Screens for piwi Suppressors in Drosophila Identify Dosage-Dependent Regulators of Germline Stem Cell Division

Corresponding author: Haifan Lin, Duke University Medical Center, Box 3709, 412 Nanaline Duke Bldg., Durham, NC 27710., h.lin{at}cellbio.duke.edu (E-mail)

Communicating editor: F. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX LITERATURE CITED

The Drosophila piwi gene is the founding member of the only known family of genes whose function in stem cell maintenance is highly conserved in both animal and plant kingdoms. piwi mutants fail to maintain germline stem cells in both male and female gonads. The identification of piwi-interacting genes is essential for understanding how stem cell divisions are regulated by piwi-mediated mechanisms. To search for such genes, we screened the Drosophila third chromosome (36% of the euchromatic genome) for suppressor mutations of piwi² and identified six strong and three weak piwi suppressor genes/sequences. These genes/sequences interact negatively with piwi in a dosage-sensitive manner. Two of the strong suppressors represent known genes—serendipity- and similar, both encoding transcription factors. These findings reveal that the genetic regulation of germline stem cell division involves dosage-sensitive mechanisms and that such mechanisms exist at the transcriptional level. In addition, we identified three other types of piwi interactors. The first type consists of deficiencies that dominantly interact with piwi² to cause male sterility, implying that dosage-sensitive regulation also exists in the male germline. The other two types are deficiencies that cause lethality and female-specific lethality in a piwi² mutant background, revealing the zygotic function of piwi in somatic development.

STEM cells possess the unique abilities to self-renew and to produce numerous differentiated daughter cells. Endowed with these properties, stem cells play a central role in the proliferative and regenerative capabilities of tissues. Understanding mechanisms that control stem cell division is of fundamental biological and medical significance.

Germline stem cells in the Drosophila ovary have been an excellent model for studying the mechanism of stem cell division via combined genetic, cell-biological, and molecular approaches. The Drosophila ovary is composed of 16–18 functional units called ovarioles. Each ovariole contains two or, less frequently, three germline stem cells located at its apical tip in a specialized structure called the germarium (Fig 1A, Fig B, and Fig C'). These stem cells have been positively identified by genetic, laser ablation, and cell-biological analyses (WIESCHAUS and SZABAD 1979 ; LIN and SPRADLING 1993 ; DENG and LIN 1997 ; Fig 1A, Fig B, and Fig C'). They are in direct contact with key somatic signaling cells known as the cap cells (Fig 1A, Fig B, and Fig C'). Each stem cell divides in parallel to the germarial axis such that the daughter stem cell remains apposed to the cap cells while the differentiating daughter cell, the cystoblast, is displaced one cell away from the cap cells (Fig 1A and Fig B). The cystoblast then undergoes four rounds of mitosis with incomplete cytokinesis to form a germline cyst containing 16 interconnected cells called cystocytes (Fig 1A and Fig C'). One cystocyte subsequently differentiates into the oocyte, while the others differentiate into nurse cells. This differentiating germline cyst is enveloped by the somatic follicle cells to form an egg chamber, which then leaves the germarium and joins the existing linear array of egg chambers to form an ovariole (Fig 1A, Fig C, and Fig C'). Thus, the Drosophila ovariole, composed of a row of stem cell products whose positions delineate their birth order, represents an effective model for stem cell research.

View larger version (47K): In this window In a new window Download PPT slide   Figure 1. Early oogenesis in Drosophila and the piwi2 mutant phenotype. The Drosophila ovary is made up of 16–18 functional units called ovarioles, each starting with a germarium and followed by a string of developing egg chambers. (A) A drawing of a wild-type germarium. There are usually two to three germline stem cells (gsc) adjacent to the terminal filament (tfc) and in contact with cap cells (cc). These stem cells divide asymmetrically to form a new germline stem cell and a differentiated daughter cell, the cystoblast (cb). The cystoblast further undergoes four sets of division with incomplete cytokinesis to produce a germline cyst containing 16 interconnected cystocytes. These mitotic events occur in region I. In region II, intracyst transport of certain RNA, proteins, and organelles occurs, leading to the differentiation of 15 nurse cells and an oocyte within each cyst. In region III, a differentiated germline cyst becomes completely enveloped by a monolayer of follicle cells to form an egg chamber. The egg chamber then leaves the germarium as it develops further. As each egg chamber is produced from one cystoblast, it represents the result of one germline stem cell division (modified from KING 1970 ). (B) A schematic of region I of the germarium showing the topology of the germline stem cell region and the expression pattern of genes relevant to this study. The terminal filament and cap cells express Yb, piwi, and hh. Yb- and piwi-mediated signaling is required for germline stem cell maintenance, although hh may also play a somewhat redundant role. In germline stem cells, spectrosomes containing spectrins, Hts, and Bam-F reside in the apical region of the cytoplasm both at interphase and during mitosis, apposed to the signaling somatic cells. During mitosis, the spectrosome anchors one pole of the spindle so that the divisional plane is approximately perpendicular to the apico-basal axis of the germarium. As a result, the daughter germline stem cell remains in contact with cap cells while the cystoblast becomes one cell away from the somatic cells (modified from LIN 1998 ). (C–D') Images of ovarioles stained with anti-Vasa (green) and anti-m1B1 (red) antibodies. Note that the wild-type ovariole, pictured at two different magnifications in C and C', has a germarium (G) full of germline cells, including two germline stem cells containing spectrosomes (S) as well as postgermarial egg chambers. This indicates that germline stem cells are functional and maintained. In contrast, the piwi2 mutant ovariole, also pictured at two different magnifications in D and D', has a rudimentary germarium (G) with little, if any, remaining germline. There is no spectrosome, but only one malformed egg chamber (1). This indicates that the germline stem cells have been depleted. The bar in C represents 100 µm for C and D; the bar in C' represents 10 µm for C' and D'.

Research in recent years has revealed both inter- and intracellular mechanisms involved in regulating the division of germline stem cells in the Drosophila ovary (reviewed in LIN 2002 ). Genetic analyses have identified genes that are expressed in somatic signaling cells for germline stem cell maintenance, such as fs(1)Yb (Yb), piwi, and decapentaplegic (dpp; COX et al. 1998 ; XIE and SPRADLING 1998 ; KING and LIN 1999 ; Fig 1B, Fig D, and Fig D'). Furthermore, combined genetic and cell-biological analyses have revealed intracellular mechanisms involving a spectrin-rich cytoplasmic organelle called the spectrosome (Fig 1B, Fig C, and Fig C') as well as genes such as pumilio (pum), piwi, nanos (nos), and bag of marbles (bam) that are differentially expressed in stem cells and cystoblasts (MCKEARIN and OHLSTEIN 1995 ; DENG and LIN 1997 ; LIN and SPRADLING 1997 ; OHLSTEIN and MCKEARIN 1997 ; FORBES and LEHMANN 1998 ; BHAT 1999 ; PARISI and LIN 1999 ; COX et al. 2000 ). Among these genes, pum is required cell-autonomously for germline stem cell maintenance (LIN and SPRADLING 1997 ; FORBES and LEHMANN 1998 ; PARISI and LIN 1999 ). nos appears to play a similar role (BHAT 1999 ; Z. WANG and H. LIN, unpublished data), while bam is required for cystoblast differentiation (MCKEARIN and OHLSTEIN 1995 ; OHLSTEIN and MCKEARIN 1997 ). The piwi gene is expressed in both germline and somatic cells in the ovary (COX et al. 1998 , COX et al. 2000 ). Loss of piwi function leads to failure in the maintenance of germline stem cells. Interestingly, piwi function is required in the somatic signaling cells for germline stem cell maintenance, whereas its expression in the germline plays only a dispensable role in promoting stem cell division. Therefore, in this study, we focus on investigating how piwi functions as an essential somatic signaling component in germline stem cell maintenance. We do so by identifying piwi-interacting genes whose mutations restore the self-renewing ability of germline stem cells in the piwi mutant.

