Abstract

Homologous chromosomes must pair and establish stable connections during prophase I of meiosis to segregate reliably from each other at anaphase I. In most organisms, the stable connections, called chiasmata, arise from crossovers. In Drosophila males, homologs pair and segregate without crossing over. Chiasmata are replaced by a homolog conjunction complex that includes the Stromalin in Meiosis (SNM) and Modifier of Mdg4 in Meiosis (MNM) proteins. MNM is one of 31 alternative splice products of mod(mdg4), all of which share a common 402-amino-acid N terminus and differ at their C termini. Previous data demonstrated that an MNM-specific exon is required for homolog conjunction, but did not address whether the N-terminal common region, which includes a BTB domain that can mediate coalescence of protein-DNA complexes, is also required. Here we describe a mutation in the common region of mod(mdg4), Z3-3401, that causes qualitatively similar phenotypes as the MNM-specific alleles but disrupts X–Y segregation much more drastically than autosomal segregation. The mutant MNM protein in Z3-3401 is expressed throughout prophase I in spermatocytes but the protein is confined to the cytoplasm, suggesting that the Z3-3401 mutation disrupts a signal required for nuclear localization or retention. Z3-3401 fails to complement a large battery of lethal and semilethal alleles in the common region for meiotic nondisjunction, including an allele containing an amino acid substitution at a conserved residue in the BTB/POZ domain, consistent with a general requirement for the mod(mdg4) common region in homolog segregation.

THE segregation of homologous chromosomes at meiosis I is an essential step in sexual reproduction and must be accomplished accurately to prevent the generation of aneuploid gametes. Missegregation of homologs is a major cause of spontaneous abortion and genetic illness in humans (Hassold and Hunt 2001).

Segregation of homologs at anaphase I depends upon their prior alignment and pairing during early prophase I (Page and Hawley 2003; McKee 2004). In most eukaryotes, the initial homologous pairing interactions are quickly followed by the formation of elaborate homolog linking structures known as synaptonemal complexes (SCs) and by the onset of meiotic recombination (Roeder 1997; Page and Hawley 2003, 2004). The crossovers that occur between homologous chromatids during this stage are in turn essential for generation of chiasmata, the stable linkers that connect homologs throughout late prophase I and metaphase I and that enable the homologs to segregate reliably from one another at anaphase I (Hawley 1988; Carpenter 1994).

Homolog pairing is essential for segregation even in variant forms of meiosis that do not involve recombination and chiasmata (Wolf 1994). In Drosophila males, homologs are intimately paired throughout the first half of meiotic prophase but do not recombine or form SCs. Pairing is lost in midprophase I but homologs remain together in discrete nuclear territories until the onset of prometaphase I when they condense into tight “achiasmate” bivalents, which then segregated with exceptional fidelity at anaphase I (Vazquez et al. 2002). The central role of pairing in this process has been well documented for the X and Y chromosomes, which pair only within a discrete heterochromatic region encompassing the rDNA. X heterochromatic deletions that remove all of the rDNA prevent pairing of the X and Y and lead to their random assortment at anaphase I (McKee and Lindsley 1987; Park and Yamamoto 1995; McKee 1996). Moreover, transgenic rDNA insertions on such heterochromatically deficient X chromosomes substantially restore both pairing and disjunction of the X–Y pair (McKee and Karpen 1990; McKee 1996).

The means by which achiasmate homologs in Drosophila remain stably connected until anaphase I despite the absence of synaptonemal complexes and chiasmata has been an enigma. Recently, however, the two proteins Modifier of Mdg4 in Meiosis (MNM) and Stromalin in Meiosis (SNM) were shown to be essential for stable connections between achiasmate homologs. mnm and snm mutations cause high frequencies of univalents and random segregation of homologs during meiosis I (Thomas et al. 2005). Ectopically expressed, GFP-tagged MNM was shown to suppress the meiotic phenotypes of the two mnm mutations and to localize to meiotic chromosomes throughout prophase I and metaphase I. MNM–GFP colocalizes with native SNM protein to nucleoli of prophase I spermatocytes, where the rDNA genes are sequestered, and to the pairing region of condensed X–Y bivalents during prometaphase I and metaphase I. Both proteins disappear at the onset of anaphase I, strongly implying that they play a structural role in maintaining homolog connections. Mutations in a third gene [teflon (tef)] cause similar phenotypes but affect only the autosomes (Tomkiel et al. 2001).

Despite these recent advances, several key issues related to the mechanism of achiasmate homolog segregation remain unresolved. Perhaps the most important is the molecular basis for homolog conjunction. SNM is a distant homolog of the SCC3 family of cohesin proteins, raising the possibility that achiasmate homologs are connected by a cohesin complex of some kind. However, MNM and SNM do not visibly colocalize with the cohesin protein SMC1 on male meiotic chromosomes (Thomas et al. 2005).

An alternative mechanism is suggested by the domain structure of MNM. MNM is encoded by the complex mod(mdg4) locus, which is thought to produce 31 distinct chromosomal proteins with a common 402-amino-acid N terminus but different C termini encoded by alternatively spliced exons in the variable region (VR) (see Figure 1A). The common region (CR) includes an N-terminal BTB/POZ domain, and most of the VR C termini, including that of MNM, contain a C2H2 motif (Dorn and Krauss 2003; Labrador and Corces 2003). BTB/POZ domains are strong protein interaction domains found in many transcriptional regulatory proteins, where they function in mediating homodimerization and multimerization (Bardwell and Treisman 1994; Zollman et al. 1994; Igarishi et al. 1998; Muller et al. 1999; Melnick et al. 2000; Gause et al. 2001; Stogios et al. 2005). The BTB domain of mod(mdg4) is most similar to that of the Drosophila GAGA factor, an abundant transcription regulator required for chromatin remodeling of many developmentally regulated promoters and for pairing-dependent silencing (Granok et al. 1995). Indeed, the BTB domain of Mod(mdg4) can substitute for that of GAGA with little loss of function (Read et al. 2000). GAGA utilizes a C-terminal C2H2 zinc-finger motif along with its N-terminal BTB domain to bind cooperatively to DNAs containing multiple GAGAA sequences, forming large multimeric complexes held together by BTB–BTB interactions (Espinas et al. 1999; Katsani et al. 1999). Mod(mdg4) proteins also form multimers and both MNM and Mod(mdg4)67.2, which is required in Drosophila somatic cells for the function of gypsy insulators (Gerasimova et al. 1995), form prominent nuclear foci that presumably arise via coalescence of multiple chromosome sites bound by Mod(mdg4)-containing complexes (Gerasimova and Corces 1998; Gerasimova et al. 2000; Gause et al. 2001; Ghosh et al. 2001; Thomas et al. 2005). Moreover, comparisons of polytene chromosome localization patterns of different Mod(mdg4) proteins indicate that the variable C termini specify distinct localization patterns (Buchner et al. 2000). Thus, a plausible mechanism for MNM-mediated conjunction would involve binding to chromosome pairing sites via its C-terminal C2H2 motif and coalescence of bound sites on homologous chromosomes via BTB-mediated multimerization.

The first step in this scenario has experimental support. Both of the mnm mutations disrupt the C2H2 motif of MNM—Z3-5578 truncating MNM upstream of this motif and Z3-3298 substituting a Y for the upstream H (Figure 1, A and B)—and both mutations abrogate localization of MNM (and its partner, SNM) to meiotic chromosomes (Thomas et al. 2005). However, there is no direct evidence as yet for a role of the mod(mdg4) BTB domain in homolog conjunction. Indeed, since the two mnm mutations affect only the unique C terminus of MNM, it is not known whether any part of the 402-amino-acid N-terminal CR of mod(mdg4) is required for conjunction. Although the transgene rescue data and localization patterns of MNM–GFP described above are consistent with the scenario outlined, those data do not prove that the N-terminal sequences of MNM must be present for the MNM-specific domain to mediate conjunction, nor do they establish whether or not those sequences play a direct role in the conjunction process.

Here we provide genetic and cytological evidence that the common region of mod(mdg4) is required for homolog conjunction. We describe a new mod(mdg4) allele that causes meiotic phenotypes very similar to those of the mnm alleles but maps to the common region. We also demonstrate that a large number of mutations in the CR disrupt meiotic homolog segregation, including one that involves substitution of a conserved residue in the BTB domain. These findings set the stage for mechanistic studies of the role of the BTB domain and other domains of Mod(mdg4) in meiotic conjunction.

MATERIALS AND METHODS

Fly stocks, special chromosomes, and Drosophila culture methods:

The Zuker-3 (Z3) collection consists of >6000 EMS-mutagenized third chromosomes balanced over TM6, Tb e and maintained by C. Zuker (Koundakjian et al. 2004). Z3-3401 and the mnm and snm lines used in this study were identified in a screen of the Zuker-3 collection for mutations that cause paternal loss of chromosome 4 (Wakimoto et al. 2004) and were kindly provided by B. Wakimoto. mod(mdg4) alleles were obtained from V. Corces (John Hopkins University, Baltimore), Rainer Dorn (Institute of Genetics, Martin-Luther-University, Halle, Germany), M. Frasch (Mount Sinai School of Medicine, New York), and the Bloomington Stock Center at the University of Indiana.

