Dynamic localization of DNA topoisomerase I and its functional relevance during Drosophila development

Abstract DNA topoisomerase I (Top1) maintains chromatin conformation during transcription. While Top1 is not essential in simple eukaryotic organisms such as yeast, it is required for the development of multicellular organisms. In fact, tissue and cell-type-specific functions of Top1 have been suggested in the fruit fly Drosophila. A better understanding of Top1’s function in the context of development is important as Top1 inhibitors are among the most widely used anticancer drugs. As a step toward such a better understanding, we studied its localization in live cells of Drosophila. Consistent with prior results, Top1 is highly enriched at the nucleolus in transcriptionally active polyploid cells, and this enrichment responds to perturbation of transcription. In diploid cells, we uncovered evidence for Top1 foci formation at genomic regions not limited to the active rDNA locus, suggestive of novel regulation of Top1 recruitment. In the male germline, Top1 is highly enriched at the paired rDNA loci on sex chromosomes suggesting that it might participate in regulating their segregation during meiosis. Results from RNAi-mediated Top1 knockdown lend support to this hypothesis. Our study has provided one of the most comprehensive descriptions of Top1 localization during animal development.


Introduction
DNA topoisomerase I (Top1) in eukaryotic organisms belongs to type I topoisomerases that produce a transient single-stranded DNA break to relieve torsion generated during transcription and other processes (Wang 1985;Pommier et al. 1998). Top1 inhibitors have been widely used in cancer therapies, most of them cause cytotoxicity by trapping the cross-linked DNA-Top1 intermediate, thus interfering with processes such as DNA replication and transcription (Thomas and Pommier 2019). Therefore, a better understanding of Top1's in vivo function carries medical significance.
Loss of Top1 function is tolerated in yeast with elevated instability at the rDNA locus (Christman et al. 1988;Andersen et al. 2015), perhaps the most active locus in the genome, suggesting that Top1 is needed to prevent the accumulation of recombinogenic intermediates at rDNA during transcription. How Top1 is recruited to transcriptionally active loci has been extensively studied in many different organisms. At the nucleolus, where rDNA transcription by RNA polymerase I (Pol I) happens, Top1's accumulation depends on its ability to interact with Pol I (Rose et al. 1988;Christensen et al. 2004). At other active loci such as the heat shock genes in Drosophila, Top1's recruitment might involve interaction with RNA Pol II or directly with the transcribed region (Gilmour et al. 1986;Gilmour and Elgin 1987;Shaiu and Hsieh 1998). In addition to interactions with the transcriptional machineries, other mechanisms of Top1's recruitment have been implicated, such as interacting with supercoiled DNA (Muller 1985;Zechiedrich and Osheroff 1990) or with specific DNA sequences (Bonven et al. 1985;Christiansen et al. 1987). Therefore, our understanding of the enzymology and of the recruitment of Top1 by transcriptional activities is extensive, yet that of Top1's in vivo functions, particularly those in the developmental context or independent of transcription, is less so in comparison.
In contrast to simpler eukaryotes, loss of Top1 does not support life in complex organisms such as the fruit fly Drosophila (Lee et al. 1993) or the worm Caenorhabditis elegans . Several functions of Top1 that might be specific to higher eukaryotes have been shown or proposed in different organisms. In early embryos of Drosophila, in the absence of ongoing transcription genome wide, a reduction of Top1 level leads to mitotic catastrophe (Zhang et al. 2000). A fly Top1 mutation suppresses seizure by regulating neuronal cell death (Song et al. 2007). Top1's function in the meiotic segregation of nonexchanged chromosomes in Drosophila has been proposed (McKee et al. 1992), and its activity is weakly required for pairing of homologous chromosomes in mouse meiosis (Cobb et al. 1997). In addition, Top1 purified from human and Drosophila cells possesses kinase activities toward some of the SR proteins that participate in splicing regulation (Rossi et al. 1996;Juge et al. 2010), suggesting an additional mode of Top1 action in the regulation of gene expression.
A better understanding of Top1 in complex organisms may require a thorough description of its localization in different tissue and cell types. Previously, studies of Top1 localization in Drosophila cells have been limited to polytene cells from larval salivary glands (Gilmour and Elgin 1987;Shaiu and Hsieh 1998;Zobeck et al. 2010) and cells in the earliest divisions (Zhang et al. 2000). Here we conducted a more thorough investigation into the cellular localization of Top1 in developmental stage and cell-type-specific manners, focusing primarily on interphase cells. We uncovered new features of Top1 distribution that may have important implications for its function.

Drosophila stocks
The Top1 mutation was generated as described below. The Top1 40 and Top1 26 alleles were used in this study. The Top1 40 and Top1 26 alleles carry an 8-bp and a 1-bp deletion of the Top1 coding region, respectively, both leading to a frame-shift mutation with prematured stop codons downstream of the mutation. A stock with the Top1 CC01414 allele and one with a marked Y chromosome (B S Yy þ ) were obtained from the Bloomington Stock Center. Fly lines carrying the bam-gal4 and nos-gal4 drivers were gifts from Prof. Xin Chen at Johns Hopkins University, and topi-gal4 was a gift from Prof. Christian Lehner at the University of Zurich.

