vret is a gonad-specific, Tudor-domain protein required for germline differentiation and ovarian soma organization. (A) vret encodes a protein of 691 amino acids that includes two Tudor domains in the C terminus. vret stop codon mutations are indicated above and point mutations below the diagram of the protein. (B) In vret^148-60 heterozygotes, as in wild type, two to three GSCs with round spectrosomes are located at the anterior tip of the germarium (arrows), adjacent to the somatic niche, which is composed of terminal filaments, cap cells and ISCs (arrowheads). A cystoblast (CB) and a differentiating cyst (dashed line) are indicated. The germline is labeled with Vasa and somatic cell membranes and spectrosomes with 1B1. 1B1 staining in the GSC and CB marks single spectrosomes whereas in differentiating cysts it stains the fusome that connects the germ cells. (C) vret³⁹ heterozygote shows a normal egg chamber with an oocyte specified at the posterior, labeled by Orb, and surrounded by a somatic follicle cell epithelial layer. (D) Germarium filled with single, undifferentiated germ cells in vret^148-60 ovarioles. (E) Defective egg chambers with mislocalized oocytes and incomplete encapsulation by follicle cell layers in vret³⁹, a weaker vret allele. (F) vret³⁹ mutant exhibits an abnormal number of nurse cells and a mislocalized oocyte. (G) Multilayered follicle cells partially envelope nurse cells in vret^148-60 mutants. (H) Vret protein is detected in an immunoblot of wild-type adult ovary and testis lysates. β-Tubulin was used as loading control. (I) Vret antibody shows specificity in ovarian lysates of vret mutant alleles. Vret full-length protein (around 80 kDa) is not detectable in vret^148-60. A weak but specific band of smaller molecular weight is detected in vret^148-60, consistent with molecular data. A lower, non-specific band is observed in all samples (arrow). α-Tubulin was used as loading control.

vret is required in both soma and germline for the production of progeny. (A,B) vret is not required autonomously in germ cells for GSC differentiation. vret^148-60 mutant clones are marked by the absence of GFP. (A) A vret mutant GSC clone produced normally differentiating progeny (outlined), even 5-7 days after clone induction. (B) A budding vret germline cyst (dashed outline in germarium) shows a properly specified and localized oocyte (indicated by Orb staining). A stage 6 egg chamber with a large vret follicle cell clone (outlined) shows normal follicle cell morphology and polar cell specification (arrow), marked by anti-Fas III in addition to a properly specified oocyte. (C,D) Oogenesis defects are rescued in vret^148-60 mutants by expression of a UASp-vret-myc transgene using the somatic driver c587-Gal4 (D), compared with vret^148-60 mutant ovaries in the absence of the driver (C). GFP expressed by a UASp-gfp transgene was used to mark c587-Gal4 somatic cell populations. (E,F) Programmed cell death is detected by cleaved Caspase-3 staining in ovaries. Somatic (and possibly also germline) cells show increased Caspase-3 staining in vret mutant germaria (F) compared with heterozygotes (E).

vret is required for suppressing transposon expression in both the germline and somatic cells of the ovary. (A) A weak vret allelic combination, vret³⁹/Df, shows germline differentiation defects and abnormal somatic morphology. (B) flam^BG/flam^KG mutants show defective germline and soma differentiation phenotypes, similar to vret mutants. (C,D) In flam^BG/flam^KG mutant ovaries, somatic cells die as revealed by Caspase-3, compared with flam heterozygotes. (E,F) Cell death is also detected in piwi¹/piwi² mutant ovaries. (G,H) gypsy expression is silenced in vret heterozygote (G), whereas in vret mutants (H), gypsy is derepressed in somatic cells, marked by β-gal. (I) Venn diagram showing overlap between transposable elements regulated in vret, piwi and aub mutant ovaries. (J) Profile of transposons regulated in vret mutant ovaries compared with piwi and aub mutant ovaries, shown as a heat map. All homozygous or transheterozygous mutant samples were normalized to their respective heterozygotes. Note that for some transposons, more than one probe was represented on the microarray (indicated with a superscript number).