The marked Y chromosome [Dp(1;Y)BSYy+ = BSYy+] carries two transposed segments from the X chromosome with the markers Bs and y+ appended to the ends of the left and right arms, respectively (FlyBase 2006). C(1)RM, y2 su(wa) wa, C(4)RM, ci eyR, and C(2)EN, b pr are attached chromosomes consisting of two genetically complete copies of the chromosome (X, 4, or 2) attached to a single centromere (FlyBase 2006). Dp(1;1)scV1 contains a small duplication from the tip of XL carrying the y+ allele appended to the small heterochromatic right (XR) arm (Rasooly and Robbins 1991). The attached-XY chromosome used in the recombination crosses was YSX.YL In(1)EN, y B (X^Y, y B) (FlyBase 2006).

Unless otherwise specified, the males being tested were crossed singly to two or three females in shell vials. Crosses were incubated at 23° on cornmeal–molasses–yeast–agar medium. Parents were removed from the vial on day 8 and progeny were counted between day 13 and day 22.

Mapping of Z3-3401:

Z3-3401 was mapped to the mod(mdg4) region by its failure to complement Df(3R)GC14 (93D7; 93E1) (Mohler and Pardue 1984) for X–Y nondisjunction (NDJ). More detailed mapping was carried out by complementation against a battery of deletions, transposon insertions, and EMS mutations described in Table 1 using the same assay. To aid in this analysis, the breakpoints of several small deletions in the mod(mdg4) region were molecularly mapped, as described below. Z3-3401 failed to complement all deletions that encompass part or all of the CR and all transposon insertions and mutations in the CR of mod(mdg4).

View this table:
TABLE 1

Results of complementation tests between Z3 alleles and mod(mdg4) alleles

Determination of deletion breakpoints:

Deletion breakpoints were mapped relative to polymorphisms between ORiso3 and CSiso3, which are wild-type lines with isogenic third chromosomes derived from the Oregon-R and Canton-S stocks. The DNA sequences of ORiso3 and CSiso3 differ from each other at multiple sites, some of which have been identified (Hoskins et al. 2001). ORiso3 and CSiso3 were crossed with iso-third chromosome stocks carrying the mod(mdg4) deficiencies B2, T16, eGP4, 142Δ10, 142Δ29, 142Δ33, 142Δ49, and Df(3R)GC14. Genomic DNA was prepared from F1 adult heterozygotes as described above. Fragments of ∼500–800 nt within and beyond the mod(mdg4) locus were amplified by polymerase chain reaction (PCR) from these DNAs and purified and sequenced as described below. Sequences were analyzed for SNPs and double peaks. The logic of the assay is that sequence differences between ORiso3 and CSiso3 that lie within the deleted region will result in different single peaks on the DNA sequence electropherograms for the ORiso3/Df and CSiso3/Df samples (e.g., G vs. A at a specific nucleotide position), whereas sequence differences outside of the deleted region will result in a double peak on the electropherogram for at least one of the two samples (e.g., a G/A double peak in one sample and a G peak in the other). This method enabled us to map the relevant breakpoint of each deletion (Table 2, Figure 1A) with respect to 12 SNPs within the mod(mdg4) locus. The molecular coordinates and associated primers of the flanking SNPs are available upon request.

View this table:
TABLE 2

Mapping deletion breakpoints in the mod(mdg4) region

Molecular identification of mutations in Z3-3401, mod(mdg4)324, and mod(mdg4)340:

To identify the Z3-3401 mutation, genomic DNA was extracted from adult flies homozygous for Z3-3401 and for the Zuker-3 progenitor chromosome using the Wizard genomic DNA purification kit (Promega, Madison, WI). Known and conceptual exons of mod(mdg4) were amplified from the genomic DNAs using primer pairs complementary to intronic sequences immediately flanking the exons and, where necessary, exon-internal primers. The PCR parameters were 1 min at 94°, 35 cycles of 94° for 1 min, 55° for 1.5 min, and 72° for 2 min in a Perkin-Elmer (Norwalk, CT) thermocycler. Reaction mixtures contained 0.2 μm of each primer, 50 ng Drosophila genomic DNA (Zuker-3 or y w), 1.5 mm MgCl2, 0.2 mm dNTP mix, and 2.5 units Taq DNA polymerase (Promega) in a total volume of 50 μl. The amplicons were sequenced directly using an ABI 373 sequencer. We identified a single nucleotide substitution (C to T) in exon 4 of Z3-3401 predicted to result in substitution of cysteine (C) for arginine (R) at residue 224 (R224C).

To identify the mutations in mod(mdg4)324 and mod(mdg4)340, genomic DNA from mod(mdg4)324/Zuker-3 and mod(mdg4)340/Zuker-3 flies was extracted and analyzed as above. Both mutations were identified as double peaks on the resulting DNA sequence electropherograms. The mod(mdg4)324 mutation is a G-to-A substitution predicted to result in replacement of the glycine at residue 92 with aspartic acid (G92D). The mod(mdg4)340 mutation is a G-to-T substitution predicted to result in replacement of the codon for glutamine 177 (Q177) with a nonsense codon. Additional double peaks were present on the electropherograms of the CR sequences derived from both of the mod(mdg4)/Zuker-3 DNA samples, but all except the mutations cited above proved to represent synonymous substitutions.

Measuring X–Y NDJ:

+/BsYy+ males were crossed singly to two X/X y w (yellow1 white1118) females in shell vials. The X, Y, XY, and O sperm classes yield + (y w/+) females, w Bs (y w/BSYy+) males, BS (y w/+/BSYy+) females, and y w (y w/O) males, respectively. The NDJ frequency (percentage of X–Y NDJ) = 100 × (BS females + y w males)/N. N is the number of progeny scored.

Measuring chromosome 4 loss frequencies:

Males were crossed singly to two C(4)RM, ci eyR (4^4/O) females and the progeny scored for the recessive cubitus-interruptus (ci) and eyeless (eyR) markers. 4^4/O females generate only 4^4 and nullo-4 (O) eggs, which, when fertilized by regular sperm carrying a wild-type fourth chromosome, yield only ci+ ey+ progeny (viable triplo-4 and poorly viable Minute haplo-4 progeny). Nullo-4 (O) sperm from paternal NDJ or chromosome 4 loss yield viable disomic 4^4/O, ci ey progeny. Paternal NDJ generates 44 sperm as well but these yield only ci+ ey+ progeny that cannot be distinguished from the regular progeny. The percentage of chromosome 4 loss (% 4-loss) = 100 × (ci ey)/N.

Measuring second chromosome NDJ:

Males were +/BSYy+; bw/+; Z3/(Df or +). Sibling mutant (Z3/Df) and control (Z3/+) males were crossed either to C(2)EN, b pr (2^2/O) females or to y w (2/2) females with unattached second chromosomes at a ratio of two males to four females or one male to two females, respectively. Males and females were left together for 6–8 days, and then the females were transferred to fresh vials every 3–6 days and allowed to continue laying eggs until fertilized eggs were exhausted. Males were transferred to vials with fresh virgin females and the procedure was repeated as long as the males remained fertile. Progeny were counted to completion and scored for relevant markers. 2^2/O females generate eggs that are nullosomic (O) or disomic (2^2) for chromosome 2, so the only viable progeny are the products of paternal chromosome 2 NDJ (22 and O sperm). Overall, second chromosome NDJ was estimated from progeny per male in the 2^2/O and 2/2 crosses [F(2^2/O) and F(2/2)]. The formula is the percentage of NDJ = 100 × (2 × f × F(2^2)/[2 × f × F(2^2) + F(2/2)]. F(2^2/O) is doubled in the formula to compensate for the deaths of one-half the nondisjunctional progeny in the 2^2/O cross, due to fertilization of the wrong eggs (e.g., O sperm fertilizing O eggs). “f” is a fertility correction (1.8 in these data), based on an independent estimate of the relative fertility of 2^2/O and 2/2 females in crosses to males of like genotype (2^2/O × 2^2/O and 2/2 × 2/2), again doubling the progeny from the 2^2/O cross to account for the loss of 50% of the aneuploid fertilization products.

The frequency of sister-chromatid NDJ relative to homolog NDJ (percentage of sis-2 NDJ) for the second chromosomes was estimated from the ratio of brown-eyed (bw) progeny to bw+ (red-eyed) progeny. Both bw and bw+ progeny carry two paternal second chromosomes; progeny from fertilization of 2^2 eggs by O sperm are black body, purple eyes (b pr). Since the paternal genotype is heterozygous bw/bw+, the 22/O progeny can be bw/bw, bw+/bw+, or bw/bw+, the former two genotypes resulting from sister-chromatid NDJ and the latter from homolog NDJ. The formula for % sis-2 NDJ is 100 × 2 × bw/(bw + bw+). (The bw progeny are doubled to account for a presumed equal number of bw+ progeny that are homozygous for the bw+ chromatid.)