Cas-9-mediated mutagenesis of Top1
We generated new Top1 alleles with CRISPR-Cas9-mediated mutagenesis using a transgenic approach in which both the Cas9 protein (expressed from a vasa promoter) and gRNA (expressed from a U6 promoter) were produced from transgenes inserted into the Drosophila genome (Port et al. 2014). The target gRNA was designed with an online tool: http://tools.flycrispr.molbio. wisc.edu/targetFinder/. A target sequence was chosen that has the sequence of 5 0 -GGCGCGCAAGAAGGTTAAGAAGG with the PAM sequence in bold. Mutations were verified by genomic PCR and sequencing using DNA samples from homozygous mutant larvae.
Construction of Top1-gfp and Top1 Y932F -gfp transgenic lines DNA from the Top1 locus is carried by the BAC Clone #48D17 and was subcloned into the pUAST-attB (Bischof et al. 2007) vector by gap repair based on the technique of "recombineering" (Venken et al. 2006;Wesolowska and Rong 2013), and the gfp gene was subsequently inserted at the end of Top1 coding region before the stop codon by recombineering (Zhang et al. 2014). Top1-gfp fly line was generated by PhiC31 integrase-mediated germline insertions on chromosomes II (25C) and chromosome III (86F). To test the ability of Top1-gfp to rescue the lethality of Top1 mutant animals, female flies heterozygous for the X-linked Top1 frameshift mutations were mated to wild-type males carrying a single copy of the Top1-gfp transgene inserted on an autosome. Male progeny that carried the Top1 mutation and Top1-gfp survive while those carrying Top1 mutation alone were not recovered. For progeny counts from this rescuing cross, see Supplementary Table S1 in Supplemental Materials. Both chromosome II and III insertions of Top1-gfp rescued the lethality. The insertion on chromosome II was used primarily for localization studies.

Live GFP imaging
For live imaging, wing discs and salivary glands from third instar larvae, and testes and ovaries from adult flies were dissected in PBS, placed on a coverslip in a drop of PBS, squashed gently by the weight of the slide lowered on top before imaging.

Immunostaining
Embryo collections were performed on grape juice plates for 2 h, dechorionated with 50% bleach and washed with embryo wash buffer (0.7% NaCl, 0.05% TritonX-100), fixed with a 1:1 mixture of freshly diluted 3.7% formaldehyde in PBS and heptane for 30 min, and devitellinized in methanol and heptane (1:1) following the slow formaldehyde fix method described by Sullivan et al. (2000). Purified TALE-mCherry proteins were kindly provided by Prof. Kai Yuan of the Central South University in China and used as an antibody in localizing the 359 satellite. A mouse MPM-2 antibody from Abcam was used at 1:1000.
Larval or adult tissues were dissected in fresh PBS, fixed with a 1:1 mixture of freshly diluted 3.7% formaldehyde in PBS and heptane for 30 min at room temperature. Tissues were then washed three times in PBST (1Â PBS containing 0.1% of Tween 20) for 15 min each, blocked in 3% bovine serum albumin (BSA) in PBST for 1 h at room temperature, followed by an overnight incubation with a rabbit anti-Fib (1:400, Abcam) in 3% BSA in PBST at 4 C. After three washes as above, the samples were incubated in Alexa Fluor 555 conjugated goat antirabbit IgG (1:200, Thermal) for 1 h at room temperature, followed by three washes in PBST and stained with DAPI for 10 min, and mounted in VECTASHIELD.
Confocal microscopy was performed on an Olympus FV1200 (equipped with VIS, UV, and IR lasers). Images were processed with photoshop and Illustrator (CS6; Adobe).

Actinomycin D and heat shock treatments of larval salivary glands
Salivary glands were dissected from third instar larvae in 1Â PBS and then incubated with 0.05, 0.5, 1, or 10 mg/ml of actinomycin D (ActD) solution for 1.5 h at room temperature. Heat shock treatment of salivary glands was done by treating the glands in Schneider's Drosophila Medium for 30 min at 37 C in a water bath. Finally, the salivary glands were stained with 2 mg/ml of Hoechst 33342 for 10 min. Fluorescence observation was carried out under a Zeiss microscope with a 40Â objective.

Data availability
Strains and plasmids are available upon request. The authors affirm that all data necessary for confirming the conclusions of the article are present within the article, figures, and tables.
Supplementary material is available at G3 online.

Results and discussion
A gfp-tagged Top1 transgene for the study of Top1 localization The best approach for tagging a protein with GFP is to knock-in a gfp fragment to the endogenous locus. Unfortunately, our repeated attempts at the Top1 locus turned out unsuccessful. To monitor Top1 protein localization, we constructed a transgene carrying the endogenous regulatory elements of the Top1 locus and expresses Top1 protein fused with GFP at its C-terminus ( Figure 1). To clone the relatively large size of the Top1 locus, we used the method of gap repair by recombineering (Venken et al. 2006;Wesolowska and Rong 2013) and successfully inserted a 13kb genomic fragment of the Top1 locus with about 1 kb each of upstream and downstream sequences ( Figure 1). Also using the recombineering method, we inserted the coding sequence for GFP just upstream of the stop codon of Top1 (Zhang et al. 2014).
To test the suitability of this tagged Top1 transgene in reflecting the in vivo behavior of the endogenous untagged Top1, we introduced it into Top1 frameshift mutations that we generated by CRISPR-mediated mutagenesis (see Materials and Methods) and discovered that it rescued the lethality caused by the mutations so that flies carrying both the Top1 mutation and a single copy of the Top1-gfp transgene on an autosome were recovered normally (Supplementary Table S1). This suggests that our Top1-GFP likely assumes functional localization similar to the endogenous Top1 proteins. To assess the influence of Top1 enzymatic activity on its cellular localization, we constructed an identically tagged Top1 transgene with the Tyrosine residue of the active site mutated to a Phenylalanine residue (the Top1 Y932F mutation). As expected, this mutated form of Top1 was not able to rescue the lethality caused by Top1 frameshift mutations. Nevertheless, under an otherwise wild-type background, it displays nuclear GFP signals similar to that of Top1-GFP (see later sections).
