

expected flanking regions in the 20A cluster, while no additional

insertions were identified

corroborating res

ults of the chromosome hybridization (Fig.

S3B

Finally, we employed CRISPR/Cas9 to generate a deletion that removes sequences flanking the

insertion site in

the

20A

X line. We verified the deletion and concomitant loss of the reporter

sequence from the g

enome by genomic qPCR and loss of GFP expression (Fig. 3C

). Thus, the

20A

X line contains a reporter insertion in the single genomic site

in the

same position as other

20A lines.

We employed

whole

genome DNA

seq and

qPCR to analyze reporter copy numbe

r in the

genome, which

showed

that

while

other 20A reporters harbor

a single

copy of

the

reporter

sequence

20A

X contains

copies (Fig. 3D

, E

, S3D

). In addition,

both approaches

revealed

that, unlike 20A[+] and 20[

] lines, the 20A

X insertion contains

plasmid backbone sequence

that

used for recombinase

mediated cassette exchange

and is

normally removed during this

process (Fig.

1A,

, S3D

). Furthermore, we also found multiple DNA

seq reads indicating three

unexpected junctions: (1) between the SV40 3’ UTR and the GFP sequence, (2) between the

SV40 3’ UTR and the plasmid backbone and (3) between

the

attB site and

the

plasmid backbone

(Fig.

). We confirmed all junctions by genomic PCR and Sanger sequencing

. As in situ

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

hybridization on polytene chromosomes, DNA

seq and CRISPR/Cas9 deletion all indicate a

single insertion site in the genome, all reporter copies are located in

single genomi

c site. Taken

together, our results indicate that the 20A

X line contain multiple, rearranged copies of the

reporter sequence in a single site within the 20A cluster. Together, analysis of 20A

X indicates

that

unlike single

copy reporters in 20A, repetiti

ve sequences inserted in the same site generate

piRNAs that induce repression in the germline.

To explore if

the

‘host’ 20A cluster is required for 20A

X to function as dual

strand piRNA

cluster we employed CRISPR/Cas9 to delete

the

promoter of

the

20A cl

uster (Fig. 3G)

In wild

type flies, deletion of

the

putative promoter eliminated expression of long RNA (piRNA precursors)

from

the

20A cluster

indicating that deletion disrupts its function.

However, d

eletion of

the

20A

cluster promoter in 20A

X flies d

id not release reporter silencing (Fig. 3G)

indicating that piRNA

dependent repression of

the

20A

X locus does not require activity of

the

‘host’ 20A cluster.

20A

X is inserted in the same site as 20A[+] and 20[

] reporters but differ

s in

its

repetitive

nature as well as

its

structural rearrangements. Particularly, 20A

X harbors inversions that

are

expected to generate dsRNA upon their transcription. To explore if the presence of transcribed

inverted repeats is sufficient to create

functional dual

strand piRNA cluster, we generated

dsGFP reporter

which

consists

of the same sequence fragments as the original reporter, but

harbors inverted GFP sequence

that would form dsRNA upon transcription (Fig. 3H). We

obtained flies with inser

tion of

the

dsGFP reporter into the same site as other 20A reporters as

well as in

control (non

piRNA cluster) region of the genome (chr

3L:

575

013). Analysis of

small RNAs in ovaries of flies carrying dsGFP constructs revealed the presence of abundan

t 21

nt siRNA

generated from

the

inverted GFP repeat, but no other portion of the construct (Fig. 3H).

In contrast to 20A

X, dsGFP insertions produce only miniscule amount of larger RNA species and

these RNA

lack

bias. There were no significant diffe

rences between small RNAs generated

from the

dsGFP reporter inserted into

the

20A cluster in either orientation and in

the non

cluster

control genomic region. Overall, these results shows that inverted repeat

trigger generation of

siRNA

, but not piRNA

indicating that the presence of inverted repeats might be necessary, but

not sufficient to make 20A

active dual

strand piRNA cluster and

that the

multi

copy nature

and

/or the

extended lengths of the locus might play

critical role.

piRNA

induce

d repression of the 20A

X reporter depends on maternal transmission of

cognate piRNAs

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Previously we and others have shown that the activity of artificial dual

strand piRNA clusters

in the progeny requires cytoplasmic inheritance of piRNAs from

the

mother

(de Vanssay et a

l.,

2012; Hermant et al., 2015; Le Thomas et al., 2014b)

, and proposed that all dual

strand clusters

might depend on trans

generationally inherited piRNAs to maintain their activity

(de Vanssay et

al., 2012; Le Thomas et al., 2014b)

. Therefore, we explored expression of the 20A

X reporter

the progeny

after paternal and maternal inheritance. Flies that inherited the 20A

X insertion from

their mothers (maternal t

ransmission) showed

–

similar to their mothers

–

GFP expression in

follicular cells, but strong

GFP

repression in

the

germline with sense reporter RNA restricted to

the nucleus. In contrast, females that inherited the 20A

X reporter from their fathers (paternal

transmission) had robust GFP protein expression and cytoplasmic localization of GFP mRNA in

the germline (Fig. 4

A). We also observed GFP repression in the germline of males that inherited

the 20A

X locus maternally, but not when they inherited it paternally. In agreement with FISH and

IF results, RT

qPCR showed

fold increase of GFP RNA after paternal transmission

(Fig. 4B).

Thus, repression of the 20A

X reporter in both

in the

male and female germline requires maternal

inheritance of the reporter.

