Nuclear Structure and Chromosome Segregation in Drosophila Male Meiosis Depend on the Ubiquitin Ligase dTopors

Abstract

In many organisms, homolog pairing and synapsis at meiotic prophase depend on interactions between chromosomes and the nuclear membrane. Male Drosophila lack synapsis, but nonetheless, their chromosomes closely associate with the nuclear periphery at prophase I. To explore the functional significance of this association, we characterize mutations in nuclear blebber (nbl), a gene required for both spermatocyte nuclear shape and meiotic chromosome transmission. We demonstrate that nbl corresponds to dtopors, the Drosophila homolog of the mammalian dual ubiquitin/small ubiquitin-related modifier (SUMO) ligase Topors. We show that mutations in dtopors cause abnormalities in lamin localizations, centriole separation, and prophase I chromatin condensation and also cause anaphase I bridges that likely result from unresolved homolog connections. Bridge formation does not require mod(mdg4) in meiosis, suggesting that bridges do not result from misregulation of the male homolog conjunction complex. At the ultrastructural level, we observe disruption of nuclear shape, an uneven perinuclear space, and excess membranous structures. We show that dTopors localizes to the nuclear lamina at prophase, and also transiently to intranuclear foci. As a role of dtopors at gypsy insulator has been reported, we also asked whether these new alleles affected expression of the gypsy-induced mutation ct⁶ and found that it was unaltered in dtopors homozygotes. Our results indicate that dTopors is required for germline nuclear structure and meiotic chromosome segregation, but in contrast, is not necessary for gypsy insulator function. We suggest that dtopors plays a structural role in spermatocyte lamina that is critical for multiple aspects of meiotic chromosome transmission.

ASSOCIATIONS between chromosomes and the nuclear envelope during meiotic prophase are a widely conserved phenomenon important for proper chromosome transmission. In many species, such interactions are required for bouquet formation, an arrangement in which telomeres cluster in association with the nuclear envelope to facilitate homolog pairing and synapsis (reviewed in Scherthan 2007). In Caenorhabditis elegans, analogous interactions between nuclear envelope proteins and chromosomes are mediated by zinc finger proteins that connect chromosome-specific pairing sites to integral nuclear membrane proteins (Phillips et al. 2009). These connections establish bridges across the nuclear envelope and allow for interactions between meiotic chromosomes and cytoskeletal actin. Chromosome movements dependent on these connections are important for homolog pairing (Sato et al. 2009).

The association of meiotic chromosomes with the nuclear periphery is particularly striking in Drosophila males, in which paired homologs occupy discreet domains closely apposed to the nuclear membrane. The relevance of this organization to meiotic chromosome segregation in this organism, however, has not been explored. Drosophila males have an unconventional meiosis in which homologs pair but do not assemble synaptonemal complex (Meyer 1960) and do not undergo crossing over. The mechanism of pairing in unknown. Pairing sites have been defined for the sex chromosomes (Mckee and Karpen 1990); however, they have not been demonstrated to associate with the nuclear envelope.

Toward understanding the relationship of nuclear structure to meiotic chromosome segregation in this organism, we have examined mutations in nbl. Mutant nbl males exhibit defects in spermatocyte nuclear shape and also show a high frequency of fourth chromosome loss (Wakimoto et al. 2004). This suggested that further characterization of nbl might reveal aspects of nuclear organization important for meiotic chromosome transmission. Here we show that the nbl gene corresponds to dtopors, the fly homolog of Topors, a mammalian tumor suppressor that has both ubiquitin and small ubiquitin-related modifier (SUMO) ligase activities (Rajendra et al. 2004; Saleem et al. 2004; Pungaliya et al. 2007). Our results reveal a requirement for dtopors in both male germline nuclear envelope structure and in events associated with meiosis I chromosome segregation. We suggest that both requirements may be related to a role at the nuclear lamina. As dtopors has been implicated in gypsy insulator function, we also examine the effects of these new alleles on gypsy insulator function and find that in contrast to meiosis, dtopors is not required at gypsy insulators.

Materials and Methods

Drosophila stocks

Fly stocks were cultured on standard cornmeal molasses medium at 25°. The Z mutations were provided by C. Zuker laboratory at the University of California at San Diego and are described in Wakimoto et al. (2004). All other stocks were obtained from the Bloomington Stock Center and are described in FlyBase (Tweedie et al. 2009).

Mapping nbl mutations

The nbl mutations were mapped to the cytogenetic interval 56A1–B2 by failure of both Df(2R)Exel6068 (56A1–B5) and Df(2R)PC66 (55D2–56B2) to complement the fourth chromosome loss and the spermatocyte nuclear phenotypes. This localization was consistent with recombination mapping of nbl to the region between pr and px. Additional complementation tests revealed that Df(dtopors^AA), a small deletion that removes dtopors and the adjacent gene prod (Secombe and Parkhurst 2004), failed to complement the nuclear and chromosome segregation defects of nbl. Furthermore, an insertional mutation in dtopors, P(Bac{WH}Topors^f05115, failed to complement nbl, whereas an insertional mutation in prod, P{lacW}prod^k08810, fully complemented.

Genetic assays of meiotic chromosome segregation

To test whether dtopors mutations caused sex chromosome and fourth chromosome nondisjunction, y/y+Y; dtopors; spa^pol males were crossed to y w sn; C(4)RM ci ey/0 females. In these tests, normal progeny were y+ w sn sons and y daughters, whereas sex diplo-exceptional progeny that received both the X and Y from their fathers were distinguished as y+ females, and sex nullo-exceptional progeny that failed to receive a paternal sex chromosome were distinguished as y w sn males. Diplo-XY exceptions are indicative of nondisjunction at meiosis I, whereas nullo-XY exceptional progeny could be a consequence of loss or nondisjunction at either MI or MII. Fourth chromosome nondisjunction was monitored simultaneously in these crosses. Diplo-4 exceptional sperm gave rise to spa progeny, and nullo-4 exceptional sperm gave rise to ci ey progeny. These crosses did not allow distinction between MI and MII fourth chromosome segregation errors. The relative frequencies of nullo- vs. diplo-exceptional progeny, however, are an indication of whether errors were largely due to nondisjunction (in which case a 1:1 nullo:diplo ratio is expected) or chromosome loss (in which case the frequency of nullo-exceptions would exceed that of diplo-exceptions).

To directly assess MII nondisjunction, yw /y+Y; dtopors; spa/+ males were crossed to C(1)RM y v/Y; C(4)RM ci ey/0 females. From these crosses, progeny produced from diplo-XY sperm resulting from MI nondisjunction (y+ w sons), diplo-XX sperm resulting from meiosis II nondisjunction (y w daughters), and nullo-XY sperm resulting from either MI or MII nondisjunction (y v daughters) could be distinguished from normal y w sons and y+ v daughters. Simultaneously, we could detect MII fourth chromosome diplo-exceptions as spa progeny, and MI or MII nullo-exceptions as ci ey progeny. For both MI and MII nondisjunction tests, a minimum of 100 males of each genotype were tested.