A gfp-tagged Top1 allele (Top1 CC01414 ) has been recovered previously in a transposon-mediated gene tagging project (Morin et al. 2001). In this allele, a P element was inserted in the first intron of Top1. A pre-mRNA splicing reaction using the splicing donoracceptor sites in the P element results in an in-frame fusion of gfp with the rest of the Top1 coding region. Indeed, this allele has been used previously to monitor Top1 localization using live GFP fluorescence in larval polytene cells (Zobeck et al. 2010). Our gfptagged Top1 transgene potentially has given us two modest advantages over the P element tagged allele. First, it simplified our introduction of the active site mutation (Top1 Y932F ) and other Top1 mutations in the future. Secondly, since GFP-tagged Top1 protein expressed from the Top1 CC01414 allele has to be produced by an ectopic splicing reaction, alternative splicing using the endogenous splicing sites of exon 1 would generate Top1 proteins lacking GFP. Indeed, multiple Top1 isoforms are predicted by genomic annotation (flybase.net) even though molecular analyses of Top1 transcripts have yet to lend support for their existence (Brown et al. 1998). Nevertheless, as shown in later sections, whenever we compared localization patterns of GFP-tagged Top1 produced from either experimental set up (transgene vs "tagging by splicing"), we obtained very similar if not identical results. This series of "control" experiments further validated our transgenic approach for studying Top1 localization.

Nucleolar enrichment of Top1 in metabolically active polyploid cells
One of the most important function of Top1 is in transcription regulation, particularly in the regulation of rRNA synthesis by RNA Pol I. Consistently, Top1 is enriched in the nucleolus of eukaryotic cells (Fleischmann et al. 1984;Muller et al. 1985;Zhang et al. 2000;Christensen et al. 2004;Juge et al. 2010;Cha et al. 2012). We investigated whether Top1 is similarly distributed in metabolically active but nondividing cells in the female germline and observed nuclear GFP signal in ovaries of flies carrying the Top1 gfp , Top1 Y932F , or Top1 CC01414 alleles individually ( Figure 2). More specifically, Top1 can be seen on chromatin in polyploid nurse cells but with a prominent enrichment at a "chromatin-poor" region that likely corresponds to the nucleolus. In somatic follicle cells, which are also polyploid in later egg chambers, Top1 has a similar bipartite mode of distribution: weaker GFP fluorescence on chromatin and a strong enrichment in the presumed nucleolus ( Figure 2, B and C). To confirm the region with the strongest Top1-GFP signal does represent the nucleolus, we used Fibrillarin (Fib) as a nucleolar marker in antibody staining of ovaries expressing Top1-GFP and observed colocalization of GFP and anti-Fib signals ( Figure 2B). Interestingly, Top1-GFP localization in the nucleolus is not uniform but displays very strong puncta ( Figure 2B) that might represent the "Fibrillar Centers" previously identified as Top1-rich Christensen et al. 2004). To investigate whether the overproduction of Top1 proteins, because of the presence of the gfp-tagged Top1 transgene, might have led to abnormal localization of Top1-GFP, we studied its localization in the female germlines that are also homozygous for the Top1 frameshift mutation on the X chromosome. The distribution pattern of Top1-GFP in these "rescued" germlines is indistinguishable from that in the "over-expressed" germlines ( Figure  2, A and B).
Besides somatic follicle cells and germline nurse cells, which are metabolically active to support oogenesis, we also observed Top1-GFP in the nucleus of the oocyte ( Figure 2C). Interestingly, Top1-GFP forms a large body inside the oocyte nucleus that is separated from the bulk of chromatin. A large "extra-chromatin" body inside the oocyte was previously observed for Top1 (Liu et al. 2006a) and Nopp140, another nucleolar protein (McCain et al. 2006). The biological function of this large structure remains obscure.
We also investigated Top1-GFP localization in another type of metabolically active cells: the giant polytene cells from the salivary glands of a third instar larva. As shown in Figure 3, A and B, Top1-GFP or its mutant form Top1 Y932F is highly enriched at the nucleolus but is nonuniform. Since not all rDNA units are transcribed (Franz and Kunz 1981;Muscarella et al. 1987), we suggest that the GFP-bright regions in individual nucleoli might represent rDNA units actively being transcribed by Pol I.

