piwi

Regulation

Piwi and transgene silencing

Two types of transgene silencing were found for the Alcohol dehydrogenase (Adh) transcription unit. Transcriptional gene silencing (TGS) is Polycomb dependent and occurs when Adh is driven by the white eye color gene promoter. Full-length Adh transgenes are silenced posttranscriptionally at high copy number or by a pulsed increase over a threshold. The posttranscriptional gene silencing (PTGS) exhibits molecular hallmarks typical of RNA interference (RNAi), including the production of 21-25 bp length sense and antisense RNAs homologous to the silenced RNA. Mutations in piwi, which belongs to a gene family with members required for RNAi, block PTGS and one aspect of TGS, indicating a connection between the two types of silencing (Pal-Bhadra, 2002).

Despite the fact that posttranscriptional silencing appears to be a matter of RNA metabolism, some indications of chromatin modification of the homologous endogenous gene under certain circumstances have emerged. (1) Transgene copies of viral genes present in the nucleus only become methylated upon infection of the plant by the homologous virus, which has a double-stranded RNA genome and which does not enter the plant nucleus. (2) In plants and rodent cells, the introduction of DNA constructs into cells can trigger the RNA degradation reaction. (3) Mutations in Arabidopsis selected for reduced DNA methylation, ddm1, a SWI2/SNF chromatin component, and met1, the major DNA methyltransferase, will relieve gene silencing, including a stochastic reversal of posttranscriptional silencing. (4) Transgene arrays in the C. elegans germline are desilenced and appear less condensed in mutant backgrounds for mut-7 and rde-2, which are both required for RNAi. (5) The ectopic transcription of promoter sequences or their introduction to the plant cell in a virus will trigger transcriptional silencing of another reporter construct in the same cell with a homologous promoter, which also becomes hypermethylated. Evidence is presented for a link between posttranscriptional and transcriptional modes of gene silencing in Drosophila (Pal-Bhadra, 2002 and references therein).

The silencing of the promoter-reporter construct white-Alcohol dehydrogenase (w-Adh) and full-length Adh transgenes occurs in Drosophila . The w-Adh effect is modulated by mutations in the Polycomb group (Pc-G) of chromatin-repressive proteins, and the silenced transgenes are associated with the Pc-G complex, whereas single highly expressed copies are not. The endogenous Adh gene is drawn into the silencing pool and is capable of further extending the effect to an Adh-w construct, which has no portion in common with w-Adh but which does share homology with the endogenous Adh 5' sequences. The silenced Adh-w construct also accumulates the Pc-G complex. Deletion of the endogenous Adh gene eliminates the silencing interaction between the two nonhomologous, reciprocally constructed transgenes, w-Adh and Adh-w (Pal-Bhadra, 2002 and references therein).

The full-length Adh transgene silencing operates posttranscriptionally as determined by nuclear run-on transcription assays that directly assess the distribution of transcriptionally-engaged Pol II. In contrast, the w-Adh and Adh-w silencing is transcriptional. The posttranscriptional silencing of Adh correlates with the appearance of 21-25 bp sense and antisense RNAs as occurs with virus and PTGS silencing in plants and RNAi in flies (Pal-Bhadra, 2002).

Mutations have been recovered in C. elegans that are defective for RNAi. Homologs to these genes exist throughout the plant and animal kingdoms. One of these mutations, rde1, has several related gene family members in flies. A homozygous viable member of this group was tested for its impact on transgene silencing. The piwi mutation drastically reduces the magnitude of posttranscriptional silencing. Surprisingly, piwi also had a strong impact on the transcriptional silencing of Adh-w by w-Adh. This result indicates that under certain circumstances the two types of silencing can be mechanistically related (Pal-Bhadra, 2002).

When one to five copies of full-length Adh transgenes are introduced in the genome, the steady-state RNAs accumulate in direct correlation with copy number. However, at higher dosage (six to ten), Adh RNA levels depart dramatically from a linear relationship with transgene copy number. The transgene analyzed contains all the Adh sequences required for normal function. Five single insert strains showing minimal positional effects were selected. By combining these insertions via genetic crosses, a series of Adh stocks was generated that carry one to ten copies. Each stock is also homozygous for the endogenous transformation recipient allele, Adh^fn6. This allele is defective at the 3' splicing acceptor of the first intron, resulting in a longer transcript. The endogenous allele produces only 5%-10% of the steady-state level of a normal Adh gene (Pal-Bhadra, 2002).

To determine whether the endogenous gene was silenced in concert with the transgenes, an RNase protection assay was used that distinguishes the two RNAs. The normal full-length Adh RNA protects two fragments, 142 and 160 bp, while the Adh^fn6 RNA protects a 355 bp fragment due to defective splicing. To correct for loading differences, a ß-tubulin probe was included, which protects a 70 bp fragment. The level of each protected fragment (142/70 or 160/70 ratio) was reduced at higher dosage. The amount of endogenous Adh^fn6 transcripts did not show any significant difference in one to five copies, suggesting an equal expression. However, these transcripts followed a similar trend as those from the transgenes at higher dosage (Pal-Bhadra, 2002).

To test for any positional influence on Adh transgene silencing, separate sets of one to nine copy stocks were examined. A similar level of Adh expression at any one dosage suggests that the silencing is not significantly affected by the insertion sites. RNA in situ analyses in embryos the Adh^fn6 strain, as well as with one, five, and seven Adh transgene copies, indicates a linear increase of Adh expression to five copies. Silencing of the seven copy stock was already evident during blastoderm, the stage where Adh RNA is accumulated initially (Pal-Bhadra, 2002).

To determine whether the Adh transgene silencing is transcriptional or posttranscriptional, nuclear run-on assays were performed. ß-tubulin was included as an internal control in order to determine the relative amount of Adh transcription, while lacZ acted as a negative control (Pal-Bhadra, 2002).

Transcription levels were estimated in adult flies carrying zero to ten copies of full-length Adh transgenes. The results reveal that the endogenous Adh^fn6copies, present in each stock, are transcribed similarly to two normal copies. In the one to ten copy series, the amount of transcription increases proportionally to the transgene copy number relative to ß-tubulin (Pal-Bhadra, 2002).

A similar experiment was performed using flies that contain selected genotypes from zero to six copies of the w-Adh hybrid constuct. In the absence of transgenes, the Adh^fn6 allele is transcribed at the normal level. When one copy of the w-Adh transgene is present, the transcript level is increased as expected. In contrast, two copies of w-Adh exhibit a plateau of transcript accumulation. In the presence of more w-Adh copies (four to six), total Adh transcription is reduced progressively. The transcribed RNA in four to six copy stocks is below the level produced by the Adh^fn6 alleles alone. The amount of Adh transcription in males is always greater than in females with equal dosage (Pal-Bhadra, 2002).

Adh transcriptional level (as detected using run-on assays), produced by multiple full-length Adh or w-Adh transgenes, was compared to steady-state mRNA levels produced by the same genotypes, as determined in Northern analyses. In the Adh series, the amount of mRNA is proportional to dosage from one to five, while in six to ten copies, the steady-state levels depart from linearity. This difference from the run-on assays indicates that silencing of full-length Adh transgenes is posttranscriptional. On the contrary, with the w-Adh dosage series, a similar curve was found when the transcription level and steady-state RNA were compared. A parallel pattern of reduction of both nascent and mature RNA suggests that w-Adh silencing, which includes an effect upon the endogenous Adh gene, is transcriptional. The reason for the different modes of silencing of the two types of transgenes is not known (Pal-Bhadra, 2002).

The silencing of w-Adh and the endogenous Adh locus can also be extended to an Adh-white (Adh-w) transgene that is the reciprocal construct of w-Adh. Run-on transcription analysis of the Adh-w transgene in a w deficiency background with varying numbers of w-Adh indicates that this type of silencing also occurs on the transcriptional level. A comparison of the run-on data with a previous Northern analysis of Adh-w with increasing w-Adh copies shows a similar relationship. The Adh-w + 1 w-Adh and Adh-w + 4 w-Adh genotypes have nearly identical levels of reduction in gene expression in the two assays. There is a residual amount of transcription in the Adh-w + 2 w-Adh genotype but no detectable RNA in the Northern analysis, which might suggest that a combination of PTGS and TGS is operating. However, these data points lie near the limit of detection for the two techniques and likely differ by chance due to exposure time because the same genotype assayed subsequently by both methods shows a similar low level of expression. Thus, at the developmental stage examined (adults), the silencing is predominantly, if not exclusively, transcriptional (Pal-Bhadra, 2002).

Biochemical experiments on RNA interference (RNAi) in Drosophila have shown that double-stranded RNAs in the form of 21-25 nt fragments are generated and these fragments are used in the sequence-specific degradation of mRNA. Therefore, the in vivo existence of such RNAs was tested in the case of Adh silencing. The 21-25 nt antisense Adh RNAs were strongly accumulated in the stocks that contain six to ten copies of the Adh transgene. At the lower doses, the same RNA was not detected or was present at very low amounts in the four and five copy stocks. Using an antisense probe for the detection of sense fragments, a similar-sized RNA was found in the silencing doses (Pal-Bhadra, 2002).

Whether transcriptional silencing induced by w-Adh transgenes was associated with small RNAs was tested by analyzing stocks containing zero to six copies of w-Adh. Hybridization with sense or antisense Adh probes revealed that low molecular weight RNA was not found in abundance at any w-Adh dosage (Pal-Bhadra, 2002).

To investigate the nature of PTGS, it was reasoned that an induced hsp70-Adh construct with four Adh transgenes would bring the Adh expression over the silencing threshold. The hsp70-Adh construct contains a heat shock promoter and the Adh reporter gene. Adh RNA from adult flies was measured relative to ß-tubulin controls in three different circumstances: no heat, heat shock (37°C for 45 min), and 20 hr after heat shock. In flies with five full-length Adh copies, the amount of RNA in heat shock and 20 hr after heat shock did not change relative to that of the untreated flies (Pal-Bhadra, 2002).

The heat treatment of adult flies carrying two copies of the hsp70-Adh gene increased their RNAs as expected. In contrast, the heat treatment of a stock that carries four copies of the Adh transgene and in addition an hsp70-Adh gene causes a rapid loss of Adh mRNA. A similar response was found in embryos. The heat shock induced increase in Adh RNA appears to surpass the threshold limit that triggers silencing. A time course sampling of RNA after heat shock showed that Adh mRNA levels were partially recovered 20 hr subsequent to heat treatment. Interestingly, the transient silencing in this case is in contrast to the systemic spread and prolonged silencing that occurs in C. elegans and plants following a localized induction of posttranscriptional silencing (Pal-Bhadra, 2002).

Whether pulsed threshold induced RNA degradation is correlated with the small species of Adh RNA was tested. The small RNAs were found at a high level in flies with the combination genotype after heat shock. Treatment of the control stock composed only of two hsp70-Adh copies does not produce the small RNAs. This result suggests a threshold level must be exceeded to initiate the synthesis of the small RNAs, which in turn participate in the destruction of the homologous mRNA (Pal-Bhadra, 2002).

Because Polycomb and Polycomb-like (Pcl) mutations have a significant effect on w-Adh transcriptional silencing and Pc-G proteins are bound at the sites of the transgenes under silencing conditions, binding of Pc-G proteins at the Adh transgene insertion sites was tested cytologically. Those sites that do not overlap (1CD and 53B) normal positions of Pc labeling were examined for evidence of binding in the single insert stocks and for the same insert in the ten copy larvae. The Pc proteins were not observed at these sites in either case. Therefore, Pc protein recruitment is not a consequence of posttranscriptional silencing of Adh (Pal-Bhadra, 2002).

A group of genes has been characterized that affects PTGS or RNAi in fungi, plants, and animals. Some of the mutations show sequence similarity to members of the piwi/sting/eIF2C/argonaute gene family conserved from plants to vertebrates. Members of this family, including piwi, have been characterized in Drosophila . To determine whether this mutation has any effect on PTGS and TGS of Adh, the role of this mutation on posttranscriptional silencing was tested. Two or four copies of fully functional Adh genes (two normal endogenous copies ± two transgenes) plus a hsp70-Adh transgene in combination with either heterozygotes or homozygotes of piwi were examined. In the stocks analyzed, a normal Adh allele, rather than Adh^fn6, resides on both the balancer chromosome present in heterozygotes and on the chromosome carrying the piwi mutations. These endogenous Adh copies contribute greater amounts of mRNA to the total pool than Adh^fn6 (Pal-Bhadra, 2002).