piwi is the founding member of the piwi family (a.k.a. argonaute family, PPD family) of genes whose function in stem cell maintenance is well conserved in both animal and plant kingdoms (reviewed in BENFEY 1999 ; CERUTTI et al. 2000 ). In addition to the essential role of piwi in germline stem cell maintenance in Drosophila, piwi related gene-1 (prg-1) and prg-2 are required for germline stem cell maintenance in Caenorhabditis elegans (COX et al. 1998 ). A human homolog of piwi called hiwi is expressed in spermatogenic cells, with its overexpression highly correlated to seminomas—cancers due to the malignant proliferation of germline stem cells or their progenitors (QIAO et al. 2002 ). In the hematopoietic system, hiwi is specifically expressed in human CD34⁺ hematopoietic stem/progenitor cells and not in more differentiated populations (SHARMA et al. 2001 ). In Arabidopisis thaliana, argonaute (ago1) and zwille/pinhead are both necessary for the maintenance of meristem cells (BOHMERT et al. 1998 ; MOUSSIAN et al. 1998 ; LYNN et al. 1999 ).

The piwi family genes are also involved in other developmental processes. In Drosophila, a gene closely related to piwi, aubergine/sting, is involved in pole cell formation and the translational regulation of such downstream targets as oskar and gurken (WILSON et al. 1996 ; HARRIS and MACDONALD 2001 ). Some other piwi family genes in a variety of organisms are also implicated in translation (eIF2C, ZOU et al. 1998 ; KOESTERS et al. 1999 ) as well as in post-transcriptional gene silencing (a.k.a., RNA interference; rde-1, TABARA et al. 1999 ; ago2, HAMMOND et al. 2001 ; qde-2, COGONI and MACINO 1997 , reviewed in CATALANOTTO et al. 2000 ; ago1, FAGARD et al. 2000 ). Most recently, the piwi family genes have also been implicated in epigenetic modification (reviewed in STEVENSON and JARVIS 2003 ) and even in genomic rearrangement (MOCHIZUKI et al. 2002 ).

Since piwi family genes are so widely conserved and play essential roles in stem cell division and other developmental processes, identification of piwi-interacting genes is a key step toward understanding the regulation of stem cell division and other piwi-mediated mechanisms. At present, only Yb has been linked to piwi (KING et al. 2001 ; Fig 1B). Yb coordinately regulates the division of germline and somatic stem cells via regulating hh and piwi expression in the terminal filament and cap cells—somatic signaling cells in the Drosophila ovary (KING et al. 2001 ; Fig 1B). However, little is known about the regulatory relationship among the other genes known to be involved in germline stem cell division. In addition, many more genes are expected to take part in the inter- and intracellular regulation of germline stem cells. To discover some of these missing genes, we screened the Drosophila third chromosome for piwi interactors.

We report here the identification of six strong suppressor and three weak suppressor genes of piwi. Two of these strong suppressors represent known genes: serendipity- (sry-) and similar. The sry- and similar genes both encode transcription factors (PAYRE et al. 1994 ; NAMBU et al. 1996 ). Thus, this study implicates the involvement of previously unknown transcriptional mechanisms in the regulation of germline stem cells. In addition, we recovered enhancers of piwi that cause lethal phenotypes, which reveals the requirement of the zygotic piwi activity for somatic development.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX LITERATURE CITED

Drosophila strains and culture:
All Drosophila strains were grown at room temperature on yeast-containing cornmeal/molasses medium unless otherwise indicated. Canton-S and Oregon-R strains were used as wild-type flies. The hh-lacZ flies carry an enhancer trap insert in hh as initially characterized by MOHLER and VANI 1992 . Its ovarian expression pattern was characterized by FORBES et al. 1996 and by KING et al. 2001 .

piwi mutants:
A P-element insertional mutation of piwi, piwi², was described in COX et al. 1998 . It was used as a stock, piwi²/CyO; MKRS/TM6B. To facilitate easier screen design, piwi² If stocks were generated by introducing the irregular facets (If) marker to the piwi² chromosome by standard genetic crosses and were used in the following stocks: piwi² If/CyO; Ly¹/+ and piwi² If/CyO; TM8/+. piwi¹² is an excision allele of piwi⁴ that contains only a 42-nucleotide remnant of the P{w, lacZ} P element and that still retains the germline stem cell defect of piwi⁴. Both piwi⁴ and piwi¹² were generated in an independent screen in Tulle Hazelrigg's lab and were gifts from her. piwi¹² was used in this study as a piwi¹²/CyO; MKRS/TM6B stock.

Individual suppressor mutants:
The three sry- alleles tested in this study, sry-¹², sry-^SF1, and sry-^SF2, were kindly provided by Alain Vincent and are described in CROZATIER et al. 1992 . The putative similar allele, P{w^+mC = lacW}l(3)j11B7^j11B7, was described in SPRADLING et al. 1999 and obtained from the Bloomington Drosophila Stock Center (BDSC, stock no. 12162). The tango allele, tango⁵ st¹ p^p, was described in EMMONS et al. 1999 and kindly provided by Steve Crews. Two P-element stocks, ry506 P{ry^+t7.2 = SRS3.9}3 (DE CICCO and SPRADLING 1984 ) and P{w^+t11.7ry^+t7.2 = wA}4-4, were provided by the BDSC (stock nos. 10395 and 10833, respectively). In addition, a number of lethal mutant stocks provided by the BDSC were also tested for their candidacy as suppressors. They are described in WARMKE et al. 1989 , and a partial list of them is given below, with the BDSC stock number in parentheses: e^s? ro¹ ca¹ l(3)99Da¹ (4417); l(3)99Ea¹ (4425); l(3)99De¹ (4419); l(3)99Dg³ (4422); l(3)99Db¹ (4426); l(3)99Dc¹ (4423); l(3)99Dd² (4427); l(3)99Dh¹ (4418); and l(3)99Di⁴ (4421).