Top1 distribution upon transcription perturbation
We applied conditions that are known to affect transcription to probe the regulation of Top1-GFP localization in larval polytene cells. First, we treated polytene cells with the known transcriptional inhibitor, ActD (Christensen et al. 2004). ActD is a general transcription inhibitor that affects both RNA Pol I and II polymerases. However, based on mammalian cell culture studies, the sensitivity of Pol I to ActD is one to two orders of magnitude Figure 1 Top1 alleles used in the study. The wt Top1 locus is shown at the top with exons denoted as black boxes, which are connected with lines that indicate introns. The two point mutations (Top1 26 and Top1 40 ) are shown below with the approximate position of the small deletions indicated with an asterisk. The approximate position of the insertional site of the GFP tag is indicated for Top1 gfp . In the Top1 Y932F allele, the Y to F change is denoted by an asterisk and the insertion of gfp at the Cterminus is also indicated. The Top1 CC01414 allele has been described in Morin et al. (2001) and FlyBase, and the approximate position of the insertion site of the P element is indicated. higher than that of Pol II (Perry and Kelley 1970;Bensaude 2011). We tested a series of ActD concentrations for their effects on Top1-GFP localization and arrived at a concentration of 10 mg/ml as the lowest concentration that yields a consistent effect (Supplementary Figure S1 in Supplemental Materials). At this concentration, ActD abolished Top1-GFP localization to the nucleolus with little effect on Top1-GFP localization on chromosomes ( Figure 3C). We also tested two other compounds that have been used for Pol I inhibition in mammalian cells: BMH-21 (Colis et al. 2014) and CX-5461 (Rossetti et al. 2018), even though their efficacy in transcription inhibition has not been demonstrated in Drosophila. As shown in Supplementary Figure S1, neither drug elicited a change of Top1-GFP localization at a concentration significantly higher than that used in the mammalian studies and was not pursued further. Nevertheless, our results from ActD treatment are consistent with that Top1 localization to the nucleolus depends on Pol I activities. Our results also suggest that the non-nucleolar localization of Top1 is independent of Pol I activities, but could be via other activities such as those maintained by Pol II.
We first applied the known Pol II inhibitor DRB to probe Top1-GFP localization regulation. As shown in Supplementary Figure  S1, we did not observe a discernable change of Top1-GFP distribution at a DRB concentration (100 mM) similar to ones previously used (Petesch and Lis 2008; Teves and Henikoff 2011). Our second way of altering Pol II transcription was by applying a heat shock treatment (HS). HS induces a global repression of transcription that includes a reduction of paused Pol II (Teves and Henikoff 2011) and a dissociation of Pol II from chromatin in Drosophila and mammalian cells (Jamrich et al. 1977;Hieda et al. 2005). As shown in Figure 3C, a 30-min HS at 37 C resulted in a dramatic "clearing" of Top1-GFP from chromosomal sites. Our results are consistent with the scenario in which a general inhibition of Pol II transcription on genes other than the heat shock loci leads to a reduction of chromatin-bound Top1 except those at the nucleolus. We did observe a few chromosome sites with persistent Top1-GFP signals (arrows in Figure 3C) and suggest that these represent the loci encoding the heat shock proteins, which have been shown previously to accumulate Pol II and Top1 upon HS (Fleischmann et al. 1984;Gilmour and Elgin 1987;Shaiu and Hsieh 1998;Zobeck et al. 2010).
When we combined the ActD and HS treatments, we made an interesting but unexpected observation. As we showed earlier, HS leads to a dramatic clearing of Top1-GFP from chromosomal sites (the "HS" panel in Figure 3C). This clearing was halted if HS was preceded with an ActD treatment (the "ActDþHS" panel in Figure  3C), while Top1's accumulation at the nucleolus was again greatly reduced. In essence, the effect on Top1-GFP localization from the "ActD only" treatment is very similar to that from an "ActDþHS" treatment [compare the "ActD(10)" panel with the  An embryo at cycle 9 was used as a representative for displaying Top1-GFP localization in embryos of the "Early stage." GFP signal was overexposed to show the lack of Top1-GFP foci in interphase nuclei. In the embryos at Cycles 12, 13, or 14, nuclear Top1-GFP foci were observed. Top1-GFP is absent from mitotic chromosomes. The quantification of the number of Top1-GFP focus per interphase nucleus is shown in the chart to the right of the images. The embryos were sexed based on the number of the X-linked 359 satellite locus in the embryos using the method described in Supplementary Figure S1 and (C) in this figure. Scale bar represents 10 mm. (B) Top1-GFP foci in postblastoderm embryos. At the top is a lower magnification image of an area from a postblastoderm embryo showing uneven nucleus distribution. Scale bar represents 50 mm. Higher magnification images are shown in the bottom three panels with Top1-GFP forming one to two prominent foci per nucleus. Scale bar represents 10 mm. (C) The TALE-light method for determining the positional relationship between Top1-GFP focus and the 359 satellite. The top panels show images of syncytial embryos expressing Top1-GFP (in green) that have been stained with TALE-light (in red) for labeling of the 359 satellite. The "merged" image shows GFP and TALE-light signals. Scale bar represents 5 mm. The sex of the embryos is indicated to the left of the images and was determined based on the criteria described in Supplementary Figure S1. DAPI signals were used to estimate the boundary of a nucleus, which is marked with a circle. We identified four classes of nuclei, which are labeled from 1 to 4 next to the nuclear circle. In a class "1" nucleus, the single green focus juxtaposes a red focus. In a class "2" nucleus, the single green focus does not juxtapose any of the red focus. In a class "3" nucleus, one of the two green foci juxtaposes a red focus. In a class "4" nucleus, none of the two green foci juxtaposes a red focus. Higher magnification pictures of the four classes of nuclei are shown to the right of the main image panels. Scale bar indicates 1 mm. Below the image panels is the quantification for the four classes in male (left chart) and female (right chart) embryos. (D) Top1-GFP colocalization with the MPM-2 epitope. In the merged image, only Top1-GFP and anti-MPM-2 signals are shown. n ¼ 30 embryos. Magnified images of the areas marked with dashed rectangles are shown as inserts. Scale bar indicates 5 mm. A line-scan analysis from eight nuclei is provided as the bottom and error bars indicate SEM.
