A Genetic Analysis of the Drosophila mcm5 Gene Defines a Domain Specifically Required for Meiotic Recombination

Abstract

Members of the minichromosome maintenance (MCM) family have pivotal roles in many biological processes. Although originally studied for their role in DNA replication, it is becoming increasingly apparent that certain members of this family are multifunctional and also play roles in transcription, cohesion, condensation, and recombination. Here we provide a genetic dissection of the mcm5 gene in Drosophila that demonstrates an unexpected function for this protein. First, we show that homozygotes for a null allele of mcm5 die as third instar larvae, apparently as a result of blocking those replication events that lead to mitotic divisions without impairing endo-reduplication. However, we have also recovered a viable and fertile allele of mcm5 (denoted mcm5^A7) that specifically impairs the meiotic recombination process. We demonstrate that the decrease in recombination observed in females homozygous for mcm5^A7 is not due to a failure to create or repair meiotically induced double strand breaks (DSBs), but rather to a failure to resolve those DSBs into meiotic crossovers. Consistent with their ability to repair meiotically induced DSBs, flies homozygous for mcm5^A7 are fully proficient in somatic DNA repair. These results strengthen the observation that members of the prereplicative complex have multiple functions and provide evidence that mcm5 plays a critical role in the meiotic recombination pathway.

THE minichromosome maintenance (MCM) family is a set of highly conserved proteins that have complex roles in many essential biological processes. As their name implies, these proteins were first identified for having a role in the maintenance of plasmids (minichromosomes) in proliferating cells of Saccharomyces cerevisiae (Maine et al. 1984). In part, they function to ensure the faithful transmission of the genome from one generation to the next; however, in addition to a critical role in DNA replication, members of this family are now thought to be multifunctional and also play roles in transcription, cohesion, condensation, and recombination (Forsburg 2004).

All MCM members belong to the AAA+ ATPase family, which has a distinct ATPase domain that spans ∼200 bases (Walker et al. 1982). This domain, referred to as the MCM box, consists of a Walker A ATPase motif, a Walker B ATPase motif, and an arginine finger motif (R-finger). Conserved sequences within the Walker B motif (IDEFDKM) and R-finger (SRDF) define the MCM family (Koonin 1993). Six of these members are conserved in all eukaryotes and form a heterohexameric complex known as Mcm2-7, which has been studied extensively for its role in DNA replication (Kearsey and Labib 1998; Tye 1999). Mcm2-7 is required for licensing and initiating origins of replication, and it acts during elongation as a helicase at the replication forks (Labib et al. 2000; Patel and Picha 2000). Because of this function and studies in yeast, Arabidopsis and Drosophila, members of the Mcm2-7 complex, are thought to be essential. In addition, two other MCM family members, Mcm8 and Mcm9, have recently been identified and are thought to be a distinct subgroup of MCM proteins (Maiorano et al. 2006). Mcm8 has been reported in vertebrates and Drosophila, but not in fungi and nematodes, and although it retains some sequence similarities in the Walker B and R-finger, its Walker A ATPase motif contains sequences more like the canonical ATPases (Gozuacik et al. 2003; Blanton et al. 2005). Mcm9 is also found in similar organisms with the exception that it is missing in Drosophila, and it is unique to the family in that it lacks the carboxy-terminal ATPase domain including the Walker B motif (Blanton et al. 2005; Lutzmann et al. 2005).

Recently, studies have indicated that in addition to the role in DNA replication, certain members within the Mcm2-7 complex, as well as other MCM family members, have functions outside of DNA replication (Forsburg 2004). Specifically, some of these functions include a role in transcriptional activation (Yankulov et al. 1999; DaFonseca et al. 2001), chromosome condensation (Christensen and Tye 2003), cohesion (Ryu et al. 2006), and recombination (Blanton et al. 2005; Shukla et al. 2005). The existence of multiple functions is consistent with studies in yeast, which showed that MCM proteins are far more abundant than would likely be required for the number of replication origins that exist, and this abundance cannot explain the fact that slight decreases in amounts of MCM proteins lead to the inability to complete S-phase and progress through the cell cycle (Liang et al. 1999; Bailis and Forsburg 2004). Moreover, in addition to the heterohexameric Mcm2-7 complex, subcomplexes of MCM family members have been identified (Su et al. 1996; Lee and Hurwitz 2000), which fuel the speculation that these complexes could be functionally distinct subgroups that possess functions beyond those involved in DNA replication.

Limited functional studies have been done on the Drosophila orthologs of the Mcm2-7 complex. Although genes for each of these members have been identified in Drosophila (Feger 1999), only mcm2 (Treisman et al. 1995), mcm4 (Feger et al. 1995), and mcm6 (Schwed et al. 2002) have been shown to be required for mitotic DNA replication. Null alleles of each inhibit proliferation of cells of the central nervous system (CNS) and imaginal discs, which leads to a reduction in brain size and lack of discs within the developing larvae. These larvae begin pupariation but never develop into adults. In addition to a role in mitotic DNA replication, two other functions for mcm6 have been identified that were not observed in mcm2 or mcm4 mutants. Mcm6 is required for endo-reduplication which is a process of reoccurring rounds of DNA replication in the absence of cell division that occurs within the developing larvae and is responsible for most of the larval growth and is also required for chorion gene amplification (Schwed et al. 2002).