Deficiencies:
Laurel Raftery kindly provided the Df(3R)KpnA,ca¹,awd^K deficiency stock. The following deficiency stocks were provided by the BDSC (with the stock number in parentheses): Df(3R)B81 P{ry^+t7.2 = RP49}F2-80A e¹ (3546); Df(3R)faf^BP st¹ (1011); Df(3L)Aprt-1 ru¹ h¹ (600); Df(3L)R-G5 rho^ve-1 (2400); Df(3L)Aprt-32 (5411); Df(3L)R (57); Df(3L)R-G7 rho^ve-1 (2400); Df(3R)L127 (3547); Df(3R)X3F P{ry^+t7.2 = RP49}A3-84F e¹ (2353); Df(3R)tll-e ca¹ (5415); Df(3R)04661 (5415); Df(3L)ri-79C (3127); Df(3L)rdgC-co2 th¹ st¹ in¹ kni^ri-1 p^p (2052); Df(3L)31A (3627); Df(3L)66C-G28 (1541); Df(3L)pbl-X1 (1420); Df(3L)kto2 (3617); Df(3L)VW3 (3000); Df(3L)XS533 (5126); Df(3L)XS572 (5583); Df(3R)ea kni^ri-1 p^p (383); Df(3R)P115 e¹¹ (1467); Df(3R)C4 p* (3071); Df(3R)DG2 (4431); Df(3R)2-2 (3688); Df(3R)p712 red¹ e¹ (1968); Df(3R)23D1 ry⁵⁰⁶ (2586); Df(3R)p40 red¹ e¹ (1945); Df(3R)ry506-85C (1534); Df(3R)ry27 Dfd¹ cu¹ kar¹ (480); and Df(3R)ry615 (3007). Unless otherwise mentioned, most non-piwi² interacting deficiencies in the Appendix are also from the Bloomington Drosophila Stock Center, included in the deficiency kit. All of the stocks mentioned in this article have been previously mentioned in FlyBase (http://flybase.bio.indiana.edu).

Deficiency screens:
A five-generation scheme was used in pilot screens to produce piwi²/piwi²; Df/+ females, which were then tested for fertility and ovarian germline stem cell phenotypes, as described below. In these screens, neither the piwi nor the deficiency chromosomes were marked with dominant mutations; thus both the second and third chromosomes needed to be balanced or marked in both stocks to identify piwi²/piwi²; Df/+ females. This resulted in very few lines reaching the F₄ cross and those crosses yielding very few progeny of the target phenotype. To achieve a more efficient scheme, the piwi² chromosome was recombined with b cn If to produce a piwi² chromosome dominantly marked with If. These stocks were tested for retention of the piwi² mutation by phenotypic and Southern analysis (described below). No change was seen in the piwi² mutant phenotype, suggesting that If did not interact with the piwi² mutation. Therefore, the piwi² If/CyO line designated W was then used for all further crosses (unless otherwise specified), reducing the cross scheme to two generations (Fig 2).

View larger version (11K): In this window In a new window Download PPT slide   Figure 2. The main cross scheme for identifying piwi2 suppressors. In the P generation cross for each deficiency or mutant, 12 piwi2 If heterozygote females were crossed to 10 males carrying the balanced deficiency or mutant in two vials. In the F1 cross for each deficiency or mutant, 10 F1 piwi2 If heterozygote and deficiency or mutant heterozygote males were crossed to 12 piwi2 If heterozygote females in two vials. Finally, in the F2 cross, homozygous piwi2 If females were counted, and homozygous piwi2 If females with heterozygous deficiency or mutant chromosomes were tested for fertility and analyzed for suppression. If, irregular facets; CyO, balancer; Ly, Lyra; TM8, balancer; +, wild-type chromosome; Df, deficiency.

The screen in Fig 2 started with two vials for each experimental stock, with each containing 5–6 piwi² If/CyO; Ly¹ or TM8/+ virgin females crossed to 5–6 deficiency/balancer males ordered from the deficiency kit at the Bloomington Drosophila Stock Center. In the F₁ generation we collected 10–12 piwi² If/+; deficiency/Ly¹ or TM8 males and divided them into two vials each with 5–6 piwi² If/CyO; +/+ virgin females. The parents in these two vials were transferred many times to optimize the numbers of F₂ progeny. We collected 1- to 2-day-old piwi² If/piwi² If; deficiency/+ females and crossed them to 5 wild-type males to check fertility. After 2–3 days we dissected the target females and checked for suppression of the piwi² germline stem cell maintenance defect.

P-element excision:
To create a stronger deletion allele of P{ry^+t7.2 = SRS3.9}3, we mobilized the P element. This was accomplished by crossing ry⁵⁰⁶ P{ry^+t7.2 = SRS3.9}3 females to ry⁵⁰⁶ Sb¹ P{ry⁺ 2-3}[99B]/TM6 Ubx¹³⁰ e¹ males and collecting ry⁵⁰⁶ P{ry^+t7.2 = SRS3.9}3/ry⁵⁰⁶ Sb¹ P{ry⁺ 2-3}[99B] males. These single males (150) were then crossed to D¹ red¹ e¹/TM3 ry^RK Sb¹ Ser¹ females. From each cross, two Sb and phenotypically ry^- males were collected and made into 300 stocks. The P-excision chromosomes in these stocks were examined for recessive lethality and sterility.

Genomic Southern blot analysis:
Standard molecular biology techniques, such as DNA preparation, cloning, and Southern blotting, were performed as described in SAMBROOK et al. 1989 . Genomic DNA from flies was digested with EcoRI, separated on 0.7% agarose gels, and transferred to GeneScreen hybridization transfer membranes (New England Nuclear Research Products, DuPont, Boston). To confirm the presence of the piwi² mutation in the piwi² If chromosome, blots were probed with gel-purified 5' and 3' genomic fragments and a 7.2-kb fragment containing the ry sequence. To check for precise or imprecise excision of P{ry^+t7.2 = SRS3.9}3, blots were probed with gel-purified 5' and 3' genomic inverse PCR fragments and the ry sequence.

Inverse PCR mapping of the P-element insertion sites:
DNA preparation, cloning, inverse PCR, and other standard molecular biology techniques were performed as described in SAMBROOK et al. 1989 . Genomic DNAs from P{ry^+t7.2 = SRS3.9}3 and P{ w^+t11.7ry^+t7.2 = wA}4-4 homozygous fly stocks were digested with Sau3AI. They were then circularized by ligation and digested further with HindIII. These genomic DNAs were then used as templates for PCR using two different sets of primers, one specific to the 5' end of the P elements and one to the 3' end. A primer to the inverted repeats of P elements, designated IR, was used as one of the primers for both the 3' and 5' ends of both target P elements. Its sequence is 5'-CGATCGGGACCACCTTATGTTATTTCATCAT-3'. The second primer for the 5' end of the P elements had the sequence 5'-CTGTGCGTTAGGTCCTGTTCATTG-3'. The second primer for the 3' end of the P elements had the sequence 5'-TTAATCTCCCATAGAGCTTCGTTA-3'. The PCR products were either cloned and sequenced or directly sequenced at the Duke Comprehensive Cancer Center sequencing facility.

We found that parts of the P{ry^+t7.2 = SRS3.9}3 flanking genomic DNA sequence are identical to parts of the 5' ends of at least 13 independently isolated expressed sequence tags (ESTs) identified by the Berkeley Drosophila Genome Project (BDGP) and clustered with at least 27 other ESTs to form clot 2858 (RUBIN et al. 2000 ). This clot has been associated by BDGP with the mRNA transcript CT7108 and the predicted gene CG7816.