"ActD(10)þHS" panel in Figure 3C]. It is intriguing that Top1 failed to vacate from chromatin when both rDNA transcription by Pol I and Pol II transcription genome wide were inhibited. It is possible that the ActD treatment resulted in the immobilization of Top1-GFP on chromatin as suggested by prior results in mammalian cells (Trask and Muller 1988). However, we observed similar dynamics for the active site mutated Top1-GFP (Top1 Y932F , Figure  3C lower panels) suggesting that normal Top1 enzymatic activity was not involved in the presumed Top1 immobilization by ActD.
When we reversed the sequence of the two treatments (HS followed by ActD), neither the clearing of "chromosomal" Top1-GFP nor the accumulation of Top1-GFP in the nucleolus could be reversed by ActD (Supplementary Figure S1B). These results suggest that cells treated under the current conditions might have entered an arrested state thus are incapable of responding to further stimulations.

Top1 enriched loci in syncytial embryos
Animals with no intact copy of the Top1 gene die as early larvae possibly due to a block of tissue proliferation (Lee et al. 1993;Zhang et al. 2000), which was similarly observed for our Top1 alleles. Therefore, determining Top1's function in early embryonic development requires sophisticated genetic manipulations. Based on a clever design using the heat shock promoter to drive Top1 expression, Zhang et al. (2000) supplied sufficient Top1 only to somatic tissues during oogenesis so that early embryos with a reduced level of Top1 could be studied. They discovered that these Top1-insufficient embryos suffered mitotic catastrophe and arrested early in development. However, several issues remain unresolved if one considers that Top1's primary role is in transcription regulation, in particular transcription originated from the highly active rDNA locus. Early cycles in syncytial embryos are run on maternal contributions without zygotic transcription. Falahati et al. (2016) showed that the rDNA locus is among one of the first zygotically active loci where rDNA transcripts can be first detected at cycle 11 and become prominent at cycle 12. Therefore, one might expect that the maternal Top1 proteins start to accumulate at rDNA at the time of its activation. However, results from Top1 immunostaining experiments of early embryos suggest a broad distribution in interphase nuclei and the disappearance of Top1 from mitotic chromosomes (Zhang et al. 2000). In addition, some of the embryos deficient for Top1 arrest earlier than cycle 12 with mitotic defects, suggesting that Top1 might be needed specifically for early cell divisions but function independently of its role in transcription since major zygotic transcription does not start until cycle 14.
To gain possible insights into Top1's role in early embryonic cycles, we studied the localization of Top1-GFP in syncytial embryos collected 0-2 h after egg laying. As shown in Figure 4A, nuclear Top1-GFP signal is undetectable in embryos cycling before cycle 12 (n ¼ 7 embryos), while in those cycling at or after cycle 12 (n ¼ 48) nuclear Top1-GFP is clearly visible and forms 1-2 bright foci per interphase nucleus (quantification in Figure 4A). Consistent with earlier results (Zhang et al. 2000), Top1 is not enriched on the mitotic chromosomes of the embryos ( Figure 4A). In postblastoderm embryos, Top1 foci become more intense but again forming 1-2 foci per nucleus ( Figure 4B).
The interesting coappearance of rDNA transcription and Top1-GFP foci in and around cycle 12 prompted us to investigate whether Top1-GFP foci are located at the rDNA loci. Drosophila has two rDNA gene clusters, one near the centromere of the X chromosome and the other on the short arm of the Y chromosome. Next to the X-linked rDNA is the 359 satellite that consists of about 11 Mb of middle-repetitive elements (Tang et al. 2017). Yuan et al. (2014) developed the "Tale-Light" method in which a TALE DNA-binding domain was engineered to specifically recognize a sequence in the repeated unit of the 359 satellite. When fused with the mCherry fluorescent protein and purified from bacteria, this TALE-Light protein allows visualization of the 359satellite in fixed samples. The invention of the 359-TALE-light thus offers a unique way to approximate the localization of Top1-GFP foci in relationship with rDNA on the X vs the Y chromosomes since only the X-linked one is situated next to 359. For representative images showing the relative locations between 359 satellite and the rDNA locus in syncytial embryos, see Tang et al. (2017).
We used fixed syncytial embryos in which green fluorescence from Top1-GFP was detectable under a confocal microscope. These embryos have also been incubated with TALE-mCherry recombinant proteins similar to a primary antibody used in a traditional immunostaining experiment. When we estimated the number of TALE-mCherry foci per nucleus in a given embryo, we observed two classes that we suggest represent males and females. Representative images are shown in Supplementary Figure S2 in Supplemental Materials. In one class, presumably the female class about 10-20% of the nuclei have two TALE-mCherry foci (quantification can be found in Supplementary  Table S2 in Supplemental Materials). The cases of a single focus in these embryos likely represent "pairing" of the two homologous Xs at the 359 satellite. In the other class, presumably the male class, more than 99% of the nuclei had one TALE-mCherry focus. The rare cases of two foci in these embryos are likely the result of sister chromatid separation at the 359 regions.
