Article
DNA probes were hybridized with Perfect Hyb Plus (Sigma) at 42 °C
overnight. The primer sequences for genomic PCR, real-time PCR and
DNA probes for Northern blotting are listed in Supplementary Table 1.
Immunofluorescence
Stable cells expressing GFP-tagged BmArmi were treated with 100 nM
Mitotracker Red CMXRos (Cell Signaling) at 27 °C for 1 h. After fixing
with 4% paraformaldehyde at room temperature for 10 min, the cells
were permeabilized with 0.3% Triton X-100 for 5 min and incubated with
PBS supplemented with 1% BSA (Sigma) and 0.1% Triton X-100 at room
temperature for 1 h. Then, the cells were incubated with anti-BmGasz
antibody (1:400) in PBS supplemented with 1% BSA (Sigma) and 0.1%
Triton X-100 at 4 °C overnight. Alexa Fluor 647 donkey anti-rabbit IgG
antibody (Thermo Fisher/Invitogen) was used as the secondary anti-
body. Images were captured using Olympus FV3000 confocal laser
scanning system with a × 60 oil immersion objective lens (PLAPON
60XO, NA 1.42, Olympus) and processed FV31S-SW Viewer software
and Adobe Photoshop Elements 10.
Construction of plasmid-based randomized sequence library
and sequencing
A DNA fragment containing an N35 random sequence embedded
between two piRNA target sites was amplified by PCR using 3rand-const
primers and synthetic N35-containing DNA oligos (IDT) (Siwi/BmAgo3-
3endRAND50-double) as the template (Supplementary Table 1). The
PCR products were digested with BamHI and HindIII, and cloned into the
BamHI/HindIII sites of pIEx-4 vector (Millipore/Novagen). The plasmid-
based library expresses 35-nt of random sequence flanked by target
sites for an abundant BmAgo3- (Fig. 4a, Extended Data Fig. 5d left, and
Supplementary Note 2) or Siwi-dominant piRNA (Extended Data Fig. 5d
right)^31. The library was transfected into (1) Tri-KO cells (mock) or Tri-
KO cells overexpressing either (2) wild-type BmZuc and BmArmi, to
enhance BmZuc-mediated cleavage, or (3) the catalytic mutant BmZuc
(HN), to repress BmZuc-mediated cleavage. Transcripts derived from
the library are expected to be sliced by complementary piRNAs bound
to BmAgo3/Siwi, loaded into Siwi/BmAgo3 via the ping-pong pathway,
and cleaved by BmZuc within the downstream randomized region.
We sequenced 20−50 nt small RNAs bearing the common sequence
in their 5′ region, and restored the original sequences downstream of
the obtained small RNAs by using the sequence data of the plasmid
library. The variation of the randomized region was estimated to be
215,879 for Siwi and 178,735 for BmAgo3 based on the number of dis-
tinct sequences with RPM > 0.25 in the reference libraries. The rand-
omized sequence libraries for reference were constructed by PCR using
a common reverse primer (plasmid-Hind-randR) and a specific forward
primer (plasmid-Siwi-randF-index12 or plasmid-Ago3-randF-index19,
shown in Supplementary Table 1). The libraries were sequenced by the
Illumina Hiseq 3000 platform to obtain 100-nt paired-end reads using
a custom primer containing the consensus sequence (random plasmid
sequence primer, shown in Supplementary Table 1) and an index read
sequence primer provided by the manufacturer.