Immunocytochemistry and immunofluorescence microscopy:
Wild-type and mutant ovaries from adult females were dissected, fixed, and stained as described in LIN et al. 1994 . For immunofluorescence staining, anti-Vasa antibodies were used to specifically mark germ cells at 1:2000 dilution (HAY et al. 1990 ). The monoclonal 1B1 antibody was used at 1:1 dilution to mark spectrosomes, fusomes, and the cell cortex (DING et al. 1993 ; ZACCAI and LIPSHITZ 1996A , ZACCAI and LIPSHITZ 1996B ) and was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the Department of Biological Sciences at the University of Iowa (Iowa City, IA). The monoclonal 9E10 anti-myc antibody was used at 1:50 dilution (EVAN et al. 1985 ) to mark the myc-Piwi protein, as described in COX et al. 2000 . All the fluorescence-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratory (West Grove, PA) and were used at 1:200 dilution. Immunofluorescently labeled samples were also counterstained with the DNA-specific dye 4',6-diamidino-2-phenylindole (DAPI) as described in LIN and SPRADLING 1993 . The immunologically labeled samples were examined by Nomarski and epifluorescence microscopy under a Zeiss Axioplan microscope equipped with a Star-1 cooled CCD camera (Photometrics, Tucson, AZ). Images were collected by IPLab software and processed by the Adobe Photoshop program.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX LITERATURE CITED

Six regions on the third chromosome suppress the piwi² phenotype:
Using a two-generation cross scheme (Fig 2), we screened deficiencies on the third chromosome for genes that interact with piwi² and are involved in germline stem cell division in the Drosophila ovary (see MATERIALS AND METHODS). Each of these deficiencies on average deletes <1% of the genome, corresponding to 100 or fewer genes. We limited our search to deficiencies that were suppressors of piwi² for two reasons. First, the piwi² mutant itself fails to maintain germline stem cell division, rendering an enhancer screen for a more severe germline stem cell phenotype impractical. Second, the suppressor screen, advantageously, avoids nonspecific enhancers. We did, however, identify a few deficiencies that led to sterility and haplo-lethality in piwi² mutant backgrounds; we further describe those interactions in three of the following sections.

Pilot screens were used to test the validity of the piwi² allele as an appropriate starting point for our piwi dominant suppressor screen (see MATERIALS AND METHODS). We chose this allele for three reasons. First, while both piwi¹ and piwi² cause strong defects in germline stem cell division in adult female ovaries, the piwi¹ mutant has additional effects on the initial organization of the third instar gonad (COX et al. 1998 ). Because we are interested in the regulation of germline stem cell division, not gonadogenesis, piwi² was the better choice. Second, piwi² as a weaker allele may be more easily suppressed. Since the nature of the screen already strictly limited the type of suppressors that can interact with piwi in a dosage-dependent, dominant, and negative manner, increasing this pool by using a somewhat weaker allele could be advantageous for such a stringent screen. Third, piwi is required for germline stem cell maintenance, yet the piwi² mutation does not affect male fertility. This provides an added opportunity to discover potential enhancers of piwi by identifying mutations that interact with piwi² to cause male sterility. Identification of specific suppressor deficiencies in the pilot screens (see MATERIALS AND METHODS) confirmed that the piwi² allele was indeed a suitable allele for a suppressor screen.

We judged suppression of piwi² by the ability of the deficiency (or, later, of the mutation) to restore germline stem cell division in the ovary. To examine this restoration, we used Nomarski optics and antibodies that mark germline cells and their subcellular structures. Since germline stem cell division gives rise to a germline cyst in the germarium, which subsequently develops into an egg chamber in the ovariole, the number of germline cysts and egg chambers per ovariole indicates the number of germline stem cell divisions that have been restored (Fig 1A and Fig C). Because the two or three germline stem cells in a piwi mutant germarium frequently differentiate without renewal, the piwi² mutant ovarioles often contain a rudimentary germarium plus only one egg chamber (Fig 1D and Fig D'). If an ovariole contains an apparently normal-looking germarium and at least three egg chambers, it suggests that the germline stem cell division defect in this ovariole is significantly suppressed. The following markers were used to characterize suppression more precisely. DAPI was used to mark nuclei, anti-Vasa antibodies (HAY et al. 1990 ) were used to identify germline cells, and the anti-1B1 antibody (ZACCAI and LIPSHITZ 1996A , ZACCAI and LIPSHITZ 1996B ) was used to visualize spectrosomes and fusomes. The spectrosome is a germline-specific organelle that exists in germline stem cells and their immediate daughter cells, the cystoblasts (LIN and SPRADLING 1995 ), and the fusome is present in early germline cysts (LIN et al. 1994 ). By this method, even the early products of stem cell division in the germarium can be identified. Thus, if an ovariole has a germarium with multiple spectrosome- and fusome-containing germline cells and cysts, as well as at least three egg chambers, this wild-type-looking ovariole is strongly indicative of the restoration of germline stem cell division (i.e., strong expressivity of suppression). For each fly, every ovariole in each ovary was examined; the germarium was analyzed and the number of egg chambers was counted. We consider the germline stem cell defect in a fly to be suppressed if >50% of its ovarioles were suppressed. For each deficiency or mutation, the total percentage of flies with the suppressed phenotype was used to define the overall level of suppression. Stocks with <25% of suppressed flies were classified as not suppressed, 26–40% as weakly suppressed, and 41–100% as strongly suppressed.

In total, we screened 109 deficiencies, which uncovered 87% of the third chromosome, or 36% of the whole euchromatic genome, for dominant suppression of piwi² (Fig 3; Appendix). Within this 36% of the genome, we identified six regions defined by deficiencies as suppressing (Fig 3). Further analysis of these deficiencies is reported below (Table 1; Fig 3).

View larger version (39K): In this window In a new window Download PPT slide   Figure 3. Summary of third-chromosome deficiency suppressor screens. Each deficiency tested for suppression of piwi2 If (by cross scheme in Fig 2 or similar schemes) is represented by one bar that shows the region uncovered by that particular deficiency. Approximately 87% of the euchromatic region in the third chromosome has been tested. The deficiencies that interact with piwi2 as described in the text are identified and labeled. All other deficiencies are mentioned in the Appendix with cytology included for identification. Also labeled are individual mutations that suppress piwi2, with strong suppressing mutations labeled by boldface type. No such distinction is made for the deficiency suppressors. Other mutants that do not suppress piwi2 are also included in the Appendix.

Df (3R)L127: This deficiency (99B5–6 to 99E4–F1) suppresses the germline stem cell defect of piwi² in 43% of the females (Table 1; Fig 3). Because this deficiency overlaps with another suppressing deficiency, Df(3R)B81 (99C8–100F5), it is likely that the region responsible for suppression is 99C8–99E4/F1. We then tested available mutations in 14 genes (Table 1; Fig 3; Appendix) in this region for their suppression of the piwi² phenotype. Mutations in 5 genes, P{SRS3.9}3, l(3)99Da, l(3)99Ea, sry-, and l(3)j11B7 (similar), showed strong suppression, and mutations in two others, l(3)99De and l(3)99Dg, displayed weak suppression (Table 1; Fig 3). It is interesting to note that, in this case, one suppressor region yielded multiple piwi-suppressing genes involved in the self-renewing division of germline stem cells in the Drosophila ovary.