To assay for autosomal nondisjunction, crosses were performed between dtopors males and females bearing either Compound 2 or Compound 3 chromosomes. In these females, autosomal (A) homologs are attached to a single centromere, resulting in the production of diplo-A and nullo-A eggs. Because aneuploidy for either of the two major autosomes (chromosome 2 or 3) is lethal, progeny of C(A) females mated to chromosomally normal males survive only if produced from aneuploid sperm. We used a metric of “progeny per male” to determine whether dtopors males produced a greater number of viable progeny than their heterozygous siblings, which would indicate an increased incidence of aneuploidy for the major autosomes.

For assays of male fertility of dtopors; mnm allelic combinations, we also determined progeny produced per male. Crosses were performed using single y/y+Y; spa males and five y w sn; C(4)RM ci ey/0 females. Parents were allowed to reproduce for 8 days before removal, and progeny produced were scored until 18 days after the cross was initiated. Sex and fourth chromosome nondisjunction were monitored as above.

PCR and DNA sequencing

Genomic DNA was PCR amplified from Drosophila genomic DNA using the following primers:

• 5′ CTGTTAACATGGCGGAGGAGAATCCC 3′,

• 5′ CCAGATCTATACGGCAGTAGTCCCTGAT 3′,

• 5′ CCAGATCTACGCCTCACAATGTGGTAACG 3′,

• 5′ CCAGATCTGTAACCGTTTATGTCATACGG 3′,

• 5′ CCAGATCTTGCGGCCTCCAGCGAATAGGC 3′,

• 5′ CTGTTAACATGCCCAGGTACACGCCGCTGGTG 3′,

• 5′ CTGTTAACATGGATCATGTGGTGCAGTATTCG 3′,

• 5′ CTGTTAACATGATCGATGTAGTTGGCGAATCA 3′

PCR products were purified using QIAquick columns (Qiagen) and sent for commercial sequencing to Eurofins, MWG Operon.

Indirect immunofluorescence and confocal microscopy

Testes were dissected in Schneider’s insect tissue culture medium (Invitrogen) and gently squashed under silanized coverslips. Coverslips were removed after freezing in liquid nitrogen and tissues fixed 10 min in cold methanol. Primary antibodies were used at the following dilutions in phosphate buffered saline (PBS) + 1.0% bovine serum albumin, fraction V (Fisher Scientific): rabbit anti-GFP 1:1000 (Invitrogen), mouse anti-dTopors, 1:1000 (Secombe and Parkhurst 2004), mouse anti–β-tubulin 1:500 (E7, Iowa Hybridoma Bank), mouse anti–γ-tubulin 1:2000 (GTU-88, Sigma Scientific), mouse anti-lamin Dm0 1:10 (LC28), and mouse antilamin C (ADL67) 1:100 (Iowa Hybridoma Bank), rabbit anti-pH 3 (Sigma 06570) 1:500, anti-centrosomin 1:500 (Li et al. 1998), anti-dTACC 1:2000 (Gergely et al. 2000), anti-Aurora A 1:1000 (Berdnik and Knoblich 2002), and anti-HP1 (CA19) 1:500 (Iowa Hybridoma Bank). Secondary antibodies used were Alexa Fluor 546 goat anti-mouse, 1:1000 (Molecular Probes) and Alexa Fluor 488 goat anti-rabbit (Molecular Probes). Primary incubations were 1 hr at room temperature (rt) followed by 3 × 5 min washes in PBS. Secondary incubations were 45 min at rt followed by 3 × 5 min washes in PBS. Slides were stained by 1-min incubation in 1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI, Sigma) and mounted in 50% glycerol. For each antigen examined, a minimum of 200 MI cells at prophase and 30 MI cells between prometaphase and telophase I from at least five individuals were observed.

Condensation assay

Testis squashes were prepared as above and stained with DAPI and anti–γ-tubulin antibodies. To minimize variations in staining, all slides were processed simultaneously in identical solutions. Twenty S6 cells from at least five different males per genotype were selected on the basis of staining with DAPI and antibodies to γ-tubulin, as above. Images were captured using standardized settings using an Olympus FV500 confocal laser scanning microscope and Fluoview software. The nuclear area of each DAPI image was selected and pixel intensity histograms generated using Adobe photoshop software. Pixels with an intensity value of <10 (on a scale of 0–250) were considered negative and were used to determine the area of each nucleus devoid of chromatin.

Western blotting

Testes were dissected from 1-day-old adults, sonicated in sample buffer (1% SDS, 10% glycerol, 0.01% bromophenol blue, 5% β-mercaptoethanol, 250 mM Tris pH 6.8) plus 2 μm MG132, 0.2 mm IAA, 80 mm NEM, and proteinase inhibitor cocktail P2714 (Sigma) and then separated by SDS–PAGE. After transfer to Immobilon membrane (Millipore), blots were probed with anti-Drosophila SUMO (N terminal), (AP1287a, Abgent), diluted 1:500 in Tris buffered saline, pH 7.4, mouse anti-lamin C (1:100), mouse anti–β-tubulin (1:500) or mouse anti-lamin Dm0 (1:10), and detected using goat antimouse HRP-conjugated secondary antibodies (1:10,000) (Jackson Immunoresearch) with Supersignal West Pico chemiluminescence reagents (Thermo Scientific). Signals were captured and analyzed using a Bio-Rad Chemidoc XRS image acquisition system. Comparisons of antigen abundance were based on the average signal intensities relative to control β-tubulin signals from triplicate blots.

Electron microscopy

Spermatocytes were dissected in Schneider’s tissue culture media (Sigma), centrifuged onto Thermanox plastic coverslips (Nunc), and immediately fixed in 2% glutaraldehyde, 0.1 M cacodylate, 3% sucrose, and 3 mm CaCl. Fixed tissues were stained in 1 μm DAPI and examined using fluorescence microscopy to identify cells at stages of interest. Selected cells were then fixed in reduced osmium (McDonald 1984) and stained en bloc with 2% aqueous uranyl acetate, dehydrated in graded ethanol, and embedded in Eponate-12 (Ted Pella). Thin sections were examined using a Hitachi 7600 TEM at an accelerating voltage of 80 kV. All images were captured with an Advanced Microscope Techniques digital camera (AMT 16000-S).

Results

nuclear blebber corresponds to dtopors

Two mutations in nuclear blebber (nbl) had been isolated in a screen for viable recessive male chromosome loss (mcl) mutations that caused increased incidence of fourth chromosome loss among progeny of mutant males (Wakimoto et al. 2004). The original designations were nbl^Z1837 and nbl^Z4522. In addition to the chromosome loss phenotype, nbl males display a striking nuclear dysmorphology in primary spermatocytes (Figure 2; Wakimoto et al. 2004). We mapped the nbl to the dtopors locus by recombination and deficiency mapping (see Materials and Methods). Henceforth, we will refer to these mutations as dtopors^Z1837 and dtopors^Z4522.

Figure 2.

Abnormal meiosis I in dtopors. Wild-type and dtopors^f05115 meiocytes at the indicated meiotic stages stained with DAPI (red) and (A) anti–γ-tubulin antibodies (green) or (B) anti–β-tubulin antibodies. In A, the stage of wild-type cells is indicated above, and the stage of dtopors cells in shown below each panel. Arrows point to chromatin bridges. Bar, 10 μm. (C) Percentage of meiosis I cells in dtopors^f05115 vs. dtopors^f05115/+ classified by stage (according to Cenci et al. 1994).