vret is required for Piwi and Aub protein accumulation. (A-D′) vret ^148-60 germline and somatic mutant clones were generated at L3 and marked by the absence of GFP. (A-B′) Germline and somatic Piwi protein expression is dramatically reduced in vret mutant clones. (B-B′) An egg chamber with all follicle cells mutant for vret shows a dramatic reduction in somatic Piwi expression (outlined). (C-C′) Aub germline expression is severely affected in vret germline clones, whereas AGO3 expression (D-D′) is unchanged. GFP and the proteins tested are shown in separate channels for clarity. (E) Protein levels of Vret, Piwi, Aub, AGO3, Armi and Vasa are shown for wild-type (w¹¹¹⁸), vret^148-60/TM6 and vret^148-60 mutant ovaries. α-Tubulin was used as loading control.

Vret interacts physically with Piwi proteins and regulates their stability. (A-D) Post-transcriptional effects of Vret on Piwi and Aub proteins tested by expression of piwi and aub transgenes containing the UASp heterologous promoter. (A,C) Piwi and Aub proteins were expressed from UASp-HA-piwi and UASp-aub-gfp transgenes, respectively, in the germline. (B,D) Piwi and Aub protein levels are strongly reduced in vret mutant ovaries, shown by endogenous and transgenic expressions. (E) Physical interaction of Vret with piRNA components. nos-Gal4-VP16 was used to express a UASp-vret-myc transgene and immunoprecipitation performed with Vret in ovarian lysates using an anti-Myc antibody. β-Tubulin was used as control. (F) Ovarian cell compartments can be separated by subcellular fractionation. Fibrillarin, a nuclear marker, and Orb, a cytoplasmic marker, were used as controls for efficient fraction separation. Homogenate (H), cytosolic (C) and nuclear (N) fractions are indicated. In vret^148-60 heterozygous ovaries, Vret appears in the cytosolic fraction whereas Piwi localizes to both nuclear and cytosolic fractions. In vret homozygotes, Piwi nuclear fraction is more affected than the cytosolic one.

Vret is required for primary piRNA biogenesis. (A) piRNA production is reduced in vret mutant ovaries. Size profiles of `repeat, antisense mapping' annotated small RNAs from vret<sup>148-60</sup> heterozygous and vret<sup>148-60</sup> homozygous ovaries are shown. Note that increase in siRNAs derives from the transposon mdg1 (light red) (for explanation, see text). (B,C) piRNAs generated from both germline and somatic-expressed clusters are severely reduced in vret mutant ovaries. (B) Density map showing normalized peaks of genome-unique piRNAs from vret heterozygote and mutant ovarian small RNA libraries across the 42AB (also called Cluster 1) and flam piRNA clusters. (C) Cluster-derived, normalized, genome-unique piRNAs remaining in vret mutant ovarian libraries, shown as a percentage of normalized vret heterozygote levels. Clusters are indicated by their cytological band location, their name, and predominant expression in either `germline', `soma' or `both' types of ovarian tissues. (D) Size distribution of piRNAs bound by Piwi proteins. (E) Piwi-bound piRNAs are preferentially lost in vret mutant ovaries. The change in average size of ovarian, transposon-derived piRNAs, due to mutation in either aub (aub^HN2/aub^QC42), piwi (piwi¹/piwi²)or vret (vret^148-60) is shown. Individual aub, piwi or vret heterozygote libraries were used to determine wild-type average piRNA size. (F) piRNA `ping-pong' signal, indicated by a peak at 10 nt, from 23-29 nt reads mapping to the Batumi and Roo transposons in aub, piwi or vret mutant ovarian small RNA libraries. (G) Density map showing normalized peaks of piRNAs from vret heterozygote or mutant ovarian small RNA libraries, with sense mappers above the x-axis and antisense below. Close-ups show the remaining piRNA `ping-pong' peaks (green arrows) in vret mutant ovaries compared with heterozygotes.