Df(3R)faf^BP: This deficiency (100D–100F5) suppresses the germline stem cell defect of piwi² in 47% of the females (Table 1; Fig 3). This deficiency also overlaps with the suppressing deficiency Df(3R)B81 (99C8–100F5) as well as Df(3R)04661 (100D2–100F5; Table 1; Fig 3). However, it is separable from the previous suppressor region of 99C8–99E4/F1, because Df(3R)tll-e (100A2–100C2/3) does not suppress piwi² (see Appendix). We tested 17 mutations (Table 1; Fig 3; Appendix) that are uncovered by Df(3R)faf^BP and found that a P-insertional mutation, P(wA)4-4, strongly suppressed piwi² (Table 1; Fig 3 and Fig 4). Thus, we narrowed down this suppressor region to a single P-disrupted sequence.

View larger version (41K): In this window In a new window Download PPT slide   Figure 4. Suppressor mutations of piwi2 that represent novel genes. Ovarioles in all panels are double-stained with anti-Vasa (green) and anti-m1B1 antibodies (red). A, A' and B, B' show an unsuppressed (A, A') and suppressed (B, B') ovariole of the piwi2/piwi2; P{SRS3.9}3/+ genotype at low and high magnifications. The unsuppressed ovariole (A, A') has a rudimentary germarium (G) and one malformed egg chamber (1) but no spectrosome. However, the suppressed ovariole (B, B') consists of a germarium containing germline stem cells (marked by the spectrosome, S) and multiple germline cysts as well as several postgermarial egg chambers. C, C', and D, D' show an unsuppressed (C, C') and suppressed (D, D') ovariole of the piwi2/piwi2; l(3)99Da1/+ genotype. E, E' and F, F' show an unsuppressed (E, E') and suppressed (F, F') ovariole of the piwi2 If/piwi2 If; l(3)99Ea1/+ genotype. G, G' and H, H' show an unsuppressed (G, G') and suppressed (H, H') ovariole of the piwi2 If/piwi2 If;P{wA}4-4/+ genotype. The bar in A represents 100 µm for A–H; the bar in A' represents 10 µm for A'–H'.

Df(3L)66C-G28: This deficiency (66B8/9–66C9/10) suppresses the germline stem cell defect of piwi² in 42% of the females (Table 1; Fig 3). Because an adjacent deficiency, Df(3L)pbl-X1 (65F3–66B10), did not suppress piwi², the suppressor interval should be from 66B10 to 66C9/10 (Fig 3). This suppressor interval was not tested further for its suppression activity.

Df(3L)kto2: This deficiency (76B1/2–76D5) restores germline stem cell division in 27% of piwi² mutant females (Table 1; Fig 3). Thus, we classified Df(3L)kto2 as a weak suppressor of piwi². Ten individual mutations and overlapping deficiencies, Df(3L)VW3 (76A3–76B2), Df(3)XS-533 (76B4–77B), and Df(3L)XS572 (76B6–77C1), were tested; none of them suppressed the piwi² defects in germline stem cell maintenance (Appendix). Despite this, the suppressor in this deficiency may still reside in 76B2–76B4. Alternatively, the suppression could be caused by the effects of the Df(3L)kto2 breakpoints or an unrelated mutation elsewhere on the Df(3L)kto2 chromosome.

Df(3L)Aprt-1: This deficiency (62A10/B1–62D2/5) suppressed the piwi² defect in 50% of the females with almost complete penetrance at the ovariolar level (Table 1; Fig 3). To narrow down the region of suppression, we tested four overlapping deficiencies [Df(3L)R-G5 (62A10/B1–62C4/D1), Df(3L)Aprt-32 (62B1–62E3), Df([3L)R (62B7–62B12), and Df(3L)R-G7 (62B8/9–62F2/5)] and six P-element insertional mutations (see Appendix) in the region. None of them suppressed the piwi² mutation. Since these deficiencies collectively uncover the region defined by Df(3L)Aprt-1, the suppression effect of Df(3L)Aprt-1 may be caused by one of the two breakpoints of Df(3L)Aprt-1 or due to a background mutation outside of the deficiency. Alternatively, it could suggest that the enhancing effect of many haploid genes in these deficiencies cancels the effect of a specific suppressor (see DISCUSSION).

Df(3L)ri-79C: This deficiency (77B/C–77F/78A) restores germline stem cell division in 44% of the piwi² mutant females (Table 1; Fig 3). Overlapping deficiencies uncovering 77A1–77D1 (Df[3L]rdgC-co2) and 78A–78E (Df[3L]31A) do not suppress piwi², narrowing down the potential suppression region to 77D1–78A (Fig 3). None of the five tested mutations (see Appendix) within this region showed suppression. Further mutagenesis in this region may allow the identification of a potential suppressor gene(s), as may also be the case with Df(3L)Aprt-1 and Df(3L)kto2 (see above).

Deficiencies that cause the sterility of heterozygous piwi² males:
In our screen, we found enhancers of the piwi² phenotype by identifying deficiencies that cause the sterility of piwi²/+ males. We identified four adjacent deficiencies within cytological regions 88–90 [Df(3R)ea (88E7/13–89A1), Df(3R)P115 (89B7/8–89E7/8), Df(3R)C4 (89E–90A), and Df(3R)DG2 (89E1/F4–91B1/B2)] that cause such trans-heterozygous male sterility (Fig 3; also see Fig 2 for the generation of trans-heterozygous F₁ males). These deficiencies may thus contain genes that interact with piwi in a dosage-dependent mechanism and are positively required for germline stem cell maintenance at least in males.

Deficiencies that are haplo-lethal in a piwi² mutant background:
We also identified three deficiencies that are haplo-lethal in the piwi² mutant background: Df(3R)2-2 (81F–82F10/11 or 83A), Df(3R)p40 (84E8/9–85B6), and Df(3R)23D1 (93F–94F; Table 2; Fig 3). Of >100 piwi² mutant progeny from the suppressor cross to each of the deficiencies (Fig 2), none were found to contain the deficiency (Table 2). The haplo-lethality of Df(3R)p40 in the piwi²/piwi² background was confirmed by its overlapping deficiency, Df(3R)p712 (84D4/6–85B6). Three additional deficiency crosses yielded only very few deficiency-carrying progeny and these deficiencies were thus classified as semi-haplo-lethal in a piwi² mutant background (Table 2; Fig 3). These two classes of deficiencies may uncover genes that interact with piwi in a dosage-sensitive manner required for somatic development. Although many selected individual mutations in these six regions were checked for lethality and suppression of piwi² (Appendix), the available overlapping deficiencies and individual mutations did not allow us to identify individual genes responsible for the haplo-lethal interactions. Regardless of the cause of the haplo-lethal interaction, the discovery of these deficiencies reveals the involvement of the zygotic piwi activity in somatic development.