The dtopors genes were PCR amplified from flies homozygous for either dtopors^Z4522, dtopors^Z1837, or the wild-type progenitor cn bw second chromosome, and DNA sequences determined. Comparison of wild-type and mutant sequences revealed a single-base-pair difference from the progenitor chromosome for each dtopors allele. Each EMS allele is predicted to encode a dTopors protein bearing a single-amino-acid substitution. In dtopors^Z4522, this substitution lies in an 84-amino-acid region of high conservation (41% identity, 56% similarity) between mammalian Topors homologs, but for which no particular function has been ascribed (Figure 1). The substitution in the dtopors^Z1837 allele is 124 amino acids closer to the N terminus, in the same general region. Neither mutation affects a residue conserved in mammals, but the Z4522 mutation alters a residue conserved among Drosophilids. Overall, human Topors and dTopors share only limited homology outside of the RING domain and this region. For the full-length proteins, only 122/1038 of the fly amino acids are identical in the human homolog.

Figure 1.

Structure of dTopors. Positions of EMS-induced mutations and a piggyBac insertion in dtopors are shown on a diagram of the dTopors protein (modified from Secombe and Parkhurst 2004). Numbers below each mutation indicate positions of the substituted base pair in the gene and the substituted amino acid in the protein. Conceptual translation of the piggyBac insertion allele predicts a protein truncated after K47. PEST sequences involved in protein degradation are indicated by solid bars, a RING domain essential for ubiquitin and/or SUMO ligase function is indicated by gray shading, an arginine/serine-rich region (RS) is indicated by diagonal hatch marking and a region of four consensus bipartite nuclear localization domains by underlining. A region of high conservation between mammalian and fly homologs is indicated by horizontal striping, corresponding to amino acids 281–364 in dTopors. The sequence of this region is shown below with the position of the dtopors^Z4522 mutation indicated.

Mutations in dtopors disrupt segregation of all chromosomes at MI in males

As the original mcl screen monitored only fourth chromosome loss from mutant males, we performed genetic tests to determine (1) whether dtopors mutations affect the meiotic transmission of other chromosomes, (2) at which meiotic division defects in chromosome segregation occur, and (3) whether the effects are male specific.

In addition to the EMS-induced alleles, we made use of an existing deletion, Df(dtopors^AA) (Secombe and Parkhurst 2004), and a piggyBac insertion allele, dtopors^f05115. Conceptual translation of the piggyBac insertion allele sequence predicts that it encode a protein truncated after amino acid 47 and therefore is likely a null allele. Homozygous dtopors males bearing each possible combination of the four alleles showed near random segregation of both sex and fourth chromosomes (Table 1). The frequencies of progeny that resulted from simultaneous sex and fourth nondisjunction were similar to predictions based on independence, suggesting that the two sets of chromosomes were not influencing each other’s behavior. Consistent with a severe chromosome segregation defect, the fecundity of dtopors males was greatly reduced, producing on average fewer than five progeny per male. We were unable to assay nondisjunction in dtopors^Z1837 homozygotes, as these males were essentially sterile, producing only two progeny from over 500 matings. This allele also had a weak dominant effect, as we observed low frequencies of exceptional progeny from heterozygous dtopors^Z1837/+ fathers. All other alleles were recessive. As predicted, dtopors^f05115 was similar to that of the deletion, producing random segregation of both sex and fourth chromosome homologs, suggesting that it is a null allele. The dtopors^Z4522 homozygotes produced significantly fewer exceptional progeny than Df(dtopors^AA) homozygotes, and frequency of exceptional progeny from transheterozygous dtopors^Z4522/Df(dtopors^AA) was intermediate. These data indicate that the dtopors^Z4522 allele is hypomorphic. As in null mutants, sex and fourth chromosome segregation were equally perturbed in dtopors^Z4522 males, suggesting that the two pairs of chromosomes are equally sensitive to the dtopors defect.

Table 1. Sex and fourth chromosome disjunctional data from crosses of y/y⁺Y; dtopors; spa^pol males to y w sn; C(4)RM ci ey/0 females.

Recovered male gametes:	Y; 4	X; 4	0; 4	X/Y; 4	X; 0	Y; 0	X; 4/4	Y; 4/4	0; 0	0; 4/4	X/Y; 0	X/Y; 4/4
Paternal genotype
dtoporsZ1837/SM1, Cy	185	299	5	6	6	3	5	2	3	0	0	0
dtoporsZ4522	216	273	229	52	52	53	63	63	9	5	27	21
dtoporsZ4522/SM1, Cy	503	979	0	1	0	1	0	0	0	0	0	0
dtoporsf05115	33	50	48	31	10	4	11	8	24	5	4	17
dtoporsf05115/SM1, Cy	76	210	2	1	0	0	1	0	0	0	0	0
Df(dtoporsAA)	16	15	12	12	4	7	2	10	15	8	2	2
Df(dtoporsAA)/SM1, Cy	206	342	0	0	0	0	0	2	0	0	0	0
dtoporsZ1837/Df(dtoporsAA)	48	61	75	32	22	9	27	16	40	18	3	20
dtoporsZ4522/Df(dtoporsAA)	48	67	32	25	8	5	13	12	8	9	6	10
dtoporsf05115/Df(dtoporsAA)	28	35	53	20	11	3	8	7	27	6	6	7
dtoporsZ1837/dtoporsZ4522	226	270	324	153	60	24	46	46	65	46	21	37
dtoporsZ1837/dtoporsf05115	26	32	46	33	15	16	15	12	14	16	3	19
dtoporsZ4522/dtoporsf05115	61	77	67	30	13	8	9	9	9	11	4	9

% Nondisjunction	Nullo XY	Diplo XY	Nullo 4	Diplo 4	XY	4	XY + 4a
dtoporsZ1837/SM1, Cy	1.9	1.2	2.3	1.4	3.1	3.6	0 (0)
dtoporsZ4522	22.9	9.4	13.3	14.3	32.3	27.6	5.8 (8.9)
dtoporsf05115	31.4	21.2	17.1	16.7	52.6	33.8	20.4 (15.9)
Df(dtoporsAA)	33.3	15.2	26.7	21.0	48.5	47.7	25.7 (23.1)
dtoporsZ1837/Df(dtoporsAA)	35.8	14.8	19.9	21.8	50.6	41.7	21.8 (21.1)
dtoporsZ4522/Df(dtoporsAA)	20.2	16.9	11.1	18.1	37.1	39.2	13.6 (10.8)
dtoporsf05115/Df(dtoporsAA)	40.8	15.6	22.3	13.3	56.4	35.6	21.8 (20.0)
dtoporsZ1837/dtoporsZ4522	33.0	16.0	12.9	13.3	49.0	26.2	12.8 (12.8)
dtoporsZ1837/dtoporsf05115	30.8	22.3	19.4	25.1	53.1	44.5	21.1 (23.6)
dtoporsZ4522/dtoporsf05115	28.3	14.0	11.1	12.4	32.3	23.5	10.7 (7.6)

^aFrequencies of simultaneous sex and fourth chromosome nondisjunction. Observed and (expected on the basis of independence).

