The piRNA production-related proteins used in this study

The 1:1 dimer structure prediction by AlphaFold2 for piRNA-related proteins

(A) Heatmaps of the prediction confidence scores (pcScore, Green), pTM values (Blue), and ipTM values (Red) provided by AlphaFold2. The 20 types of proteins are aligned from top to bottom and left to right in the same order. Boxes on diagonal line represent homodimers.

(B) Scatter plot of the pcScores. The scores from first and second predictions for each heterodimer pair are plotted on X and Y axis, respectively.

(Ci ∼ xii) The predicted 3D structures (top panels) and the Predicted Aligned Error (PAE) plots (bottom panels) for each candidate heterodimers scoring above 0.6. The PAE plot displays the positional errors between all amino acid residue pairs, formatted in a matrix layout.

(D) Co-immunoprecipitation assays using tagged proteins to verify interactions between specific pairs: Spn-E_Squ (i), Aub_Vret (ii), Spn-E_BoYb (iii), BoYb_Shu (iv), and Me31B_Vret (v). Single transfected cells expressing only Myc-tagged but not Flag-tagged proteins are used as negative controls for each set. Box and whisker plots show the intensity ratio between immunoprecipitated and input bands (n=3).

The screening for the interacting proteins (prediction confidence score, pcScore > 0.6)

Interaction between Spn-E and Squ

(A) Schematic of Spn-E domain structures defined in SMART44. Boxes (α-helix: orange) and arrow (β-sheet: green) for Squ structure. The predicted interacting regions between Spn-E and Squ are indicated in gray boxes. Tej interaction site of Spn-E is also shown16.

(B) The predicted five models of heterodimer of Spn-E (in gray) and Squ (in magenta). Spn-E molecules in all five models are superimposed.

(C) 3D structure of the Spn-E_Squ dimer colored by Spn-E domains as indicated in (A), with Squ in magenta. The enlarged image of the interface indicated by box is also shown.

(D) The predicted salt bridges at the interface, with Spn-E in gray and Squ in magenta. The residues forming salt bridges are depicted in stick model.

(E) Co-immunoprecipitation assay using S2 cell lysate to examine the interaction between Myc-Spn-E and Flag-Squ mutant (4A) whose salt bridge-forming residues are mutated to Ala. S2 cells expressing Myc-Spn-E alone is used as a control. The ratios of the band intensity (IP/input) are shown in a box and whisker plot (n=3).

(F) The heterotetramer model of Spn-E_RNA_Squ_Tej predicted by AlphaFold3. Spn-E is shown as a space filled model in gray, Squ in magenta, Tej in cyan, and RNA in yellow. The model on the left is rotated 180° in the Y axis to produce the image on the right.

Spn-E and Squ interact in Drosophila ovary

(A) Western blotting analysis using anti-Squ antibody reveals a specific band at the expected size (approximately 28 kDa) for endogenous Squ in Drosophila ovarian lysates of the heterozygous control. This band is absent in the transheterozygoute, squPP32/HE47.

(B) Immunostaining of Drosophila egg chambers with anti-Squ antibody and anti-mKate2 (mK2) antibody demonstrates colocalization of Squ and Spn-E-mK2 in nuage, a perinuclear granule in germline cells. The enlarged images of nuclei are shown in the panels below. Scale bars: 10 μm (top row), 2.5 μm (enlarged images)

(C) Immunoprecipitation of the endogenous Squ from ovarian lysate revealed the interaction with Spn-E protein. Proteins were detected by western blotting analysis using the specific antibody for each protein. The negative control was performed without anti-Squ antibody (beads only).

Squ- and Tej-interacting proteins predicted by AlphaFold2

(A i-iii) The predicted dimer structures (top) and Predicted Aligned Error (PAE) plots (bottom) of Mei-W68 in blue and Squ in magenta (i), CSN3 in green and Squ in magenta (ii), Pka-C1 in orange and Tej in cyan (iii). The PAE plot displays the positional errors between all amino acid residue pairs, formatted in a matrix layout.

(B i-iii) Co-immunoprecipitation assays using tagged proteins to verify interactions between specific pairs: Mei-W68_Squ (i), CSN3_Squ (ii), and Pka-C1_Tej (iii). Single transfected cells expressing only Myc-tagged but not Flag-tagged proteins are used as negative controls for each set. Box and whisker plots show the intensity ratio between immunoprecipitated and input bands (n=3).

The binding candidates predicted by AlphaFold2

Screening for Piwi-interacting proteins in Drosophila proteome

(A) Pie chart displaying the distribution of pcScores from the AlphaFold2 screening for Piwi-interacting proteins among those encoded by Drosophila genome.

(Bi∼ v) The predicted dimer structure (top) and PAE plots (bottom) for the Piwi and the binding candidates in red: Arx (i), Hen1 (ii), CG33703 (iii), Twf (iv), and Brn (v). Piwi is shown in the same colors as Supplemental Fig. S5A.

(C) Co-immunoprecipitation assays using tagged proteins to verify interactions between Piwi and the binding candidates, Twf and Brn. Single transfected cells expressing only Flag-Piwi is used as negative control. Box and whisker plots show the intensity ratio between immunoprecipitated and input bands (n=3).

Piwi-interacting proteins predicted by AlphaFold2 (Socre >= 0.75).