Female-specific haplo-lethal deficiency in a piwi² mutant background:
The deficiency Df(3R)ry506-85C (87D1/2–88E5/6) caused female-specific haplo-lethality in the piwi² mutant background (Table 2; Fig 3). Of 319 piwi² mutant progeny, 146 contained the deficiency, all of which were males. We then tested seven smaller or overlapping deficiencies, and only Df(3R)ry27 (87D1/2–87F1/2) showed female lethality in a piwi² mutant background (Table 2; Fig 3). As deficiencies Df(3R)ry506-85C and Df(3R)ry27 were initially recovered separately, it is unlikely that they share common background mutations responsible for the female haplo-lethal effect. Thus, the gene(s) responsible for the female haplo-lethality likely resides between 87D1/2 and 87F1/2. Since an overlapping deficiency, Df(3R)ry615 (87B11/13–87E8/11), shows no lethality or suppression, the gene(s) responsible for female-specific lethality probably resides between 87E8/11 and 87F1/2 (Fig 3; Appendix), even though this gene(s) is not represented by the tested individual mutations in the region (Appendix). The discovery of this haplo-insufficient female-specific lethal interaction with piwi² further implicates the existence of a female-specific mechanism that interacts with piwi in a dosage-sensitive manner required for somatic development.

Individual suppressor mutations define novel genes or sequences that interact with piwi:
By systematically testing individual mutations uncovered by the deficiency suppressors for their ability to suppress the piwi² ovarian phenotype, we identified four novel genes/sequences whose mutations strongly suppress the defects in germline stem cell division in the piwi² mutants. Two such genes are defined by recessive lethal EMS alleles, while two other genes/sequences are defined by P-element insertions that do not display any detectable phenotype as homozygotes. This study reports their high levels of penetrance in suppressing the piwi² phenotype.

The P{SRS3.9}3 gene: The P{SRS3.9}3 (99D1) mutation restores germline stem cell division in 58% of piwi² mutant females (Table 1; Fig 3). This suppression exceeds that of the original suppressing deficiencies (Df[3R]L127 and Df[3R]B81; Table 1; Fig 3). In fact, all the individual genes that have been identified as suppressors of piwi² share this characteristic of stronger suppression (see below and DISCUSSION). Fig 4A&NDASH;B', shows examples of both the unsuppressed (Fig 4A and A') and the suppressed (Fig 4B and Fig B') ovarioles. The unsuppressed ovariole displays a typical piwi germline stem cell depletion phenotype, which contains only a rudimentary germarium and an egg chamber. In contrast, the suppressed ovariole contains a rescued germarium as well as multiple egg chambers. In the rescued germarium particularly, the presence of multiple spectrosome-containing cells in the germarium suggests the existence of germline stem cells and/or cystoblasts, while the presence of fusomes suggests the existence of developing germline cysts. The presence of multiple germline cysts and egg chambers in the ovariole indicates that the germline stem cells have undergone multiple rounds of divisions. Thus, the germline stem cells and their divisions are significantly restored in these females.

Although P{SRS3.9}3 strongly suppresses the germline stem cell defect of the piwi² mutant, it has no detectable phenotype of its own. To verify that the P{SRS3.9}3 insertional mutation is responsible for the suppression and to generate potential new alleles with phenotype, we conducted P-element excision experiments (see MATERIALS AND METHODS). Out of 150 independent P-excised lines, a few lines displayed lethal phenotype, but the lethality did not map to the insertion site. The rest were viable and fertile and showed no apparent phenotype, as did the original P{SRS3.9}3 line. We tested two such lines for their ability to suppress the piwi² phenotype. Neither line showed suppression. Genomic Southern blotting analysis of these two excision lines (see MATERIALS AND METHODS) shows that they are precise excisions (data not shown). These data support the assertion that the original P{SRS3.9}3 P-element insertion was the agent of piwi² suppression.

To verify that the suppression is not due to the effect of If, either alone or in combination with piwi², we used a longer crossing scheme to generate piwi²/piwi²; P{SRS3.9}3/+ females and examined their ovarian phenotype (see MATERIALS AND METHODS). The strong suppression was maintained in the new crosses without If (Fig 4B and Fig B'), suggesting that the If mutation does not contribute to the suppression effect.

To further test whether the suppression is due to the effect of the P{ry11} sequence in the piwi² allele or to the genetic background associated with the piwi² chromosome, we tested the suppression effect of P{SRS3.9}3 toward piwi¹², a P-excision allele derived from a completely different P-insertional mutagenesis screen that retains only a 42-nucleotide remnant of the original P{w, lacZ} insertion (see MATERIALS AND METHODS). The P{SRS3.9}3 mutation effectively suppresses the germline stem cell defect of the piwi¹² mutant ovaries (data not shown). Thus, P{SRS3.9}3 is most likely a genuine suppressor of the piwi mutations.

The same verification screens were also conducted for piwi suppressors l(3)99Da, P{wA}4-4, similar, and sry-. These experiments demonstrated that the screen using the piwi² If chromosome was valid for identifying dosage-sensitive and non-allele-specific suppressors of piwi.

To precisely map the P{SRS3.9}3 insertion site to the genome, we cloned the surrounding genomic DNA via inverse PCR and sequenced it (see MATERIALS AND METHODS; Fig 5). Matching sequence data against the Drosophila genome sequence at the BDGP revealed that P{SRS3.9}3 disrupts the 5' untranslated region of the predicted CG7816 gene defined by mRNA CT7108 (RUBIN et al. 2000 ; Fig 5). CG7816 encodes a possible transmembrane domain. Sequences similar to CG7816 are found in predicted genes in Saccharomyces cerevisiae, C. elegans, Mus musculus, and Homo sapiens, suggesting the evolutionary conservation of this gene.

View larger version (14K): In this window In a new window Download PPT slide   Figure 5. The location of P{SRS3.9}3 and l(3)j11B7 suppressor mutations. (A) The P-element P{SRS3.9}3 is inserted in the 5' untranslated region of a predicted transcript of the putative gene CG7816. (B) The P-element l(3)j11B7j11B7 is inserted in the fourth intron of the gene similar. The P elements are not drawn to scale. Drawings in this figure are modified from the Berkeley Drosophila Genome Project (RUBIN et al. 2000 ).

It has been shown that piwi and hh represent two parallel somatic signaling pathways regulated by Yb in maintaining the stem cell fate in the germline (KING et al. 2001 ; Fig 1B). Overexpression of hh can rescue the self-renewing division of germline stem cells in piwi mutants (KING et al. 2001 ). To examine whether P{SRS3.9}3 suppresses the germline stem cell defect of the piwi² mutant by upregulating hh expression, we examined the expression of a knock-in hh-lacZ gene (MOHLER and VANI 1992 ; FORBES et al. 1996 ) in the P{SRS3.9}3/+ vs. the +/+ background. The hh-lacZ expression was not elevated by the P{SRS3.9}3 mutation, suggesting that P{SRS3.9}3 does not suppress the piwi² stem cell defect by upregulating hh expression (data not shown). This is also the case for four other suppressors, l(3)99Da, P{wA}4-4, similar, and sry- (data not shown).