The frequencies of chromosome 4 nullo- and diplo-exceptions among progeny of dtopors males were roughly equal, consistent with nondisjunction rather than loss as the primary consequence of the dtopors defect. Among sex chromosome exceptions, nullo-exceptions significantly outnumbered diplo-exceptions. This likely reflects meiotic drive, which favors the recovery of sperm with lower chromatin content (Peacock et al. 1975), rather than chromosome loss.

We also tested for meiosis II nondisjunction, using both null Df(dtopors^AA) and hypomorphic dtopors^Z4522 males (see Materials and Methods). While we observed frequencies of MI exceptions comparable to previous tests (Table 1), we observed no evidence for MII nondisjunction (Table 2). This suggests that the effects of dtopors mutations are specific to meiosis I.

Table 2. Sex and fourth chromosome disjunctional data from crosses of yw/y⁺Y; dtopors; spa^pol/+ males to C(1)RM y v/Y; C(4)RM ci ey/0 females.

Recovered male gametes:	Y; 4	X; 4	0; 4	X/Y; 4	Y; 0	Y; 4/4	X; 0	X; 4/4	0; 0	X/X; 4	XX; 0	XX; 4/4
Paternal genotype
dtoporsz4522	66	26	29	7	10	0	2	0	2	0	0	0
dtoporsz4522/SM1, Cy	517	194	3	0	2	0	0	1	0	0	0	0
Df(dtoporsAA)	58	31	34	21	17	0	12	0	20	0	0	0
dtoporsAA/SM1, Cy	336	172	0	0	1	0	0	0	0	0	0	0

% Nondisjunction
Sperm genotype	Nullo XY	Diplo XY	Diplo XX	Nullo 4	Diplo 4
Errant division	MI or MII	MI	MII	MI or MII	MII
dtoporsz4522	21.8	4.9	0	9.9	0
Df(dtoporsAA)	27.8	11.3	0	25.8	0

To ask whether transmission of the major autosomes was also affected in dtopors males, we crossed mutant males and their heterozygous siblings to females bearing compound autosomes [C(A)] and counted the viable progeny per male produced as an indication of autosomal nondisjunction (see Materials and Methods). From crosses involving either C(2) or C(3) females, we observed significantly more progeny per male from dtopors vs. dtopors/+ males (Table 3, P < 0.05). Paternal nullo-A and diplo-A exceptional progeny were observed for both chromosomes 2 and 3, indicating that dtopors causes nondisjunction of both major autosomes. As for the sex chromosomes, an excess of autosomal nullo-exceptions was observed, likely as a consequence of meiotic drive, which also occurs as a result of autosomal missegregation (Dernburg et al. 1996).

Table 3. Results of crosses of 50 dtopors males to 100 C(2)EN, bw sp or C(3)EN, st cu e females.

Maternal genotype	C(2)EN,	bw sp	F1/male	C(3)EN,	st cu e	F1/male
Sperm genotype	Diplo-2	Nullo-2		Diplo-3	Nullo-3
dtoporsZ4522	49	230	5.58	54	108	3.24
dtoporsZ4522/SM1, Cy	0	4	0.08	4	0	0.08

In contrast to males, we detected no increase in chromosome transmission errors in dtopors females. We crossed females of each dtopors allelic combination to In(1)EN, X.Ys, y f/0; C(4) ci ey/0 males and scored at least 1000 progeny from each cross (data not shown). In each case, we observed <0.5% sex chromosome exceptions and <1% fourth chromosome exceptions, and the frequencies of each did not significantly differ from those observed among progeny of control dtopors/+ mothers. Nor did we observe any significant reductions in female fertility. We conclude that the effects of dtopors on meiotic chromosome transmission are limited to meiosis I in males.

dtopors causes precocious centriole separation and anaphase I bridges

To determine the cause of the dtopors meiosis I chromosome segregation defects, we compared the progression of meiosis I in wild type and dtopors mutant spermatocytes. To assess spermatocyte stage, chromosome morphology and microtubule organization were monitored in fixed spermatocytes using the DNA-specific dye DAPI and antibodies to γ- and β-tubulin, as described by Cenci et al. (1994). Males homozygous for each of the four dtopors alleles were examined, and similar phenotypes were observed for each allele. The dtopors^f05115 allele was characterized most extensively, as this allele was expected to be a null, based on the predicted protein product (Figure 1).

In wild-type meiosis, the centriolar pair is duplicated in prophase of MI, and segregates as a perpendicular pair to opposite poles to establish the MI spindle. When stained with anti–γ-tubulin antibodies, it appears as a “V” shape. At telophase of MI, each centriolar pair separates without duplication, and the single centrioles migrate to opposite poles to establish two parallel MII spindles (Li et al. 1998).

Staining with anti–γ-tubulin antibodies revealed an abnormal separation of centrioles at meiosis I in dtopors^f05115 males (Figure 2A). This defect occurred rarely at late prophase S6 (2/501), but was seen much more frequently at prometaphase (9/32). Of 43 cells examined at meiosis I metaphase through telophase, all were abnormal with respect to γ-tubulin staining. In 22 or these, centrioles were slightly separated at each pole, giving the appearance of telophase I. Examination of spindles and chromosomes, however, suggested these cells were at metaphase to early anaphase. In the remaining cells, the chromosomes also appeared to be at metaphase or early anaphase I, yet the centrioles had an MII configuration, resulting in tetrapolar spindles. In 30/31 cells at mid to late anaphase I, chromatin bridges were observed connecting the separating masses of chromatin (Figure 2, A and B). Because our genetic data indicated MI homolog nondisjunction, and we failed to observe abnormal connections between nonhomologs at prometaphase (e.g., Figure 2A), we conclude that these bridges likely result from the failure to separate homologs at anaphase I.

To verify that the extra γ-tubulin foci resulted from centriole separation rather than an additional round of duplication, we examined EM serial sections of 10 MI dtopors^f05115spermatocytes. None of 10 cells contained more than four centriole halves, and centriole morphology was normal. Antibody staining of centrosomal antigens centrosomin, dTACC, and Aurora A showed wild-type patterns of associations of these antigens with each γ-tubulin focus (data not shown).

We also noted a subtle defect in chromatin condensation in dtopors spermatocytes. DAPI staining appeared normal from early–mid prophase stages S1–S4, but at late prophase stages S5 and S6 the chromosomes appeared slightly less condensed than normal. This mild condensation defect was apparent until stage M1a (Figure 2A). Pairing appeared grossly normal, as paired homologs occupied territories adjacent to the nuclear periphery at this stage. No differences between dtopors and wildtype were detected in antibody staining patterns for HP1 (n = 133 and 136, respectively) or S10-phosphorylated H3 (n = 179 and 145, respectively).

