Knock down analysis reveals critical phases for specific oskar noncoding RNA functions during Drosophila oogenesis

Abstract The oskar transcript, acting as a noncoding RNA, contributes to a diverse set of pathways in the Drosophila ovary, including karyosome formation, positioning of the microtubule organizing center (MTOC), integrity of certain ribonucleoprotein particles, control of nurse cell divisions, restriction of several proteins to the germline, and progression through oogenesis. How oskar mRNA acts to perform these functions remains unclear. Here, we use a knock down approach to identify the critical phases when oskar is required for three of these functions. The existing transgenic shRNA for removal of oskar mRNA in the germline targets a sequence overlapping a regulatory site bound by Bruno1 protein to confer translational repression, and was ineffective during oogenesis. Novel transgenic shRNAs targeting other sites were effective at strongly reducing oskar mRNA levels and reproducing phenotypes associated with the absence of the mRNA. Using GAL4 drivers active at different developmental stages of oogenesis, we found that early loss of oskar mRNA reproduced defects in karyosome formation and positioning of the MTOC, but not arrest of oogenesis. Loss of oskar mRNA at later stages was required to prevent progression through oogenesis. The noncoding function of oskar mRNA is thus required for more than a single event.


Introduction
Drosophila oskar (osk) mRNA is required, independent of Osk protein, during oogenesis (Jenny et al. 2006). Mutants in which osk mRNA levels are substantially reduced or eliminated exhibit myriad defects. Most conspicuous of these is female sterility, a consequence of arrested oogenesis. Even before the arrest, more subtle defects appear: egg chambers often have too many nurse cells, the condensation of oocyte chromosomes to form the karyosome is incomplete, positioning of the microtubule organizing center (MTOC) within the oocyte is abnormal, certain ribonucleoprotein particles are disorganized, and proteins normally restricted to the germline cells of the ovary can be detected at low levels in surrounding somatic follicle cells (Jenny et al. 2006;Kanke et al. 2015;Kenny et al. 2021). Discovery of additional defects seems likely, awaiting only the application of appropriate assays. The diversity of osk RNA null mutant phenotypes, with no discernible shared dependence on a particular cellular pathway downstream of osk itself, might suggest the existence of multiple different noncoding functions. However, each phenotype listed above is sensitive to mutation of the same defined functional sequences in the osk mRNA 3' UTR: a tightly clustered combination of Bruno1 (Bru1) binding sites, the osk noncoding element (ONCE), and A-rich sequences (ARS) (Vazquez-Pianzola et al. 2011;Kanke et al. 2015;Kenny et al. 2021). Thus, these elements either mediate multiple events or interactions, or they perform a single function with multiple pathways affected by disruption of that event. An extreme version of the latter option would be an initial "gateway" event with an immediate outcome, such as karyosome formation or proper MTOC organization, being a prerequisite for later outcomes, such as progression through oogenesis.
One approach to explore the relationship of different osk RNA null defects is to disrupt osk ncRNA activity at different stages of oogenesis, and ask if specific defects can be induced separate from others. Classical application of this strategy has often relied on temperature-sensitive mutants (e.g., Jarvik and Botstein 1973). A more recent take on this concept relies on inducible gene activation or silencing (e.g., Gossen and Bujard 1992;Gossen et al. 1995;Kumar et al. 2009;Fenno et al. 2011). In Drosophila, a useful method is gene knockdowns (KDs), relying on transgenic shRNAs under UAS/GAL4 control (https://fgr.hms. harvard.edu/fly-in-vivo-rnai) and a variety of GAL4 drivers (Brand and Perrimon 1993) with different temporal and/or spatial patterns of expression.
Here, we have taken the KD approach, commencing with the development of effective osk-shRNA transgenes to address a disabling limitation of the existing versions. We provide evidence of different critical periods for different osk ncRNA roles, and show that initial defects from early loss of osk mRNA do not have lasting effects if osk mRNA is restored.
The TRiP transgenic strain for KD of osk was TRiP.GL01101, Bloomington stock #36903, and for KD of bru1 was TRiP.GL00314, Bloomington stock #35394.
The osk syn transgene was based on a rescuing genomic transgene (Kim-Ha et al. 1991) in which the portion of the osk gene from just 3' to the start of the second exon to the stop codon (coordinates 8935984-8937367 in release r6.40 of the D. melanogaster genomic sequence) was replaced with a synthetic sequence retaining the wild type intron sequences and encoding the wild type Osk protein sequence, but with synonymous codons to maximize differences in the DNA sequence (see Supplementary File S1). Drosophila genomic sequence coordinates were obtained from FlyBase (Larkin et al. 2021).
Egg laying assays were performed as described ).