As shown in Figure 4C, the majority (70-80%) of Top1-GFP foci are not situated in close proximity to a 359 satellite whether it is in a male or female embryo. This result applies to all Top1-GFP expressing stocks that we tested, and on two independent trials. Therefore, Top1-GFP focus in about 70% of the female nuclei is not at either of the rDNA loci on the X. In male embryos, on the other hand, we cannot rule out that the majority of Top1-GFP foci were actually associated with the rDNA locus on the Y chromosome therefore apart from the 359 satellite on X. This possibility would imply sex-specific mechanisms for Top1 recruitment so that Top1 is rDNA-associated in male but not female embryos. Warsinger-Pepe et al. (2020) reported previously that differential transcriptional activities from the X and Y rDNA loci remain insignificant in pregastrulating male embryos, similar to those that we examined. Therefore, a plausible sex-specific mechanism remains missing for localizing Top1 to the Y-linked rDNA locus but not the X-linked copy. Therefore, the most parsimonious explanation for our results would be that the single Top1-GFP focus in a male embryo is not at rDNA.
We therefore speculate that in a significant portion of the nuclei in preblastoderm embryos, the strongest Top1-GFP foci do not colocalize with the rDNA locus. Another nuclear body forms during early embryonic development is the histone locus body (HLB) that colocalizes with the histone gene cluster (Liu et al. 2006b;White et al. 2007). We tested the possibility that Top1-GFP accumulates at HLB by using an antibody that recognized the MPM-2 epitope, a marker for HLB in early embryos (White et al. 2007). Remarkably, Top1-GFP colocalizes with the strongest MPM-2 signal ( Figure 4D). Although this result suggests that Top1-GFP accumulates at HLB during early development and would be consistent with Top1's enrichment at a transcriptionally active region of the genome, the final confirmation requires colocalizing Top1-GFP with the histone locus and other HLB markers such as the U7 snRNP important for Histone mRNA processing (Liu et al. 2006b). We note that there are minor MPM-2 positive foci colocalizing with Top1-GFP signals. The molecular nature and functional significance of these secondary foci require future investigations.

Top1-GFP may mark both rDNA loci in larval diploid cells
We used cells in larval imaginal discs as another cell type to study Top1-GFP localization in diploid cells. In particular, these cells are primarily at interphase, different from those in early embryos that are highly active in cell divisions. We focused our live analyses on cells from the wing discs of third instar larvae. As shown in Figure 5A and quantified in Figure 5B, in 94.7-99.6% of the nuclei Top1-GFP forms 1-2 foci per nucleus. Our hypothesis was that these major foci are localized at rDNA. Testing of the hypothesis requires means to simultaneously identify rDNA and Top1-GFP loci in disc cells.
When we performed immunostaining with anti-Fib in these cells, the Top1-GFP signal overlaps with that of Fib antibody suggesting that at least some of the Top1-GFP is at the nucleolus ( Figure 5C). However, these results from immunostaining were unsatisfactory as GFP signals are too diffuse to define as clear a focus as we were able to do in live analyses. Moreover, although Fib is a common marker for the nucleolus, which is derived from active rDNA transcription, Fib could not indicate the location of the inactive rDNA locus, which is important to identify in males as discussed later. In addition, we attempted but unsuccessfully to apply the TALE-light method described before in larval imaginal discs. Without the ability to simultaneously identify both rDNA and Top1-GFP loci in discs, we nevertheless investigated the distribution pattern of Top1-GFP focus number using larvae carrying different combinations of the sex chromosomes: XX, XY, and XO to extract any relationship between Top1-GFP and rDNA, but without the a priori assumption that Top1-GFP is located at the rDNA loci.
As third instar larvae can be sexed visually (Kerkis 1931), we first investigated whether the number of Top1-GFP focus correlates with the sex of the animal. As shown in Figure 5B, we observed a significant increase of nuclei with two foci in male vs female larvae (P < 0.000001 from a 2X2 contingency test). Since the only difference in chromosome composition between the two sexes lies in the sex chromosomes, the change in Top1-GFP focus number was likely brought about by the same mechanism, suggesting that Top1-GFP foci are likely on one of the two sex chromosomes. In addition, the increase of the two-focus class in males is inconsistent with that all of the Top1-GFP foci are on the X chromosome since males have one less X than females. Therefore, at least some of the foci must be on the Y chromosome. This would be consistent with Top1-GFP foci at rDNA as it is a homologous locus between the sex chromosomes.
Although a female nucleus has two rDNA loci on each of its two Xs, the somatic pairing would reduce the average closer to one when observed cytologically. As the X and the Y are not homologous and hence not paired, a male nucleus with two foci might represent Top1-GFP accumulation on both of its sex chromosomes. To provide additional support for this hypothesis, we measured the number of Top1-GFP focus in wing disc cells from XO larvae. Again, we observed predominantly one or two foci per nucleus ( Figure 5B). The distributions of nucleus between one and two foci are not statistically different between XO and XX discs (P ¼ 0.07), consistent with our suggestion that the number of rDNA locus in a female nucleus is effectively one due to somatic pairing. The two-foci class in an XO nucleus could be the result of transient sister chromatid separation at rDNA. Interestingly, the portion of XY nuclei with two Top1-GFP foci is significantly greater than that of XO nuclei (P < 0.000001). Since this increase must have been brought about by the presence of the Y chromosome, we suggest that at least in some of these two-foci nuclei both rDNA loci were marked with Top1-GFP accumulation.
An important alternative explanation of the above results is that one or both of the Top1-GFP foci are located at places unrelated to rDNA, such as the HLB similar to what we observed in embryos ( Figure 4D). However, this model would have to correlate the number of Top1-GFP focus with changes in the number (one or two) and composition (X or Y) of the sex chromosomes, and we find that difficult since the histone cluster is on chromosome 2.