To understand if derepression of GFP after paternal transmission is caused by changes in

piRNA expression, we clon

ed small RNA libraries from ovaries of progeny that inherited the locus

maternally and paternally. Upon paternal transmission

piRNA level

was 10.1

fold

reduced (Fig.

4C). Thus, piRNA generation from the 20A

X locus requires its maternal inheritance and

repression of GFP upon paternal reporter transmission is explained by the dramatic decrease

in reporter

targeting piRNA. Next

we determined enrichment of

the

H3K9me3 mark and Rhino

protein on chromatin of 20A

X reporter in

the

progeny

upon paternal or mat

ernal inheritance of

the locus. Both Rhi and H3K9me3

were

reduced on 20A

X chromatin after paternal transmission

to levels comparable

those

detected

at the

control euchromatic region (Fig. 4D). Thus

loss of

piRNA upon paternal transmission correlates w

ith loss of H3K9me3 and Rhino

from the 20A

locus

Progeny that inherit

the 20A

X locus from their mothers receive two distinct contributions.

First, the locus itself might have different chromatin imprints when inherited maternally or

paternally. Secon

d, mothers deposit piRNAs into the oocyte, while paternal progeny do not inherit

piRNA from their fathers

(de Vanssay et al., 2012; Hermant et al., 2015; Le Thomas et al., 2014b)

Therefore, it is important to discriminate if the parent

origin effects of 20A

X depend on

inheritance of the genomic locus or cytoplasmic transmission of piRNAs to the next generation.

To discriminate between these possibilities, we designed two diff

erent crosses: in both crosses

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

the progeny inherited 20A

X paternally, however, in one cross the mothers also carried a copy of

the 20A

X locus which, however, was not transmitted to the progeny (Fig. 4E). The presence of

the

20A

X locus in mothers caused

GFP repression and piRNA generation in the progeny even

though the locus itself was not transmitted to the offspring. Indeed, progeny that inherited the

20A

X locus from their fathers but received cognate piRNA

from their mothers had similar level

of 20A

X piRNA

as progeny that simply inherited the 20A

X locus maternally.

It is worth noting

that maternally inherited cognate piRNAs were not able to convert the 20A[+] and 20A[

] loci to

piRNA

producing clusters, nor did

they

change the expression of GFP fro

m these reporters

(

data

not shown

)

These

result

indicate that

the

activity of 20A

X as a piRNA

generating locus requires

both the extended, multi

copy nature of the locus and

cytoplasmic inheritance of cognate piRNAs

through the maternal germline.

piRNA cluster

established over several generations

The finding that maternal inheritance of piRNAs is required for the activity of

the

20A

X locus

in the progeny prompted us to re

examine the observation that initially,

upon

establishment

the

20A

X transgenic flies

by RMCE, GFP was expressed in the germline but got repressed in later

generations (Fig. 5A). First, we established that the age of flies did not influence GFP silencing,

as repression was similar in young (5

day

) and old (30

day

) f

emales of the 14 months old stock

(Fig. 5A). Next, we analyzed small RNA profiles in ovaries of 20A

X flies 3, 11 and 21 months

after establishment of the stock. piRNAs and siRNAs derived from 20A

X were already present

at 3 months, but their abundance inc

reased 4

fold and 3.7

fold, respectively, by 11 months (Fig

5B). At 11 months 20A

X piRNAs also showed stronger sign of ping

pong processing as

measured by complementary piRNA pairs that overlap by 10nt (Z

scores at 3 months and 11

months

were

and 4.

, respectively) (Fig.

5C,

). No further increase in abundance of 20A

X piRNAs and ping

pong processing was observed when comparing 11 and 21 months

old stocks.

Thus, transgene

derived piRNA abundance is increasing over multiple generations after

transg

enesis and this increase correlates with repression of GFP in the germline. The findings

that 20A

X requires maternally

supplied piRNAs in order to generate piRNAs suggests that 20A

X was not active as a piRNA cluster in the first generation after establis

hment of this stock.

The loss of 20A

X’s ability to generate piRNAs upon paternal inheritance provides a unique

opportunity to explore

de novo

establishment of a piRNA cluster.

fter paternal transmission

the

progeny

(G0) generated very few piRNAs and siRN

As with levels similar to those of 20A[+] and

20A[

] flies (after normalization to transgene copy numbers)

(Fig. S5C)

. We monitored whether

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

20A

X can recover its ability to generate piRNAs in future generations upon continuous maternal

transmission (Fig.

5D). While no GFP repression occurred in

the

germline of G0, each subsequent

generation showed

decreased GFP expression,

until complete repression was observed in G8

(Fig. 5D

5B). Establishment of GFP repression over multiple generations also occurred i

the

male germline. Interestingly, ovaries of flies of intermediate generations (G2

G5) showed large

variation in the extent of repression between individual egg chambers (Fig. 5D, E

S5B

). For

example, in G2 almost equal fractions of flies showed normal

expression, complete silencing or

mixed

phenotype

(Fig. 5E)

We profiled small RNAs in ovaries of G3 flies after separating them into three groups (active,

mixed

and silenced) based on GFP expression as well as from ovaries of G0

and

G8 flies.

Reporter piRNA

levels

increased

over generations. Importantly,

the

three groups of

ovaries

with different levels of GFP repression had proportionally different levels of

reporter piRNAs

indicating that repression correlates with piRNA abundance (Fig. 5F).