Detection of RNAs and proteins
RNA levels were determined by qPCR as described . In situ hybridization to detect osk mRNA made use of tiled short DNA oligonucleotides (smFISH) 3'-end labeled with Quasar 670 fluorophore (LGC Biosearch Technologies) and used at 1.5 nM. Assays were performed as described (Abbaszadeh and Gavis 2016). Immunodetection of proteins in ovaries was as described (Ryu and Macdonald 2015), with a different fixation protocol for gamma-tubulin . TO-PRO-3 Iodide (Invitrogen) (1:1000) or DAPI (1:4000) were used to stain DNA. Antibodies for imaging were mouse anti-1B1 which detects Adducin (Hts) protein (Developmental Studies Hybridoma Bank), diluted 1:1, mouse anti-Lamin Dm0 (ADL84.12)(Developmental Studies Hybridoma Bank), diluted 1:100, and rabbit anti-gammatubulin (Sigma T-0950), diluted 1:2000. Samples were imaged with a Nikon C2þ laser scanning confocal microscope. For all imaging experiments, the samples were obtained from at least five flies, typically many more. Quantification of in situ hybridization data was done in FIJI, with germaria or egg chambers traced and maximum pixel intensity measured. Measurements of oocyte areas bounded by gamma-tubulin foci were performed as described ).

Statistical methods
Reproducibility was confirmed by performing independent experiments. Imaging experiments involved examination of multiple individual egg chambers in each experiment. Here, repetition served to reveal any technical problems, and the large number of individual egg chambers scored in each experiment ensured consistency and reproducibility. Egg laying experiments were performed at least three times. Rates of egg laying often show circadian variation, but because each experiment consisted of egg collections over several days, this source of variation was minimized. The experiments were not randomized, and no statistical method was used to predetermine sample size. One-way ANOVA was used to ask if there were significant differences among data sets with three or more variables. For all analyses Shapiro-Wilk tests rejected normality, and Wilcoxon rank-sum tests were used for post hoc analysis.

Development and validation of new osk KD reagents
One approach to defining the critical phase or phases of osk ncRNA activity is to remove osk RNA by knock down (KD), relying on GAL4 drivers with different temporal activities to express a transgenic shRNA that targets osk mRNA. However, the maternal triple driver (MTD-Gal4) in combination with the osk shRNA transgenic reagent for germ-line expression (TRiP.GL01101) from the Transgenic RNAi Project (TRiP) is not effective at mimicking the most dramatic osk RNA null phenotype, arrest of oogenesis, as egg laying is not substantially reduced (Liu and Lasko 2015). Similarly, the matalpha4-GAL-VP16 driver (henceforth referred to as MAT; abbreviations used for other drivers are given in MATERIALS AND METHODS) in combination with TRiP.GL01101 only modestly reduced the level of ovarian osk mRNA ( Figure 1B) and failed to substantially reduce egg laying ( Figure 1C).
Possible explanations for the weak KD include insufficient driver activity or limited efficacy of the shRNA during oogenesis. Notably, the osk shRNA target site overlaps substantially with a binding site for Bru1 protein ( Figure 1A), which acts in translational regulation of osk mRNA (Kim-Ha et al. 1995;Reveal et al. 2010). Bru1 is present from the earliest stages of oogenesis (Webster et al. 1997), and when bound to the osk mRNA might interfere with binding of the shRNA, thus limiting the KD effect.
Three new transgenes for expression of shRNAs targeting other sites in osk mRNA were made and tested (osk-shRNAs#1, #2, and #3; Figure 1A). Using the MAT driver for expression, each new osk-shRNA dramatically reduced osk mRNA levels ( Figure 1B) and largely eliminated egg production ( Figure 1C). To confirm specificity for osk mRNA the KD with osk-shRNA#3 was tested in the presence of an osk transgene-osk syn -with synonymous codon changes that alter the osk-shRNA#3 target site ( Figure 1A): egg laying was restored ( Figure 1C). The osk-shRNA#3 transgene was used for all subsequent osk KD experiments.
The contrasting properties of the original osk-shRNA and the novel osk-shRNAs strongly support the notion that Bru1 can disrupt binding of an shRNA when their targets in an mRNA are overlapping. It would not be surprising if a similar effect underlies a subset of other ineffective shRNAs, and compilations of shRNA effectiveness could potentially be mined for information contributing to the identification of recognition sites for other RNA binding proteins.