Recently, the phenomenon of "nucleolus dominance" was described in Drosophila melanogaster in that somatic cells of males express rRNA from the Y chromosome, while the rDNA locus on the X is silenced (Greil and Ahmad 2012;Zhou et al. 2012;Warsinger-Pepe et al. 2020). As Top1 recruitment to chromatin has been frequently linked with transcription, we suggest that in an XY nucleus with a single Top1-GFP focus, it is likely associated with the active Y rDNA, while a two-foci nucleus has both rDNA loci marked with Top1. Our data thus suggest that rDNA on the X chromosome in males is also capable of attracting Top1 even when it is transcriptionally silenced. It is possible that a transcription-independent recruitment mechanism exists for Top1 as what has been reported in Tetrahymena in which a sequence fragment contained in the intergenic sequence (IGS) of Tetrahymena rDNA displays a high affinity for Top1-binding in vitro (Bonven et al. 1985). A highly homologous sequence also exists in the IGS of Drosophila rDNA. Unfortunately, we failed repeatedly to introduce a DNA fragment containing 32 copies of the 240-bp IGS into the Drosophila genome. This precluded us from testing the hypothesis that sequence-specific binding of Top1 to IGS serves as another mechanism for Top1 recruitment in Drosophila. We note that Thomas and McKee (2007) succeeded in introducing an eight-copy IGS fragment carried on a transgene.
Top1 marks the X-Y bivalent in meiotic cells and potentially function to ensure X-Y disjunction Top1's function in transcription regulation has been extensively studied during mitotic growth. Its meiotic function is less clear. We set out to investigate Top1-GFP distribution in testicular cells out of consideration of a special relationship between rDNA and meiotic chromosome segregation in Drosophila males [for a review of Drosophila male meiosis, see McKee et al. (2012)]. The landmark study by McKee and Karpen (1990) established that the rDNA loci on the sex chromosomes mediate their conjunction to ensure their segregation during meiosis I. It was subsequently proposed by McKee et al. (1992) that Top1, known to be enriched at the rDNA loci in somatic cells, might serve to regulate X-Y disjunction in Drosophila. In addition, the Top1-interacting dTopors protein regulates chromosome segregation in Drosophila males (Matsui et al. 2011).
As shown in Figure 6A, Top1-GFP forms foci in cells from the mitotic compartment of the testis. In particular, Top1-GFP foci are the largest in size in spermatocytes ( Figure 6, A and B), consistent with the high transcriptional activity in these cells. Using phase-contrast microscopy, we observed colocalization between the strongest Top1-GFP foci and the nucleolus in these cells ( Figure 6B). These Top1-GFP foci juxtapose the X-Y bivalent when they become visible after the initiation of meiotic chromosome condensation ("S4-5 stage" in Figure 6A). Interestingly, even in cells with highly condensed meiotic chromosomes, Top1-GFP are prominently located on regions of chromosomes ("S6 stage" in Figure 6A).
To test whether Top1 participates in regulating X-Y disjunction, we used RNAi to reduce Top1 level in the testes as Top1 knockout mutations cause lethality. As shown in Figure 6, B and C, our Top1-GFP can be effectively knocked down (KD) by RNAi reagents driven by the germline-specific bam-gal4 or nos-gal4 drivers. Therefore, using Top1-GFP as a reporter, we demonstrated that Top1 could be effectively KD, at least in the mitotic compartment of the testis.
Using a marked Y chromosome (B S Yy þ ), we investigated whether Top1 KD in the male germline induces X-Y nondisjunction (NDJ). Normally, the Y chromosome from the father is transmitted solely to its male progeny. In the event of an X-Y NDJ, progenies of two exceptional classes are recovered: sons bearing no Y chromosome (XO) and daughters bearing a Y chromosome (XXY). As shown in Table 1, when Top1 was KD, X-Y NDJ at a rate about 10-fold higher than the normal rate was indeed observed. To confirm that the exceptional classes of progenies indeed carry the expected sex chromosome compositions, we performed additional tests as followed. First, all suspected XO sons were confirmed to be sterile by test-mating with wild-type females, and they were subsequently dissected to reveal the presence of needle-shaped crystalline in primary spermatocytes ( Figure 6D), a cellular phenotype consistent with the absence of the Y chromosome (Hardy et al. 1981). Secondly, some of the XXY offspring were mated to wild-type males and some of their progenies were subjected to karyotyping in which neuroblasts from female third  Cenci et al. (1994). Three alleles of Top1-GFP are used with the genotypes indicated to the right of the image panels. The potential rDNA foci are indicated by arrows. For the Top1 gfp allele, magnified images are also provided and organized similarly according to stages of development. In the "Spermatogonia" panel, a mitotic nucleus is marked with an arrowhead. Scale bar represents 20 mm. (B) Top1-GFP KD in primary spermatocytes. Images from phase-contrast microscopy (Phase) and live fluorescent microscopy (GFP) are shown for primary spermatocytes from a wild-type male (top) expressing Top1-GFP. Magnified images of a single spermatocyte nucleus are shown as inserts. The position of the nucleolus is indicated by an arrow. Images from DAPI staining and live fluorescent microscopy (GFP) are shown for a male experiencing Top1-GFP KD driven by nos-Gal4 (bottom). Scale bar indicates 10 mm. instar larvae were subjected to a mitotic chromosome squash assay. Some of the squashes clearly display the presence of the extra Y chromosome ( Figure 6D). The presence of the extra Y in any of the daughters of a suspected F1 female confirms that this exceptional female in question was indeed XXY. Importantly, it is the presence of these XXY exceptional progenies that indicates that these exceptional events were the result of meiosis I NDJ.