The heat shock and nonheat shock RNAs of the respective stocks were compared in quantitative Northern blots. The stocks that are heterozygous for the piwi ¹ or piwi ² alleles with two copies of the normal Adh gene plus hsp70-Adh exhibit an increase of RNA after heat incubation. In contrast, a sharp reduction of the Adh RNA was found in the five copy (four Adh + hsp70-Adh) stock following heat shock. In piwi homozygotes, the presence of two endogenous genes + hsp70-Adh has a similar pattern of expression as in heterozygotes following heat shock. However, the Adh mRNA level in the four Adh + hsp70-Adh stocks that are homozygous for piwi alleles is restored to nearly normal after heat shock. An almost equal level of Adh expression in nonheat shock and heat shock flies suggests that the piwi mutation disrupts the threshold-based posttranscriptional Adh silencing. The two separate piwi alleles examined as well as their heteroallelic combination, which was used to minimize any influence of linked modifiers, produced the same results (Pal-Bhadra, 2002).

Examination of 21-25 nt RNAs in the piwi-segregating classes shows that the small RNAs are present in the heat shocked class with four Adh genes plus a hsp70-Adh construct when the flies are heterozygous for piwi. In contrast, these RNAs are diminished in the piwi homozygotes that carry the same constellation of Adh genes and that were subjected to heat shock. This result suggests that piwi acts before or during the formation of these small RNAs (Pal-Bhadra, 2002).

As a first measure of the effect on transcriptional silencing, the interaction between the reciprocal w-Adh and Adh-w transgenes was examined because the expression of the latter can be observed phenotypically. One copy of w-Adh reduces and two copies nearly eliminate the expression of Adh-w. The w-Adh transgenes among themselves participate in silencing and include endogenous Adh. The w-Adh/Adh-w silencing interaction is eliminated when the endogenous Adh is deleted from the genome, suggesting that it mediates the reaction via mutually homologous sequences. The expression of the Adh-w transgene can be assayed on the RNA level by probing for white RNA because the transgene is present in a stock with the normal white gene deleted. In the same RNA populations, the silencing of w-Adh and its effect on endogenous Adh can be assayed by probing for total Adh messenger RNA (Pal-Bhadra, 2002).

piwi mutations were introduced into a background with an Adh-w and a w-Adh transgene. The eye color of Adh-w flies, which is reduced in the interaction with one w-Adh copy, is restored to a normal level in piwi¹ or piwi² homozygotes. The heteroallelic combination piwi¹/piwi² shows a similar level of restoration. These combinations of piwi alleles did not have any effect on the eye color of Adh-w in the absence of w-Adh, indicating that their effect is due to a relief of silencing (Pal-Bhadra, 2002).

w mRNA was measured in genotypes with Adh-w and zero to two copies of w-Adh segregating for the piwi alleles. The accumulation of white transcripts in the Adh-w/Y flies without a w-Adh transgene was not significantly different in the presence of the mutations. As expected, the w transcripts from Adh-w were reduced from the normal level in the presence of one copy of w-Adh. However, the presence of the homozygous mutations restored the levels significantly but not completely to the normal amount. Moreover, the w transcripts from Adh-w were significantly restored in mutant homozygotes carrying two w-Adh copies (Pal-Bhadra, 2002).

The same RNA blots were hybridized using an Adh probe to estimate the effect on w-Adh/endogenous Adh silencing. The data revealed that the chromosomes carrying either allele of piwi slightly increased the Adh mRNA in the Adh-w flies without w-Adh copies compared to the balancer chromosomes. The effect of the mutant-bearing chromosome may reflect variation at linked modifiers or at Adh itself. In the Adh-w/Y;w-Adh/+ and Adh-w/Y;w-Adh/w-Adh flies, the total Adh expression is significantly reduced as expected when piwi is heterozygous. This Adh expression is minimally increased in piwi homozygotes, although the increase in these classes is likely due to the variation on chromosome 2, rather than a release from silencing. The magnitude of this slight increase is similar with Adh-w alone and in the presence of one or two w-Adh copies, suggesting that the mutations do not interfere with silencing at this step. These data suggest that piwi has no effect on w-Adh/Adh silencing in contrast to the effect on Adh-w (Pal-Bhadra, 2002).

Run-on analysis was conducted on heterozygous and homozygous genotypes to determine the influence of piwi on the transcriptional silencing of Adh-w. The heteroallelic mutant combination has no impact on Adh-w expression in the absence of w-Adh transgenes. In the heterozygous genotypes, introduction of one or two w-Adh copies causes a progressive reduction of Adh-w expression. The magnitude of reduction in expression in the transcriptional assay is quite similar to the Northern analysis, again indicating that the silencing of Adh-w at the adult stage does not appear to have a posttranscriptional component. In the heteroallelic piwi mutant flies, Adh-w transcription is only slightly diminished by the addition of one or two copies of w-Adh. These results indicate that piwi mutations interfere with the transcriptional silencing of Adh-w (Pal-Bhadra, 2002).

Thus, a single transcription unit, namely Alcohol dehydrogenase, can experience two types of transgene silencing: transcriptional and posttranscriptional. The w-Adh/Adh/Adh-w silencing is transcriptional as might have been anticipated from the involvement of the Polycomb chromatin complex. In contrast, the full-length Adh transgene silencing is posttranscriptional. The molecular features of this silencing follow those established from biochemical studies of RNAi. In other words, the specific loss of Adh messenger RNA is accompanied by the appearance of small sense and antisense 21-25 nt length RNAs (referred to as small interfering RNAs, siRNAs). The appearance of these RNAs and the degradation of Adh messenger RNA require a certain threshold over which the silencing is triggered. When these in vivo data are considered together with in vitro analyses of RNAi, it is reasonable to suggest that at a certain concentration of Adh messenger RNA, a double-stranded RNA moiety is transiently formed, presumably by a form of RNA-dependent, RNA polymerase activity. These molecules would then be cleaved to form the siRNAs by an RNase type III nuclease and subsequently incorporated into a larger RNase complex (RISC) to specifically target the homologous mRNA for enzymatic destruction. It is noted, however, that until further study is performed, it remains a formal possibility that the siRNAs form by an alternative pathway that does not involve double-stranded RNA (Pal-Bhadra, 2002).

The piwi gene is a member of a family including the RNAi defective 1 (rde1) gene of C. elegans, which was isolated for its failure to support RNAi. Family members contain a PAZ domain (Cerutti, 2000) that is characteristic of several gene products involved with RNAi. This study demonstrates that piwi mutation blocks the posttranscriptional silencing of Adh, including the production of siRNAs. In contrast, rde1 does not inhibit siRNA formation during RNAi. Another member of this gene family, the aubergine locus, interferes with the germline silencing of the Stellate genes present on the X chromosome (Aravin, 2001). This silencing must occur for male fertility and is accomplished by repeated genes on the Y chromosome that generate siRNAs in conjunction with Stellate. The product of another family member, the argonaute2 gene, is associated with the RISC complex. Thus, several members of this gene family have been implicated in PTGS at various steps (Pal-Bhadra, 2002).

It was of interest to determine whether genes involved with posttranscriptional silencing might also have an impact on transcriptional silencing. Several lines of evidence from plant research have suggested a connection between silencing via RNA in the cytoplasm and changes in the nucleus, although no previous data in animal species have indicated such a relationship. For example, cDNAs of plant viruses (or other sequences incorporated into the virus) transformed into the nucleus are undermethylated until infection by the corresponding virus. The presence of the virus in the cytoplasm undergoing silencing causes a hypermethylation of the sequences in the nucleus. These DNA modifications are coincident in length with the homologous portion carried in the virus. While the mechanism of this RNA-directed DNA methylation is unknown, it is presumed to involve an RNA-DNA interaction. In addition, ectopic transcription of promoter sequences will cause transcriptional silencing of transgenes with a homologous promoter driving another reporter gene (Pal-Bhadra, 2002).

The piwi mutations inhibit the transcriptional silencing of Adh-w. Two aspects of the Adh-w silencing can be assayed by probing for either white or Adh messenger RNA. The silencing of w-Adh and its effect on endogenous Adh can be monitored by examining the amount of Adh RNA. With increasing dosage of w-Adh, the total Adh RNA declines. This trend is not affected by the mutations. The silencing of Adh-w itself can be determined phenotypically and by measuring the amount of w RNA. Typically, the expression of Adh-w declines with increasing dosage of w-Adh, but this response is strongly diminished in homozygotes of piwi (Pal-Bhadra, 2002).

While the evidence presented here suggests a relationship between posttranscriptional and transcriptional silencing, the basis for this connection is unknown. The step in the silencing of Adh-w that is affected by piwi requires the presence in the nucleus of the 5' regulatory sequences of Adh. These sequences are not known to be transcribed under normal circumstances (Pal-Bhadra, 2002).

There are several possibilities to draw a connection between an RNAi-like mechanism and this transcriptional silencing. First, there may be undetected, transient transcription of the Adh regulatory sequences, which in turn form homologous siRNAs. These may act in the nucleus to trigger a chromatin change at the complementary regulatory regions in an analogous fashion to experimental transcription of promoters in tobacco. The piwi mutations might block such a mechanism by inhibiting the formation of the siRNAs homologous to the regulatory sequences. Considering this scenario, in the case of the w-Adh/Adh-w interaction, one must postulate that increased dosage of w-Adh would increase the amount of ectopic transcription of the endogenous Adh regulatory sequences and that the Polycomb complex becomes targeted to Adh-w in conjunction with a homologous interaction between the siRNAs and the regulatory region of Adh-w (Pal-Bhadra, 2002).

If small RNAs trigger transcriptional silencing, it is of interest why w-Adh/Adh silencing is unaffected by piwi. One potential explanation is that the RNA involvement is needed only for the establishment of transcriptional silencing but not its maintenance through to the adult stage. The initiation of w-Adh/Adh silencing in early embryogenesis precedes that of Adh-w. It is possible that maternal contributions of piwi product in early embryos are sufficient for initiation of w-Adh/Adh silencing but are depleted to effective levels by the time of the establishment of Adh-w silencing (Pal-Bhadra, 2002).

Another explanation suggests that the piwi gene product may play dual roles in an RNAi-like mechanism and transcriptional silencing. One possibility is that it could affect some aspect of gene-to-gene association that might trigger transcriptional silencing. Pairing of transgenes or other intranuclear interactions appears to have an impact on transcriptional gene silencing. Indeed, the w-Adh transgene is silenced much more effectively when paired between homologs than when two dispersed copies are present in the nucleus. If the piwi product participates in some aspect of sequence recognition, nucleic acid associations or protein-nucleic acid interactions, its elimination might block certain steps of both posttranscriptional and transcriptional silencing. It remains a possibility that gene-to-gene associations might trigger silencing as well as dsRNA-to-gene interactions with the two acting independently but by a related mechanism. Finally, the connection between the two types of silencing could be indirect, with piwi blocking a function secondarily removed from the transcriptional silencing but that is required nevertheless (Pal-Bhadra, 2002).

Certainly, members of this gene family have diverse roles in the cell. A previously identified function of the nucleoplasmic product of piwi indicates its requirement for germline stem cell renewal. The aubergine product is required for dorsoventral patterning of the early embryo, which is mediated by an enhancement of the translation of the oskar mRNA. The aubergine product is also implicated with the expression of the Stellate (Ste) repeats and Suppressor of Stellate [Su(Ste)] genes on the X and Y chromosomes. The argonaute1 gene is required early in Drosophila embryogenesis for proper development. In mammals, a member of this gene family, eIF2C, functions in translation initiation via tRNA association with messenger RNA, while in Arabidopsis the argonaute mutation affects the functions of the meristem. In C. elegans, Drosophila, and human cells, related genes play a role in the maturation of small regulatory RNAs involved in the temporal control of development. The data presented here suggest that they can also play a direct or indirect role in transcriptional regulation. Clearly, many aspects of development and cellular metabolism use these gene products, raising the possibility that the transposon and virus defense functions may have been co-opted early in evolution from genes involved with other regulatory processes (Pal-Bhadra, 2002).

Regulatory relationship among piwi, pumilio, and bag-of-marbles in Drosophila germline stem cell self-renewal and differentiation

The transition from a Drosophila ovarian germline stem cell (GSC) to its differentiated daughter cell, the cystoblast, is controlled by both niche signals and intrinsic factors. piwi and pumilio (pum) are essential for GSC self-renewal, whereas bag-of-marbles (bam) is required for cystoblast differentiation. This study demonstrate that Piwi and Bam proteins are expressed independently of one another in reciprocal patterns in GSCs and cystoblasts. However, overexpression of either one antagonizes the other in these cells. Furthermore, piwi;bam double mutants phenocopy the bam mutant. This epistasis reflects the niche signaling function of piwi because depleting piwi from niche cells in bam mutant ovaries also phenocopies bam mutants. Thus, bam is epistatic to niche Piwi, but not germline Piwi function. Despite this, bam⁻ ovaries lacking germline Piwi contain approximately 4-fold fewer germ cells than bam⁻ ovaries, consistent with the role of germline Piwi in promoting GSC mitosis by 4-fold. Finally, pum is epistatic to bam, indicating that niche Piwi does not regulate Bam-C through Pum. It is proposed that niche Piwi maintains GSCs by repressing bam expression in GSCs, which consequently prevents Bam from downregulating Pum/Nos function in repressing the translation of differentiation genes and germline Piwi function in promoting germ cell division (Szakmary, 2005).