Finally, we determined enrichment of H3K9me3 mark and Rhino protein on chromatin of

the

20A

X reporter in ovaries of G0 and G8 progeny. While Rhi and H3K9me3

were

lost

after paternal

transmission in G0, both Rhi and H3K9me3

were

enriched on 20A

X chromatin in G8 (Fig. 5G).

Overall, our results indicate that 20A

X gradually establishes its ability to generate piRNAs over

eight generations if maternal 20A

X piRNAs are tran

smitted to the progeny in each generation.

Maternal siRNA triggers activation of piRNA cluster

in the progeny

How

loci

like 20A

X start to function as dual

strand cluster and generate piRNA

remain

unclear. In addition to piRNA

, all studied dual

strand clusters generate siRNA. Although we

found that

transgenes containing

simple

inverted repeats produce exclusively siRNA

and no

piRNA

, it is still possible that

the

presence of cognate siRNA

might provide

the

initial trigger to

activate piRNA production.

To test the role of siRNA

piRNA cluster

activation

we crossed males carrying

the

20A

locus with heterozygous females carrying dsGFP constructs

harboring simple inverted repeat

that generate siRNA

(Fig. 6A). As

seen befor

, paternal transmission of 20A

X led to release of

GFP repression in

the

germline of

the

progeny (Fig. 6B). However, GFP remained repressed in

progeny that carried maternally

inherited dsGFP constructs. Remarkably,

similar level of GFP

repression was al

so observed in sibling progeny that did not inherit

the

dsGFP construct from

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

their mothers

indicating that the presence of cognate siRNA

the

mothers was sufficient to

activate repression in progeny (Fig. 6

To further explore the role of siRNA

we abrogated

the

siRNA pathway either in

the

mothers

or in the progeny using Ago2 mutation

(Fig. 6

)

In flies

Ago2 is required for

the

stability and

function of siRNA

and

its

mutation

completely disrupts

the

siRNA pathway

(Okamura et al., 2004)

GFP repression was strongly disrupted in the progeny of Ago2

deficient mothers

confirming that

it requires trans

generational cytopla

smic transmission of siRNA

(Fig. 6

)

. In contrast, Ago2

deficient progen

that inherited siRNA from their heterozygous mothers show

strong

GFP

repression.

Taken together, these results indicate that initiation of 20A

X repression in

the

progeny requires tr

ans

generational inheritance of cytoplasmic siRNA

while

the

siRNA pathway

is dispensable for maintenance of the repression.

To further analyze

the

effect of siRNA on activation of 20A

X we cloned and sequenced

small RNAs. As expected, progeny that inherited

the

dsGFP construct

had

high level of siRNA

targeting GFP (

fold increase compared to progeny with only paternal 20A

X) (Fig. 6

These abu

ndant siRNAs were restricted to the GFP sequence

which forms inverted repeat

the

dsGFP construct. Interestingly, sibling progeny that inherited

the

balancer chromosome

instead of dsGFP also show

moderate (3~7

fold) increase in GFP siRNA

level

compar

ed to

flies with

paternal 20A

only

Remarkably

both

the

progeny that inherited

the

dsGFP construct

and their siblings that inherited

the

balancer chromosome had elevated levels of piRNA

mapping

the 20A

X reporter

(Fig.

piRNAs mapping to the

GFP

sequence

were 15

fold

abundant

progeny that inherited

the

dsGFP constructs

and 5

fold more in

progeny with

the

balancer chromosome

when

compared to flies that only had

the

paternal 20A

However, even

more remarkably, both progenies that inherited dsGFP and those with

the

balancer chromosome

had

similar,

3 to 8

fold elevated levels of piRNA produced from regions of 20A

X that are not

targeted by GFP siRNA

This means that

maternal

contributed

GFP siR

that target a

portion of the 20A

X locus

were

sufficient to

induce piRNA generation from the entire 20A

X locus

in the progeny

. Furthermore,

using

ChIP

qPCR

we found

that cytoplasmic inheritance of GFP

siRNA

from the mother was sufficient to trigger a

ccumulation of H3K9me3 and Rhi

on chromatin

of paternally

inherited 20A

X (Fig. 6

). These results suggest that siRNA

are able to provide

the

initial trigger that converts

the

20A

X locus into

dual

strand piRNA cluster.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Discussion

siRNAs can provide initial trigger to activate piRNA biogenesis

As a system that protects the genome against selfish genetic elements, the piRNA pathway

has to be able to adapt to target new invader elements. Previous studies

have

revealed

mechanisms to s

tore information about genome invaders in piRNA clusters and to maintain piRNA

biogenesis through a feed

forward loop that involves trans

generational cytoplasmic transmission

of piRNAs. These inherited piRNAs guide deposition of the RDC chromatin complex

on clusters,

which is required for efficient piRNA expression in the progeny. However, the question of how the

pathway adapts to new TEs and starts creating piRNAs against novel threats remained

unresolved. One possibility for adaptation is integration of

new transposons into pre

existing

piRNA clusters, which would lead to generation of novel piRNAs, a process that has been

modeled experimentally

(Le Thomas et al., 2014b; Muerdter et al., 2012)

and observed naturally

(Khurana et al., 2011; Zhang et al., 2020)

. However, other studies suggest that entire new piRNA

generating regions can arise in evolution, providing another mechanism for acqu

iring immunity

against novel elements. Indeed, some piRNA clusters are active only in one but not other

D. virilis

strain

(Le Thomas et al., 2014a)

, suggesting their recent formation. Recent comprehensive

evolutionary analysis showed that piRNA clus

ter regions are extremely labile in

Drosophila

evolution, suggesting frequent acquisition and loss of piRNA clusters

(Gebert et al., 2021)

. Finally,

spontaneous formation of novel piRNA clusters from transgenic sequences has been observed

(de Vanssay et al., 2012)

. Though these findings suggested that new piRNA

producing genomic

regions that contain no sequence homology to pre

existing piRNAs can arise,

the conceptual

framework to explain this process was lacking.