Effectiveness of GAL4 drivers in reducing osk ncRNA activity
Several GAL4 drivers active in the female germ line were tested with osk-shRNA#3 to monitor effects on osk ncRNA activity, focusing initially on egg laying (Figure 2A). For comparison, KDs of bru1 were also performed. Females mutant for strong alleles of bru1 produce no eggs and display an early arrest of oogenesis, the expected phenotype for an effective KD.
Three of the drivers tested, NGV, MAT, and 34C10, substantially reduced egg laying in the osk KDs; each must be active at or before the stage when osk ncRNA activity is required for progression through oogenesis. When used for bru1 KDs each of these drivers was also effective at arresting oogenesis. One driver, NGT, caused arrest of oogenesis in bru1 KDs, but did not interfere with progression through oogenesis in osk KDs. Although the osk and bru1 KDs are similar in causing arrest of oogenesis, the nature of the arrest is quite different. Whereas the osk KD allows progression through the early stages, the bru1 KD arrest is associated with overproliferation of germ cells in pseudo egg chambers (Supplementary Figure S1), as observed in mutants lacking bru1 activity (Parisi et al. 2001).
Each of these drivers was used in combination with a UAS-GFP transgene to compare periods of activity during oogenesis. Identical imaging settings were used for all genotypes. To detect lower levels of activity, the original images ( Figure 2B) were adjusted identically in the green (GFP) channel to increase sensitivity ( Figure 2B'). Only those drivers with substantial effects on egg laying in the osk KD were strongly active for UAS-GFP expression during the middle stages of oogenesis.
Both MAT and 34C10 drivers produced GFP at substantial levels from stage 3 though at least stage 8 (the endpoint of this analysis). In principle, the continual appearance of GFP could arise from an initial phase of expression and persistence of the protein later in development, or from continual expression. Two observations suggest that persistence of GFP protein did not make a substantial contribution to the observed patterns. First, the half life of GFP appeared to be short relative to the time required for progression through oogenesis: GFP produced by the NGT driver in the germarium did not persist ( Figure 2B'). Second, the intensity of the GFP signal from the MAT and 34C10 drivers remained high and even increased from stages 3-8 (Figure 2, B and B'), despite a more than 10-fold increase in the volume of the egg chamber in that developmental period (King 1970). If there was only a short period of expression, the resulting GFP should have become progressively more dilute as volume increased.