The l(3)99Da gene: The EMS-induced homozygous lethal mutation l(3)99Da¹ (99D3–E1) suppresses the germline stem cell defect of 50% of the piwi² mutant females (Table 1; Fig 3; Fig 4, C–D'). To confirm that the l(3)99Da¹ mutation was responsible for piwi² suppression, the l(3)99Da¹ chromosome was recombined with a ry⁵⁰⁶ chromosome to within 1 MU of l(3)99Da¹. Five such independently derived "clean" l(3)99Da¹ chromosomes were tested for suppression of piwi². All five lines continued to show strong suppression, confirming that the l(3)99Da¹ mutation indeed suppresses piwi² (data not shown).

The l(3)99Ea gene: A second EMS-induced homozygous lethal mutation, l(3)99Ea¹ (99D9–E3), suppresses the germline stem cell defect of 42% of piwi² mutant females (Table 1; Fig 3; Fig 4, E–F'). A second allele of l(3)99Ea, designated l(3)99Ea⁴, also strongly suppressed the piwi² phenotype, confirming the l(3)99Ea¹ suppression of piwi² (data not shown). The suppression effect of l(3)99Ea and l(3)99Da is unlikely to be due to their common background, since other mutations derived from the same screen (see below and Appendix) show either different or no interaction with piwi². This conclusion is supported by the retention of the suppression effect of the l(3)99Da¹ mutation after "cleaning the chromosome" (see above).

The P{wA}4-4 suppressor: This P-element insertion, P{wA}4-4 (100F), is the strongest suppressor of piwi², as judged by penetrance both at the individual female level (72%) and at the ovariolar level (Table 1; Fig 3; Fig 4, G–H'). However, P{wA}4-4 is homozygous viable and fertile. Inverse PCR and sequencing the genomic DNA flanking the precise site of insertion of the P element have yielded a sequence that matches subtelo-meric heterochromatic repeats at 100F on chromosome 3R in the BDGP database, a result consistent with the previous mapping by LEVIS et al. 1993 .

Weak suppressors of piwi²: The recessive lethal mutations l(3)99De¹ (99D3–9) and l(3)99Dg³ (D3–9) suppress the germline stem cell defect of piwi² in 36 and 27% of the mutant females, respectively (Table 1; Fig 3). This likely reflects that these are weak alleles and, possibly, that these two genes do not strongly interact with piwi. Both mutations are EMS-induced mutations derived from the same screen as l(3)99Da¹ and l(3)99Ea¹, yet have different interactions with piwi² (see above). In addition, five other lines from the same screen show no suppression of piwi² [l(3)99Db¹, c¹, d², h¹, and i⁴ complementation groups], strengthening the conclusion that piwi² suppression by l(3)99Da¹, l(3)99Ea¹, l(3)99Ea⁴, l(3)99De¹, and l(3)99Dg³ is not due to their shared background.

Individual suppressor mutations define known genes that interact with piwi:
The l(3)j11B7^j11B7 (similar) and tango genes: The P-element insertion l(3)j11B7^j11B7 (99E1) is a recessive lethal mutation and a strong suppressor of piwi², with 50% of the mutant females displaying the suppressed phenotype (Table 1; Fig 3; Fig 6A&NDASH;B'). This P element is inserted in an intron of the similar gene (NAMBU et al. 1996 ; RUBIN et al. 2000 ; Fig 5). l(3)j11B7^j11B7 may thus represent the first mutation of similar.

View larger version (46K): In this window In a new window Download PPT slide   Figure 6. Suppression of piwi2 by similar and tango mutations. Ovarioles in all panels are double-stained with anti-Vasa (green) and anti-m1B1 antibodies (red). A, A' and B, B' show an unsuppressed (A, A') and suppressed (B, B') ovariole of the piwi2/piwi2; l(3)j11B7j11B7 (similar)/+ genotype. The unsuppressed ovariole (A, A') has a germlineless germarium (G) and one malformed egg chamber (1) but no spectrosome. However, the suppressed ovariole (B, B') consists of a germarium containing germline stem cells (marked by the spectrosome, S) and multiple germline cysts as well as several postgermarial egg chambers. C, C' and D, D' show an unsuppressed (C, C') and suppressed (D, D') ovariole, respectively, of the piwi2 If/piwi2 If; tango5/+ genotype. The bar in A represents 100 µm for A–D; the bar in A' represents 10 µm for A'–D'.

Since the Similar protein is a member of the bHLH-PAS family transcription factors that are known to function as heterodimers (NAMBU et al. 1996 ; CREWS 1998 ; CREWS and FAN 1999 ), we examined whether its likely partner, Tango, will also suppress the piwi² phenotype. Tango binds to Similar in a yeast two-hybrid assay (SONNENFELD et al. 1997 ). If Tango heterodimerizes with Similar in regulating germline stem cell division, it may also be a suppressor of piwi. To test this possibility, we introduced a copy of tango⁵, a strong allele of tango, into the homozygous piwi² mutant background. As expected, tango⁵ suppresses the piwi² oogenic phenotype, albeit at a somewhat lower level of penetrance (33% of females were suppressed; Table 1; Fig 3; Fig 6, C–D'). This suggests that Similar and Tango act as heterodimeric partners in interacting with piwi in a direct or indirect fashion.

The sry- gene: Three alleles of sry- (99D5) were tested in a piwi² mutant background. sry-^SF1 showed no interaction with piwi², sry-^SF2 showed a strong suppression (50% penetrance), and sry-¹² showed a weak suppression (20% penetrance; Table 1; Fig 3; Fig 7A&NDASH;B'). Thus sry- is an allele-specific suppressor of piwi with the strength of the suppression of a sry- allele correlating well to its phenotype (see DISCUSSION).

View larger version (37K): In this window In a new window Download PPT slide   Figure 7. Interactions between sry- and piwi alleles. A, A' and B, B' show an unsuppressed (A, A') and suppressed (B, B') ovariole of the piwi2 If/piwi2 If; sry-SF2/+ genotype stained with anti-Vasa (green) and anti-m1B1 (red) antibodies. The unsuppressed ovariole (A, A') has a rudimentary germarium (G) and one malformed egg chamber (1) but no spectrosome. However, the suppressed ovariole (B, B') consists of a germarium containing germline stem cells (marked by the spectrosome, S) and multiple germline cysts as well as several postgermarial egg chambers. C and C' show a G38-2B (myc-piwi transgene)/CyO ovariole stained for anti-Vasa (green) and anti-Myc antibodies (red). D and D' show an escaper ovariole of the G38-2B (myc-piwi transgene)/CyO; sry-SF2/sry-SF2 genotype also stained for anti-Vasa (green) and anti-Myc antibodies (red). Ovarioles in C, C' and D, D' show no significant difference in the Myc-Piwi expression, implying that sry- does not regulate piwi expression. The bar in A represents 100 µm for A–D; the bar in A' represents 10 µm for A'–D'.