This study investigates the regulatory relationships between Piwi, Bam, and Pum, three key regulators of GSC versus cystoblast fates. Among them, Pum and Bam are intrinsic factors, whereas Piwi is expressed both in niche cells as an essential component of niche signaling and in GSCs to promote its division. Pum was originally identified as a maternal effect protein that heterodimerizes with Nanos (Nos) to bind and suppress the translation of its target hunchback mRNA in the posterior of the Drosophila embryo. In addition, Pum and Nos have important germline development zygotic roles, including their cell-autonomous function for GSC maintenance. In contrast to this function of Pum and Nos, Bam is necessary and sufficient in promoting GSC differentiation, even though its molecular activity is not known. bam encodes two protein isoforms: the cytoplasmic (Bam-C) and the fusomal (Bam-F) forms, with Bam-C specifically present in cystoblasts and differentiating cysts but absent in GSCs. Finally, Piwi is the founding member of the evolutionarily conserved Piwi protein family (a.k.a. Argonaute family) involved in stem cell division, RNA interference, transcriptional gene silencing, and other developmental processes. In the Drosophila ovarian germline, Piwi is a nuclear protein that is preferentially expressed in GSCs but is only weakly expressed in cystoblasts and mitotic cysts, consistent with its germline function (Szakmary, 2005).

To investigate the regulatory relationship between piwi and bam, the reciprocal expression pattern was confirmed by double immunofluorescence microscopy of wild-type germaria for Piwi and Bam-C. A fully functional myc-tagged Piwi is expressed at high levels in GSCs and is downregulated in cystoblasts and early mitotic cysts. In contrast, Bam-C is absent from GSCs but accumulates in most cystoblasts and mitotic cystocytes in germarial region 1. The downregulation of Piwi coincides with the zone of Bam-C expression. In a few cases, germ cells were observed expressing both Piwi and Bam-C in cystoblast positions. These cells might represent the transitional stage from GSCs to cystoblasts. At a very low frequency, cystoblast-like cells express low levels of Piwi, but no detectable Bam-C. On the basis of piwi;bam double mutant analysis, these cystoblast-like cells are likely to be undifferentiated or potentially apoptotic. Overall, the reciprocal expression pattern of Piwi and Bam-C proteins supports the opposing functions of piwi and bam genes (Szakmary, 2005).

To determine whether this reciprocal expression pattern is a result of mutually negative regulation toward each other's expression, Bam expression was analyzed in piwi mutants and vice versa. Because piwi mutant ovarioles typically contain germaria that are depleted of germline cells, it is difficult to assay Bam-C expression in them. For this reason, piwi¹ GSC clones were generated with the FLP-DFS (flipase-mediated dominant female sterile) technique. Bam-C is expressed normally in cystoblasts and early mitotic cysts in germaria that contain only piwi¹ germline cells. Moreover, no ectopic Bam expression was detected in GSCs. Therefore, proper Bam-C expression in the adult germline during oogenesis does not require piwi⁺ function in the germline. To address whether piwi expression in apical somatic cells affects bam expression in GSCs, Piwi was eliminated in somatic niche cells. This was achieved by using Yb mutations that eliminate Piwi expression specifically in niche cells. Yb mutants are phenotypically very similar to piwi mutants. However, if examined within the first day of eclosion, Yb mutant germaria still contain germ cells. Bam-C expression is unaffected in adult Yb mutant ovaries, suggesting that there is no specific requirement for YB or Piwi in niche cells for proper Bam-C expression or localization in the germline. Taken together, the above analyses indicate that neither niche nor germline piwi is required for Bam-C expression (Szakmary, 2005).

Whether the absence of Bam affects Piwi expression was examined. A P[myc-piwi] transgene was introduced into a bam null mutant background to monitor Piwi expression. Ovaries were dissected from these females and stained with Myc antibody to monitor the Piwi expression and with Vasa antibodies to label germ cells. Piwi is present in all bam null germline cells. In addition, Piwi is also strongly expressed in apical somatic cells that correspond to terminal filament, cap cells, and inner sheath cells in the wild-type germarium. Therefore, bam⁺ function is dispensable for Piwi expression in the germline and the apical somatic cells (Szakmary, 2005).

Whether piwi or bam negatively regulates the expression of the other was tested. Overexpression of Piwi in apical somatic cells increases the number of GSC-like cells. These ectopic stem cell-like cells fill regions 1 and 2a of the germarium and, thus, displace Bam-C-expressing cells to region 2b. If Piwi and Bam expression are mutually antagonistic, the prediction would be that expanding Bam expression to GSCs would downregulate Piwi expression there during oogenesis. To express Bam-C protein ectopically in GSCs, a heat shock-inducible bam transgene was used that places the bam cDNA under the control of the hsp70 promoter. Flies carrying a P[myc-piwi] and a hs-bam transgene were subjected to heat shock twice daily for 3 days after eclosion. Ovaries were subsequently dissected and stained for Bam-C and Myc to monitor ectopic Bam-C expression and its effects on Piwi expression. As predicted, ectopic expression of Bam in GSCs diminishes Piwi expression specifically in these cells. Interestingly, ectopic Bam expression in both somatic cells and other germline cells within and beyond the germarium has no effect on Piwi expression in these cells. Particularly, Piwi expression in apical somatic cells (i.e., cap cells and inner sheath cells) of the germarium is unaffected by ectopic Bam expression. Thus ectopic Bam expression may specifically downregulate the germline Piwi expression (Szakmary, 2005).

The mutually independent expression of Piwi and Bam does not rule out their regulatory relationship in GSC cell fate, whereas the suppression of piwi in GSCs by ectopic bam expression suggests that these two genes interact antagonistically. To further define the interaction between piwi and bam, females were constructed lacking both piwi and bam function and the double mutants' ovaries were examined. In contrast to the piwi mutant phenotype, in which ovarioles typically contain a germlineless germarium and 2-3 egg chambers, the double mutant ovaries are characterized by 'tumorous' germaria filled with hundreds of undifferentiated germ cells. Moreover, there is no apparent egg chamber development in the double mutant ovary. This phenotype is qualitatively similar to the tumorous phenotype observed in bam mutant ovaries, which can contain up to thousands of undifferentiated germ cells. The piwi;bam double mutant phenotype therefore indicates that bam is epistatic to piwi. Given the opposing functions of piwi and bam, these results suggest that piwi acts upstream of bam to repress its function in promoting GSC differentiation (Szakmary, 2005).

Although the piwi;bam double mutant shows a bam-like phenotype, there is a difference between the defect of the double mutant and that of bam alone. The bam mutant typically contains 300-1000 undifferentiated germ cells, whereas the piwi;bam double mutants contain only 50-300 germ cells. One possible explanation for this difference is the absence of the mitosis-promoting, germline cell autonomous piwi function in the double mutant. The cell autonomous function of piwi in GSCs is to promote mitosis, resulting in a 4-fold increase of mitotic rates. In bam mutants, 'tumorous'germ cells are more mitotic because of the presence of piwi⁺ function, whereas in piwi;bam double mutants, 'tumorous' germ cells are less mitotic because of the absence of piwi⁺ function. Therefore, these analyses suggest that, whereas bam is epistatic to the niche function of piwi, the cell autonomous function of piwi is epistatic to bam (Szakmary, 2005).

To verify the complex epistasis between bam and distinct somatic versus germline functions of Piwi, the effect of specifically removing Piwi protein from either the germline or the somatic niche cells of bam mutants was investigated. The piwi (somatic);bam double mutant was achieved by generating Yb;bam double mutants because Yb specifically eliminates piwi expression in niche cells. The piwi (germline);bam double mutant was achieved by driving transgenic piwi expression specifically in the niche cells of a piwi;bam double mutant background (Szakmary, 2005).

Yb;bam double mutant ovaries display a clear bam phenotype. This phenotype, however, is not as attenuated as in piwi;bam double mutants, but rather appears to be as pronounced as in bam single mutants. This result supports the assumption that the epistasis of bam over piwi reflects the somatic piwi function, and the attenuated bam phenotype of the double mutant reflects the germline cell autonomous piwi function (Szakmary, 2005).

To further verify this hypothesis, the phenotype of piwi (germline);bam double mutants was analyzed. The piwi (germline) mutant was generated with an en-gal4 transgene to drive the expression of piwi^EP to produce specific expression of fully functional Piwi in niche cells. Because piwi^EP is inserted into the piwi locus, it is therefore a piwi mutant allele in the absence of gal4 expression. The en-gal4 piwi¹/piwi^EP transheterozygotes were generated in bam mutant and wild-type backgrounds. The piwi/piwi^EP;bam⁺ ovaries display the expected piwi mutant phenotype. The en-gal4 piwi¹/piwi^EP;bam^Δ86~/TM3 Sb ovaries appear wild-type, aside from a mild reduction in size, and give rise to females capable of laying eggs. This finding directly confirms that Piwi expression in niche cells is sufficient for GSC maintenance, whereas the observed reduction in ovary size may reflect the absence of germline piwi function in promoting GSC mitoses. As expected, the en-gal4 piwi¹/CyO;bam^Δ86/bam^Δ86 flies display typical bam mutant ovarioles. Also as expected, piwi¹/piwi^EP;bam^Δ86/bam^Δ86 and en-gal4 piwi¹/piwi^EP;bam^Δ86/bam^Δ86 ovaries display the phenotypes of piwi (somatic);bam double mutants and the piwi (germline);bam double mutants, respectively. These analyses further verified that bam is epistatic to somatic niche piwi function, yet germline piwi is epistatic to bam function (Szakmary, 2005).

The fact that Piwi expression in somatic cells has a downregulating effect on Bam-C expression in GSCs raises the question of how this signal may be relayed. The reciprocal expression patterns of Piwi and Bam-C in the germline closely resemble those of Pum and Bam-C. Pum maintains GSC self-renewal during oogenesis, whereas Bam promotes GSC differentiation. GSCs are depleted in pum mutants but overproliferate in bam mutants. This raised the possibility that Piwi may exert its functions by acting on Pum. Pum expression was therefore examined in piwi¹ mutants and Piwi expression in pum¹⁶⁸⁸ and pum²⁰⁰³ mutants. The expression of one gene was not detectably altered in the mutant background of the other, suggesting that neither gene regulates the other's expression. However, pum encodes two distinct protein isoforms (156 kDa and 130 kDa). Either isoform is sufficient for maternal function, but both are required for zygotic function, including GSC maintenance. Because pum¹⁶⁸⁸ and pum²⁰⁰³ eliminate the expression of the 156 kDa and 130 kDa Pum isoforms, respectively, these results could suggest that either the 156 kDa or the 130 kDa isoform of Pum alone is sufficient for proper germline Piwi expression. Even if this is the case, the niche expression of piwi is independent of pum because pum is not required somatically to maintain GSCs (Szakmary, 2005).

To more definitively determine the regulatory relationship between Pum and the Piwi-Bam-C pathway, pum,bam double mutants were constructed and analyzed. If somatic Piwi acts through Pum to regulate Bam-C, then pum,bam double mutants should resemble piwi;bam double mutants. This, however, was not the case. In pum^ET1,bam^Δ86/pum^ET9,bam^Δ86 double mutant flies, in which both pum^ET1 and pum^ET9 are null alleles, the majority of germaria were devoid of germ cells. Only a minority of germaria (<10%) contained a number of undifferentiated germ cells with restricted proliferation. This range of defects is indistinguishable from that of the phenotype of typical pum mutant ovaries. These results suggest that pum is epistatic to bam and that the proliferation of germ cells in bam mutants requires Pum function (Szakmary, 2005).

In summary, these results show that somatic niche Piwi function antagonizes Bam-C, which in turn antagonizes Pum and germline Piwi. The niche function of Piwi in downregulating BAM function appears to converge with the Dpp signaling pathway that is also required for GSC maintenance. This is based on three observations. (1) the expression of dpp does not require piwi. Therefore, Dpp is not a downstream signal of piwi. (2) dpp overexpression does not rescue piwi mutant defects. Therefore, Dpp and niche Piwi are functionally parallel. (3) The dpp signaling pathway directly represses bam transcription. Likewise, piwi niche signaling also downregulates bam expression because bam is epistatic over piwi and because overexpression of Piwi in germarial somatic cells causes overproliferation of GSCs and displaces Bam-C expression beyond region 1 and 2a of the germarium. Taken together, these results indicate that these two signaling pathways must converge at some point to regulate Bam function. The convergence point could be in niche cells, where piwi directly affects Dpp signal production by aiding in its modifications, stability, and/or secretion. Alternatively, it could be in GSCs, where Piwi suppresses a Dpp agonist(s) or perhaps even the Bam/Bgcn complex. This scenario would require that Piwi produce an intercellular signal independent of Dpp. At present, these two possibilities cannot be distinguished (Szakmary, 2005).