Assaying homologous pairing in spermatogonia and spermatocytes:

Pairing was assayed by counting GFP spots in spermatogonia and spermatocytes from males homozygous for a chromosome 2 transgene carrying a 256mer tandem array of lacO repeats and heterozygous for a transgene (also on chromosome 2) expressing a GFP-LacI chimeric protein under control of the hsp83 promoter (Robinett et al. 1996; Straight et al. 1996; Vazquez et al. 2002; Thomas et al. 2005). Testes were dissected from third instar larvae, pupae, or young adults in testes buffer (183 mm KCl, 47 mm NaCl, 10 mm Tris–HCl, 1 mm EDTA, 1 mm PMSF) and gently squashed in testes buffer. Spot frequencies were compared between Z3-3401/Df(3R)T16 and sibling control (Z3-3401/+ and Df(3R)T16/+) males. The mean distance among the four GFP spots (spot dispersion) during late prophase I was determined by measuring and averaging the four shortest pairwise distances using Metamorph. The resulting values were averaged over the N nuclei scored to obtain the mean spot dispersion.

Transgene rescue crosses:

Two insertions of [hs∷MNM-GFP] (Thomas et al. 2005), one each on chromosomes 2 and 3, were used in the rescue experiments. For the chromosome 2 transgene, +/BsYy+; [hs∷MNM-GFP]2, bw+/bw; Z3-3401, st/ Z3-3401, st males and their +/BsYy+; bw/bw; Z3-3401, st/ Z3-3401, st brothers (scarlet vs. white eyes, respectively) were subjected to zero to two heat shocks (39° for 1 hr) during larval stages and then crossed as adults to y w females to assay X–Y NDJ or their testes were dissected and analyzed cytologically. Similar methods were used to test a third chromosome insertion of [hs∷MNM-GFP] and a second chromosome insertion of [CR-7.5] (= P[w+ 7.5-kb BamHI] (Buchner et al. 2000)).

Testis immunostaining:

Testes were dissected from third instar larvae, pupae, or young adults. For anti-α-tubulin/DAPI experiments, testes were fixed according to Cenci et al. (1994). Before incubation with antibodies, slides were rinsed twice in PBS and blocked in PBS, 1% BSA for 45 min. Testes preparations were incubated overnight at 4° with FITC-conjugated monoclonal anti-α-tubulin (Sigma, St. Louis) diluted 1:150 in PBT (PBS with 0.1% Triton X-100) containing 1% BSA. Slides were rinsed twice with PBT, once with PBS, stained with DAPI (1 μg/ml) for 5 min, rinsed twice in PBS, and mounted with Vectashield mounting medium.

For the anti-SNM staining, the procedure of Thomas et al. (2005) was followed. Briefly, testes were fixed according to Gunsalus et al. (1995). Before incubation with antibodies, slides were washed twice in PBT plus DOC (PBS with 0.3% Triton X-100 with 0.3% sodium deoxycholate) for 15 min and once in PBT (PBS with 0.1% Triton X-100) for 10 min and blocked in TNB (0.1 m Tris–HCl, pH 7.5, 0.15 m NaCl, 0.5% blocking reagent (Perkin-Elmer) for 30 min. Testes preparations were incubated overnight at 4° with FITC-conjugated monoclonal anti-α-tubulin (Sigma) diluted 1:150 with either undiluted anti-SNM N terminus antibody or anti-SNM C terminus diluted 1:250. Slides were rinsed three times with TNT (0.1 m Tris–HCl, pH 7.5, 0.15 m NaCl, 0.05% Tween-20) and then incubated with Alexa Fluor 647 goat anti-rabbit IgG (H+L) diluted 1:500 in TNB for 30 min. Slides were rinsed three times with TNT, stained with DAPI (1 μg/ml) for 5 min, rinsed two more times with TNT, and mounted with Vectashield mounting medium. Anti-ModC (Buchner et al. 2000; Thomas et al. 2005) was diluted 1:4000 in PBS and visualized using Alexa Fluor 546 goat anti-rabbit IgG (H+L) diluted 1:5000.

Microscopy:

All testis preparations were examined with an Axioplan (Zeiss) microscope equipped with an HBO 100-W mercury lamp for epifluorescence and with a scientific grade cooled charge-coupled device (Roper). Grayscale digital images were collected, pseudocolored, and merged using Metamorph Software (Universal Imaging, West Chester, PA).

Measuring recombination and NDJ in female meiosis:

To measure sex chromosome NDJ and recombination, Dp(1;1)scV1, y pn cv m f · y+/y females were crossed with YSX.YL In(1)EN, y B/Y (X^Y, y B/Y) males. The regular progeny are B females and B+ males; X–X NDJ yields B+ females and B males. The percentage of NDJ = 100 × 2 × (B+ females + y B males)/(N + B+ females + y B males). Recombination was scored in the regular (B+) sons. Since both X chromosomes carry mutant y alleles at the native y locus, the duplicated y+ allele on the X chromosome right arm in the y pn cv m f · y+ homolog serves as a centromere marker. pn, which is <1 cM from the XL tip, is the distal-most marker. Recombination on chromosomes 2 and 3 was measured as described in the legend to Table 7. Map distances were calculated by standard formulae and expressed in centimorgans.

Analyzing pigmentation of y2 flies:

The body, wing, and bristle colors of adult flies were scored visually at ×20 magnification; a minimum of 20 flies of each genotype were scored.

Analysis of MNM RNAs:

Total RNA was isolated using RNAwiz (Ambion, Austin, TX) from the following genotypes: Z3-5578/Z3-3401, Z3-5578/Df(3R)GC14, and Z3-3401/Df(3R)GC14. For each RNA, oligo(dT) and gene-specific primer reverse transcription (RT) reactions were performed using the SuperScript first-strand synthesis system for RT–PCR (Invitrogen, San Diego). Primers for gene-specific primer RT are 5′-gattgttagatgtcttatgg-3′ and 5′-tgtaagcctatgacgcatcc-3′. The three RT reactions were combined into a cocktail and were used in PCR. To determine if trans-splicing occurs, PCR was performed using the RT cocktail from trans-heterozygotes (Z3-5578/Z3-3401). As a control for template switching, PCR was performed on the combined cocktails of hemizygotes [Z3-5578/Df(3R)GC14 and Z3-3401/Df(3R)GC14]. The primers used in PCR are 5′-tgaaatggctacatatgtgg-3′ and 5′-cggcatctgagtgaacatct-3′. PCR products were run on an agarose gel, gel purified using the QIAquick gel extraction kit (QIAGEN, Chatsworth, CA) and TA cloned (Invitrogen). Minipreps were performed on individual colonies and the DNA was sequenced using standard techniques.

The parental (P) and recombinant (R) RNA frequencies in the trans-heterozygous sample were estimated from the frequencies of the corresponding parental (P1 + P2) and recombinant (R1 + R2) cDNAs by correcting for the observed frequency of template switching (0.16) in the control reaction, as follows. The frequency of template switching for both P and R templates in the trans-heterozygous reaction was assumed to be the same as for P templates in the control reaction. Therefore, R1 + R2 cDNAs originate from R templates at a frequency of (1 − 0.16)R and from P templates at a frequency of 0.16(P). Since P = 1 − R, R1 + R2 = (1 − 0.16)R + 0.16(1 − R) = 0.16 + 0.68R. Plugging in R1 + R2 = 0.40, we obtain R = (0.40 − 0.16)/0.68 = 0.35 with a 95% confidence interval of ±1.96[R(1 − R)/N]½.

RESULTS

An amino acid substitition in the mod(mdg4) CR that causes elevated X–Y NDJ:

The meiosis-specific mod(mdg4) mutations described previously, Z3-5578 and Z3-3298 (Thomas et al. 2005), were recovered in a screen for paternal chromosome loss mutants among the Z3 collection of EMS-mutagenized autosomes (Koundakjian et al. 2004; Wakimoto et al. 2004). That screen yielded an additional viable mutation, Z3-3401, that caused high frequencies of X–Y NDJ in male meiosis (Table 3) but no NDJ of sister X chromatids (data not shown), similar to the phenotypes of the mnm alleles, Z3-5578 and Z3-3298. Deficiency complementation (Table 1, analyzed more fully below) and DNA sequence analysis of Z3-3401 led to the identification of a single base-pair substitution in exon 4 of the mod(mdg4) CR predicted to result in a R224C substitution in all Mod(mdg4) proteins (Figure 1B). This mutation lies downstream of the BTB domain in a region of the protein rich in T residues and of unknown function (Bardwell and Treisman 1994; Zollman et al. 1994; Buchner et al. 2000; Dorn and Krauss 2003).

View this table:
TABLE 3

Sex chromosome NDJ in Z3-3401 and mnm males

Figure 1.—

Structure of the mod(mdg4) locus and molecular identification of Z3-3401. (A) Map of the mod(mdg4) locus, modified from Dorn et al. (2001). Rectangles represent exons (open for untranslated exons, solid for translated exons). The CR contains four exons common to all Mod(mdg4) transcripts, transcribed left to right and cis spliced. The VR contains C-terminal exons specific for each isoform on the same strand (below the line) or the opposite strand (above the line) from the CR. Diagonal lines represent splice sites. Cross-hatched rectangles represent C-terminal exons of MNM. Mapped mutational lesions are indicated by triangles (transposon insertions), stars (point mutations), or dark lines below the exons (deletions). Endpoints with diamonds represent deletion breakpoints mapped within mod(mdg4); endpoints with arrows represent deletion breakpoints outside mod(mdg4). The dashed line (CR-7.5) depicts the 7.5-kb genomic rescue fragment used in the experiments in Table 8 (Buchner et al. 2000). See Table 1 legend for references for mod(mdg4) alleles. (B) Sequence of MNM protein, based on conceptual translation of a cDNA isolated from y w testis RNA (Thomas et al. 2005). Altered residues in Z3 mutations and in mod(mdg4)324 and mod(mdg4)340 are indicated by boldface type with mutant residue above wild-type residue. The BTB domain is shaded, the FLYWCH domain is underlined, and the C2H2 motif is enlarged.