Sry- is a Cys-2/His-2 zinc-finger transcriptional activator expressed in the nuclei of germline and somatic cells of the ovary and the testis, as well as in many other tissues and stages of development (PAYRE et al. 1989 ). To examine the function of sry- in germline stem cell division, we examined whether the sterile sry- hemizygous and intra-allelic escapers display germline stem cell defects during oogenesis and spermatogenesis. The ovaries of all allelic or hemizygote combinations appeared to have wild-type levels of germline stem cell division, as indicated by long ovarioles with apparently normal germaria and numerous egg chambers (data not shown). However, the testes are considerably smaller, containing fewer bundles of sperm (data not shown). This testicular phenotype is very similar to the piwi male-sterile phenotype (LIN and SPRADLING 1997 ). Thus, the sry- escapers may have early, piwi-like, stem-cell-related defects in spermatogenesis. Such a function in oogenesis may not be revealed by the existing sry- alleles or may be somewhat redundant in oogenesis.

To examine whether sry- mutations suppress the piwi² phenotype by enhancing Piwi expression, we examined the expression of a fully functional transgenic myc-piwi gene (COX et al. 2000 ) in the sry- escapers (see MATERIALS AND METHODS). The expression of Myc-Piwi was unchanged in the sry- escapers (Fig 7, C–D'). This implies that sry- does not suppress piwi² by affecting Piwi expression.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX LITERATURE CITED

We have screened 36% of the euchromatic genome for suppressors of piwi² and identified six strong and three weak piwi²-suppressing mutations. These suppressors reveal the role of specific transcription factors that interact with piwi in a negative and dosage-sensitive manner in regulating germline stem cell division. In addition, we found deficiencies that dominantly interact with piwi² to cause male sterility, as well as two other types of piwi² interactors that cause haplo-lethal and haplo-female-specific lethal types in a piwi mutant background. The latter two types of interactors reveal the somatic function of piwi during development.

Germline stem cell division is regulated by dosage-sensitive mechanisms:
It is amazing that germline stem cell defects in the piwi mutants can be rescued by the removal of one copy of another gene (i.e., by 50% reduction in the activity). The identification of six dominant suppressors of piwi at the deficiency level, as well as six strong and three weak suppressors at the individual gene/sequence level, shows that one or more dosage-sensitive mechanisms must negatively interact with piwi in regulating germline stem cell division in the Drosophila ovary. The fact that each of these suppressors can restore the self-renewing ability of germline stem cells in piwi mutants suggests the importance of the dosage-sensitive mechanisms. The existence of multiple suppressors further implies that such negatively interacting and dosage-sensitive mechanisms may be a significant component of the molecular machinery that regulates germline stem cell division. Finally, four deficiencies in the 89E–91B region dominantly interact with piwi² to cause male sterility. This implies that dosage-sensitive mechanisms may also exist in the male germline.

Such dosage-sensitive mechanisms may not be manifested as the solo act of an individual gene. The P{wA}4-4 suppressor mutation is inserted in the subtelomeric heterochromatic repeats of chromosome 3R, but does not interrupt TART or HeT-A elements in that region (see RESULTS and LEVIS et al. 1993 ). This suggests that the insertion could affect either the transcriptional or the transpositional activity of the retrotransposons or the epigenetic state of that chromosomal region. The latter possibility in turn would suggest that epigenetic effect in the subtelomeric region is involved in regulating germline stem cell division via the piwi-mediated mechanism.

Germline stem cell division requires dosage-sensitive transcriptional regulation:
The discovery of sry-, similar, and tango as suppressors of piwi further suggests that the dosage-sensitive mechanism operates at least at the transcriptional level. Sry- is a Cys-2/His-2 zinc-finger DNA-binding protein that is present in both germline and somatic cells in the ovary and the testis, as well as in many other tissues and stages of development (PAYRE et al. 1989 ). It is known to act as a homodimer to activate the transcription of bicoid during oogenesis (PAYRE et al. 1994 , PAYRE et al. 1997 ; RUEZ et al. 1998 ). Although sry- mutants are homozygous late embryonic lethal (PAYRE et al. 1989 ; CROZATIER et al. 1992 ), hemizygous and intra-allelic escapers are sterile (CROZATIER et al. 1992 ), indicating sry-'s function during oogenesis. The three sry- alleles used in this study (sry-¹², sry-^SF1, and sry-^SF2) are all single-amino-acid changes in the third zinc-finger domain of the protein (CROZATIER et al. 1992 ). However, they are not equivalent mutations (CROZATIER et al. 1992 ). For example, sry-^SF2 hemizygotes show many more gonadal defects than the other alleles do, while sry-^SF1 shows a lower escaper rate than sry-^SF2 does (CROZATIER et al. 1992 ). This phenotypic difference leads us to suspect that the sry-^SF2 mutation tends to perturb a subset of downstream effectors necessary for gonadal function, while the sry-^SF1 mutation tends to disrupt more general downstream factors, leading to higher lethality (CROZATIER et al. 1992 ). Consistent with this speculation, sry-^SF2 is the strongest suppressor of the piwi² phenotype. It would be interesting to conduct genomic screens to identify the target genes of sry- whose transcription is selectively affected by sry-^SF1 but not by sry-^SF2 mutation. Such target genes would likely be involved in germline stem cell division and gonadogenesis.

Like Sry-, Similar and Tango play a key role in the dosage-sensitive regulation of germline stem cell division. Similar is homologous to a large group of heterodimerizing transcriptional activators (CREWS 1998 ; CREWS and FAN 1999 ). It shows closest homology to the human hypoxia inducible factor-1 (HIF-1) and has been shown to function in hypoxic response in Drosophila (LAVISTA-LLANOS et al. 2002 ). HIF-1 binds to HIF-1ß to drive transcription of downstream genes (CREWS and FAN 1999 ). As Tango is the only Drosophila homolog of HIF-1ß (SONNENFELD et al. 1997 ), it is likely to be a partner of Similar. Indeed, Tango interacts with Similar in the yeast two-hybrid system. However, Tango is also known to bind to two other Drosophila bHLH-PAS family proteins, Single-minded and Trachealess, to mediate the transcription of their downstream targets (reviewed in CREWS 1998 ; CREWS and FAN 1999 ). By showing that tango suppresses the germline stem cell phenotype of piwi², this study suggests that Tango heterodimerizes with Similar in the dosage-sensitive transcriptional activation of genes involved in germline stem cell division.

piwi as a regulator of gene expression:
Although the biochemical properties of the Piwi family proteins have not been systematically characterized, this family of proteins has been extensively implicated in RNA-related processes. Piwi itself is necessary for both transcriptional and post-transcriptional gene silencing in Drosophila (PAL-BHADRA et al. 2002 ). The aubergine (a.k.a. sting) gene, a Drosophila homolog of piwi, functions in the regulation of the stellate transcript (SCHMIDT et al. 1999 ), as well as in the translational regulation of oskar and gurken (WILSON et al. 1996 ; HARRIS and MACDONALD 2001 ). ago2 in Drosophila, rde-1 in C. elegans, ago1 in A. thaliana, and qde-2 in Neurospora crassa are necessary for post-transcriptional gene silencing (COGONI and MACINO 1997 ; TABARA et al. 1999 ; FAGARD et al. 2000 ; HAMMOND et al. 2001 ). Most recently, the piwi family genes have also been implicated in epigenetic modification (reviewed in STEVENSON and JARVIS 2003 ) and even in genomic rearrangement via microRNA-mediated mechanisms (MOCHIZUKI et al. 2002 <