Consequences of removing osk ncRNA activity during the early stages of oogenesis
The NGT driver was active only very early in oogenesis: expression of the UAS-GFP reporter was detected in the germarium and in stage 2 egg chambers, in a developmental profile roughly complementary to MAT activity ( Figures 2B' and 3A). Despite the early activity of NGT, there was no effect on egg laying in the osk KD ( Figure 2A). Thus, either NGT is not effective for the osk KD, or osk mRNA is not required at early stages for the progression through oogenesis. To address the effectiveness of the NGT KD of osk, we monitored osk mRNA levels by in situ hybridization. In the germarium, including stage 1 egg chambers not yet budded off, osk mRNA was largely eliminated (Figure 3, B and C). By stage 4 the osk mRNA levels had begun to recover: a substantial fraction of egg chambers still had very little osk mRNA while others had levels approaching that of the w 1118 control. By contrast, the osk KD with MAT caused only a modest reduction of osk mRNA in the germarium, while stage 4 egg chambers were substantially depleted (Figure 3, B and C). We conclude that osk ncRNA activity is not required during the earliest stages of oogenesis for progression through oogenesis. Moreover, the results with the NGT driver further narrow the definition of the critical phase for the requirement of osk ncRNA activity in progression through oogenesis. Because a significant fraction of stage 4 egg chambers in the osk KD with NGT still have extremely low levels of osk mRNA, yet there is no reduction in egg laying, osk ncRNA activity must be required only later. We conclude that at some point between stage 5 and the time of arrest at stage 6/7 osk mRNA performs the function that allows further progression through oogenesis. Arrest of oogenesis is only one manifestation of absence of osk mRNA. Other osk RNA null phenotypes, such as altered organization of the MTOC  and defective formation of the karyosome (Jenny et al. 2006;Kanke et al. 2015), can appear prior to the time of arrest. These processes are expected to be sensitive to loss of osk mRNA early in oogenesis from KD. Indeed, the distribution of osk mRNA, when present in stage 4 egg chambers after osk KD with NGT ( Figure 3B), suggests an MTOC defect. In wild-type oocytes the MTOC from stages 2 to 6 is positioned at the posterior of the oocyte (Theurkauf et al. 1992), and this distribution underlies the posterior enrichment of multiple mRNAs including osk itself (Clegg et al. 1997). By comparison to wild-type stage 4 oocytes, in the osk KD with NGT the posterior enrichment of osk mRNA appeared to be reduced.
MTOC organization was monitored using gamma-tubulin as a marker, initially focusing on stage 3 and 4 egg chambers. Posterior enrichment is lost in the absence of osk mRNA, with multiple foci of gamma-tubulin present throughout the oocyte   (Figure 4A). A similar defect was found in the early osk KD with the NGT driver (Figure 4, A and B). The loss of osk mRNA from the NGT KD did not persist as oogenesis progressed. Consequently, we could ask if MTOC organization recovered as osk mRNA reappeared from ongoing transcription. In wild type stage 5 and 6 egg chambers, the clusters of gamma-tubulin transitioned into one or two predominant foci ( Figure 4C). In the absence of osk mRNA, the abnormal distribution of gamma-tubulin persisted at later stages of oogenesis ( Figure 4C). By contrast, in the early osk KD with the NGT driver the normal pattern was restored, with no significant difference from wild type by stages 5/6. We also monitored karyosome formation at the same early and mid stages of oogenesis. At stage 2 of oogenesis, chromosomes appear throughout the oocyte nucleus. With the completion of meiotic recombination the oocyte chromosomes condense to form the karyosome, a single compact cluster within the nucleus (King 1970) ( Figure 5). Karyosome formation is defective in the absence of osk ncRNA activity, with the chromosomes partitioned into two or more foci (Jenny et al. 2006;Kanke et al. 2015) (Figure 5). KD of osk mRNA with the NGT driver produced the same defect ( Figure 5A), statistically indistinguishable from the osk 0 mutant ( Figure 5C).
The karyosome defects from the early osk KD with the NGT driver were substantially corrected as oogenesis proceeded and osk mRNA levels increased, although a significant fraction of egg chambers still had abnormal karyosomes at stages 5/6 ( Figure 5, B and D; the example shown in Figure 5B is one in which the karyosome remains abnormal).
Because the initial failure to form the karyosome or to correctly position the MTOC did not prevent at least partial recovery when osk mRNA levels began to recover, the critical period for osk ncRNA activity in these two processes is ongoing and the role played by osk mRNA must occur repeatedly and not only once. The differing extents to which these defects were corrected could reflect different modes of action for osk ncRNA activity in each process, or different sensitivities to loss of osk mRNA. By either explanation, it seems unlikely that osk mRNA performs a single initial event, which then feeds into the different pathways. Each egg chamber is oriented with posterior to the right. Left: complete egg chambers stained for DNA (red) and lamin (green) to outline the nuclei. Scale bar is 10 mm. Right: enlargements of the oocyte nuclei from the left images. (B) Representative stage 5-6 egg chambers from females with the genotypes indicated. Presented as in panel A, except that the scale bar is 20 mm. (C) Proportion of stage 3-4 oocytes of the indicated genotypes with normal karyosomes, defined as a single cluster of DNA in the oocyte nucleus. The number (n) of oocytes analyzed for each genotype is indicated. The P-values were derived from the Wilcoxon rank-sum test: *** P < 0.01. (D) Proportion of stage 5-6 oocytes of the indicated genotypes with normal karyosomes, defined as a single cluster of DNA in the oocyte nucleus. The P-values were derived from the Wilcoxon rank-sum test: *** P < 0.01.

Data availability
The data underlying this article are available in the article and in its online supplementary material. Supplementary material is available at G3 online.