Z3-3401 disrupts segregation of sex chromosomes more severely than autosomes:

As shown in Table 3, the X–Y NDJ frequency in Z3-3401/Df males was similar to the frequencies in Z3-5578/Df and Z3-3298/Df males. However, unlike the mnm alleles that behave as genetic nulls, the X–Y NDJ frequency in Z3-3401/Z3-3401 males, although substantial (∼30%), was significantly lower than in Z3-3401/Df males, suggesting that Z3-3401 may be a hypomorphic allele.

Both mnm and snm mutations disrupt autosomal segregation as strongly as X–Y segregation. To ascertain if Z3-3401 also disrupts segregation of the autosomes, fourth chromosome loss (production of nullo-4 sperm) was assayed by crossing Z3-3401 males carrying wild-type fourth chromosomes to females carrying C(4)RM, an attached fourth chromosome marked with the recessive mutations ci and eyR (Table 4). Z3-3401/Df males generated ci ey progeny [C(4)RM/0], indicative of nullo-4 sperm, at frequencies of 5.2 and 7.9% over two different mod(mdg4) deficiencies, frequencies much lower than those of hemizygous mnm males (22–28%). An even lower chromosome 4 loss frequency (3.3%) was exhibited by Z3-3401 homozygotes, consistent with the above suggestion that it is a hypomorphic allele. The difference between the chromosome 4 loss and X–Y NDJ frequencies suggests that Z3-3401 affects sex chromosomes more strongly than autosomes or, alternatively, large chromosomes more strongly than small ones (the fourth chromosome being ∼4% the size of the X chromosome and 2% that of the second or third chromosome).

View this table:
TABLE 4

Fourth chromosome loss in Z3-3401 and mnm males

To assess the effects of Z3-3401 on segregation of a large autosome pair, Z3-3401/Df males were crossed to females carrying the attached second chromosome, C(2)EN, b pr (Table 5). Since all eggs in this cross are either diplo-2 or nullo-2, the only viable progeny result from fertilization by NDJ sperm (diplo-2 or nullo-2), so that “progeny per male” is a rough measure of NDJ. Crosses with wild-type males typically produce <0.5 progeny/male, whereas snm and mnm males generate 20–25 progeny each (Table 5; Thomas et al. 2005). Z3-3401/Df males generated 15 progeny/male, indicative of a significant amount of chromosome 2 NDJ. The distribution of paternal chromosome 2 markers among the NDJ progeny indicated that virtually all of the NDJ occurred at meiosis I.

View this table:
TABLE 5

Chromosome 2 NDJ in Z3-3401 and mnm males

To obtain quantitative estimates of the frequency of chromosome 2 NDJ, parallel crosses of Z3-3401/Df sibling males to C(2)EN females (2^2/O) and to females with normal second chromosomes (2/2) were conducted by a procedure designed to exhaustively sample sperm (see materials and methods). Under these conditions, the numbers of progeny produced per male (F) in the 2^2 and 2/2 crosses are proportional to the numbers of NDJ and regular sperm, respectively, produced by the tested males.

Using this method, we estimated the second chromosome NDJ frequency for Z3-3401/Df males at 17%, a moderate frequency more comparable to that of the fourth chromosome than to the X–Y pair. By comparison, similar crosses for the two mnm alleles (Table 5) yielded chromosome 2 NDJ frequencies ranging from 43 to 45%, very near to random assortment. These values are in excellent agreement with NDJ estimates from cytological assays (data not shown), thus providing a validation for the parallel cross method. We conclude that Z3-3401 is significantly more disruptive of sex chromosome than autosomal segregation, unlike mnm and snm mutations, which disrupt segregation of all chromosome pairs to roughly the same degree.

Z3-3401 causes bivalent instability and missegregation of univalents at meiosis I:

To gain further insight into the nature of the meiotic anomaly in Z3-3401 males, we examined testis squashes stained with DAPI or acetic orcein to visualize DNA (Figure 2, A and C). Inspection of spermatids from Z3-3401 squashes revealed considerable variability in size of spermatid nuclei in Z3-3401 (Figure 2A), an indicator of NDJ. The cause of the high NDJ in Z3-3401 males was evident from inspection of meiosis I spermatocytes (Figure 2, A and C). In wild-type spermatocytes, meiotic chromosomes are first clearly resolved shortly before prometaphase I as three large and one small condensing mass of chromatin, corresponding to the three large bivalents (X–Y and second and third chromosomes) and the small fourth chromosome pair, arrayed around the nuclear periphery. The chromosomes subsequently congress to form a single compact clump of chromatin on the meiosis I spindle (Cenci et al. 1994). In Z3-3401 spermatocytes, although the chromosomes condensed normally, the homologs were frequently unpaired at prometaphase I and metaphase I. There was no sign of chromosome fragmentation or of breakdown of univalents into their constituent chromatids. Anaphase I was disorganized, with chromosomes migrating to poles asynchronously (Figure 2C). Meiosis I poles frequently (54%) exhibited nuclei of unequal sizes, indicative of meiosis I NDJ (Figure 2, A and B). Although the meiosis I phenotypes of Z3-3401/Df males were qualitatively similar to those of mnm males, normal metaphase I configurations and equal telophase I poles were observed at considerably higher frequencies in Z3-3401 males than in Z3-3298/Df males (Figure 2B), consistent with the genetic evidence that autosomes segregate more regularly in Z3-3401/Df than in Z3-3298/Df.

Figure 2.—

Cytological analysis of male meiosis in Z3-3401. (A) Representative figures from various stages of meiosis in wild-type and Z3-3401/Z3-3401 males stained with DAPI to visualize the DNA and an antitubulin antibody to visualize the spindles. The round spermatid nuclei in the far right panels are unequal in size in Z3-3401, indicative of NDJ. Prometaphase I and metaphase I cells exhibit too many DNA clumps, suggestive of conjunction failure, and anaphase I cells exhibit unequal poles, indicative of missegregation of homologs at meiosis I. Meiosis II chromosome morphology and segregation appear normal. Bar, 5 μm. (B) Quantification of meiosis I abnormalities in Z3-3401 spermatocytes. Data from wild-type and Z3-3298 spermatocytes (Thomas et al. 2005) provided for comparison. MI, metaphase I; AI, anaphase I; TI, telophase I. (C) Meiosis I chromosomes from wild-type and Z3-3401 homozygous males stained with acetic orcein to visualize DNA and observed by phase-contrast microscopy. There are four large univalent (unpaired) chromosomes (red arrows) and one bivalent at prometaphase I and six large univalent chromosomes at metaphase I in Z3-3401 spermatocytes. Note the lagging chromosomes at anaphase I in Z3-3401. Bar, 5 μm. (D) Representative meiotic figures from hemizygous Z3-3401/Df(3R)T16 males with or without one copy of [hs∷MNM-GFP]3. Both groups of males were heat-shocked twice during larval development. Red arrows point to univalents in late prophase I; unpaired X and Y chromosomes are in the top right. Bar, 2 μm.

No meiotic abnormalities unique to Z3-3401 were observed. As in Z3-5578 and Z3-3298 spermatocytes, both spindle structure and kinetochore function appeared normal at both divisions. The second meiotic division appeared to proceed normally, consistent with the genetic data (Figure 2A). Pairing of homologous sequences during early prophase I as well as in premeiotic gonial cells, as revealed by fusion of fluorescent spots resulting from recruitment of LacI-GFP proteins to transgenic lacO arrays at homologous sites in autosomal euchromatin (Robinett et al. 1996; Straight et al. 1996; Vazquez et al. 2002), also appeared completely normal (Figure 3), consistent with the results of similar analyses of pairing in Z3-5578 and Z3-3298 mutants (Thomas et al. 2005). Moreover, in Z3-3401/Df males, as previously reported for wild-type and mnm males, pairing of homologous and sister chromatids is lost simultaneously during stage S2b, corresponding to midprophase I, and four distinct GFP spots, corresponding to the four chromatids of the chromosome 2 bivalent, are seen from then until the conclusion of the meiotic divisions. Finally, as described previously for both mnm and snm alleles, the separate GFP spots are farther apart, on average, in Z3-3401/Df than in wild-type males during late prophase I (Figure 3B), a phenotype referred to as “territorial expansion” (Thomas et al. 2005).

Figure 3.—

Homologous pairing and territory expansion in Z3-3401. Chromosomal GFP-LacI foci were imaged in germ-cell nuclei from wild-type (Z3-3401/+ or Df(3R)T16/+) and Z3-3401/Df(3R)T16 males homozygous for a single lacO array on chromosome arm 2R and hemizygous for [hsp83∷GFP-LacI](2) (Vazquez et al. 2002). (A) Homologous pairing in spermatogonia (gonia) and early PI spermatocytes. Nuclei with one spot were scored as paired; nuclei with more than one spot were scored as unpaired. (B) GFP-LacI spot dispersion in late prophase I spermatocytes. Mean Nuc. Diam., mean nuclear diameter (in micrometers) ±SD; Mean Disp., mean spot dispersion (in micrometers) ±SD. N, number of nuclei scored. Bars, 5 μm.