Small RNA library preparation
Small RNA libraries were prepared from 20−50 nt total, Siwi-bound,
or BmAgo3-bound RNAs, according to the Zamore lab’s open pro-
tocol (https://www.dropbox.com/s/r5d7aj3hhyaborq/)^37 with some
modifications. The 3′ adapter was conjugated with amino CA linker
instead of dCC at the 3′ end (GeneDesign) and adenylated using 5′
DNA adenylation kit at the 5′ end (NEB). To reduce a ligation bias,
four random nucleotides were included in the 3′ and 5′ adapters
[(5′-rAppNNNNTGGAATTCTCGGGTGCCAAGG/amino CA linker-3′) and
(5′-GUUCAGAGUUCUACAGUCCGACGAUCNNNN-3′)] and the adapter
ligation was performed in the presence of 20% PEG-8000^38 , except for
Fig. 1d and Extended Data Fig. 2l, 3c, 5a. After the 3′ adapter ligation at
16 °C for ≥ 16 h, RNAs were size-selected by urea PAGE. In Fig. 1d, and
Extended Data Fig. 2l, 3c, 5a, the 3′ and 5′ adapters without the four
random nucleotides were used. In Fig. 4e and Extended Data Fig. 5e,
the 5′ adapter without the four random nucleotides was used. For RNA
extraction from polyacrylamide gel, ZR small-RNA PAGE Recovery Kit
(ZYMO Research) was used. For small RNA library preparation from the
randomized sequence library (Fig. 4e, Extended Data Fig. 5a, e), specific
forward primer (piRNA-Siwi or BmAgo3-randF, shown in Supplemen-
tary Table 1) was used in PCR to selectively amplify the plasmid-derived
transcripts. Small RNA libraries were sequenced using the Illumina
HiSeq 4000 platform to obtain 50-nt single-end reads.Sequence analysis of endogenous small RNAs
After removal of adapter sequences by cutadapt^39 , 20−45 nt reads with-
out any ambiguous bases were mapped to sequences of defined piRNA
loci^3 with Bowtie^40 allowing one mismatch. Sam files were converted
to bam files by SAMtools^41 and then to bed files by BEDTools^42. Length
and 5′-end position for each piRNA were obtained from bed files using
custom R programs. To determine Siwi- and BmAgo3-dominant loci,
Siwi-immunoprecipitated and BmAgo3-immunoprecipitated libraries
treated with NaIO 4 from naive BmN4 cells were compared and defined
Siwi-dominant piRNA loci (RPM (Siwi-IP) > RPM (BmAgo3-IP), n = 1,946)
and BmAgo3-dominant piRNA loci (RPM (BmAgo3-IP) > RPM (Siwi-
IP), n = 1,259) (Extended Data Fig. 3a). For mouse small RNA analysis,
piRNA loci were defined from the deep sequencing data by Gainetdinov
et al.^2. In brief, 1st–23rd sequence of each read was extracted and the
frequency of each 1st–23rd sequence was calculated in each library
(SRR7760309, SRR7760310, SRR7760317, SRR7760318, SRR7760321,
SRR7760322, SRR7760343, SRR7760344, SRR7760347, SRR7760348,
SRR7760369, SRR7760370, SRR7760373, SRR7760374, SRR7760377,
SRR7760378). Sequences that are abundantly found (RPM >10) in at
least one library, 36,431 in total, were selected and mapped to the mouse
genome (GRCm38.p5) to define representative piRNA loci in pachytene
spermatocytes.Sequence analysis of randomized plasmids and small RNAs
Sequences in the randomized region of the plasmids were extracted
and the frequency of each distinct sequence was calculated. Those
sequences with RPM >4 (80,966 species for Siwi and 70,442 species for
BmAgo3) were used as references. For analysing small RNAs derived
from the randomized libraries, adapters were trimmed by cutadapt^39
and 20−45 nt reads without any ambiguous bases but with a 15-nt com-
mon sequence from the plasmids were mapped to the randomized
sequence references with Bowtie^40 allowing no mismatch. Sam files
were converted to bam files by SAMtools^41 and then to bed files by BED-
Tools^42. Length and 5′-end position for each small RNA were obtained
from bed files using custom R programs. To analyse the nucleotide
frequencies relative to the 3′ end of small RNAs, we selected sequences
with peak lengths of 31−44 nt, corresponding to the size range of type-E
pre-piRNAs.Calculation of the similarity scores with weighted BmZuc motifs
The top 15 highest-frequency nucleotides in the randomized Siwi-
immunoprecipitated or BmAgo3-immunoprecipitated libraries
(Fig. 4e) were chosen to define the ‘weighted BmZuc motif ’ of 17-nt
long (from –12 to +4, 0 = predicted BmZuc cleavage site) for Siwi or
BmAgo3 (Extended Data Fig. 5g, upper), with each nucleotide having
a score of the log 2 value of the nucleotide frequency at that position,
normalized to the randomized sequence references. The similarity
score was calculated by summing up the weighted BmZuc motif if a
nucleotide at a given position matches to the nucleotide at the cor-
responding position in the defined BmZuc motif for Siwi or BmAgo3
(Extended Data Fig. 5g, lower), by sliding the 17-nt window on each 27-nt
sequence in the extracted genomic sequence pools or the control shuf-
fled pool, using custom R programs. The control shuffled sequences
have the average nucleotide composition of the silkworm genome