One possible cause for the appearance of the above-described exceptional offspring is mitotic NDJ event, involving sister chromatids for example, which could be brought about by chromosome exchanges between the rDNA loci. This would result in germ cells that are XO, XXY, XYY, or XXYY. We suggest this to be unlikely based on the following considerations. First, an XO spermatogonium would be nonunctional due to the absence of the Ylinked fertility factors. It is therefore difficult to account for the XO exceptional male offspring based on mitotic NDJ alone. Secondly, the exceptional progenies were all from different fathers inconsistent with the expected "clustering" of mitotic events. The caveat here is that the number of exceptional offspring was much smaller than the number of male parents tested in our experiments. Lastly, a physical linkage due to an aberrant exchange between the rDNA loci on the X and Y could have forced them to mis-segregate in meiosis I. We suggest that this would be resolved in the offspring leading to the formation of Xand Y-derived chromosomes, similar to those derived from rDNA exchanges induced by the I-CreI endonuclease (Maggert and Golic 2005). Our sampling of the XXY females by karyotyping their offspring offers no support for the existence of such aberrant chromosomes ( Figure 6D).
Data shown in Table 1 appear to suggest that not all gal4 drivers were equally effective in inducing X-Y NDJ. However, although differences in NDJ frequency between Top1 KD germlines and the non-RNAi controls reach statistical significance except for the cases of bam-driven KD, differences between bam-driven KD vs nos-driven or topi-driven KD are not statistically significant (P ¼ 0.12 for the THO2218.N siRNA and P ¼ 0.19 for THU0318). We nevertheless speculate that different stages of Gal4 expression might affect meiotic chromosome segregation to different degrees. The expression of the bam gene is limited to the mitotically amplifying cells in the spermatogonium (Schulz et al. 2004). Importantly, bam expression is very low or absent in spermatocytes right before meiosis. Consistently, we observed normal X-Y disjunction in bam-gal4 driven RNAi KD germline. Importantly, many of these males were sterile or semisterile, consistent with that Top1's reduction in mitotic cells likely causes germ cell loss. Interestingly, the nos-gal4 has a wider expression than bam including detectable expression in spermatocytes (Schulz et al. 2004). Consistently, the nos-gal4 driven Top1 KD induced NDJ at a significant rate. These males also suffer reduced fertility possibly owning to Top1 knockdown in proliferating cells. topi-gal4 has a narrower window (just prior to meiosis) and lower level of expression than either bam-gal4 or nos-gal4 (Raychaudhuri et al. 2012). Consistently, it is not very efficient in directing Top1 KD in the premeiotic compartment ( Figure 6C), and the "KD" males had normal fertility. Remarkably, topi-Gal4 directed Top1 KD is as effective in inducing X-Y NDJ as the more potent nos-Gal4, consistent with our proposition that the timing of Top1 KD is important in inducing NDJ.
Although our RNAi-mediated Top1 knockdown resulted in an X-Y NDJ rate (0.3%) significantly above the baseline level, it remains two orders of magnitude lower than that observed for male hemizygous for the rDNA locus (essentially random X-Y segregation, hence 50%). Other known mutants affecting X-Y segregation have a range of NDJ of 20-50% (e.g., Thomas et al. 2005;Matsui et al. 2011), again significantly higher than what we observed here. This could be due to that Top1 is not critically involved in directing X-Y disjunction. Alternatively, the small effect of Top1 KD on X-Y disjunction could be due to the ineffectiveness of our current system in reducing Top1 level.
The potential role of Top1 in mediating meiotic chromosome segregation, as proposed by McKee et al. (1992), was not meant to be limited to the sex bivalent, but rather applicable to chromosomes undergoing achiasmatic pairing. In our analyses, Top1-GFP signals are clearly present on autosomes ("S6 stage" in Figure  6A), consistent with a boarder role of Top1 proposed earlier. A well-timed elimination of Top1 molecules in premeiotic cells might nevertheless be essential to efficiently impair chromosome conjunction in meiosis. We plan to continue developing tools to test this hypothesis. Until then we suggest that it remains possible that Top1 is required for meiotic chromosome disjunction in Drosophila males.

Concluding remarks
In this study, we conducted a comprehensive characterization of the localization of the important chromosomal protein Top1 during the complex developmental program of a metazoan. Our results confirm a prominent enrichment of Top1 at the rDNA locus and its derived structure, the nucleolus. In addition, we The genotypes of the test males are shown in the leftmost column. THO2218.N and THU0318 are two independent Top1 RNAi lines. The four classes of progeny were scored for each testcross according to their sex chromosome compositions. The frequency of nondisjunction ("NDJ%") was calculated by dividing the sum of exceptional progenies (XO and XXY) by the total number of progenies. "N" indicates the number of fertile male parents tested for NDJ. a A 2 Â 2 contingency test with the non-RNAi control yield P < 0.01. b A 2 Â 2 contingency test with the non-RNAi control yield P < 0.05. uncovered Top1 localization seemingly unrelated to ongoing rDNA transcription particularly in diploid cells, such as Top1 foci not situated at rDNA in preblastoderm embryos, Top1 foci on both the active (Y-linked) and inactive (X-linked) rDNA loci in cells from larval wing discs, and Top1 foci in transcriptionally quiescent meiotic cells in the testis. These results suggest the existence of additional Top1 recruitment mechanisms that might be related to its function in DNA metabolism.