Thus, we find no genetic or cytological evidence that Z3-3401 has any meiotic phenotypes qualitatively different from those previously described for mnm (and snm) mutants, namely disruption of homologous chromosome territories during late prophase I and premature dissociation of bivalents, leading to random assortment of the resulting univalents at anaphase I. However, Z3-3401 differs from mnm and snm alleles in that it affects sex chromosome segregation more strongly than autosomal segregation.

Does ectopically expressed MNM–GFP fully complement the meiotic phenotypes of Z3-3401?:

Previous data showed that ectopic, heat-shock-driven expression of MNM–GFP fully complements the meiotic phenotypes of both Z3-5578 and Z3-3298, thus verifying that the mutations in the mnm exons of those two alleles are responsible for their mutant phenotypes (Thomas et al. 2005). However, the R224C mutation in Z3-3401 should affect the sequences of all 31 Mod(mdg4) isoforms, the functions of most of which are unknown. Ectopic expression of transgenic MNM–GFP would be expected to fully complement the meiotic phenotypes of Z3-3401 only if those phenotypes are caused by disruption of the MNM isoform alone.

To test for rescue of Z3-3401, +/BSYy+; Z3-3401/Df males carrying a single copy of the [hs∷MNM-GFP] transgene on chromosome 2 or 3 were generated, exposed to variable numbers of heat shocks during development, and tested for X–Y NDJ. The results were clear-cut: both the chromosome 2 and chromosome 3 insertion suppressed X–Y NDJ below 1% when two heat shocks were given (Table 6). In fact, the chromosome 2 transgene afforded full rescue even in the absence of heat shock. In addition, DAPI-stained spermatocytes from transgenic, heat-shocked males and from nontransgenic males were compared. Spermatocytes from the heat-shocked, transgene-bearing males appeared indistinguishable from wild-type spermatocytes with respect to organization and uniformity of spermatid nuclei, absence of univalents at prometaphase I and metaphase I, and equality of telophase I poles, whereas their nontransgene-bearing brothers exhibited the expected array of meiotic anomalies (Figure 2D).

View this table:
TABLE 6

X–Y NDJ in Z3-3401 males with and without P[hs∷MNM-GFP]

These results confirm that the R224C mutation in exon 4 is responsible for the male meiotic phenotypes seen in Z3-3401 males. More interestingly, however, they indicate that the male meiotic phenotypes of Z3-3401 are due entirely to the effects of the R224C mutation on functioning of the MNM isoform, even though the mutation should be present in all of the 31 Mod(mdg4) isoforms. These findings could indicate that MNM is the only mod(mdg4) isoform involved in homolog conjunction in male meiosis. However, they do not rule out the possibility of other meiotic isoforms of Mod(mdg4) that are not functionally disrupted by the R224C substitution.

The Z3-3401 mutation disrupts a nuclear localization or retention signal:

To gain additional insight into how the Z3-3401 mutation affects conjunction, we used antibodies against SNM and against the CR of Mod(mdg4) (anti-ModC) to examine the localization of MNM and SNM proteins in primary spermatocytes (Buchner et al. 2000; Thomas et al. 2005). We previously showed that these antibodies colocalize to multiple nucleolar foci throughout prophase I and to a prominent dense focus associated with the X–Y bivalent during prometaphase I and metaphase I in wild-type spermatocytes (Thomas et al. 2005). Although the anti-ModC antibody is not specific for MNM, the dense anti-ModC signal on the X–Y bivalent reflects MNM-specific staining since it, along with the anti-SNM signal, is completely abolished in spermatocytes from Z3-5578 and Z3-3298 males. Very similar results were observed in Z3-3401 spermatocytes at prometaphase I and metaphase I (Figure 4). No detectable staining of the X–Y bivalent with either anti-SNM or anti-ModC was observed in Z3-3401/Df spermatocytes during prometaphase I or metaphase I, although robust staining was evident in the Z3-3401/+ sibling controls. Thus, Z3-3401 abolishes staining of the X–Y pairing structure as thoroughly as do snm and mnm alleles. This finding suggests that the Mod(mdg4) CR as well as the MNM-specific VR exon might be required for chromosomal localization of MNM and SNM. It is important to note that the failure to observe staining of autosomal bivalents at prometaphase I in Z3-3401 spermatocytes does not imply that MNM and SNM are absent from autosomal bivalents at this stage (which would be difficult to reconcile with the relatively mild disruption of autosomal segregation in Z3-3401 males) because neither antibody yields detectable staining of autosomes at this stage in wild-type spermatocytes either (Thomas et al. 2005).

Figure 4.—

Localization of MNM and SNM proteins to the X–Y bivalent is disrupted by the Z3-3401 mutation. Spindles and DNA were stained with antitubulin (green) and DAPI (blue), respectively, in right panels. Both anti-MNM (A) and anti-SNM (B) antibodies stain a compact structure associated with the X–Y bivalent at prometaphase I in wild-type spermatocytes (top panels). No staining of condensed meiotic chromosomes with anti-MNM or anti-SNM is detected in Z3-3401 spermatocytes (bottom panels). Bar, 5 μM.

The absence of anti-MNM staining on condensed X–Y bivalents in Z3-3401 spermatocytes was unexpected since the mutation in Z3-3401 lies far upstream of the C-terminal domain that we have postulated to be responsible for chromosome binding. To gain further insight into the basis for the Z3-3401 phenotype, we analyzed anti-ModC staining patterns throughout prophase I, prior to chromosome condensation.

The anti-ModC staining pattern in wild-type spermatocytes exhibited a complex temporal pattern (Figure 5). In very young primary spermatocytes (stage S1; Cenci et al. 1994), strong anti-ModC staining was concentrated at the nuclear periphery although the chromosomes were distributed generally throughout the nucleus. However, by stage S3 when the three major bivalents have mostly separated from one another and throughout the remainder of prophase I, anti-ModC staining was largely confined to the nucleolus and to the DAPI-stained chromosomes. Within chromosomes, anti-ModC staining was not uniform but rather enriched in particular regions. The anti-ModC staining pattern in mid- and late prophase I is similar to that of MNM–GFP (Thomas et al. 2005), albeit somewhat less punctate. Thus, unlike condensed autosomal bivalents at prometaphase I or metaphase I, uncondensed autosomes stain robustly with the anti-ModC antibody throughout prophase I.

Figure 5.—

Mod(mdg4) proteins in early–mid (A) and late (B) prophase I spermatocytes in wild-type (WT) and Z3 mod(mdg4) mutants. Dissected testes were stained with anti-ModC antibody, which reacts against the common region of Mod(mdg4) proteins; antitubulin to stain the spindles; and DAPI to stain DNA. Nomenclature for stages follows Cenci et al. (1994): S1–S2, early prophase I; S3–S4, midprophase I; S5–S6, late prophase I. Arrows indicate nucleoli. Bar, 5 μm.

In Z3-5578 and Z3-3298 spermatocytes, early prophase I staining appeared to be relatively normal both in intensity and localization, with the bulk of the signal distributed around the nuclear periphery (Figure 5A). However, by midprophase I (not shown) and more clearly at late prophase I (Figure 5B), diffuse anti-ModC staining was distributed throughout the nucleus. Some of this staining overlapped with the DAPI-stained chromosome territories and might represent other chromosomal Mod(mdg4) isoforms present in spermatocytes in addition to MNM; however, the staining failed to show the prominent enrichment on chromosomes and in nucleoli that we observed in wild type. These findings suggest that the mutant MNM proteins encoded by Z3-5578 and Z3-3298 are stable and able to gain entry into the nucleus but are unable to localize properly either to chromosomes or to the nucleolus.

Z3-3401 spermatocytes exhibited a different pattern from the other two mutants. Strong anti-ModC staining was apparent throughout prophase I but was restricted primarily to the cytoplasm at all stages (Figure 5, A and B). This contrasts sharply with the staining patterns in wild-type and mnm spermatocytes, in which little or no cytoplasmic anti-ModC staining was detected at any stage. These results suggest that the Z3-3401 mutation might disrupt a nuclear localization or retention signal essential for nuclear localization of MNM. Late prophase I nuclei from Z3-3401 males did exhibit faint granular staining in the nucleus (Figure 5B), consistent with the presence of small amounts of mutant MNM protein or of other, less abundant Mod(mdg4) isoforms in the nucleus. We conclude that the mutant MNM proteins encoded by all three Z3 alleles are stable throughout prophase I but improperly localized to the nucleoplasm in Z3-5578 and Z3-3298 mutants and to the cytoplasm in Z3-3401.

Is Z3-3401 specific for male meiosis?

Z3-3401 differs from most other mutations in the CR in that it does not affect viability. This could be either because it is hypomorphic or because it disrupts a domain that is required only for meiotic homolog conjunction and not for other functions carried out by other Mod(mdg4) proteins. To assess the specificity of Z3-3401, we evaluated its effects on two other processes previously shown to be affected by mod(mdg4) mutations: female fertility and chromatin insulator function.