The finding that siRNAs can activate piRNA biog

enesis provides an explanation for how

immunity to new transposons can be established through initial detection of new element by the

siRNA pathway, which then triggers a stable piRNA response. In contrast to the piRNA pathway

that relies on genetic and ep

igenetic memory

–

in form of active piRNA clusters

–

to recognize its

targets, the siRNA pathway uses a simple rule for self/non

self discrimination. Unlike normal

genes, transposons often generate both sense and antisense transcripts that form dsRNAs, whi

are recognized by Dicer and processed into siRNAs. Indeed, in addition to sense transcription

from their own promoters, transposons are often transcribed in antisense orientation from host

genes

’

promoters, for example if a TE is inserted into an intron

or the 3’UTR in opposite orientation

than the host gene orientation. In addition, recursive insertion of transposons into each other

creates inverted repeats that generate hairpin RNAs. The presence of sense and antisense

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

transcripts leads to dsRNA and si

RNA formation, providing a simple yet efficient mechanism to

discriminate mobile genetic elements from host genes. Indeed, the siRNA pathway has well

established functions in recognizing and suppressing both endogenous (transposons) and

exogenous (viruses)

invader genetic elements in all eukaryotic lineages, especially in TE

rich

plant genomes. In contrast, the piRNA pathway is restricted to Metazoa, suggesting that it is a

more recent evolutionary innovation. As both pathways target foreign genetic element

s, siRNAs

provide an ideal signal to activate piRNA biogenesis against novel invaders. Activation of piRNA

response by siRNAs can be compared to stimulation of the robust and long

lasting adaptive

immune response by the first

line innate immune systems

(Med

zhitov, 2007)

Our results indicate that siRNAs are important to jump

start piRNA cluster activity, but are

dispensable later on. Indeed, cytoplasmically

inherited siRNAs are sufficient to trigger piRNA

biogenesis in the progeny, while the zygotic siRN

A pathway is completely dispensable for this

process (Fig

. 6

). Consistent with siRNAs being dispensable for maintenance of piRNA biogenesis,

a previous study showed that the siRNA

are not required for the activity of an artificial piRNA

cluster (T1/BX2)

(de Vanssay et al., 2012)

On the other hand, increase in the level of antisense

transcripts was proposed

to be linked to spontaneous activation of BX2

(Casier et al., 2019)

suggesting a general role of antisense

RNA (and hence the siRNA pathway) in activation of piRNA

clusters. Overall, our results and the previous findings suggest a two

step model of cluster

activation (Fig

. 7

): during the first step siRNAs activate piRNA generation from cognate genomic

regions,

while during the second step continuous generation and maternal inheritance of piRNAs

reinforces piRNA biogenesis making siRNAs dispensable for cluster maintenance.

The precise molecular mechanism by which siRNAs trigger piRNA biogenesis remains to be

derstood. In yeast, siRNAs guide a methyltransferase complex to their genomic targets leading

to H3K9me3 deposition on chromatin

(Verdel et al., 2004)

. In

Drosophila

, this modification is

required for deposition of the RDC complex

and thus robust piRNA biogenesis

(Le Thomas et al.,

2014b; Mohn et al., 2014; Yu et al., 2015a; Zhang et al., 2014)

. However, in

flies siRNA repression

seem

to be restricted to target RNA cleavage in the cytoplasm, suggesting that they induce

piRNA biogenesis through a different mechanism. The cleavage of complementary transcripts by

siRNAs creates aberrant RNAs with 5’

monophosoho

rylated ends, which are good substrates for

the cytoplasmic piRNA processing machinery. Thus, we propose that siRNA

induced cleavage of

complementary transcripts generates substrates for piRNA processing. Cytoplasmic piRNA

processing, in turn, generates pi

RNAs that are loaded into the three piwi proteins, including the

nuclear Piwi protein

(Huang et al., 2017; Pandey et al., 2017; Rogers et al., 2017)

. As the nuclear

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Piwi/piRNA complex guide

s establishment of the H3K9me3 mark, this model explains how

cytoplasmic siRNAs are capable of inducing chromatin changes that are associated with piRNA

cluster activation.

Genomic requirements for piRNA cluster function

While our results indicate that siRNAs can trigger piRNA biogenesis from cognate genomic

regions, we and others also found that simple inverted repeats generate exclusively siRNAs and

not piRNAs. Thus, not every region that generates self

targeting siRNAs

turns into a piRNA

cluster, indicating that siRNAs might be necessary but not sufficient to convert a region into a

piRNA cluster. Although 20A

X is composed of the same sequences as the simple inverted repeat

construct, it contains ~ 10 copies of the tran

sgenic sequence. The only other artificial piRNA

clusters described, T1/BX2, also contain multiple tandem sequences. Thus, the extended length

and repetitive nature seems to be important for

de novo

establishment of clusters. These features

might be linked

to the important role that chromatin organization plays in piRNA cluster function.