To ask whether progression through meiosis was altered in dtopors, we assessed the frequency of MI stages in dtopors^f05115 vs. dtopors^f05115/+ siblings (n = 435 and n = 467, respectively, Figure 2C), on the basis of DAPI staining and phase contrast imaging of spindle morphology. A greater proportion of cells was observed at prometaphase (M1a, M1b, and M2) in dtopors^f05115 males, suggesting a delay is associated with the defect in condensation. In addition, a greater proportion of cells was observed in anaphase in the mutant, consistent with a delay in anaphase progression owing to bridge formation. These observations raise the possibility that the centriole separation defect may be a consequence of continued progression of the centriole cycle concomitant with a block or delay in the chromosome cycle. We do not know, however, whether centriole defects are limited to cells that are delayed in meiosis. Determination of whether dTopors directly or indirectly affects the centriole cycle will require real-time observations of chromosome and centriole behavior at meiosis.

In summary, our observations suggest that there are two main defects leading to chromosome missegregation: an asynchrony between the centrosome and chromosome cycles, leading to tetrapolar MI spindles, and inappropriate or unresolved connections between homologs at anaphase I.

Epistasis tests between dtopors and mod(mdg4)

In somatic cells, dTopors has been demonstrated to physically interact with components of gypsy chromatin insulators, including the chromatin protein Modifier of mdg4 [Mod(mdg4)] (Capelson and Corces 2005). Sumoylation of one isoform, Mod(mdg4)2.2, has been shown to be negatively regulated by dTopors, which correlates with enhanced insulator function and formation of insulator bodies (Capelson and Corces 2006). This suggested a model in which interactions between insulators could be modulated by the activity of dTopors, leading to the formation of insulator bodies.

The mod(mdg4) locus is complex, producing 31 protein isoforms, some of which are produced via trans-splicing (Dorn et al. 1993). One of these isoforms, named Mod(mdg4) in meiosis (MNM), is required for male meiotic chromosome segregation. Mutations in mnm precociously disrupt conjunction between paired homologs at MI in males, leading to homolog nondisjunction. MNM has been proposed to form a pairing complex with two other proteins, Stromalin in meiosis (Thomas et al. 2005) and Teflon (Arya et al. 2006), to mediate bivalent stability. All Mod(mdg4) isoforms share a BTB protein interaction domain at their carboxyl termini and may form homodimers through self-interactions of this domain (Dorn et al. 1993). By analogy to its role at insulators, we wondered whether dTopors might also regulate germline Mod(mdg4). Specifically, the lack of dTopors might alter the MNM-dependent pairing complex formation and/or stability. The anaphase I bridges and nondisjunction in dtopors mutants might be a consequence of prolonged associations between homologs resulting from misregulation of MNM-dependent conjunction. To test this, we examined meiosis in males singly or doubly mutant for null alleles of dtopors (dtopors ^f05115) and mnm (mnm^Z5578).

At anaphase/telophase I, chromosome segregation appeared normal in dtopors ^f05115/+; mnm^Z5578/+ (n = 30). In dtopors ^f05115/+; mnm^Z5578, individual lagging chromosomes were observed, as was reported for mnm mutants (Thomas et al. 2005), but no anaphase I bridges were seen (n = 34). In both dtopors ^f05115; mnm^Z5578/+ and dtopors ^f05115; mnm^Z5578 males, nearly all anaphase/telophase figures had bridges (52/56 and 19/21, respectively). This demonstrates that the connections between chromosomes that lead to anaphase bridge formation in dtopors mutants are not dependent on MNM.

Homozygous dtopors males are nearly sterile, and we used this near sterility as a second assay for interactions between mnm and dtopors. Although flies homozygous for mnm exhibit nearly random segregation of autosomes, they are significantly more fertile than dtopors flies. We predicted that if the dtopors-induced anaphase bridges were dependent on MNM that we might see an increased fertility of dtopors; mnm doubly mutant flies relative to dtopors; + flies. No increase in fertility was observed, nor did frequencies of nondisjunction of sex or fourth chromosomes indicate an interaction (Table 4). Consistent with our cytological observations, these results suggest that the dtopors anaphase bridges are not dependent on MNM.

Table 4. Results of y/y+Y; spa males of the indicated dtopors and mnm genotypes crossed to y w sn; C(4)RM ci ey/0 females.

Paternal genotype	Males tested	Progeny per male	% XY ND	% 4 ND
dtoporsf05115/+; +	59	15.0	0.3	0.1
dtoporsf05115/+; mnmZ5578/+	39	16.7	0.2	0.1
dtoporsf05115/+; mnmZ5578	42	4.8	39.9	49.3
dtoporsf05115; +	135	1.6	46.8	37.0
dtoporsf05115; mnmZ5578/+	63	2.2	40.4	37.5
dtoporsf05115; mnmZ5578	59	0.9	38.9	48.1

We also compared chromosom2e morphology at late prophase in singly and doubly mutant males (Figure 3). Both chromosome condensation and homolog associations appeared normal at S6 in dtopors^{f 05115}/+; mnm^Z5578/+ males. In dtopors^f05115/+; mnm^Z5578 males, the previously described mnm phenotype of precocious separation of homologs was observed (Thomas et al. 2005) and chromosomes were noticeably more diffuse. A mild condensation defect was also apparent in dtopors ^f05115; mnm^Z5578/+ cells. In doubly mutant males, the condensation defect was more severe than either single mutant, suggesting that the effects of the two mutants were additive and therefore that the genes act in distinct pathways. To quantify these differences, we compared the distribution of chromatin within nuclei of 20 S6 stage spermatocytes of each genotype, matched as closely as possible on the basis of centriole morphology and position as revealed by anti–γ-tubulin staining. Confocal images of DAPI-stained nuclei were analyzed to determine the percentage of pixels below an arbitrary low intensity level (i.e., black pixels), as an estimate of nuclear space between chromosomes, which was expected to reflect the degree of condensation. Values obtained were consistent with our qualitative impressions (mean percentage ± SD: dtopors/+; mnm/+ 16.7 ± 1.5, dtopors; mnm/+ 13.6 ± 3.0, dtopors/+; mnm 7.7 ± 3.9 and dtopors; mnm 1.7 ± 0.8; P < 0.05 for each pairwise comparison).

Figure 3.

Chromosome morphology in DAPI-stained dtopors and mnm single and double mutants. The dtopors genotypes are indicated to the left of each row; mnm genotypes are indicated above each column. The top four panels depict S6 spermatocytes, staged by anti–γ-tubulin staining showing paired centrioles adjacent to the nucleus (arrows). The lower panels depict meiocytes at late anaphase I/telophase I. Bar, 10 μm.

Germline nuclear lamina assembly or stability is disrupted in dtopors mutants

As dTopors physically interacts with lamin Dm0 (Capelson and Corces 2005), we wondered whether the nuclear blebbing phenotype might reflect a defect in the nuclear lamina. To test this, we used EM to examine serially sectioned whole-mounted testes from dtopors and wild-type males (Figure 4). We selected S5–S6-stage cells on the basis of nucleolar morphology, which breaks down at S6, position of the centrioles, which complete their migration from a position adjacent to the cell membrane to a perinuclear location by S5, and the presence of condensed chromatin. In nine wild-type cells examined (5 S5 and 4 S6), spermatocyte nuclei were not perfectly round as in flattened preparations, but had a generally round shape, with only minor involutions and protrusions of the nuclear membrane. The perinuclear space between the inner and outer nuclear membranes was very regular around the entire nucleus in all cells. In contrast, in 11 S5 and 13 S6 Df(dtopors^AA) homozygous spermatocytes examined, nuclear membranes were highly involuted in each cell. In cross-section, these involutions often appeared as finger-like projections (Figure 4). These protrusions may represent “weak” spots in the nuclear structure that result in the blebs observed in flattened live preparations. In addition, in each of these cells the perinuclear space was irregular. The distance between the inner and outer nuclear membrane varied as much as 10-fold at different positions (Figure 4, inset). These results suggest that interaction between dTopors and Lamin Dm0, or lack thereof, has structural consequences on nuclear lamina assembly in the germ line.