How does Bam function as a converging target in promoting GSC differentiation? It has been suggested that benign gonial cell neoplasm (Bgcn) is an obligatory partner for Bam-C as a differentiation factor. Bgcn is expressed in GSCs, but not in somatic cells. This may explain why ectopic bam expression only downregulates Piwi in GSCs, but not in somatic cells (Szakmary, 2005).

How is Pum involved in the Piwi-Bam pathway? Piwi and Pum do not affect one another's expression, yet pum is clearly epistatic to bam. This precludes the possibility that Piwi exerts its effect on Bam-C via Pum. The epistasis of pum to bam is best explained by ascribing a translational repressing function of Pum/Nos in the germline toward mRNAs that promote differentiation. This repression is released by Bam/Bgcn. In GSCs, Bam-C is itself transcriptionally silenced; therefore, Pum and Nos are active in suppressing differentiation. In cystoblasts, Bam/Bgcn are expressed, thereby antagonizing Pum/Nos function. This allows differentiation-promoting mRNAs to be translated. Bgcn is related to the DexH-box family of RNA-dependent helicases. Recently, it has been suggested that the majority of RNA helicases function by displacing proteins from RNA strands rather than by unwinding RNA. It is therefore conceivable that the Bam/Bgcn complex displaces Pum/Nos from their target RNAs (Szakmary, 2005).

A model is proposed for switching between self-renewal and differentiation of GSCs in the Drosophila germarium. The niche cells signal to GSCs by secreting Dpp/Bmp and possibly other proteins. The Dpp signal is received by GSCs through its receptors Punt and Thick Veins (TKV), and it is transduced by pMad to silence bam transcription in these cells. This is achieved via the direct binding of Smads to a discrete silencing element in the bam gene. Piwi in niche cells has an essential and cooperative functional involvement in this signal. Piwi and Dpp signaling pathways converge at some point upstream of bam, in either niche cells or GSCs. The absence of Bam allows Pum and Nos to be active, which suppresses the translation of differentiation genes, thus maintaining the stem cell fate. In the cystoblast and differentiating germline cysts, the Dpp signal is no longer effective, thereby relieving the transcriptional repression of bam. The Bam/Bgcn complexes then repress Pum/Nos function, allowing these cells to differentiate. Therefore, Pum/Nos can be considered a switch between self-renewal and differentiation, whereas niche signaling through Bam/Bgcn regulates this switch at a single cell level (Szakmary, 2005).

Gene circuitry controlling a stem cell niche

Many stem cell populations interact with stromal cells via signaling pathways, and understanding these interactions is key for understanding stem cell biology. In Drosophila, germline stem cell (GSC) maintenance requires regulation of several genes, including dpp, piwi, pumilio, and bam. GSCs also maintain continuous contact with cap cells that probably secrete the signaling ligands necessary for controlling expression of these genes. For example, dpp signaling acts by silencing transcription of the differentiation factor, bam, in GSCs. Despite numerous studies, it is not clear what roles piwi, primarily a cap cell factor, and pumilio, a germ cell factor, play in maintaining GSC function. With molecular and genetic experiments, it is shown that piwi maintains GSCs by silencing bam. In contrast, pumilio is not required for bam silencing, indicating that pumilio maintains GSC fate by a mechanism not dependent on bam transcription. Surprisingly, it was found that germ cells can differentiate without bam if they also lack pumilio. These findings suggest a molecular pathway for GSC maintenance. dpp- and piwi-dependent signaling act synergistically in GSCs to silence bam, whereas pumilio represses translation of differentiation-promoting mRNAs. In cystoblasts, accumulating Bam protein antagonizes pumilio, permitting the translation of cystoblast-promoting transcripts (Chen, 2005).

dpp-dependent silencing of bam transcription defines a key -- probably the primary -- mechanism for maintaining GSCs. By repressing bam transcription in the germ cells attached to cap cells, dpp signaling prevents these cells from forming cystoblasts and assigns them as GSCs. It is speculated that all GSC maintenance genes might act by repressing bam transcription and this prediction was tested for piwi and pumilio (Chen, 2005).

Two genetic observations suggested that piwi might negatively regulate bam expression: (1) bam was epistatic to piwi in double mutants, indicating that the piwi GSC-loss phenotype required an active bam gene; (2) Bam coexpression suppressed the formation of extra GSCs induced when piwi was overexpressed. Thus, piwi-dependent GSC formation depends on maintaining low levels of bam expression (Chen, 2005).

Both overexpression and loss-of-function phenotypes could be explained if piwi, like dpp signaling, were necessary to silence bam transcription in GSCs. This possibility was tested by scoring the expression of bam transcriptional reporters in piwi mutant GSCs. Because piwi inactivation causes GSC loss, P{bamP-GFP} reporters were assayed in piwi bgcn (benign gonial cell neoplasm) double mutant flies that preserve GSCs. GSCs lacking bgcn were GFP negative, but GSCs that lacked both piwi and bgcn were GFP positive. Thus, like dpp signaling, piwi⁺ was necessary to silence bam transcription in GSCs (Chen, 2005).

piwi⁺ action in somatic, but not germline, cells is critical for GSC maintenance, and, therefore, piwi must act indirectly to repress bam transcription. A putative piwi target (or targets) in GSCs must integrate with dpp signaling because previous work has established that the Mad:Medea binding site in bam is a sufficient silencer element. Two recent findings drew attention to the E3-ligase Dsmurf as a candidate for a germ cell piwi target: (1) Dsmurf inactivation produces extra GSCs, just as does ectopic piwi expression and (2) Dsmurf suppresses dpp signaling by targeting phosphorylated Mad for degradation (Chen, 2005).

If piwi silences bam transcription by repressing Dsmurf in GSCs, then GSCs might be restored in piwi mutants if Dsmurf were simultaneously removed. Therefore ovaries of piwi Dsmurf double mutant females were examined and it was found that most germaria contained supernumerary GSCs and a continuous supply of egg chambers. It was verified that piwi Dsmurf GSC-like cells behaved as GSCs by noting that they did not express BamC protein. In 62/80 double mutant germaria, no cells expressing BamC were detected, whereas BamC-positive germ cells were detected in 18/80 germaria. In those cases, the most apical cells, in the GSC position, were BamC negative (Chen, 2005).

Dpp signaling and piwi act as GSC maintenance factors by repressing bam transcription. pumilio (pum) is a component of an evolutionarily conserved mechanism of translational control and is also essential for ovarian GSCs. The expression of P{bamP-GFP} reporter was examined in pum mutant germ cells to determine if bam transcriptional silencing also depends on pum⁺. In contrast to piwi, the reporter was properly silenced in pum bam GSCs. For example, GSCs in 84.6% of pum^MSC bam^BG/pum²⁰⁰³ bam^Δ86 germaria were GFP negative. Because pum mutant germ cells are unstable, it was suspected that the few GFP-positive cells in the GSC position had either differentiated or were dying (Chen, 2005).

GSCs required (1) dpp⁺ and piwi⁺ signaling to repress cystoblast (CB) differentiation by silencing bam transcription and (2) pum⁺ to repress CB differentiation by a mechanism that is independent of bam silencing. Because previous work has shown that Pum forms a translational repressor complex with Nanos, it was reasoned that pum⁺ might maintain GSCs by repressing translation of CB-promoting mRNAs. One candidate target mRNA is bam itself, but, because dpp-dependent transcriptional silencing of bam fully accounts for the absence of bam from GSCs, it is unlikely that Pum sustains GSCs by repressing bam translation (Chen, 2005).

The Pum:Nos repressor complex probably blocks translation of other unidentified target mRNAs that are essential for CB differentiation. Cystoblast formation would then depend on relieving this block, and, because bam is both necessary and sufficient to induce CB differentiation, bam might antagonize or bypass translational repression. The phenotypes of double mutants can distinguish between these possibilities. If bam bypasses translational repression, pum bam germ cells would not form CBs and would resemble bam mutant gonads. If, however, bam antagonizes Pum/Nos-mediated translational repression, pum bam germ cells might differentiate (Chen, 2005).

Ovaries formed in various pum and bam genotypes were compared with several alleles of each gene. Double mutant ovaries produced a complex phenotype that was distinct from either single mutant. Staining nuclei with DNA dyes revealed a mixture of apparently undifferentiated cells and overtly polyploid cells. Indeed, in many cases the polyploid chromosomes were also thick and expanded like nurse cell chromosomes. Most remarkably, these pseudo-nurse cells were occasionally organized within an epithelial layer of follicle cells, like a cyst, although these cysts never contained a full complement of 16 cystocytes. Cells with hallmarks of post-CB differentiation occurred only in the pum bam double mutant ovaries, where they were seen in over half the ovarioles scored (see Table S2) (Chen, 2005).

The appearance of pseudo-nurse cells and even cysts suggested that double mutant germ cells had formed functional cystoblasts, remarkably bypassing the requirement for bam⁺ expression. To verify that pum bam germ cells were undergoing differentiation, the double mutant cells were examined with several additional markers of differentiation (Chen, 2005).

Orb is expressed in all germ cells, but its levels increase dramatically in the cystocytes of developing cysts. Orb protein levels remain at very low levels in bam mutant cells. Double mutant cells, however, expressed Orb at levels seen in differentiating cysts and well above the levels in bam cells. Orb accumulation revealed that many of the pum bam germ cells that did not yet have obvious pseudo-nurse cell chromosomes had, in fact, progressed well beyond the "pre-CB" stage of bam cells. Pseudo-nurse cells also had high levels of Orb expression, similar to accumulation seen in developing nurse cells (Chen, 2005).

Ring canal formation is a distinctive feature of germ cell cysts, and the multiple cell pum bam cysts contained ring canals. The incidence of these pum bam cysts was modest but reproducible in all double-mutant combinations, including those containing null alleles of bam and very strong or null alleles of pum. It is suspected that the infrequent appearance of multi-nurse cell cysts is due to a second requirement for bam⁺ to drive cystocyte divisions during cyst formation. This requirement would not have been recognized previously because bam mutations arrested cells as 'pre-CBs' (Chen, 2005).

Although they are not normal, the appearance of these cysts is a striking manifestation that CBs lacking bam could differentiate as long as they also lacked pum. Combined with previous studies showing that ectopic bam expression is sufficient to direct GSC differentiation, the pum bam phenotype strongly suggests that bam acts as a CB-promoting factor by antagonizing, rather than bypassing, pum action. The data suggests further that dpp signaling, which directly regulates bam expression, does not control pum⁺ expression. A similar conclusion has been reached about the relationship between dpp signaling and nanos expression on the basis of studies of primordial germ cell differentiation (Chen, 2005).

A unifying model is proposed to explain the gene circuitry of GSC and CB fate within the GSC niche. The results suggest that Drosophila ovarian GSCs are retained as stem cells because Pum:Nos complexes repress translation of a pool of mRNAs that induce CB differentiation (Chen, 2005).

In wild-type GSCs that contact cap cells, Pum:Nos translational repression remains active because dpp signaling from stromal cells silences bam transcription and thus blocks the formation of Bam:Bgcn complexes that would antagonize Pum:Nos translational repression. Expression of piwi in stromal cells contributes a key, but unknown, signal that stabilizes or strengthens the Dpp response and bam transcriptional silencing (Chen, 2005).

After the GSC divides, the strength of Dpp signaling falls to levels that can no longer efficiently silence bam transcription in the cell displaced to the posterior and away from cap cells. This could be due to declining Dpp levels or diminished piwi-dependent signals that lead to reduced phospho-Mad levels. As bam transcription increases, Bam:Bgcn complexes antagonize Pum:Nos action and cause derepression of CB-promoting mRNAs, initiating the events of CB differentiation. Because pumilio and nanos are evolutionarily conserved proteins, it will be important to determine if a similar 'multiple-negative' circuitry is at work in mammalian stem cell niches (Chen, 2005).

RNAi components are required for nuclear clustering of Polycomb group response elements

Drosophila Polycomb group (PcG) proteins silence homeotic genes through binding to Polycomb group response elements (PREs). Fab-7 is a PRE-containing regulatory element from the homeotic gene Abdominal-B. When present in multiple copies in the genome, Fab-7 can induce long-distance gene contacts that enhance PcG-dependent silencing. Components of the RNA interference (RNAi) machinery are involved in PcG-mediated silencing at Fab-7 and in the production of small RNAs at transgenic Fab-7 copies. In general, these mutations do not affect the recruitment of PcG components, but they are specifically required for the maintenance of long-range contacts between Fab-7 copies. Dicer-2, PIWI, and Argonaute1, three RNAi components, frequently colocalize with PcG bodies, and their mutation significantly reduces the frequency of PcG-dependent chromosomal associations of endogenous homeotic genes. This suggests a novel role for the RNAi machinery in regulating the nuclear organization of PcG chromatin targets (Grimaud, 2006).