Specifically, the extended repetitive organization might be required for maintenance of the RDC

chromatin compartment, which is essential for transcription and post

transcr

iptional processing of

piRNA precursors

(Andersen et al., 2017; Chen et al., 2016; ElMaghraby et al., 2019; Hur et al.,

2016; Klattenhoff et al., 2009; Mohn et al., 2014; Zhang et al., 2014)

. piRNA clusters are enriched

in the heteroch

romatic H3K9me3 mark and the RDC complex that binds this mark

(Klattenhoff et

al., 2009; Le Thomas et al., 2014b; Mohn et al., 2014; Yu et al., 2015a; Zhang et al., 2014)

. Rhi

is a paralog of HP1 that forms ‘classic’ heterochromatin domains, which depend on cooperative

interactions between multiple HP1 molecules and associated proteins (through interactions

between HP1 dimers b

ound to neighbo

ring nucleosomes

)

(Le Thomas et al., 2014b; Yu et al.,

2015a; Yu et al., 2018)

. Accordingly, extended length might be required for formation of a stable

HP1 chromatin compartment. Tandem repeats seem to be particularly prone to formation of

heterochromatin, though the underlying mechanisms is not completely clear

(Dorer and Henikoff,

1994)

. Similar to HP1, Rhi is capable of self

interactions through its

chromo and chromo

shadow

domains and these interactions are required for formation of RDC

compartments and the function

of piRNA clusters

(Le Thomas et al., 2014b; Yu et al., 2015a)

. Thus, extended length and tandem

repeats might be necessary to form a stable RDC chromatin compartment in the nucleus.

Establishment of such a region that is capable of maintaining RDC

rich heterochromatin might be

the first

step in developing piRNA immunity to a new element (Fig.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Acknowledgements

We thank members of the Aravin

and Fejes Toth

lab

for discussion and comments. We thank

Julius Brennecke

for providing the Rhino antibody. We thank Igor Antoshechkin (Caltech) for help

with sequencing. This work was supported by grants from the National Institutes of Health (R01

GM097363) and by the HHMI Faculty Scholar Award to AAA.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Figure legends

Figure 1.

Reporters integrated in uni

and dual

strand clusters have different

expression

and

effect

on cluster activity.

(A)

Scheme of reporter i

ntegration

into piRNA clusters

using recombinase

mediated

cassette exchange (RMCE)

to replace Minos

mediated

integration cassette

(MiMIC).

MiMIC

contain GFP sequence

but

lack

promoter. In contrast, the reporter encodes nuclear

EGFP expressed under control of the

Ubiquitin

gene promoter. RMCE can result in integration of

the reporter in either orientation

, whi

can be discriminated by genomic PCR.

(B)

rofile

the

dual

strand cluster 42AB

(top)

and uni

strand cluster 20A

(bottom)

Shown are profiles of uniquely

mapping piRNAs, positions of

the

putative promoter

(Pol II ChIP

peak)

of the

20A cluster and positions of the reporters.

20A cluster

piRNA

levels

are dramatically

reduced in ovaries of

20A[+] and 20A[

] flies. Number of piRNA reads mapped to

the

20A cluster

normalized to total piRNA

read count

(C)

Expression of reporters integr

ated in 20A and 42AB clusters.

Expression of GFP protein

and sense and antisense RNA

(by RNA FISH)

in ovaries of flies with

reporter

insertions in

the

20A and 42AB clusters.

In all reporter lines

GFP protein is expressed in somatic follicular cells,

howev

er,

GFP

is not

express

in germline (nurse

)

cells in

the

42AB reporter

line

. Sense GFP

RNA is produced in both 20A[

+] and 20A[

] flies and it is predominantly localized in the cytoplasm

of nurse cells. This cytoplasmic localization is different from

the

nuclear localization of exclusively

antisense RNA produced by

the

promoterless MiMIC sequence integrated in the same

site

(MI08972)

The

42AB reporter generates transcripts from

both

strands and both sense and

antisense RNA are localized in the nucleus, similar to transcripts produced by

the

promoterless

MiMIC sequence integrated in the same site

(MI07308)

. Scale bar is

20μm and 2μm for egg

chamber and single nurse cell

nuclei,

respectively.

(D) piRNAs and siRNA are generated

from the

reporter in

the

42AB cluster, but not

the

reporter inserted into 20A or

the

control (non

cluster) locus.

Number of the

small RNA reads

mapp

ing to the reporter were normalized to reads mapping to

the

42AB cluster. Error bars indicate

standard deviation of two biological replicates.

(E) Knockdown of

Zuc

eliminates 42AB reporter piRNA.

Shown are

piRNA

profiles

along the

reporter in ovaries of control flies (white GLKD,

black

) and upon GLKD of Zuc GLKD driven by

nos

Gal4 (

red

). Bar graph shows number of piRNA reads mapped to the reporter normalized to

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

flamenco

derived

piRNA

s, which are

not affected by

Zuc

knockdown. Error

bars indicate standard

deviation of two biological replicates.

(F)

Derepression of 42AB reporter expression upon

Zuc

GLKD.

Shown are

GFP protein

expression and FISH

signal

for

both strands of the reporter. Note appearance of sense GFP

transcripts in

the

ytoplasm upon

Zuc

GLKD.

Scale bar is 20μm and 2μm for egg chamber

and

single nurse cell

nuclei,

respectively.