Figure 4.

Spermatocyte ultrastructure. Electron micrographs of sections of S6 spermatocytes from wild-type (A) and dtopors (B) males. Note the accumulation of excess membrane surrounding the nucleus in the mutant. Enlargements of the nuclear membrane of wild type (C) and dtopors (D and E), showing an irregular perinuclear space (arrows) in the mutant. Bars, 1 μm.

In >100 dtopors testes examined, we only observed blebs in late-stage spermatocytes (S4–S6), and never in gonial cells. The frequency of blebs was highly variable and dependent of the degree of squashing and on genetic background. Among S4–S6 cells from dtopors^Z1837, dtopors^Z4522, dtopor^f05115, and Df(dtopors^AA) males, we observed 165/413, 243/760, 140/480, and 153/596 cells with blebs, respectively. This suggested that defects were limited to meiotic prophase. Alternatively, as prophase spermatocytes have a greater nuclear volume than gonial cells, and are much more abundant, our observations may have reflected either a greater susceptibility of spermatocytes to blebbing or merely an ascertainment error. In a prior examination of dtopors^Z1837 and dtopors^Z4522 spermatocytes, Wakimoto et al. (2004) reported the presence of micronuclei (i.e., blebs) in gonial cells as well. To examine this more closely, we looked at testes of males doubly mutant for dtopors and benign gonial cell neoplasm (bgcn). In bgcn males, gonial cells overproliferate but do not differentiate into spermatocytes (Gonczy et al. 1997). Nuclear blebs were observed in testes of these males, indicating that the defect is not specific to the spermatocyte stage (data not shown). We also examined somatic tissues in dtopors males, including testis sheath cells, larval neuroblasts, and the giant nuclei of the larval salivary gland cells, but failed to detect a similar nuclear defect. This suggests that dtopors may have a unique role in germ line nuclear structure.

As dTopors localizes to the nuclear lamina in both Drosophila diploid cells and polytene salivary gland cells (Capelson and Corces 2005), we asked whether we could also detect it at the nuclear lamina in spermatocytes. Using confocal microscopy and indirect immunofluorescence with anti-dTopors antibodies (kindly provided by S. Parkhurst, Secombe and Parkhurst 2004), we found that dTopors is indeed associated with the nuclear lamina in primary spermatocytes (Figure 5). Although the signal at nuclear periphery was weak, this localization was detectable as early as S1, but most easily seen in late stage (S6) spermatocytes. In addition to this lamina staining, we observed a stronger dTopors signal localized to punctate nuclear spots. These first appear at S3, become largely chromosome associated by S6, and disappear at prometaphase I. Using confocal microscopy, we counted the spots in 0.5-μm optical sections through entire nuclei for 50 S6 primary spermatocytes and found that they ranged in number from 6 to 25, with a mean of 15.6 ± 3.3 spots per cell. To verify that these staining patterns were dTopors specific, we identically stained mutant dtopors testes (Figure 5, bottom). Neither the lamina localization nor the nuclear spots were detected.

Figure 5.

Localization of dTopors in primary spermatocytes. Wild-type primary spermatocytes stained with DAPI (red) and anti-dTopors antibodies (green). Stages of development according to Cenci et al. (1994) are indicated at left. Arrows point to staining at the nuclear periphery. The bottom row shows homozygous dtopors^AA mutant S4–S5-stage primary spermatocytes stained with DAPI (red) and anti-dTopors antibodies (green). Bar, 10 μm.

To determine whether dtopors mutations disrupted the nuclear lamina by altering assembly of lamina components, we compared localizations of the two Drosophila lamins, lamin Dm0 and lamin C, in wild-type and dtopors spermatocytes. Both localized uniformly to the nuclear periphery in wild-type cells. In dtopors cells, both lamins localized to the nuclear lamina as well, even in regions included in the blebs. In addition, however, prominent lamin Dm0 and lamin C staining was observed in the internal regions of the nucleus. The lamin C staining appeared in large intranuclear foci, whereas the lamin Dm0 staining showed a finer-grained distribution. A similar distribution of lamins was observed for each allele (Figure 6). These results suggest that dtopors is required for the normal assembly of both lamin Dm0 and lamin C in the male germ line.

Figure 6.

Lamin localizations in primary spermatocytes. (A) Wild-type and Df(dtopors^AA) S5-stage primary spermatocytes stained with DAPI (red) and antilamin C or antilamin Dm0 antibodies (green). Bar, 10 μm. (B) Representative Western blot of whole testis extracts from flies of the indicated dtopors genotypes, probed with antilamin C, antilamin Dm0, and anti–β-tubulin antibodies.

In somatic tissue, dTopors has E3 ubiquitin ligase activity (Secombe and Parkhurst 2004) and also influences the sumoylation of protein components of gypsy insulator complexes (Capelson and Corces 2005). This raised the possibility that the accumulation of lamins in the nuclear interior of spermatocytes might result from either an increased abundance resulting from decreased ubiquitin-mediated degradation or improper targeting due to changes in sumoylation. To examine this, we performed Western blotting using antilamin antibodies on wild-type and dtopors testis proteins that were isolated in the presence of SUMO isopeptidase inhibitors and proteasome inhibitors. Blots were probed with antibodies to β-tubulin to control for loading. Additional control blots were probed with anti-SUMO and antiubiquitin antibodies to verify the activity of the inhibitors (data not shown). Blots were performed in triplicate and average signal intensities compared. A representative blot is shown in Figure 6B. No changes in the abundance or migration of lamin Dm0 or lamin C relative to β-tubulin were detected in dtopors vs. wild-type extracts. This suggests that dTopors may play a structural role in lamin assembly or stability at the spermatocyte nuclear lamina, rather than regulating the turnover or abundance of lamins.

New dtopors alleles affect gypsy insulators function

The two EMS-induced mutations in dtopors alter amino acids in the same general region of the protein, and we wondered whether these changes might affect meiosis-specific functions or whether they might also affect gypsy chromatin insulators. In genetic assays using gypsy-induced mutations, a heterozygous deletion of dtopors acts dominantly to disrupt the ability of gypsy insulators to block interactions between regulatory elements and gene promoters. Reduction of dTopors by in vivo RNAi expression has a similar effect (Capelson and Corces 2005). We generated males heterozygous or homozygous for each of the EMS dtopors alleles and the piggyBac insertion allele and assayed the phenotype of a gypsy-induced mutation in the cut gene (ct⁶). This mutation is caused by the insertion of gypsy between a wing margin enhancer and the promoter of the cut gene (Gause et al. 2001). When a functional gypsy insulator is established, it prevents proper expression of ct, resulting in a wing with a notched edge.