The RNAi machinery has been implicated in a wide variety of biological processes. One of these processes is the formation of heterochromatin. In S. pombe, this involves bidirectional transcription of RNA molecules from repetitive sequences and their cleavage into short interfering RNAs (siRNAs) of 21–23 nt by an RNase III enzyme called Dicer-1. siRNAs guide the RNA-induced initiation of transcriptional gene silencing (RITS) complex to homologous sequences in the nucleus (Noma, 2004; Verdel, 2004). Clr4, the homolog of the histone methyltransferase Su(Var) 3-9, is recruited along with the RITS complex to chromatin, where it methylates lysine 9 of histone H3 (H3K9). This epigenetic mark promotes the formation of heterochromatin by recruiting the heterochromatin protein Swi6, the homolog of HP1, via its chromodomain (Grewal, 2004). Consistent with these data, a redistribution of H3K9 methylation has been observed in Drosophila chromosomes in flies mutant for components of the RNAi machinery (Pal-Bhadra, 2004; Grimaud, 2006).

The RNAi machinery is also required for cosuppression, a phenomenon whereby the introduction of multiple transgenic copies of a gene phenocopies its loss of function instead of increasing its expression. In Drosophila, cosuppression can act at either the transcriptional or posttranscriptional level and involves PcG proteins as well as the RNAi machinery (Pal-Bhadra, 1997, Pal-Bhadra, 1999 and Pal-Bhadra, 2002; Grimaud, 2006 and references therein).

The Drosophila RNAi machinery includes two Dicer proteins encoded by the dicer-1 (dcr-1) and dicer-2 (dcr-2) genes. Dcr-2 is specifically required to process double-stranded RNAs into siRNAs and mediates the assembly of siRNAs into the RNA-induced silencing complex (RISC). Dcr-1 is involved in the metabolism of siRNAs as well as the processing of pre-microRNAs into microRNAs. RNA silencing also involves several highly conserved genes coding for PAZ-domain proteins. Argonaute1 (AGO1) and Argonaute2 (AGO2) are involved in microRNA biogenesis and RNA interference (RNAi). piwi is involved in cosuppression, silencing of retrotransposons, and heterochromatin formation. aubergine (aub) was first isolated based on its role in germline development but is also responsible for maintaining the silenced state of an X-linked male fertility gene locus (Stellate) via RNAi. The Aub protein is required for RNAi and RISC assembly in ovaries. In addition, homeless/spindle-E (hls) is involved in silencing of Stellate and in heterochromatin formation. This study tested whether RNAi components are involved in the PcG pathway. The results show that the RNAi machinery affects the PcG response via a novel regulatory function in nuclear organization (Grimaud, 2006).

The role of a variety of RNAi components in a specific transgenic line called Fab-X was tested. This line contains a construct carrying a 3.6 kb fragment from the Fab-7 region, cloned upstream of a mini-white reporter and inserted into the X chromosome. In the Fab-X line, the presence of the Fab-7 sequence is sufficient to induce PcG-dependent silencing, both of the mini-white eye-color reporter gene and of the endogenous scalloped (sd) gene, which is required for wing-blade morphogenesis and is located 18.4 kb downstream of Fab-7. These two repressed phenotypes are abolished in the presence of mutations in PcG genes and are not present in heterozygous females and hemizygous males, indicating that both mini-white and sd expression are subject to PSS (Grimaud, 2006).

The eye-color and wing phenotypes were used as a basis to analyze the effect of the RNAi machinery on PcG-dependent repression. Mutations in RNAi components were introduced into the Fab-X line and placed over a balancer chromosome containing a GFP marker. As the AGO1 mutant alleles involve P element insertions containing the mini-white reporter gene, they could not be tested using the eye phenotype. A null mutation in dcr-2 (dcr-2^L811fsX) decreased silencing of the mini-white reporter gene relative to the Fab-X line when in the homozygous state. Likewise, two different mutant alleles of piwi (piwi¹ and piwi²) decreased mini-white silencing, with the effect being more pronounced in piwi² mutant flies. This effect was not restricted to the Fab-X line since it was also observed when piwi² was recombined into another Fab-7-containing line. In contrast to the effects seen for dcr-2 and piwi alleles, Fab-X females homozygous mutant for hls^E1 or that carried the heteroallelic hls^E1/hls^E616 combination silenced mini-white like wt Fab-X females (Grimaud, 2006).

The sd phenotype was then analyzed in all mutant backgrounds at 28.5°C, a temperature inducing a strong wing phenotype in Fab-X. A preselection of non-GFP female larvae was carried out in order to selectively analyze homozygous or trans-heterozygous mutant adults. This analysis revealed that mutating any of the components of the RNAi machinery, except for hls and the heterozygous dcr-1 mutation, leads to a strong decrease in the sd phenotype. These data show that the RNAi machinery can affect PcG-mediated silencing. The fact that hls mutants had no effect suggests that this process might be mechanistically distinct from the role of RNAi components in heterochromatin formation (Grimaud, 2006).

While most RNAi components are not required for binding of PcG proteins to PREs, they are required to mediate long-range contacts between multiple copies of the Fab-7 element. Moreover, Dcr-2, PIWI, and AGO1 colocalize with PH in the cell nucleus, and their effect correlates with the presence of small RNAs homologous to Fab-7 sequences. Finally, in addition to their effects on transgenic Fab-7 copies, mutations in these genes also reduce the frequency of long-distance contacts between endogenous PcG target genes. Taken together, these results reveal a novel and unexpected role for the RNAi machinery in the regulation of euchromatic genes in the nuclear space (Grimaud, 2006).

The effects caused by mutations in different RNAi components suggest the existence of distinct molecular roles for these proteins in the regulation of PcG function. First, the hls gene does not seem to play a major role in silencing at the Fab-7 PRE or maintaining long-distance Fab-7 contacts. Since Hls has been shown to play a central role in heterochromatin formation (Pal-Bhadra, 2004), there may be different subtypes of nuclear RNAi machineries for heterochromatin formation and for regulating PcG function. No effect was found when a mutation in the dcr-1 gene was analyzed at the heterozygous state, but the elucidation of the function of dcr-1 in PcG-mediated silencing awaits further analysis in a homozygous mutant background. A second class of RNAi components that participate in PcG-mediated repression contains dcr-2, AGO1, and the aub gene. Loss of any of these RNAi gene products affects PcG-dependent silencing at Fab-7, although it does not impact the binding of PcG proteins to Fab-7. Only mutations in piwi affected the binding of PcG proteins to Fab-7, at least in polytene chromosomes, but even PIWI did not affect recruitment of PcG factors at endogenous genes. In S. pombe, both RNAi components as well as DNA binding proteins are involved in recruiting heterochromatin proteins to the mating-type region (Jia, 2004). At PREs, multiple DNA binding factors and chromatin-associated proteins are known to contribute to PcG protein recruitment, such as PHO, the GAGA factor, DSP1, and the CtBP proteins. Their combinatorial action might play a key role in the robust and specific chromatin tethering of PcG proteins, while, in contrast to the situation in S. pombe, the RNAi machinery might play a relatively minor role (Grimaud, 2006).

It is interesting to note that piwi mutations affected recruitment of E(Z) and PC to Fab-7 in polytene chromosomes but had no effects on PH, another PRC1 component. In the current model for recruitment of PcG proteins to PREs, histone H3 methylation by the E(Z) protein recruits PRC1 via the chromodomain of PC. The current results indicate that multiple mechanisms might be used to anchor different PRC1 components to PREs and that the loss of PC does not necessarily lead to the disintegration of the entire PRC1 complex at PREs (Grimaud, 2006).

To date, all transcriptional gene-silencing phenomena that depend on the RNAi machinery involve the production of small-RNA molecules. RNAi components were also shown to affect telomere clustering in S. pombe, although binding of Swi6 and H3K9me to individual telomeres is not affected. The production of siRNAs is believed to be essential for the nuclear clustering of telomeres since cells carrying a catalytically dead RNA-dependent RNA polymerase (which abolishes siRNA production) are defective in telomere clustering. Consistent with a role for small RNAs in mediating gene contacts, sense and antisense transcription of Fab-7 as well as small-RNA species were found in Fab-7 transgenic lines. Moreover, a mutant allele of dcr-2 producing a truncated polypeptide lacking the RNase III domain, which is required for dsRNA processing, is defective in long-range interactions of PcG target sequences as well as in accumulation of Fab-7 small RNAs. These data suggest that small-RNA species could be involved in these gene contacts. However, no small Fab-7 RNA species was detected in the wt situation, although RNAi mutants affect the contact of the endogenous Fab-7 locus with the Antp gene. This might indicate that other RNA species produced in the endogenous Hox genes could contribute to gene clustering. However, the possibility remains that RNA-independent functions of RNAi proteins contribute to the maintenance of gene contacts, in particular in the case of endogenous PcG target genes (Grimaud, 2006).

Interestingly, none of the RNAi mutants tested are defective in the establishment of long-distance chromosomal interactions. Fab-7 contacts are correctly established during embryogenesis but decay during later stages of development. This suggests that the RNAi machinery is not required to initiate contacts but rather to maintain them via the stabilization of gene clustering at specific nuclear bodies. This clustering could be important in cosuppression, where transgene silencing can occur at the transcriptional and posttranscriptional levels, both requiring the RNAi machinery (Pal-Bhadra, 2002). It is difficult, however, to understand how a relatively modest increase in transcript levels caused by an increase in the copy number of a gene could trigger a robust silencing of all copies. This is particularly puzzling considering that the transcript levels of endogenous single-copy genes can vary, e.g., during normal physiological gene regulatory processes, without triggering gene silencing. One explanation might be that cosuppressed genes are clustered in the cell nucleus. Indeed, clustering of multiple gene copies has been reported in plant cells (Grimaud, 2006).

It is proposed that the RNAi machinery, perhaps in conjunction with PcG proteins, might stabilize this gene-clustering phenomenon. Specifically, the colocalization of multiple gene copies with components of the RNAi machinery might increase the local concentration of RNA species. Once this concentration overcomes a critical threshold, double-stranded RNAs might assemble and be cleaved in situ by the enzymatic activity of the RNAi machinery. RNA molecules might contribute to hold together loci containing PcG proteins that produce noncoding transcripts encompassing PREs. This gene clustering might involve contacts with components of the RNAi machinery as well as PcG proteins assembled in the same nuclear compartments (Grimaud, 2006).

One important question is, what is the role of the RNAi machinery in the regulation of endogenous PcG target genes? The data indicate that RNAi components affect only a subset of these genes since the colocalization of PcG bodies with RNAi bodies is limited. Hox loci are characterized by extensive noncoding RNA transcription, and, recently, other PcG target genes have been shown to be associated to intergenic transcription. RNAi components might be targeted to this subset of PcG target genes, while other PcG target genes that are characterized by the absence of noncoding transcripts might be independent on RNAi factors (Grimaud, 2006).

The fact that no homeotic phenotypes are visible in RNAi mutant backgrounds suggests that the function of RNAi components can be rescued by other chromatin factors. Indeed, the decrease in the level of nuclear interaction between the homeotic complexes was incomplete in RNAi mutant backgrounds. The data suggest that, while the RNAi machinery does not act in the establishment of PcG-dependent gene silencing, RNAi factors might help stabilize silencing during development by clustering PcG target genes at RNAi nuclear bodies. Thus, in addition to its role in defending the genome against viruses, transposons, and gene duplications, the RNAi machinery might participate in fine tuning the expression of PcG target genes through the regulation of nuclear organization. Finally, it must be noted that the developmental expression profile of the components of the RNAi machinery is highly specific. The function of specific RNAi components is therefore likely to be highly variable in different cell types and as a function of time. It will be of great interest to explore this issue in the developmental context of the whole organism, in Drosophila as well as in other species (Grimaud, 2006).

Piwi, Aubergine and Ago3 bind and cleave piwi-interacting RNAs to regulate of transposon activity in Drosophila

Drosophila Piwi-family proteins have been implicated in transposon control. This study examined piwi-interacting RNAs (piRNAs) associated with each Drosophila Piwi protein; Piwi and Aubergine were found to bind RNAs that are predominantly antisense to transposons, whereas Ago3 complexes contain predominantly sense piRNAs. As in mammals, the majority of Drosophila piRNAs are derived from discrete genomic loci. These loci comprise mainly defective transposon sequences, and some have previously been identified as master regulators of transposon activity. These data suggest that heterochromatic piRNA loci interact with potentially active, euchromatic transposons to form an adaptive system for transposon control. Complementary relationships between sense and antisense piRNA populations suggest an amplification loop wherein each piRNA-directed cleavage event generates the 5' end of a new piRNA. Thus, sense piRNAs, formed following cleavage of transposon mRNAs may enhance production of antisense piRNAs, complementary to active elements, by directing cleavage of transcripts from master control loci (Brennecke, 2007).