(G)

Ovarian e

xpression of

the

20A cluster is suppressed

upon reporter integration.

(Left) 20A cluster transcripts w

ere

measured by RT

qPCR using

primers 439 bp upstream as well

as 296 bp and 1217 bp downstream of the reporter insertion site. Cluster transcripts were

normalized to

rp49

mRNA. Error bars indicate the standard deviation of three biological replicates.

(Right) Expression of 20A cluster

piRNA. piRNA reads uniquely

mapping to 20A cluster were

normalized to

total piRNA read count

. Error bars indicate standard deviation of two biological

replicates.

Figure 2.

An unusual

reporter

insertion

the

20A cluster is a dual

stand piRNA cluster

that generates piRNA

Rhi

dependent manner

(A)

The

20A

X reporter is silenced in the germline after several months of maintaining this

line.

Top:

GFP protein and sense RNA expression in ovaries of

20A[+] and

20A

flies

at different

times after establishment of lines by RMCE.

Bottom:

GFP protein and sense RNA expression in

ovaries of 20A[+], 20A[

], 20A

X and 42AB[

] flies

14 months (20A[+], 20A[

]

and 20A

X) and 1

month (42AB[

])

after establishment of lines.

Scale bar is 20μm and 2μm for egg chamber

and

single nurse cell

nuclei,

respectively.

(B)

20A

X generates piRNA

Zuc

dependent manner

. Shown

are

profiles of piRNA

mapping to

the

20A

X reporter in control (

white

GLKD) and upon

Zuc

GLKD driven by nos

GAL4.

20A

X piRNA

were normalized to piRNA reads mapping to

the

flam

cluster. Error bars indicate

standard deviation of two biological replicates.

(C)

Zuc

knockdown

rel

eases

20A

repression in

the

germline.

Shown are expression of GFP

protein and sense RNA in control (

white

GLKD) and upon

Zuc

GLKD driven by nos

GAL4. Note

that

though GFP protein is not expressed in control flies, sense RNA is present in the nucleus.

Zuc

GLKD cause

accumulation of RNA in

the

cytoplasm and

leads to GFP

protein expression.

Scale bar is 20μm and 2μm for egg chamber

and single nurse cell

nuclei,

spectively.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

(D) 20A

X, but not other 20A reporters are enriched in Rhino and

the

H3K9me3 mark.

Left

Rhino ChIP

seq profiles (mean of two biological replicates) on 20A

X, 20A[+] (black) and 20A[

]

(yellow) reporters. Right

Rhino and H3K9me3 enrichment on

20A reporter

(Ubi and GFP region)

as well as control

locus

(chr

2L: 968088

968187, dm6) were determined by ChIP

qPCR. Only

the

20A

X reporter is enriched in Rhi and H3K9me3.

ChIP signals

ere

normalized to

the

rp49

gene. Error bars indicate the standard d

eviation of three biological replicates.

(E)

Rhino knockdown reduces 20A

X piRNA and releases its silencing in the germline

. Left

Shown are piRNA

profiles over 20A

X and 42AB reporters in ovaries of control flies (

white

GLKD)

and upon

Rhi

GLKD driven by nos

GAL4. Reporter piRNA

were normalized to piRNA reads

mapping to

the

flam

cluster. Error bars indicate standard deviation of two biological repl

icates.

Right

GFP protein expression in 20A

X and 42AB ovaries upon

Rhi

GLKD. Scale bar is 20μm.

Figure 3.

The

20A

X locus contains

rearranged,

multi

copy

reporter

sequences

(A)

Determining

reporter orientation by genomic PCR.

Positions of the primers a

re shown.

20A

X has the same minus

strand orientation as

the

20A[

] reporter.

(B)

20A

X is located in a single site in

the

20A region.

DNA FISH on salivary gland polytene

chromosomes

was

performed using probes against

the

Ubi

GFP reporter (green, Cy488) an

the

20A cluster (red, Cy594).

Probes against the reporter

detected

two locations

: one co

localizes

with

the

20A cluster signal on

the

X chromosome,

the other

signal is localized on chromosome

3L where

the

native ubiquitin gene (

Ubi

p63E

) resides (chr3L:3899259

3903184, dm6).

(C)

Verification of the

20A

X insertion

site by CRISPR deletion.

The region of

the

20A cluster

that includes the

reporter insertion

site was deleted in

the

20A

X line using CRISPR/Cas9. Shown

are positions of gui

de RNAs and primers used to verify the deletion

using

genomic PCR and

Sanger sequencing

(top)

Detection of reporter sequences using qPCR of genomic DNA (left)

shows absence of reporter sequences in flies with the deletion.

No GFP expression is detected

flies with the deletion

(right)

. Scale bar is 20μm.

(D)

20A

X includes plasmid backbone sequence

(Top) Profiles of

whole

genome

DNA

seq

reads over the sequence of the plasmid used

for

RMCE and

the

flanking 1

genomic sequences

in flies with different 20

A reporters. 20A

X, but not 20A[+] and 20A[

] flies harbor plasmid

backbone sequence

, which is

normally

not integrated during RMCE. Bar

raph show

the ratio of

reads derived from the reporter to reads from flanking genome sequence

s in different reporter

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

lines

. (Bottom) piRNA profiles over

the

reporter sequence in 20A

X and 42AB reporter flies. Note

that 20A

X generates more piRNA even after normaliz

ation to sequence copy number

. Profiles

represent a mean of two biological replicates and were normalized to total number of piRNA reads

in each library.