We found a similar dominant effect in tests of each dtopors allele, resulting in a disruption of insulator function in dtopors/+ flies (Figure 7). Surprisingly, however, this effect was not observed in flies homozygous for any of the dtopors alleles. Rather, insulator function appeared normal in dtopors homozygotes, and wing phenotypes were indistinguishable from those of dtopors/+ flies. As controls for insulator-independent effects of dtopors on ct expression, we examined both strong and weak hypomorphic alleles of ct, ctⁿ, and ct^K, respectively. We observed no effect of dtopors^Z1837/+ or dtopors^Z4522/+ on ctⁿ and only a very slight suppression of the posterior notching for the ct^K allele. We conclude that dtopors regulates the activity of gypsy insulators in a dose-dependent manner, but this regulation is not essential for insulator function. Additionally, the amino acid substitutions in the EMS alleles affect dTopors function both in meiosis and at insulators, suggesting that the mutated amino acids may be part of a domain required for both germ line and somatic activities.

Figure 7.

Modification of ct⁶ phenotypes by dtopors mutations. Wing phenotypes of flies either heterozygous or homozygous for the indicated dtopors alleles and either the gypsy-induced ct⁶ mutation or control ctⁿ or ct^K hypomorphic point mutations.

Discussion

Our results indicate that dTopors is required in the Drosophila male germ line for proper nuclear lamina and envelope formation, centriole behavior, and chromosome segregation at meiosis I. We found no essential function of dTopors in mitotic cells or in female meiosis, as homozygous dtopors null flies are viable and females are fertile and do not exhibit increased errors in meiotic chromosome segregation. This contrasts with findings in mouse in which the absence of Topors leads to perinatal lethality and increased aneuploidy in mouse embryonic fibroblasts (MEFs) (Marshall et al. 2010) and in zebrafish, in which translational knockdown of Topors using morpholinos indicates an essential role in development (Chakarova et al. 2011). It is unknown whether these different requirements for Topors in flies and vertebrates reflects functional redundancy for dtopors in flies, additional targets of Topors modification in vertebrates, or both.

dTopors and the nuclear lamina

The dtopors phenotype may be a consequence of improper ubiquitination or sumoylation of multiple targets, and it is unclear whether the pleiotropic meiotic defects result from alterations of multiple pathways. In light of the male specificity, however, we favor the hypothesis that the primary defect may be disruption of interactions between meiotic chromatin and the nuclear lamina. Drosophila males have evolved an unconventional system of meiotic homolog conjunction, which may depend on chromatin–lamina interactions. As chromosomes condense during prophase I, bivalents are spatially separated from one another and are closely apposed to the nuclear lamina and retain this association until prometaphase (Cenci et al. 1994). During prophase I, associations between homologous loci change, as monitored by LacI–GFP fusion protein labeling of integrated lacO arrays. Homologous lacO loci are tightly paired and appear as a single spot early in prophase, but by midprophase they separate into distinct spots (Vazquez et al. 2002). These observations indicate that interactions between homologs are refined or redistributed while chromosomes occupy lamina-associated domains. In dtopors males, an altered lamina may disrupt prophase progression and result in the establishment or persistence of inappropriate interactions between homologs, interfering with their subsequent separation at anaphase. Thus, the observed condensation defect in prophase, anaphase I bridge formation, and meiosis I-specific nondisjunction could all be secondary consequences of an abnormal interaction between chromatin and the nuclear lamina.

A number of observations support this notion. First, dTopors is localized to the nuclear lamina in spermatocytes. Although lamina is still formed in dtopors spermatocytes, our observations of abnormal nucleoplasmic foci of both lamin Dm0 and lamin C indicate that dTopors regulates or facilitates some aspect of lamin assembly. This is unlikely to occur via regulation of lamin turnover, as we found no differences in abundance of lamins in testis. dTopors also physically interacts with lamin Dm0 in somatic cells, and this interaction is functionally significant, as dtopors mutation or knockdown disrupts gypsy chromatin insulators (Capelson and Corces 2005).

Second, there is precedence for chromosome condensation and segregation defects resulting from disruption of chromatin–lamina interactions in other organisms. Immunodepletion of lamins in mammalian cell lines alters chromosome condensation in dividing cells (Burke and Gerace 1986; Ulitzur and Gruenbaum 1989; Dabauvalle et al. 1991; Ulitzur et al. 1992) and RNAi knockdown of lamins in C. elegans results in improper mitotic chromosome movement (Liu et al. 2000). Expression of unprocessed or truncated lamin A disrupts both lamina structure and causes aneuploidy in MEFs (Liu et al. 2005). Expression of mutant forms of human lamin B receptor, a protein that physically connects chromatin to lamins via heterochromatin protein 1 (HP1), causes a phenotype similar to that seen in dtopors spermatocytes in that it results in expansion of the perinuclear space and an accumulation of excess endoplasmic reticulum (Zwerger et al. 2010). In Topors −/− MEFs, HP1α shows a diffuse nuclear distribution rather than its normal pericentric localization (Marshall et al. 2010), raising the possibility that Topors may affect chromatin–lamina links indirectly through alterations of HP1α distribution. We failed to observe any similar alterations in HP1 distribution in dtopors spermatocytes by IIF; however, we cannot rule out the possibility that dtopors may cause subtle changes in heterochromatin that impact chromatin–lamina interactions.

Third, indirect evidence also suggests that most or all aspects of the dtopors meiotic phenotype may result from a single primary defect. The dtopors phenotype is strikingly similar to that described for a no-longer extant mutation, ms(1)244 (Lifschytz and Meyer 1977), which caused a similar spermatocyte nuclear dysmorphology, multipolar MI spindles, and chromosome segregation defects. These similarities suggest either that ms(1)244 and dtopors affect a similar spectrum of downstream targets or more parsimoniously, that they affect the same single target or pathway. We failed to identify any X-linked genes with extensive homology to dtopors in homology searches, making it unlikely that ms(1)244 encodes a protein with similar biochemical properties.

An alternative, but not mutually exclusive possibility, is that some aspects of the meiotic dtopors defects result from perturbation of the nuclear envelope and a resulting disruption of protein complexes that span the inner and outer nuclear membranes. These LINC complex proteins, including SUN and KASH domain proteins, are classes of integral membrane proteins that interact in the perinuclear space and have been proposed to form a communication bridge between the cytoskeleton and nucleoplasm, which is important for coordinating cell cycle events (Padmakumar et al. 2005; Crisp et al. 2006; McGee et al. 2006). In C. elegans, meiotic pairing is facilitated by interactions between chromosome-specific zinc finger proteins and the SUN and KASH domain proteins SUN-1 and ZYG-12, and disruption of these protein complexes has been demonstrated to affect meiotic chromosome behavior (Sato et al. 2009).