Mobile genetic elements, or their remnants, populate the genomes of nearly every living organism. Potential negative effects of mobile elements on the fitness of their hosts necessitate the development of strategies for transposon control. This is critical in the germline, where transposon activity can create a substantial mutational burden that would accumulate with each passing generation. Hybrid dysgenesis exemplifies the deleterious effects of colonization of a host by an uncontrolled mobile element. The progeny of intercrosses between certain Drosophila strains reproducibly show high germline mutation rates with elevated frequencies of chromosomal abnormalities and partial or complete sterility (Bucheton, 1990; Castro, 2004; Kidwell, 1977). Studies of the molecular basis of this phenomenon (Pelisson, 1981; Rubin, 1982) linked the phenotype to transposon mobilization (Brennecke, 2007).

Hybrid dysgenesis occurs when a transposon, carried by a male that has established control over that element, is introduced into a naive female that does not carry the element. The transposon becomes active in the progeny of the naive female, causing a variety of abnormalities in reproductive tissues that ultimately result in sterility. Since the dysgenic phenotype is often not completely penetrant, a fraction of the progeny from affected females may survive to adulthood. Such animals can develop resistance to the mobilized element, although in many cases, several generations are required for resistance to become fully established (Pelisson, 1987). Immunity to transposons can only be passed through the female germline (Bregliano, 1980), indicating that there are both cytoplasmic and genetic components to inherited resistance (Brennecke, 2007).

Studies of hybrid dysgenesis have served a critical role in revealing mechanisms of transposon control. In general, two seemingly contradictory models have emerged. The first model correlates resistance with an increasing copy number of the mobile element. A second model suggests that discrete genomic loci encode transposon resistance. The first model is supported by studies of the I element. Crossing a male carrying full-length copies of the I element to a naive female leads to I mobilization and hybrid dysgenesis. The number of I copies builds during subsequent crosses of surviving female progeny until it reaches an average of 1015 per genome. At this point, I mobility is suppressed, as the initially nave strain gains control over this element. Thus, a gradual increase in I element copy number over multiple generations was implicated in the development of transposon resistance (Brennecke, 2007).

The second model, which attributes transposon resistance to specific genetic loci, is illustrated by studies of gypsy transposon control (Bucheton, 1995). Genetic mapping of gypsy resistance determinants led to a discrete locus in the pericentric b-heterochromatin of the X chromosome that was named flamenco (Pelisson, 1994). Females carrying a permissive flamenco allele (one that allows gypsy activity) showed a dysgenic phenotype when crossed to males carrying functional gypsy elements. Permissive flamenco alleles exist in natural Drosophila populations but can also be produced by insertional mutagenesis of animals carrying a restrictive flamenco allele. Despite extensive deletion mapping over the flamenco locus, no transposon repressor from flamenco has been identified. For P elements, a repressor of transposition has been identified as a 66 kDa version of the P element transposase. Expression of the repressor was proposed to correlate with increasing P element copy number, leading to a self-imposed limitation on P element mobility. However, studies of resistance determinants indicated that control over P elements could also be established by insertion of P elements into specific genomic loci, arguing for an alternative, copy number-independent control pathway. Studies of inbred lines or of wild isolates with natural P element resistance indicated that P insertions near the telomere of X (cytological position 1A) were sufficient to confer resistance if maternally inherited. Additionally, several groups isolated insertions of incomplete P elements in this same cytological location that acted as dominant transposition suppressors. Importantly, these defective P elements lacked sequences encoding the repressor fragment of transposase (Brennecke, 2007 and references therein).

Both models of transposon resistance, those determined by specific genomic loci and those caused by copy number-dependent responses might be linked to small RNA-based regulatory pathways. Copy number-dependent silencing of mobile elements is reminiscent of copy number-dependent transgene silencing in plants (cosuppression) and Drosophila. In both cases, silencing occurs through an RNAi-like response where high-copy transgenes provoke the generation of small RNAs, presumably through a double-stranded RNA intermediate. Moreover, mutations in RNAi pathway genes impact transposon mobility in flies and C.elegans. Finally, small RNAs (rasiRNAs) corresponding to transposons and repeats have been isolated from flies and zebrafish (Brennecke, 2007 and references therein).

At the core of the RNAi machinery are the Argonaute proteins, which directly bind to small RNAs and use these as guides for the identification and cleavage of their targets. In animals, Argonautes can be divided into two clades (Carmell, 2002). One contains the Argonautes, which act with microRNAs and siRNAs to mediate gene silencing. The second contains the Piwi proteins. Genetic studies have implicated Piwi proteins in germline integrity. For example, piwi mutations cause sterility and loss of germline stem cells. aubergine is a spindle-class gene that is required in the germline for the production of functional oocytes. The third Drosophila Piwi gene, Ago3, has yet to be studied. Mutation of Piwi-family genes also affects mobile elements. For example, piwi mutations mobilize gypsy, and aubergine mutations impact TART and P elements. Finally, both Piwi and Aubergine bind rasiRNAs targeting a number of mobile and repetitive elements. These complexes are enriched for antisense small RNAs, as might be expected if they were actively involved in silencing transposons by recognition of their RNA products (Brennecke, 2007 and references therein).

Recently, a new class of small RNAs, the piRNAs, was identified through association with Piwi proteins in mammalian testes. These 26-30 nt RNAs are produced from discrete loci, generally spanning 50-100 kb. Interestingly, mammalian piRNAs are relatively depleted of transposon sequences. Despite apparent differences in the content of Piwi-associated RNA populations in mammals and Drosophila, Piwi-family proteins share essential roles in gametogenesis, with all three murine family members, Miwi2 , being required for male fertility. In order to probe mechanisms of transposon control in Drosophila and to understand the relationship between Piwi protein function in flies and mammals, a detailed analysis was undertaken of small RNAs associated with Piwi proteins in the Drosophila female germline. These studies indicate that Drosophila Piwi-family members function in a transposon surveillance pathway that not only preserves a genetic memory of transposon exposure but also has the potential to adapt its response upon contact with active transposons (Brennecke, 2007).

In C. elegans, effective RNAi depends upon an amplification mechanism. Small RNAs from the primary dsRNA trigger are largely dedicated to promoting the use of complementary targets as templates for RNA-dependent RNA polymerases (RdRPs) in the generation of secondary siRNAs. In Drosophila, no RdRPs have been identified. However, the ability of Piwi-mediated cleavage to prompt the production of new piRNAs could create an amplification cycle that serves the same purpose as the RdRP-driven secondary siRNA generation systems in worms (Brennecke, 2007).

The cycle, termed Ping-Pong amplification (Aravin, 2007b; see Properties and biogenesis of piRNA) is initiated by generating primary piRNAs, sampled from the piRNA clusters that have been identified in this study. As these are composed mainly of defective transposon copies, they serve as a genetic memory of transposons to which the population has been exposed. piRNAs that are antisense to expressed, dispersed transposons would identify and cleave their targets, resulting in the genesis of a new, sense piRNA in an Ago3 complex. The Ago3 bound sense piRNA would then seek a target, probably a precursor transcript from a master control locus that contains antisense transposon sequences. Ago3-directed cleavage would then generate additional antisense piRNAs capable both of actively silencing their target element and reinforcing the cycle through the creation of additional sense piRNAs (Brennecke, 2007).

The existence of such an amplification cycle essentially permits sense and antisense piRNAs act in concert to increase production of silencing-competent RNAs in response to the activity of individual transposons. Since Argonautes act catalytically, a significant amplification of the response could be achieved by even a relatively low level of sense piRNAs in Ago3 complexes. This model predicts that piRNAs participating in this process, namely those with complementary partners, should be more abundant than piRNAs without detectable partners. In accord with this hypothesis, the most frequently cloned Aub and Ago3-associated piRNAs show an increased probability of having partners within the data set. Interestingly, Piwi-associated RNAs do not follow this pattern. Since the amplification cycle consumes target transposon transcripts as part of its mechanism, posttranscriptional gene silencing may be sufficient to explain transposon repression. However, the possibility that transcriptional silencing may also be triggered by Piwi-family RNPs cannot be ruled out (Brennecke, 2007).

The amplification cycle may not be absolutely essential for silencing of all elements, since loci such as flamenco may operate in somatic follicle cells where the absence of Aub and Ago3 forces it to act in a stoichiometric fashion. In this regard, flamenco is unusual in that the vast majority of transposon fragments within this locus exist in a common orientation, which can lead to the production of antisense primary piRNAs given a long, unidirectional, precursor transcript (Brennecke, 2007).

In contrast to flamenco, most piRNA loci appear to be both bidirectionally transcribed and contain transposon sequences in random orientation. Nevertheless, the marked asymmetry of Piwi/Aub and Ago3 complexes is conserved in piRNAs that uniquely map to clusters. Among piRNAs that match transposons, 77% and 79% of unambiguously cluster-derived Piwi- and Aub-associated RNAs are antisense, while 73% of those in Ago3 are sense. These observations strongly suggest that piRNA clusters themselves participate in choice. According to this model, the remarkable strand asymmetry in piRNA populations hinges on informative interactions between master control loci and active transposons, which by their nature produce sense RNAs. These observations identify Ago3 as the principal recipient of piRNAs derived from transposon mRNAs. Thus, as long as there is an input to the system from active transposon transcripts via Ago3 and a preferential relationship between Ago3 and Aub for generating secondary piRNAs in their reciprocal complexes, a strand bias can be maintained even if primary and secondary piRNAs can both be derived from master control loci (Brennecke, 2007).

The amplification cycle must be initiated by primary piRNAs. Presently, the biogenesis pathway that generates primary piRNAs from piRNA clusters remains obscure. The data suggest that the piRNA precursor is a long, single-stranded transcript that is cleaved, preferentially at U residues. Transcripts have been detected from piRNA loci by RT-PCR that encompass multiple transposon fragments and find numerous small RNAs that cross junctions between adjacent transposons. In the case of flamenco, P element insertions near the proximal end of the locus have a polar effect both on these long RNA transcripts and on flamenco piRNAs (Brennecke, 2007).

Equally mysterious is the generation of piRNA 3' ends. Mature piRNAs could arise by two cleavage events and subsequent loading into Piwi complexes. Alternatively, piRNAs could be created following 5' end formation and incorporation of a long RNA into Piwi by resection of their 3' ends. The latter model is attractive, since it could provide an explanation for observed size differences between RNAs bound to individual Piwi proteins, since piRNA size would simply reflect the footprint of each Piwi protein. Although de novo biogenesis mechanisms must exist, maternally inherited piRNA complexes could also serve to initiate the amplification cycle. All three Piwi proteins are loaded into the developing oocyte, and Piwi and Aub are concentrated in the pole plasm, which will give rise to the germline of the next generation. Inherited piRNAs could enhance resistance to transposons that are an ongoing challenge to the organism, augmenting zygotic production of primary piRNAs. Indeed, maternally loaded rasiRNAs were detected in early embryos, and their presence was correlated with suppression of hybrid dysgenesis in D. virilis (Brennecke, 2007).

These data point to a comprehensive strategy for transposon repression in Drosophila that incorporates both a long-term genetic memory and an acute response to the presence of potentially active elements in the genome. The model that emerges from these studies shows many parallels to adaptive immune systems. The piRNA loci themselves encode a diversity of small RNA fragments that have the potential to recognize invading parasitic genetic elements. Throughout the evolution of Drosophila species, a record of transposon exposure may have been preserved by selection for transposition events into these master control loci, since this is one key mechanism through which control over a specific element can be achieved. Evidence has already emerged that X-TAS can act as a transposition hotspot for P elements, raising the possibility the piRNA clusters in general may attract transposons. Once an element enters a piRNA locus, it can act, in trans, to silencing remaining elements in the genome, either directly through primary piRNAs or by engaging in the amplification model described above. A comparison of D. melanogaster piRNAs to transposons present in related Drosophilids shows a lack of complementarity when comparisons are made at high stringency. However, when even a few mismatches are permitted, it is clear that piRNA loci might have some limited potential to protect against horizontal transmission of these heterologous elements. The existence of a feed-forward amplification loop can be compared to clonal expansion of immune cells with the appropriate specificity following antigen stimulation, leading to a robust and adaptable response (Brennecke, 2007).

Small interfering RNAs function through the PIWI subfamily of Argonautes to ensure silencing of retrotransposons

In Drosophila, repeat-associated small interfering RNAs (rasiRNAs) are produced in the germ line by a Dicer-independent pathway and function through the PIWI subfamily of Argonautes to ensure silencing of retrotransposons. Small RNAs were sequenced associated with the PIWI subfamily member AGO3. Although other members of PIWI, Aubergine (Aub) and Piwi, associated with rasiRNAs derived mainly from the antisense strand of retrotransposons, AGO3-associated rasiRNAs arose mainly from the sense strand. Aub- and Piwi-associated rasiRNAs showed a strong preference for uracil at their 5' ends, and AGO3-associated rasiRNAs showed a strong preference for adenine at nucleotide. Comparisons between AGO3- and Aub-associated rasiRNAs revealed pairs of rasiRNAs showing complementarities in their first 10 nucleotides. Aub and AGO3 exhibited Slicer activity in vitro. These data support a model in which formation of a 5' terminus within rasiRNA precursors is guided by rasiRNAs originating from transcripts of the other strand in concert with the Slicer activity of PIWI (Gunawardane, 2007).