Effects on female fertility have been previously described for certain viable and semilethal mod(mdg4) genotypes (Buchner et al. 2000). To determine if Z3-3401 affects female fertility, sibling Z3-3401/Df and Z3-3401/+ females were crossed singly to two wild-type males. For comparison, similar crosses involving Z3-3298 and Z3-5578 were also carried out. The results (Table 7) show that Z3-3401 dramatically reduces female fertility: Z3-3401/Df females produced, on average, only 2.8 progeny each whereas their heterozygous siblings produced, on average, 42.1 progeny. This fertility reduction must reflect the effects of Z3-3401 on the functioning of other isoforms in addition to MNM because females hemizygous for either of the MNM-specific alleles exhibited fertility levels comparable to those of their heterozygous sisters (Table 7) and because ectopic expression of MNM–GFP in Z3-3401/Df females failed to signficantly improve their fertility (data not shown).

View this table:
TABLE 7

Fertility, X chromosome NDJ, and X, second, and third chromosome recombination frequencies in Z3-3401, mnm, and mod(mdg4)T6 hemizygous females

Since mutations in some genes required for female meiosis cause semisterility, we assayed for effects of hemizygosity for Z3-3401 on several meiotic parameters, including the frequency of X chromosome NDJ and the frequency of recombination in large intervals on the X, second, and third chromosomes (Table 7). Similar assays were conducted for Z3-3298/Df, Z3-5578/Df, and mod(mdg4)T6/Df females. No significant effects of any of the four tested mod(mdg4) alleles on chromosome segregation or recombination were observed. Moreover, the eggs laid by Z3-3401/Df females did not exhibit the “spindle” phenotype caused by incomplete development of the dorsal–ventral axis that is exhibited by females carrying mutations that disrupt meiotic double-strand break repair (data not shown; Ghabrial and Schupbach 1999). Finally, synaptonemal complex formation, as assayed by incorporation of the central element protein C(3)G (Page and Hawley 2001) into linear structures in pachytene oocytes was found to be normal in Z3-3401/Df females (data not shown). Thus, the fertility reduction caused by Z3-3401 does not result from disruption of female meiosis. No further analysis of the female semisterility of Z3-3401 has been carried out as yet and it is not known which Mod(mdg4) protein(s) is required for normal female fertility.

Some mod(mdg4) mutations act as enhancers or suppressors of mutations caused by insertions of gypsy transposable elements into regulatory DNA of certain loci such as yellow and cut (Georgiev and Gerasimova 1989; Georgiev and Kozycina 1996; Gause et al. 2001). These modifier effects are thought to be due to disruption of the chromatin insulator protein Mod(mdg4)67.2, which localizes, along with its partner Suppressor of Hairy Wing [Su(Hw)], to gypsy sequences and blocks activation of nearby promoters by enhancers on the opposite side of the gypsy insertion (Gerasimova et al. 1995; Gause et al. 2001). To determine whether Z3-3401 disrupts functioning of the Mod(mdg4)67.2 protein, the effects of Z3-3401 on the pigmentation pattern of y2 flies was examined. y2 flies have pigmented (black) bristles but unpigmented (yellow) bodies and wings because of a gypsy insertion in the upstream regulatory region of the yellow locus, which is required for production of the major pigment in the integument, wings, and bristles of adult flies. The gypsy insertion lies between the yellow promoter and upstream enhancers specific for the body and wings and therefore blocks activation of the yellow promoter in the body and wings. However, the bristle-specific enhancer is located in an intron downstream of both the yellow promoter and the gypsy insertion in y2 and is therefore not prevented from activating the promoter (Geyer et al. 1986; Geyer and Corces 1987). The mod(mdg4)T6 mutation, a nonsense mutation in the 67.2-specific exon, specifically disrupts functioning of the Mod(mdg4)67.2 protein, leading to loss of insulator function. The result is that all of the yellow enhancers are silenced by Su(Hw), resulting in a y phenotype (yellow bristles, wings, and bodies) (Gerasimova et al. 1995; Mongelard et al. 2002).

To determine if Z3-3401 modifies the y2 phenotype in a similar fashion, we examined the colors of wings, abdomens and bristles of y2/Y; Z3-3401/Df(3R)T16 flies and compared them to y2/Y; mod(mdg4)T6/Df(3R)T16, y2/Y; Df(3R)T16/, and y2; Z3-5578/Df(3R)T16 flies, scoring at least 20 flies of each genotype. The result was that all y2; Z3-3401/Df(3R)T16 flies were y in phenotype and indistinguishable from y2; mod(mdg4)T6/Df(3R)T16 flies, whereas the heterozygous control Df(3R)T16/+ and the y2; Z3-5578/Df(3R)T16 flies exhibited the y2 pigmentation pattern featuring black bristles (data not shown). We conclude that Z3-3401 acts as an enhancer of y2 and infer that it disrupts the function of the Mod(mdg4)67.2 isoform. Taken together with the MNM-independent effect of Z3-3401 on female fertility, these data provide strong evidence that the Z3-3401 mutation disrupts functioning of other Mod(mdg4) isoforms in addition to MNM.

Does the mod(mdg4) common region have a general role in male homolog segregation?

Our analyses of Z3-3401 indicate that the Mod(mdg4) CR is required for an indirect step in meiotic homolog conjunction and stable nuclear localization of MNM. If the CR plays additional roles in homolog conjunction, then other mutations in the CR should also disrupt homolog conjunction. Although there are numerous extant mod(mdg4) alleles (FlyBase 2006), the great majority are recessive lethals that cannot be tested directly for meiotic phenotypes. Moreover, clones homozygous for two such mutations in the germline of mod(mdg4)/+ males were inviable (B. D. McKee, unpublished data), precluding analysis of meiosis in such clones.

To address the role of the mod(mdg4) CR in meiotic conjunction, we conducted complementation tests between Z3-3401 and a collection of preexisting mod(mdg4) alleles that included nine small deletions, three transposon insertions, and six EMS-induced mutations (Table 1 and Figure 1A). Except for mod(mdg4)T6, which results from a base substitution in the C-terminal exon specific for the nonessential Mod(mdg4)67.2 protein (Gerasimova et al. 1995; Gause et al. 2001), all of the tested mod(mdg4) mutations and deletions are recessive lethals or semilethals, and nearly all have been mapped to the CR either by direct molecular identification of lesions within the CR or by failing to complement mutations with molecular lesions in the CR (Gerasimova et al. 1995; Buchner et al. 2000; Gause et al. 2001). Five of the lethal alleles carry previously identified lesions within the mod(mdg4) transcription unit: 142Δ15, a 1-kb deletion that removes the promoter and first exon of mod(mdg4); 142Δ32, a deletion that removes upstream regulatory sequences of mod(mdg4) (Azpiazu and Frasch 1993); and the three transposon insertions (2, 3, and neo129), all of which have been molecularly mapped to sites within the mod(mdg4) common region (Figure 1A; Buchner et al. 2000). We used a SNP-mapping technique to map the breakpoints of the remaining seven small deletions (Table 2) and found that three of the deletions (T16, 142Δ33, and 142Δ10) are deficient for the entire mod(mdg4) locus, two (142Δ29 and 142Δ49) are deficient for part or all of the common region but not deficient for the variable region, and two (eGP4 and B2) have fully intact mod(mdg4) common regions but are deficient for most of the variable exons, including the MNM-specific exon (Figure 1A). In addition, we sequenced two of the lethal EMS alleles, mod(mdg4)324 and mod(mdg4)340, and found that both contain base substitutions within the common region. These mutations are predicted to result in an amino acid substitution, G92D, within the BTB domain of mod(mdg4)324, and a nonsense mutation at Q177, a site downstream of the BTB domain in mod(mdg4)340 (Figure 1B).

Z3-3401 fully complemented the lethality of all of the lethal and semilethal alleles but the trans-heterozygous males exhibited a complex complementation pattern with respect to X–Y NDJ (Table 1). Z3-3401 failed to complement most mod(mdg4) alleles, yielding NDJ frequencies comparable to those of Z3-3401 hemizygotes. However, it partially or fully complemented the two deletions B2 and eGP4, which are confined to the VR, and fully complemented the viable mod(mdg4)T6 allele. Moreover, as described above, it strongly (but not completely) complemented the mnm alleles Z3-5578 and Z3-3298, the trans-heterozygotes yielding only ∼1% X–Y NDJ (Table 3). The finding that Z3-3401 and mod(mdg4)T6 complement with respect to X–Y NDJ was expected as no X–Y or fourth chromosome NDJ was observed in mod(mdg4)T6 hemizygous males (data not shown), indicating that the Mod(mdg4)67.2 isoform is not required for chromosome segregation in male meiosis.

Parallel complementation tests between the mnm alleles, Z3-5578 and Z3-3298, and the mod(mdg4) mutations (Table 1) yielded results that in many cases were opposite to those of Z3-3401. Both mnm alleles fully complemented the lethality of all mutations and deletions confined to the CR and partially or fully complemented the same mutations with respect to X–Y NDJ. However, both mnm alleles failed to complement all mutations that disrupt the MNM-specific exon, including the two deletions eGP4 and B2 that remove most of the VR exons (including the MNM-specific exon) and the three deletions that encompass the entire mod(mdg4) locus (T16, 142Δ10, and 142Δ33). In addition, both mnm alleles fully complemented mod(mdg4)T6, which affects only the Mod(mdg4)67.2-specific exon.