The precocious separation of centrioles we observed at meiosis I in dtopors spermatocytes may also be related to disruption of LINC complexes. In Drosophila, human fibroblasts, and C. elegans, altered interactions between LINC complex proteins in the nuclear membrane have consequences on both centrosome attachment (Patterson et al. 2004; Salpingidou et al. 2007; Starr 2009) and chromosome movement (Starr 2009). Alternatively, dTopors may have a more specific function at meiotic centrosomes. It has recently been shown that Topors localizes to basal bodies in ciliated mammalian retinal cells and to centrioles in retinal cell lines (Chakarova et al. 2011). Although we failed to detect dTopors at centrosomes, we cannot rule out the possibility that the amount of centrosome-associated protein is below our limits of detection or that epitopes recognized by our antibodies are masked at centrosomes.

Intranuclear dTopors foci

In addition to a lamina localization, we also observed dTopors localization to intranuclear foci. The functional significance of this is unknown. Although in mitotic and salivary gland cells, dTopors localizes to Mod(mdg4)-containing insulator bodies, the meiotic dTopors foci are unlikely to be insulator bodies. When an antibody to the common BTB domain shared by all Mod(mdg4) isoforms is used to stain late prophase spermatocytes, the only staining observed is associated with the nucleolus, the site of the XY pairing sites (Soltani-Bejnood et al. 2007). Antibodies to the insulator protein CP190 also fail to recognize intranuclear foci in late prophase spermatocytes (data not shown). As mammalian Topors is localized to promyelocytic leukemia (PML) bodies (Rasheed et al. 2002; Weger et al. 2003), and the dTopors foci we observed are similar in number to PML bodies in mammalian cells; they could perhaps represent a functional analog of these structures. The Drosophila SUMO homolog, Smt3, forms similar intranuclear foci when expressed in either fly or human cells, suggesting that structures analogous to PML bodies may form in Drosophila (Lehembre et al. 2000). Mammalian PML bodies are dynamic, chromatin-associated structures, and our observations of a redistribution of dTopors foci to condensed chromosomes at late prophase is consistent with the process of condensation of chromatin-associated foci.

Regulation of homolog separation

Both our genetic and cytological observations suggest that dTopors function is important for homolog separation. We failed to observe any gross alterations in homolog pairing, as homolog domains were clearly established in late prophase nuclei. The chromatin bridges that we observed in nearly all dtopors anaphase and telophase I cells therefore likely represent unresolved attachments between homologs. We suggest that in male flies, interactions between the chromosomes and the nuclear lamina may be critical for prophase I chromatin condensation. Our data, however, do not eliminate the possibility that such interactions are also important for homolog pairing as in other organisms, as pairing in males may precede the requirement for dtopors.

To further investigate the role of dTopors in homolog separation, we performed tests of epistasis with a null mutation in the mod(mdg4) isoform mnm. Homolog conjunction in male flies depends on three proteins, MNM, stromalin-in-meiosis (SNM) (Thomas et al. 2005) and Tef (Arya et al. 2006), that have been proposed to interact in a conjunction complex (Thomas et al. 2005). There were two suggestions that the dtopors-induced anaphase bridges might be dependent on the activity of this conjunction complex. First, dTopors physically interacts with another Mod(mdg4) isoform, 2.2, at gypsy chromatin insulators (Capelson and Corces 2005). MNM and Mod(mdg4)2.2 share a carboxyl terminal BTB protein interaction domain (Dorn et al. 1993), and thus it seemed possible that a similar interaction might also occur between dTopors and MNM in meiosis. Second, mutations in the condensin II subunit Cap-H2 result in similar anaphase bridges and condensation defects in male meiosis. The frequency of these anaphase bridges is significantly reduced in tef Cap-H2 double mutants, suggesting that their formation is dependent on the conjunction complex (Hartl et al. 2008). Condensin complex subunits in yeast have been identified as sumoylation substrates in proteomic screens (Wohlschlegel et al. 2004; Denison et al. 2005), and although it is not known whether condensins are targets of dtopors-directed modification, the similarities between the dtopors and Cap-H2 phenotypes suggested a possible relationship.

Our results did not, however, support a direct regulation of the MNM-dependent homolog conjunction complex by dTopors or a requirement for the conjunction complex for bridge formation. Removal of MNM did not abolish anaphase I bridge formation or restore fertility of dtopors males. This suggests that there are homolog connections, in addition to those mediated by the conjunction complex, that are either prevented or resolved by dTopors activity. In wild-type flies, such interactions either do not occur or they are resolved by dTopors prior to the activity of the conjunction complex, as their presence would not allow for the precocious separation of homologs observed in mnm, snm, and tef mutants.

With regard to these persistent homolog interactions, it may be relevant that in mouse, lack of Topors results in alteration of the pericentric heterochromatin (Marshall et al. 2010). In fly male meiosis, it has been proposed that homolog interactions may be relegated to the pericentric heterochromatin during late prophase, a stage when both homologous euchromatic loci and centromeres are separated (Vazquez et al. 2002). Thus, persistent homolog attachments in dtopors may reflect an inability to resolve pericentric connections as a result of an abnormal chromatin structure.

dTopors role at Gypsy insulators

While we found that dTopors activities are essential for meiosis, we were surprised to find that its role at gypsy insulators, as assessed by expression of the ct⁶ mutant phenotype, was dispensable. Gypsy insulator function is compromised in flies heterozygous for a deletion of the dtopors gene, and knockdown of dTopors by in vivo expression of RNAi results in a similar effect (Capelson and Corces 2005). These data led to a model in which dTopors facilitates coalescence of insulator complexes into functional insulator bodies by anchoring chromatin loops to the nuclear lamina. While our data do not negate a role of dTopors at insulators, they indicate that its role is not required for insulator function.

Rather, our results suggest that when present, dTopors regulates insulator function in a dose-dependent manner, but when absent, this regulation is unnecessary. This paradoxical result suggests that dTopor’s role at insulators evolved as part of a regulatory mechanism after functional insulators already existed. It might be explained if two activities of dTopors, (e.g., the localization of gypsy insulator complexes to the nuclear lamina and the negative regulation of sumoylation of insulator components Mod(mdg4)2.2 and CP190) (Capelson and Corces 2006), must be in balance to properly regulate insulator function. For example, dTopors biochemical activities may be disruptive of insulator function unless the complexes are lamina associated. The physical lamina association might be more sensitive to small changes in dTopors stochiometry, whereas the enzymatic activities may be less dose sensitive. Thus, a 50% reduction in dTopors may result in a 50% reduction in lamina association of insulators, while having no significant effect on SUMO or ubiquitin ligase activities. A complete deletion of dTopors, however, would equally ablate both activities. This model may be tested using separation-of-function mutations in dTopors or by expressing independent domains and assessing their effect on gypsy insulator function.

Interestingly, while both meiotic and insulator functions of dTopors appear to be related to its association with the nuclear lamina, a lamina localization has not been observed for mammalian Topors homologs (Haluska et al. 1999; Zhou et al. 1999; Rasheed et al. 2002; Weger et al. 2003). Given the domain conservation between the fly and human proteins, however, it seems likely that their activities are mechanistically similar. Further studies identifying the downstream targets of Topors and dTopors ubiquitination and sumoylation, and functional tests of the lamina association, will be necessary to shed light on the significance of this difference.