Small noncoding RNAs trigger various forms of sequence specific gene silencing, including RNA interference (RNAi), translational repression, and heterochromatin formation in a variety of eukaryotic organisms, commonly referred to as RNA silencing. Members of the Argonaute family of proteins are essential components of RNA silencing. In Drosophila, five genes encode distinct members of the Argonaute family: AGO1, AGO2, Aubergine (Aub), Piwi, and AGO3. AGO1 and AGO2 constitute the Argonaute (AGO) subfamily and bind microRNA (miRNA) and small interfering RNA (siRNA), respectively. Aub, Piwi, and AGO3 belong to the PIWI subfamily of the Argonaute family and are enriched in germline cells (Williams, 2002), and Aub and Piwi have been shown to play important roles in germline cell formation. They are involved in silencing retrotransposons and other repetitive elements and exhibit target RNA cleavage (slicing) activity in vitro . Both Aub and Piwi associate with repeat-associated siRNAs (rasiRNAs) (Vagin, 2006; Saito, 2006). Aub- and Piwi-associated rasiRNAs are derived mainly from the antisense strand of retrotransposons, with little or no phasing, and have a strong preference for uracil (U) at the 5' end. Small RNA processing factors such as Dicer and Drosha are known to cleave preferentially at the 5' side of U; however, rasiRNAs are thought to be produced by a Dicer-independent pathway (Vagin, 2006). The mechanisms governing rasiRNA production remain to be elucidated (Gunawardane, 2007).

Very little is known about the function of AGO3, the third member of the Drosophila PIWI subfamily. Full-length cDNA of AGO3 revealed that the AGO3 gene is 83 kb in length. Peptide sequence alignments among Drosophila Argonaute proteins revealed that AGO3 is most similar to Piwi. The Asp-Asp-His motif in the PIWI domain, originally identified as the catalytic center for Slicer activity in human AGO2, is conserved in AGO3 (Gunawardane, 2007).

Embryonic RNA expression patterns of AGO3 are very similar to those of Piwi and Aub; they are expressed maternally, but their expression disappears by embryonic stages 10 to 12. To confirm these results, a monoclonal antibody (mAb) to AGO3 was produced, that revealed that AGO3 is strongly expressed in earlier embryonic stages but decreases as development proceeds. AGO3 accumulates in the cytoplasm of germline cells including germline stem cells (GSCs), germline cyst cells, nurse cells, and oocytes at earlier stages. In testes, AGO3 is expressed in GSC, primary gonial cells, and early spermatocytes. Unlike Piwi, AGO3 expression was undetected in the hub, a tiny cluster of postmitotic somatic cells localized at the apical tip of the testis that functions as a niche for GSC). Thus, with respect to expression in germline cells, AGO3 is more similar to Aub than to Piwi (Gunawardane, 2007).

All of the other members of the fly Argonautes are specifically associated with a subset of small RNAs: siRNAs, miRNAs, or rasiRNAs. Therefore whether AGO3 also associates with small RNAs produced in the fly ovary was tested. Immunoprecipitation with AGO3 mAb from ovary lysate revealed small RNAs 23 to 26 nucleotides (nt) long. The size distribution of AGO3-associated small RNAs is similar to that of Aub-associated small RNAs; in both cases, the peak is 24 nt and the longest is 27 nt. Small RNAs associated with AGO3 are likely to lack either a 2' or 3' hydroxyl group, because they do not migrate faster after beta-elimination as opposed to a synthetic siRNA that has 2' and 3' hydroxyl groups at the 3' end, the latter being the hallmarks of Dicer cleavage. These results suggest that AGO3-associated small RNAs in the ovary are produced by a pathway similar to those involved in production of rasiRNAs that associate with Aub and Piwi (Gunawardane, 2007).

A cDNA library was constructed of small RNAs associated with AGO3 in the ovary. Of 420 clones sequenced, 410 matched Drosophila genomic sequences in a database search, and most were rasiRNAs (86%; 353 of 410), as in the case of Aub and Piwi. Like rasiRNAs associated with Aub or Piwi, rasiRNAs associated with AGO3 included various kinds of transposable elements, both LTR (long terminal repeat) retrotransposons and LINE (long interspersed nuclear element)like elements. rasiRNAs associated with Aub or Piwi in ovaries are derived mainly from the antisense strand of retrotransposons, and the 5' end is predominantly U. These characteristics were not found for rasiRNAs associated with AGO3. However, AGO3-associated rasiRNAs were derived mainly from the sense strand of retrotransposons (82%), and they showed a strong preference for adenine (A) at nucleotide 10, but no preference for U at the 5' end. These results suggest that AGO3-associated rasiRNAs belong to a subset of rasiRNAs that are distinct from Aub- and Piwi-associated rasiRNAs (Gunawardane, 2007).

Some Argonaute proteins exhibit Slicer activity that directs cleavage of its cognate mRNA target across from nucleotides 10 and 11, measured from the 5' end of the small RNA guide strand. Thus, these findings suggest a model for rasiRNA biogenesis, in which the 5' end of Aub- and Piwi-associated rasiRNAs is determined and cleaved by AGO3-rasiRNA complexes, and the 5' end of AGO3-associated rasiRNAs is determined by Aub- and Piwi-rasiRNA complexes through a similar rasiRNA-guided cleavage event. For instance, AGO3 associated with a rasiRNA with A at nucleotide 10 can target a long RNA molecule by Watson-Crick base pairing and cleave the target RNA, resulting in sliced RNAs with U at the 5' end. Similarly, when Aub or Piwi associated with rasiRNAs with U at the 5' end slices its cognate RNA target, the resulting cleaved RNA will have A at nucleotide 10 (Gunawardane, 2007).

To test this model, AGO3 was examined for Slicer activity by performing in vitro target RNA cleavage assays with glutathione S-transferase (GST)AGO3 fusions. The target RNA, luc passenger siRNA (21 nt long, 5' end labeled with 32P), was efficiently cleaved by GST-AGO3, as was the case for GST-AGO1 and GST-Aub. The size of the cleaved products (9 nt) indicated that they direct cleavage of target RNA across from nucleotides 10 and 11 as measured from the 5' end of the small RNA guide strand. Both GST-Aub and GST-AGO3 with a longer guide RNA (26 nt) were also able to cleave a long transcript (180 nt). Long precursors of rasiRNAs both in sense and antisense orientations appear to exist in fly ovaries. These results corroborate the model in which the 5' end of rasiRNAs within the precursors is determined by rasiRNAs and cleaved by members of PIWI that associate with these rasiRNAs (Gunawardane, 2007).

The model predicts that some AGO3-associated rasiRNAs should be complementary to the first 10 nt of Aub- and Piwi-associated rasiRNAs. Sequence comparison between AGO3- and Aub-associated rasiRNAs indeed revealed pairs of rasiRNAs that show complementarities at their first 10 nt. Sixteen of 353 AGO3-associated rasiRNAs had such pairs with 11 of 676 Aub-associated rasiRNAs. However, such pairings were only found between AGO3- and Aub-associated rasiRNAs, and no pairs were observed between AGO3- and Piwi-associated rasiRNAs (353 versus 330). Like Aub-associated rasiRNAs, Piwi-associated rasiRNAs arise mainly from the antisense strand and their 5' ends show a strong preference for U; thus, it is difficult to argue that Piwi is not involved in this type of rasiRNA biogenesis. One possible reason is that Piwi is nuclear, whereas AGO3 and Aub are cytoplasmic. This type of rasiRNA biogenesis may operate in the cytoplasm. Alternatively, formation of 5' ends of Piwi-associated rasiRNAs may occur only at an earlier time during germline development (Gunawardane, 2007).

RasiRNAs are involved in genome surveillance by silencing repetitive elements and controlling their mobilization in the Drosophila germ line. It was recently shown that rasiRNAs are produced by a mechanism that requires neither Dicer-1 nor Dicer-2 in flies. These data suggest that rasiRNAs in a sense orientation guide formation of the 5' end of rasiRNAs in an antisense orientation, and vice versa; as well, this cycle of mutual dependency elaborates optimal rasiRNA production. In this model, proteins of the PIWI subfamily function as Slicer for formation of the 5' end during rasiRNA biogenesis. This model requires that sliced rasiRNA precursors then be cleaved again at the 3' end by an as yet unidentified endonuclease (or nibbled by exonuclease) to produce mature rasiRNAs before or after loading of the resulting cleavage products onto another member of the PIWI. Once 'primary' complexes of rasiRNAs with proteins of PIWI are produced, these complexes will in turn function as the 'initiator' of secondary rasiRNA biogenesis, and so nascent rasiRNAs should be continuously supplied in the ovary and testis. Such a process may occur through rasiRNA germline transmission. Of the PIWI members, at least Aub is accumulated to the posterior pole in oocytes and remains in polar granules in early embryos. It is then incorporated in pole cells, the progenitor of the Drosophila germ line (10) (Gunawardane, 2007).

An epigenetic activation role of Piwi and a Piwi-associated piRNA in Drosophila

Heterochromatin, representing the silenced state of transcription, consists largely of transposon-enriched and highly repetitive sequences. Implicated in heterochromatin formation and transcriptional silencing in Drosophila are Piwi (P-element induced wimpy testis) and repeat-associated small interfering RNAs (rasiRNAs). Despite this, the role of Piwi in rasiRNA expression and heterochromatic silencing remains unknown. This study reports the identification and characterization of 12,903 Piwi-interacting RNAs (piRNAs) in Drosophila, showing that rasiRNAs represent a subset of piRNAs. Piwi promotes euchromatic histone modifications and piRNA transcription in subtelomeric heterochromatin (also known as telomere-associated sequence, or TAS), on the right arm of chromosome 3 (3R-TAS). Piwi binds to 3R-TAS and a piRNA uniquely maps to 3R-TAS (3R-TAS1 piRNA). In piwi mutants, 3R-TAS loses euchromatic histone modifications yet accumulates heterochromatic histone modifications and Heterochromatin Protein 1a (HP1a). Furthermore, the expression of both the 3R-TAS1 piRNA and a white reporter gene in 3R-TAS becomes suppressed. A P element inserted 128 base pairs downstream of the 3R-TAS1 piRNA coding sequence restores the euchromatic histone modifications of 3R-TAS and the expression of 3R-TAS1 piRNA in piwi mutants, as well as partly rescuing their defects in germline stem-cell maintenance. These observations suggest that Piwi promotes the euchromatic character of 3R-TAS heterochromatin and its transcriptional activity, opposite to the known roles of Piwi and the RNA-mediated interference pathway in epigenetic silencing. This activating function is probably achieved through interaction with at least 3R-TAS1 piRNA and is essential for germline stem-cell maintenance (Yin, 2007).

The above findings reveal the complexity of small RNA-mediated epigenetic regulation, namely that Piwi can exert opposite effects on different genomic regions. piRNAs may have a function in guiding Piwi to the target sites, yet the opposite effects of Piwi at different target sites might be mediated by the local chromatin context, which would render the selective binding of the Piwi-piRNA complex to different partners such as HP1a or JmJC domain-containing histone demethylases. The activating effect of Piwi, 3R-TAS1 piRNA and possibly other piRNAs in 3R-TAS can be explained by a heterochromatin/euchromatin counterbalance model in which the repetitive nature of 3R-TAS by default is a substrate for heterochromatization. The heterochromatic state could be established and maintained by the Polycomb group proteins or an RNAi pathway mediated by Ago proteins, or yet another mechanism. However, the association of Piwi with 3R-TAS by means of the 3R-TAS1 piRNA or P{w+,ry+}A4-4 insertion as a roughly 20-kb unique sequence counteracts the heterochromatization. The finding that Piwi is required for the epigenetic activation of the subtelomeric region starts to reveal a mechanism underlying epigenetic regulation and stem-cell maintenance (Yin, 2007).

Pimet, the Drosophila homolog of HEN1, mediates 2'-O-methylation of Piwi- interacting RNAs at their 3' ends

Piwi-interacting RNAs (piRNAs) consist of a germline-specific group of small RNAs derived from distinct intergenic loci in the genome. piRNAs function in silencing selfish transposable elements through binding with the PIWI subfamily proteins of Argonautes. This study shows that piRNAs in Drosophila are 2'-O-methylated at their 3' ends. Loss of Pimet/Hen1 (piRNA methyltransferase), the Drosophila homolog of Arabidopsis HEN1 methyltransferase for microRNAs (miRNAs), results in loss of 2'-O-methylation of fly piRNAs. Recombinant Pimet shows single-stranded small RNA methylation activity in vitro and interacts with the PIWI proteins within Pimet mutant ovary. These results show that Pimet mediates piRNA 2'-O-methylation in Drosophila (Saito, 2007; full text of article).