Thus, all of the complementation results are consistent with a pattern in which mutations or deletions that are confined to the CR complement (partially or fully) mutations or deletions that are confined to the VR. Moreover, mutations in two different variable exons complement each other. Notably, deletions and point mutations exhibited similar complementation behavior, thus ruling out the possibility that intragenic complementation results from formation of functional protein homodimers from two monomers with mutations in different domains.

MNM transcripts are generated by trans-splicing:

Intragenic complementation between mutations in the CR and in the C-terminal exon specific for the Mod(mdg4)67.2 isoform has been previously reported (Mongelard et al. 2002). Complementation in that case has been shown to be due trans-splicing between separate precursor transcripts for the CR and VR components of Mod(mdg4)67.2 expressed from opposite homologs (Labrador et al. 2001; Mongelard et al. 2002). As the common and variable regions of Mod(mdg4)67.2 are encoded on opposite genomic strands, the mature Mod(mdg4)67.2 transcript can presumably be generated only by trans-splicing. Since all MNM exons are encoded on the same strand, generation of MNM transcripts by conventional cis-splicing should be possible (see Figure 1A). However, our complementation data suggest that at least some MNM transcripts are trans-spliced and, moreover, that trans-homolog trans-splicing (THTS) (Figure 6) makes a significant contribution to the pool of MNM transcripts.

Figure 6.—

Trans-homolog trans-splicing in males trans-heterozygous for Z3-5578 and Z3-3401. (A) Possible expression patterns for components of MNM. Expression of common (black) and variable (red) exons of MNM continuously on a single precursor transcript (above each homolog) or discontinuously on two separate transcripts (below each homolog). “m” and “+” signify mutant and wild-type sites. (B) Trans- vs. cis-splicing. Both cis- and trans-splicing can generate parental-type (P) mRNAs. Only trans-splicing can generate recombinant (R) mRNAs. (C) Estimated fraction of P and R MNM RNAs in Z3-5578/Z3-3401 spermatocytes. Table gives number and frequency of each type of MNM cDNA clone from trans-heterozygous (trans-het.) (Z3-5578/Z3-3401) male testis RNA and control (an equimolar mixture of testis RNA from Z3-3401/Df(3R)GC14 and Z3-5578/Df(3R)GC14 males). Clones were obtained by RT–PCR using an MNM-specific primer pair. f(P&R), estimated frequency of P and R RNAs (see materials and methods).

To test for MNM transcripts derived from THTS, we sequenced a sample of cloned testis cDNAs from Z3-3401/Z3-5578 (a complementing genotype, Table 2) prepared by RT–PCR amplification of a segment of MNM that encompasses the sites of both mutant lesions. If splicing occurs exclusively in the cis mode, all cDNAs from these males should contain either the Z3-3401 mutation or the Z3-5578 mutation but never both (Figure 6B). However, if THTS occurs at a significant frequency, cDNAs that are wild type at both sites or mutant at both sites should be recovered. Of 137 cDNAs from Z3-3401/Z3-5578 males, 55 (40%) were either wild type at both sites or mutant at both sites, suggesting that THTS does occur at a substantial frequency (Figure 6C).

A technical complication in experiments of this type is that wild-type and doubly mutant cDNAs can arise in vitro from singly mutant transcripts by template switching during the PCR reaction (Tasic et al. 2002). To assess the frequency of such artifacts under the conditions of our experiments, we carried out a control experiment in which the RNA templates consisted of an equimolar mixture of singly mutant RNAs prepared from testes of males hemizygous for Z3-3401 or Z3-5578. In this control, wild-type or doubly mutant cDNAs can be generated only by template switching during the PCR reaction. Of 130 control cDNAs, 21 (16%) were either wild type at both sites or mutant at both sites (Figure 6C). Thus, template switching does occur at a significant frequency in this experiment and we cannot conclude that the wild-type and doubly mutant cDNAs in the trans-heterozygous sample all resulted from in vivo trans-splicing events. However, a statistical analysis shows that the difference in frequency of wild-type and doubly mutant cDNAs between experimental and control samples is highly significant (χ2 = 18.8, 1 d.f., P ≪ 0.001), thus supporting the THTS hypothesis. Taking into account the error introduced by template switching in both directions, we estimate that 35% (95% confidence interval of 27–43%) of MNM mRNAs from Z3-3401/Z3-5578 males are either wild type at both sites or doubly mutant at both sites, presumably due to THTS.

As a further test of the trans-splicing hypothesis, we evaluated the ability of a transgene [CR-7.5] located on chromosome 2 that carries a 7.5-kb genomic fragment encompassing the entire CR of mod(mdg4) but lacks MNM-specific sequences (see Figure 1A) to rescue the meiotic phenotypes of Z3-3401. This transgene partially rescues the lethality of some mutations in the CR (Buchner et al. 2000), presumably via trans-splicing between precursor transcripts for the CR and VR components of one or more Mod(mdg4) isoforms encoded by the transgene and by the native mod(mdg4) locus, respectively.

To test for complementation of Z3-3401 by the [CR-7.5] transgene, Z3-3401/Z3-3401, Z3-3401//mod(mdg4)neo129, or Z3-3401/Df(3R)T16 males (as well as control Z3-5578/Df males) with and without the transgene were tested for X–Y NDJ. In all three sets of crosses involving Z3-3401, NDJ frequencies were significantly lower in the presence of the transgene than in its absence (Table 8), whereas the transgene had no effect on NDJ frequencies in Z3-5578 males (data not shown). These results indicate that partial MNM precursor transcripts generated at distant genomic locations can trans-splice at a high-enough frequency in spermatocytes to significantly ameliorate the phenotype of a strong meiotic mutant. Presumably, the relatively weak rescue afforded by the transgene reflects decreased efficiency of trans-splicing due to spatial separation of the coding sequences for the CR and VR portions of the MNM transcript.

View this table:
TABLE 8

X–Y NDJ in Z3-3401 males with and without [CR-75]

DISCUSSION

Z3-3401 is an allele of mod(mdg4) that disrupts X–Y segregation more severely than autosomal segregation:

Our previous results demonstrated that mutations in a mod(mdg4) exon presumed to be specific for the C terminus of the MNM isoform disrupt stable conjunction and regular segregation of all four homolog pairs in male meiosis I. Here we have shown that a mutation located in the CR of mod(mdg4), which therefore should be present in all Mod(mdg4) isoforms, causes a similar spectrum of meiotic defects. Like the mnm-specific mutations, Z3-3401 disrupts the maintenance of homolog associations from midprophase I through metaphase I but has no apparent effect on the initiation of homolog pairing in early meiosis or on the stability of sister-chromatid cohesion or on any aspect of female meiosis.

However, the effects of Z3-3401 on autosomal segregation are much milder than its effects on X–Y segregation. In Z3-3401 hemizygous males, X–Y NDJ occurs at frequencies in the 40–50% range, consistent with nearly random assortment, but second and fourth chromosome NDJ frequencies are in the 10–20% range. This is not because MNM plays a less critical role in autosomal than in X–Y segregation since both mnm-specific alleles disrupt X–Y and autosomal segregation equally severely. The basis for this partial X–Y specificity is not known.

One possible explanation is that the X–Y and autosomal pairs exhibit different thresholds of sensitivity to reduction in amount of MNM protein in the nucleus, either because X–Y pairing sites need more MNM protein than autosomal pairing sites to associate stably or because MNM is loaded more efficiently on autosomal than on X–Y pairing sites when MNM is present in limiting amounts. Some functional MNM protein must be present on both autosomes and the X–Y pair in Z3-3401 spermatocytes, as indicated both by the mild autosomal segregation defect and by the fact that Z3-3401 is hypomorphic for both autosomal and X–Y segregation. Moreover, late prophase I nuclei appear to contain small amounts of Mod(mdg4) proteins, some of which may represent functional, chromosome-associated MNM. The suggestion that MNM might load more efficiently on autosomes than on the X–Y pair is consistent with our observation that MNM protein is recruited to the X–Y and autosomal pairs by different mechanisms. Both autosomal conjunction and autosomal recruitment of MNM–GFP are dependent upon tef+ function, whereas both X–Y conjunction and recruitment of MNM–GFP to the X–Y bivalent are independent of tef+ (Tomkiel et al. 2001; Thomas et al. 2005).

An alternative explanation is suggested by our finding that mutant MNM protein is present throughout prophase I in Z3-3401 spermatocytes but confined to the cytoplasm prior to the onset of prometaphase I. Although MNM and SNM are normally present in the nucleus and associated with the chromosomes and nucleoli from early prophase I on, we have no direct evidence that their presence is required for stable conjunction prior to chromosome condensation. Perhaps conjunction of at least the autosomal homologs is not stably attained until they begin condensing just prior to prometaphase I. In that case, mutant Z3-3401 protein might be able to gain access to and help conjoin the condensing chromosomes at the onset of prometaphase I when the nucleus becomes permeable to cytoplasmic proteins. The differential effect of Z3-3401 on autosomal vs. sex chromosome conjunction could reflect differential timing of the establishment of autosomal vs. X–Y conjunction or differential access of MNM and SNM to autosomal vs. X–Y pairing sites at late stages.