(E)

20A

X contains multiple copies of reporter sequence.

Different portions of the reporter

sequence w

ere measured by genomic qPCR in flies with 20A reporters as well as

control

line

in which the

reporter

was

integrated into non

cluster region (chr

3L:

575

013, dm6). qPCR

values were normalized to

the

rp49 gene region. Fold

differences compared to homo

zygous

flies

with the

control reporter are indicated above the bars. Error bars indicate standard deviation of

three biological replicates.

(F)

eporter sequence r

earrangements in 20A

Three abnormal sequence junctions were

detected in

the

20A

X sequence

by DNA

seq,

Sanger sequencing PCR

amplified

genomic

DNA

(G)

Deletion of

the

20A cluster promoter does not affect

activity of

20A

The putative

promoter of

the

20A cluster (determined by

prominent

Pol II

ChIP

seq peak) was deleted using

CRISPR/Cas9 in wild

type and 20A

X flies. RT

qPCR shows

dramatic

reduction

of 20A cluster

transcripts

in flies with

the

deletion

indicating that

disrupts

the

cluster

promoter

. RT

qPCR

primers

were

positioned 2030 and 3769 bp

downstream of

the

putative

promoter (upstream and

downstream of

the

reporter insertion, respectively). Error bars indicate standard deviation of three

biological replicates. GFP remained repressed in

the

germline of 20A

X flies

upon

promoter

deletion

indicating that piRNA

mediated

silencing remains unaffected. Scale bar is 20μm.

(H)

Inverted repeat reporters generate abundant endo

siRNA, but not piRNA.

Scheme of

the

inverted repeat dsGFP reporter. The dsGFP reporter harbors the same sequence fragments as

other reporters, however,

GFP sequence forms

inverted repeat that generate

hairpin

dsRNA after transcription from

the

ubiquitin promoter.

Shown are s

mall RNA

profile

s (19

nt)

along the

reporter sequence in ovaries of flies with

reporter

integration into

the

20A cluster in

both

orientations as well as

integration into the

control non

cluster region (

66A6,

chr

3L:

7575013,

dm6).

Shown on

the right are size distributions of reporter mapping small RNAs and nucleotide

composition of 23

nt (piRNA size range) RNA

s. Size profile and nucleotide bias of 20A

X small

RNAs are shown for comparison. dsGFP reporters generate abundant endo

siRNA

exclusively

from

the

inverted repeat sequence.

The m

iniscule amount of 23

29nt small RNA generated

from

dsGFP reporters

does

not

show a 1

bias

expected from genuine piRNAs.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Figure 4. piRNA

induced repression of the 20A

X reporter depends on maternal

transmission of cognate piRNAs

(A)

20A

X r

epression is

released

after paternal transmission.

Scheme of crosses to test

effects of

maternal and paternal transmission of 20A

X. Note that the genotype of

the

progeny

from

the

two crosses are identical. Shown are GFP protein and sense and antisense GFP RNA

expression

in ovar

ies

and

GFP protein expression in

testes of

the

progeny.

Scale bar is 20μm

and 2μm for egg chamber and single nurse cell

nuclei,

respectively,

and

20μm for testis.

(B)

GFP RNA

expression increases

after paternal transmission of 20A

. RT

qPCR of GFP

RNA in ovaries of progenies

from the

two crosses in (A) were pe

rformed with random

hexamer

and oligo dT primers and normalized to

rp49 mRNA

level. Error bars indicate the standard

deviation of three biological replicates.

(C)

20A

X piRNA

level

drops after paternal inheritance

. Shown are piRNA and siRNA profiles

along

the

reporter sequence in ovaries of

progeny from the

two crosses shown in (A). Bar graphs

on the right

show number of piRNA and siRNA reads mapping to

the

reporter normalized to total

piRNA and siRNA reads, respectively. Error bars indicate standard deviat

ion of two biological

replicates.

(D) Rhino and H3K9me3

are lost

on 20A

X chromatin after paternal transmission.

The levels

Rhino and H3K9me3 on chromatin were measured by

ChIP

qPCR using primers against

the

pUbi and GFP regions as well as

control non

cluster region (chr

2L: 968

088

–

968

187, dm6)

and normalized to

the

rp49

locus. H3K9me3 and Rhino

levels

drop after paternal transmission to

levels similar to that of

the

control region. Error bars indicate the standard deviation of three

biological replicates.

(E)

ytoplasmic

piRNA

inheritance is sufficient for repression of paternally transmitted

20A

Scheme

crosses to test

the

effect of maternal cytoplasmic

piRNA

inheritance on

repression of paternally transmitted 20A

X. Progeny of

both crosses inherit

the

20A

X locus from

their fathers and are genetically identical. The only difference is cytoplasmic inheritance of small

20A

derived

RNAs from their mothers

, which

is sufficient to restore both GFP repression and

piRNA generation in

the

germline of the

progeny. Number of piRNA reads mapping to the reporter

was

normalized to total piRNA reads mapping to the genome in each library. Error bars indicate

standard deviation of two biological replicates. Scale bar is 20μm.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

Figure 5. piRNA cluster is established over several generations.

(A)

Establishment of 20A

X repression over several generations

. Shown are GFP protein

and RNA expression in ovaries of 20A

X flies 3 and 14 months after establishment of this line by

RMCE. Yo

ung (5

days

after hatching) and old (30 days after hatching)

flies

show no difference in

expression. Scale bar is 20μm and 2μm for egg chamber and single nurse cell

nuclei,

respectively.