In Pimet mutant ovary, piRNAs associated with Aub and Piwi were not methylated at the 3' ends, most likely due to loss of Pimet expression. Whether GST-Pimet is able to methylate these piRNAs associated with the PIWI proteins from Pimet mutant ovary was investigated. Aub-piRNA complexes were immunopurified with a specific antibody against Aub and subjected to in vitro methylation assays. As a control, miRNAs associated with AGO1 were also obtained through immunoprecipitation using anti-AGO1 from ovary lysate. It was found that piRNAs were methylated even in a complex form with Aub. piRNA methylated in the assay showed resistance to oxidation and β-elimination treatment. Interestingly, miRNAs associated with AGO1 were not methylated, although these miRNAs are single-stranded in a complex form with AGO1. Confirmation that the miRNA levels were several-fold higher than those of piRNAs was provided by phosphorylation of these small RNAs. It seems that small RNA methylation by Pimet is largely influenced by the accessibility of the 3' ends of the substrates to Pimet itself. Structural analysis of Argonaute proteins suggests that the 5' end of the small guide RNA is anchored in a highly conserved pocket in the PIWI domain, whereas the 3' end of the small RNA is embedded in the PAZ domain. Taken together, these results suggest that the 3' ends of Aub-associated piRNAs are not tightly bound to the PAZ domain, but are exposed to the surface of the protein. In contrast, the 3' ends of AGO1-associated miRNAs are likely to be embedded in the PAZ domain and therefore are not exposed to the surface of the protein. Alternatively, but not mutually exclusively, it is conceivable that Pimet may interact only with PIWI proteins and not with AGO proteins, thereby methylating only small RNAs associated with PIWI proteins. To test this, whether Pimet associates with PIWI proteins was investigated. A GST pull-down assay was performed; GST-Pimet was first incubated with Pimet mutant ovary lysate, and after extensive washing the eluates were probed with PIWI protein antibodies. Aub, Piwi, and AGO3 were clearly detected in the bound fraction with GST-Pimet but not with GST itself. By contrast, AGO1 was not observed. These results indicated that Pimet is capable of physically interacting with PIWI proteins containing piRNAs that can serve as substrates for Pimet methylation. Addition of RNaseA did not affect the interaction of Pimet with Aub, suggesting that Pimet is able to associate directly with the PIWI proteins. In Drosophila, piRNA methylation may occur after matured piRNAs are loaded onto PIWI proteins. If so, it clearly differs from the case of miRNA methylation in plants, which likely occurs prior to miRNA loading onto the AGO proteins when miRNAs are still in a duplex form with the complementary miRNA* molecules (Saito, 2007).

Mutations in Arabidopsis hen1 cause reduced fertility. Thus, is the piRNA methylation by Pimet crucial in Drosophila? piRNAs function in genome surveillance in germlines in concert with PIWI proteins. Mutations in aub, piwi, and others like spindle-E (homeless) cause piRNAs not to be accumulated in gonads, and lead to germ cell malformation and sterility. This clearly indicates that piRNAs are necessary for perpetuation of organisms. However, the Pimet mutant fly seems to be viable and fertile. Steady-state levels of piRNAs in the methylation-defective mutant are also similar to those in wild type. Expression levels of retrotransposons do not seem to be changed by loss of Pimet expression. Thus, the function of 3' end methylation is currently unknown. Further investigation such as by immunohistochemistry may be required to obtain a more detailed morphology of the mutant. Extensive analyses of the mechanisms underlying piRNA methylation may also provide important clues to more fully elucidating piRNA biogenesis. Aub and AGO3, which determine and form the 5' end of piRNAs in piRNA biogenesis, were shown to be in the protein fraction associated with Pimet. Identifying more Pimet-associated proteins may reveal the factors required for formation of the 3' end of piRNAs (Saito, 2007).

Drosophila PIWI associates with chromatin and interacts directly with HP1a

The interface between cellular systems involving small noncoding RNAs and epigenetic change remains largely unexplored in metazoans. RNA-induced silencing systems have the potential to target particular regions of the genome for epigenetic change by locating specific sequences and recruiting chromatin modifiers. Noting that several genes encoding RNA silencing components have been implicated in epigenetic regulation in Drosophila, a direct link was sought between the RNA silencing system and heterochromatin components. This study shows that Piwi, an Argonaute/Piwi protein family member that binds to Piwi-interacting RNAs (piRNAs), strongly and specifically interacts with heterochromatin protein 1a (HP1a), a central player in heterochromatic gene silencing. The HP1a dimer binds a PxVxL-type motif in the N-terminal domain of Piwi. This motif is required in fruit flies for normal silencing of transgenes embedded in heterochromatin. Piwi, like HP1a, is itself a chromatin-associated protein whose distribution in polytene chromosomes overlaps with HP1a and appears to be RNA dependent. These findings implicate a direct interaction between the Piwi-mediated small RNA mechanism and heterochromatin-forming pathways in determining the epigenetic state of the fly genome (Brower-Toland, 2007).

Thus, Drosophila PIWI interacts directly with HP1a and is distributed on chromosomes in a pattern overlapping HP1a. Association of Piwi with constitutive heterochromatin domains including pericentric heterochromatin, telomeres, and part of the banded portion of chromosome 4 is consistent with the profile of Piwi-associated small RNAs, which includes sequences homologous to telomeric and centromeric repetitive elements as well as some that are overrepresented in chromosome 4. The PIWI-HP1a interaction is mediated through binding of a PxVxL-type motif by dimerized chromoshadow domains of HP1a. Furthermore, the intact PxVxL motif is required in vivo for effective heterochromatic silencing. This is consistent with the wealth of data implicating Piwi at the interface of RNA silencing mechanisms with epigenetic phenomena, in particular heterochromatin-induced silencing. Piwi is a nuclear protein required for silencing of transposons and retroviruses. Piwi is required for effective silencing of multiple copies of dispersed transgenes, of tandem repeats of the white gene at euchromatic insertion sites, and of white reporter genes in the pericentric heterochromatin or fourth chromosome. Silencing in the latter two cases is dependent on HP1a. This study shows that Piwi interacts directly with HP1a. Some Piwi signal overlaps with HP1a in polytene chromosomes, suggesting that Piwi and HP1a together regulate the epigenetic state of diverse regions in the genome. This overlap is spatially complex and could therefore represent numerous varied roles for Piwi, HP1a, and the Piwi-HP1a interaction. Heterochromatin is first assembled early, while the Drosophila embryo is still a syncytium (nuclear division cycles 10-14). This is critical for the silencing observed in PEV, which may be compromised later in development. While the HP1a-PIWI association observed on polytene chromosomes might be similar to that inferred in the embryo, it might also represent a different function (Brower-Toland, 2007).

Recently, Piwi has been shown to bind in the germline to piRNAs, many of which are from heterochromatic regions. The colocalization of Piwi and HP1a in pericentric heterochromatin may indeed reflect a simple relationship between the piRNA pathway, histone modification, and heterochromatin formation similar to the S. pombe system wherein AGO1-mediated transcriptional gene silencing locally targets H3K9 methylation to create a binding site for the HP1 homolog Swi6. In Drosophila, these data on the specific interaction between PIWI and HP1a raise the possibility of an alternate pathway to HP1a-mediated heterochromatinization: A PIWI-piRNA complex might directly recruit HP1a to piRNA-corresponding genomic sequences, which could then recruit HMTs such as SU(VAR)3-9 to effect nucleation/spreading. This would represent an H3K9me-independent mode for initial HP1 localization, an alternative but potentially equally effective means for triggering local formation of heterochromatin. Conversely, if heterochromatin formation is targeted by a different mechanism, the presence of HP1a could allow stable binding of Piwi to heterochromatin for PTGS, a process that might be necessary to maintain silencing throughout the lifetime of the fly (Brower-Toland, 2007).

Telomeric trans-silencing: an epigenetic repression combining RNA silencing and heterochromatin formation

The study of P-element repression in Drosophila led to the discovery of the telomeric Trans-Silencing Effect (TSE), a repression mechanism by which a transposon or a transgene inserted in subtelomeric heterochromatin (Telomeric Associated Sequence or TAS) has the capacity to repress in trans in the female germline, a homologous transposon, or transgene located in euchromatin. TSE shows variegation among egg chambers in ovaries when silencing is incomplete. This study reports that TSE displays an epigenetic transmission through meiosis, which involves an extrachromosomal maternally transmitted factor. This silencing is highly sensitive to mutations affecting both heterochromatin formation (Su(var)205 encoding Heterochromatin Protein 1 and Su(var)3-7) and the repeat-associated small interfering RNA (or rasiRNA) silencing pathway (aubergine, homeless, armitage, and piwi). In contrast, TSE is not sensitive to mutations affecting r2d2, which is involved in the small interfering RNA (or siRNA) silencing pathway, nor is it sensitive to a mutation in loquacious, which is involved in the micro RNA (or miRNA) silencing pathway. These results, taken together with the recent discovery of TAS homologous small RNAs associated to PIWI proteins, support the proposition that TSE involves a repeat-associated small interfering RNA pathway linked to heterochromatin formation, which was co-opted by the P element to establish repression of its own transposition after its recent invasion of the D. melanogaster genome. Therefore, the study of TSE provides insight into the genetic properties of a germline-specific small RNA silencing pathway (Josse, 2007; full text of article).

Repression of transposable elements (TEs) involves complex mechanisms that can be linked to either small RNA silencing pathways or chromatin structure modifications depending on the species and/or the TE family. Drosophila species are particularly relevant to the study of these repression mechanisms since some families of TEs are recent invaders, allowing genetic analysis to be carried out on strains with or without these TEs. In some cases, crossing these two types of strains induces hybrid dysgenesis, a syndrome of genetic abnormalities resulting from TE mobility. In D. virilis, repression of hybrid dysgenesis has been correlated to RNA silencing since small RNAs of the retroelement Penelope, responsible for dysgenesis, were detected in nondysgenic embryos but not in dysgenic embryos. In D. melanogaster, repression of retrotransposons can be established by noncoding fragments of the corresponding element (I factor, ZAM, and Idefix) and can be in some cases (gypsy, mdg1, copia, Het-A, TART, and ZAM, Idefix) sensitive to mutations in genes from the Argonaute family involved in small RNA silencing pathways. In the same species, strong repression of the DNA P TE, by a cellular state that has been called 'P cytotype', can be established by one or two telomeric P elements inserted in heterochromatic 'Telomeric Associated Sequences' (TAS) at the 1A cytological site corresponding to the left end of the X chromosome. This includes repression of dysgenic sterility resulting from P transposition. This P cytotype is sensitive to mutations affecting both Heterochromatin Protein 1 (HP1) (Ronsseray, 1996) and the Argonaute family member AUBERGINE (Reiss, 2004). P repression corresponds to a new picture of TE repression shown, using an assay directly linked to transposition, to be affected by heterochromatin and small RNA silencing mutants (Josse, 2007).

In the course of the study of P cytotype, a new silencing phenomenon has been discovered. Indeed, a P-lacZ transgene or a single defective P element inserted in TAS can repress expression of euchromatic P-lacZ insertions in the female germline in trans, if a certain length of homology exists between telomeric and euchromatic insertions. This homology-dependent silencing phenomenon has been termed Trans-Silencing Effect (TSE) (Roche, 1998). Telomeric transgenes, but not centromeric transgenes, can be silencers and all euchromatic P-lacZ insertions tested can be targets. TSE is restricted to the female germline and has a maternal effect since repression occurs only when the telomeric transgene is maternally inherited (Ronsseray, 2001). Further, when TSE is not complete, variegating germline lacZ repression is observed from one egg chamber to another, suggesting a chromatin-based mechanism of repression. Recently, an extensive analysis of small RNAs complexed with PIWI family proteins (AUBERGINE, PIWI, and AGO3) was performed in the Drosophila female germline. The latter study showed that most of the RNA sequences associated to these proteins derive from TEs. TSE corresponds likely to such a situation (Josse, 2007).

This study analyzed the genetic properties of TSE and shows that it has an epigenetic transmission through meiosis, which involves an extrachromosomal maternally transmitted stimulating component. Further, in order to investigate the mechanism behind TSE, a candidate gene analysis was performed to identify genes whose mutations impair TSE. It was found that TSE is strongly affected both by mutations in genes involved in heterochromatin formation and in the recently discovered small RNA silencing pathway called 'repeat-associated small interfering RNAs' (rasiRNA) pathway. In contrast, this study shows that TSE is not sensitive to genes specific to the classical RNA interference pathway linked to small interfering RNAs (siRNA) or to the micro RNA (miRNA) pathway. This suggests thus that TSE involves a rasiRNA pathway linked to heterochromatin formation and that such a mechanism, working in the germline, may underlie epigenetic transmission of repression through meiosis (Josse, 2007).