Do other mod(mdg4) isoforms function in meiosis?:

We have suggested that MNM might function in homolog pairing by binding specific chromosomal sites on homologous chromosomes at its C-terminal C2H2 domain and utilizing its N-terminal BTB/POZ domain to glue those sites together (Thomas et al. 2005). mod(mdg4) is thought to be capable of encoding 30 additional proteins with identical BTB/POZ domains but different chromosome localization domains. Thus an appealing possibility is that some of those additional proteins could play similar roles in homolog pairing in male or female meiosis or in somatic cells. Moreover, if Mod(mdg4) isoforms can heterodimerize—which seems likely, given that their BTB domains are identical—to form proteins with unique chromosome localization patterns, mod(mdg4) might have the potential to encode a small army of site-specific pairing proteins.

However, the data presented here provide no support for such a scenario. Analysis of a mutation in the exon specific for the abundantly expressed Mod(mdg4)67.2 isoform revealed no meiotic phenotypes in either sex. Moreover, Z3-3401, which contains an R224C substitution that should be present in all Mod(mdg4) isoforms, nevertheless exhibited no meiotic phenotypes that could not be accounted for by effects on MNM. This is not because R224 is part of a domain required only for MNM function. Z3-3401 flies are functionally mutant for at least two other Mod(mdg4) isoforms in addition to MNM: the Mod(mdg4)67.2 insulator protein and at least one unidentified isoform required for female fertility. Thus, the Z3-3401 mutation failed to reveal additional meiotic proteins encoded by mod(mdg4). However, Z3-3401 has no effect on viability whereas most mutations in the common region are recessive lethals, suggesting that some Mod(mdg4) isoforms may be unaffected by the R224C substitution. Thus, our findings do not rule out the existence of other meiotic Mod(mdg4) isoforms that are functionally unaffected by the R224C substitution [e.g., if the C-terminal exon encodes a nuclear localization sequence (NLS)]. Genetic analysis of mutations specific for other isoforms in addition to MNM and Mod(mdg4)67.2 will be needed to test definitively for the existence of other meiotic proteins encoded by mod(mdg4).

MNM is expressed predominantly by a mechanism involving trans-splicing:

At least some of the mature transcripts for Mod(mdg4) proteins are generated by an unusual mechanism in which the CR and VR exons are separately transcribed and spliced together in trans (Dorn et al. 2001; Labrador et al. 2001). For some isoforms, such as Mod(mdg4)67.2, trans-splicing is presumably essential for expression, as the common and variable regions are encoded on opposite strands. However, for most isoforms, including MNM, the CR and VR components are encoded on the same strand and the relative contributions of cis- vs. trans-splicing mechanisms in their expression have been unclear.

Our data indicate that trans-splicing is the predominant mode of expression of MNM. Complementation analysis demonstrated that mutations of all molecular types, including deletions, in the CR partially or fully complement mutations in or deletions of MNM, a pattern difficult to explain except by THTS. Sequence analysis revealed that a significant fraction, 35% ± 8%, of the mRNAs present in testis extracts from trans-heterozygous Z3-5578/Z3-3401 males were either wild type for both mutant sites or doubly mutant at both sites. We presume that all such RNAs result from THTS.

These data allow a rough estimate of the relative frequency of trans- vs. cis-splicing in the generation of MNM-encoding mRNAs. Assuming no bias for or against splicing between precursor transcripts expressed from opposite homologs vs. the same homolog, an additional 35% of transcripts should result from trans-splicing between precursors encoded on the same homolog, but would nevertheless be singly mutant and therefore indistinguishable from cis-spliced transcripts. This leads to a total estimate of the fraction of trans-spliced MNM transcripts of 70%. This figure could substantially underestimate the frequency of trans-splicing since any significant spatial separation between homologs would be expected to create a bias in favor of splicing between precursors encoded on the same homolog. Thus we conclude that a majority and perhaps all of the MNM-encoding transcripts synthesized in spermatocytes are generated by trans-splicing.

Our data provide the first example of the routine use of trans-splicing at the native mod(mdg4) locus to synthesize an isoform whose exons are all encoded on the same strand. They also extend the range of Drosophila cell types shown to be competent for trans-splicing to the male germline. Previous results have shown that embryonic and somatic isoforms of the mod(mdg4) and lola genes can be generated by trans-splicing and that somatic phenotypes controlled by these genes are subject to intragenic complementation via THTS (Dorn et al. 2001; Mongelard et al. 2002; Horiuchi et al. 2003; Gabler et al. 2005). Our data indicate that spermatogonia and/or young spermatocytes must also be competent to carry out trans-splicing.

It remains unclear how the expression of MNM is regulated. The function of MNM appears to be limited to male meiosis and MNM protein has thus far been detected only in primary spermatocytes. Our data demonstrating that MNM is expressed predominantly by trans-splicing suggest that MNM expression could be regulated either by testis-specific transcription of the C-terminal MNM exon or by regulated trans-splicing. However, even when the intronless MNM–GFP transgene is expressed from the ubiquitous hsp70 promoter under heat-shock conditions, we have failed to detect significant amounts of MNM–GFP in somatic cells of the testis or in larval salivary gland cells (M. Soltani-Bejnood, S. E. Thomas and B. D. McKee, unpublished data), suggesting that MNM protein may be unstable in cell types other than spermatocytes. Moreover, MNM cDNAs have been cloned from embryonic RNA by two different groups (Harvey et al. 1997; Buchner et al. 2000), indicating that transcription of the MNM-specific exon and splicing between the common region and MNM precursor transcripts are not exclusive to male germ cells. It will be important to determine the relative contributions of regulatory mechanisms at the level of transcription, splicing, translation, and protein stability in delimiting expression of MNM to primary spermatocytes.

The role of the Mod(mdg4) common region in MNM function:

We have previously shown that mutations in a single VR exon of mod(mdg4) disrupt homolog conjunction in male meiosis. In this study, we sought and obtained evidence that the Mod(mdg4) CR is also required for homolog conjunction. The Z3-3401 allele, which contains a missense mutation in the CR, causes meiotic phenotypes very similar to those of the mnm alleles. Moreover, males trans-heterozygous for Z3-3401 and any of 17 lethal or semilethal mod(mdg4) alleles, most of which contain lesions in the CR, exhibited similar meiotic phenotypes.

What role(s) does the mod(mdg4) CR play in homolog conjunction? We previously suggested (Thomas et al. 2005) that the BTB/POZ domain, which has been shown to mediate dimerization and multimerization (Gerasimova and Corces 1998; Gerasimova et al. 2000; Gause et al. 2001; Ghosh et al. 2001), might function to stably connect MNM-containing conjunction complexes localized to allelic sites on homologous chromosomes. Unexpectedly, immunofluoresence analysis revealed that both MNM and SMN are undetectable on condensed X–Y bivalents in Z3-3401 spermatocytes and that, although mutant MNM protein is detectable prior to chromosome condensation, the great majority of it is restricted to the cytoplasm throughout prophase I. This suggests that the Z3-3401 mutation does not disrupt conjunction directly but rather disrupts a nuclear localization or nuclear retention signal required for stable nuclear localization of MNM. Although the sequence surrounding amino acid 224 in mod(mdg4) does not fully conform to known NLS sequences, R224 is part of a two-residue cluster of basic amino acids (KR), which are critical components of classical NLS sequences (Mattaj and Englmeier 1998). It will be of interest to learn whether addition of a classical NLS to the Z3-3401 MNM protein restores nuclear localization and, if so, whether this alone suffices to restore wild-type function.

The fact that all of the lethal and semilethal mod(mdg4) alleles, including five EMS-induced alleles, fully failed to complement Z3-3401 for meiotic chromosome segregation suggests that the CR is likely to play roles in homolog conjunction in addition to nuclear localization. The most interesting allele is mod(mdg4)324, which is predicted to contain a G92D substitution. G92 lies within the BTB domain and is conserved among 32 of 44 Drosophila BTB-domain proteins, including GAGA, Tramtrack, Broad Complex, and Bric a Brac (Zollman et al. 1994; Read et al. 2000). By homology modeling, G92 is predicted to be important for formation of a β-sheet crucial for homodimerization (Read et al. 2000). Although we have not established whether the mutant MNM protein produced by mod(mdg4)324 is stable and/or localized properly in spermatocytes, a mutant GAGA protein that contains a G-to-S substitution at the equivalent residue in the BTB domain [which is highly similar to that of mod(mdg4)], as well as a second substitution at a conserved residue in the BTB domain, is expressed at normal levels in Drosophila S2 cells and localizes normally to chromosomal foci (Read et al. 2000). Thus, the mod(mdg4)324 mutation may provide a very useful tool for analysis of the role of the mod(mdg4) BTB domain in homolog conjunction.

Acknowledgments

We thank C. Zuker, B. Wakimoto, R. Dorn, V. Corces, M. Frasch, and the Drosophila Stock Center at Indiana University, Bloomington, for gifts of Drosophila stocks and antibodies. Support for this work was provided by grant no. R01 GM40489 from the National Institutes of Health to B.D.M.

Footnotes

  • Communicating editor: J. Tamkun

  • Received July 11, 2006.
  • Accepted January 30, 2007.

References

View Abstract