(B) 20A

X piRNA

level

increase

over several generations.

Shown are

varian reporter

mapping

piRNA and siRNA

profiles

3 and 11 months after establishing the

20A

line. Bar graph

shows the levels of 20A

X piRNA

and siRNA

normalized to total number of piRNA and siRNA

reads in each library. Error bars indicate standard devi

ation of two biological replicates.

(C)

ing

pong signature

of 20A

mapping piRNAs increases

over

several generations.

Shown are Z

scores indicating ping

pong signature (10 nt distance between 5’

ends of

complementary piRNA

)

20A

X piRNA

. Z

scores of

DOC transposon piRNA are shown for

comparison. Error bars indicate standard deviation of two biological replicates.

(D)

Recovery of 20A

X repression after paternal transmission.

Left

Scheme of cross

monitor 20A

X after its paternal transmission. After paternal transmission in the first cross (G0),

20A

X is inherited maternally in each subsequent generation (G1

G8)

Right

Expression of GFP

in selected generations in ovaries and testes.

Several

gen

erations (G

) show variable GFP

expression between individual flies and within each fly ovary. Scale bar is 100μm and 20μm for

ovary and testis

respectively.

(E)

Accumulation of 20A

repression

over

several generations.

In each fly

germline

GFP

expression was determined and assigned one of three values: ‘silenced’ indicates complete lack

of expression, ‘active’ indicate expression in

the

majority of germline

nuclei

, while ‘

mixed’

indicate

variable expression between individual egg chambers wi

thin

the same

ovary. Plotted

the

fraction of ovaries with corresponding expression pattern in each generation from G0 to G11.

The

experiment was repeated three times and

100 ovaries were counted in each generation

in each

replica

. Error bars indicate th

e standard deviation of three biological replicates.

(F)

Accumulation of 20A

piRNA

over

several generations.

Shown are

profiles of 20A

piRNA

in different generations. In G3

ovaries were separated in three

groups

according to GFP

expression as described in (E) and for each group independent small RNA libraries were prepared.

Bar graph

(bottom)

shows

20A

X piRNA

levels normalized to total piRNA reads

in each library.

Error bars indicate standard deviation of two bi

ological replicates.

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

(G) Accumulation of Rhino and H3K9me3 on 20A

X chromatin.

Rhino and H3K9me3

levels

on chromatin were measured by

ChIP

qPCR in ovaries of G0 and G8 generation using primers

against

the

pUbi and GFP regions as well as

control

non

clus

ter region (chr

2L: 968

088

–

968

187, dm6) and normalized to

the

rp49

locus. Error bars indicate the standard deviation of

three biological replicates.

Figure 6. Cytoplasmic

inherit

ance

siRNA

activates

piRNA biogenesis

in the progeny

(A)

Crossing scheme

to test the role of siRNA

triggering

20A

X repression

The 20A

reporter

is inherited from the father

, while mothers harbor

inverted repeat

dsGFP

construct

that generates siRNA.

rogen

y that inherited the

dsGFP locus

and those

that

did not were

compared

(B) Cytoplasmic inheritance of siRNA

trigger

20A

repression

in the progeny.

GFP

expression in progenies of each genotype was assessed by fluorescent microscopy (top) and

assigned one of the three values

as described in Fig. 5E (bottom). Two independent dsGFP loci

in 20A and in a locus in band 66A were analyzed. Maternal siRNA trig

gered repression of 20A

in the progeny independently of inheritance of siRNA

generating locus or the genomic position of

the siRNA

generating locus.

Scale bar is 20μm.

N indicates the number of ovaries analyzed for

each genotype.

GFP expression in

progen

ies

of each genotype

was assessed by fluorescent

microscopy

(

)

Crossing scheme

to test the role of Ago2 in 20A

X repression.

rosses are similar to

crosses shown in

(A)

except of

the

presence Ago2 mutation either in mothers or the progeny.

(

)

Triggering

20A

X repression

by trans

generational siRNA

depends on

functional

siRNA pathway in

the

mothers, but not

the progeny.

nalysis of

GFP expression in

progenies of crosses shown in

(

)

(

)

Cytoplasmic inheritance of siRNA

activates

piRNA

production in the progeny

Shown

are profiles of

piRNA

and siRNA

mapping to the reporter in ovaries of

progenies of crosses

shown in

(A).

Bar graph shows 20A

X piRNA and siRNA

levels

normalized to

those generated

from

the

42AB

cluster. Error bars indicate standard deviation of two biological replicates.

Statistical significance is estimated by two

tailed Student’s t

test; *p<0.05, **p<0.01.

(

)

Cytoplasmic

siRNA

inheritance is required for accumulat

ion of

H3K9me3 and

for

Rhino

recruitment.

Rhino and H3K9me3

level

on chromatin

of 20A

was

measured

by ChIP

qPCR in

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint

progenies of crosses shown in

(A)

. ChIP signal in 20A

X is normalized to

the

rp49

locus.

Error

bars indicate the standard

deviation of three biological replicates.

Figure

The model for siRNA

triggered activation of piRNA immunity

CC-BY-NC-ND 4.0 International license

available under a

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

The copyright holder for this preprint

this version posted February 9, 2022.

;

https://doi.org/10.1101/2022.02.08.479612

doi:

bioRxiv preprint