Until now, there have been no genetic studies that analyzed the roles of mcm5 in Drosophila. Although it is speculated that mcm5 is required for DNA replication in Drosophila, specific functions in this process as well as other functions it may have remain unknown. The fortuitous identification of an allele of mcm5 in a screen for new meiotic mutants stimulated us to begin a thorough genetic dissection of this gene to determine the functions of mcm5 in Drosophila.

In this study we show that the mcm5 locus is essential, in that homo- or hemizygotes for a null mutation in mcm5 die prior to eclosion. They do, however, survive to third instar larvae with rudimentary imaginal discs and small brains, suggesting a defect in facilitating mitotic DNA replication. This defect in mitotically dividing cells does not extend to endo-reduplicating tissues, since the highly polytene chromosomes of the salivary gland appear normal. These findings are similar to what has been identified for mutants in mcm2 and mcm4, but differ from findings in mcm6, which is also considered essential for endo-reduplication.

In addition to the null allele of mcm5, we have also identified an EMS-induced allele that is not required for the essential functions of Mcm5, which is to say that homo- and hemizygotes for this mutant are viable and fertile, but rather has a function in the meiotic recombination pathway. We demonstrate that the decrease in recombination is not due to a failure to form either synaptonemal complex or double-strand breaks (DSBs) or to a general inability to repair DSBs that are induced by DNA damaging agents in somatic cells. This observation suggests that we have identified a residue or domain in the Drosophila mcm5 gene that is specifically required for meiotic recombination.

MATERIALS AND METHODS

Drosophila strains:

All Drosophila strains were maintained on standard food at 25° unless otherwise noted. The wild-type strain used in this study was yw eyFLP; FRT82, which was the strain used to induce the EMS mutation (Page et al. 2007). Deficiency strains used for mapping were obtained from the Bloomington Drosophila Stock Center.

Excision:

A P-element insertion mutant, KG05409, bearing a mutation between art1 and mcm5 was obtained from the Bloomington Drosophila Stock Center (BL 14003). A null allele of mcm5 was generated by imprecise excision of the P element using the Δ-2-3-mediated mobilization standard protocol. Homozygous lethal excisions were analyzed by PCR. Genomic DNA was isolated from single males that were generated by crossing yw; excision/TM3, Sb¹ virgin females to yw/Y; P{SUP or P}KG05409 males. PCR analysis of a precise excision event with primer pair 5′-AGCATCTCCTCGTGGATCCCG-3′ (located in the art1 gene) and 5′-CACCACGCGTTCACAGAG-3′ (located in the mcm5 gene) would result in an amplicon size of 1.6 kb. This primer pair would fail to amplify the chromosome containing the P{SUP or P}KG05409. PCR analysis of mcm5^exc222 resulted in an amplicon size of ∼600 bases. The excision event removed 195 bases upstream of the 5′-UTR of mcm5, the entire 5′-UTR of 137 bases, and 650 bases of the mcm5 gene, thereby creating a null allele of mcm5.

Generation of transgenes:

The genomic rescue construct was constructed by cloning a 4-kb SacII–XbaI fragment from Bac clone RP98-29B6 (BacPac Resources). This fragment, which contains the entire gene region of mcm5 plus 578 bases 5′ to mcm5 (the 5′-UTR of Art1 and 164 bases of coding region of art1) and 809 bases 3′ to the mcm5 gene, was cloned into pCasPeR4 that was previously digested with SacII and XbaI. pCasPeR4-mcm5, denoted as p{mcm5⁺}, was sequenced and then introduced into Drosophila by standard P-element-mediated transformation protocols. Three independent transformants (P{mcm5⁺} 7, 33, and 39), which all mapped to the X chromosome, were tested for the ability to complement the mcm5^A7 meiotic phenotype by analyzing nondisjunction frequency of the X chromosome.

The germline expression construct was generated by amplifying the full-length mcm5 cDNA sequence with primers 5′-GTACGGTACCATGGAAGGCTTCGACG-3′ and 5′-GTCGCGGCCGCTTAGCAAATTCGATAG-3′ from clone RE67590 (FlyBase). The PCR product was digested with KpnI and NotI and cloned into the pUASp vector (gift from Pernille Rørth; Rørth 1998) digested with the same enzymes. Sequence was verified on both strands. p{UASp-mcm5⁺} was introduced into Drosophila by standard P-element-mediated transformation protocols. Germline expression was achieved by expressing the p{UASp-mcm5⁺} construct under the control of the nos-Gal4∷VP16 driver (Rørth 1998). Virgin females of the genotype p{UASp-mcm5⁺}; mcm5^A7/TM3, Sb¹ were crossed to either p{nos-Gal4∷VP16}/y⁺Y; mcm5^A7/TM3, Sb¹ or yw/y⁺Y; mcm5^A7/TM3, Sb¹ males. Virgin females of the appropriate genotype were crossed individually to tester males and the frequency of X nondisjunction was determined (see below).

Larval analysis:

Non-GFP-expressing third instar larvae from the genotype yw/X; mcm5^exc222/TM3, P{GAL4-twi.G}2.3, P{UAS-2xEGFP}AH2.3, Sb, Ser or yw/X; mcm5^A7/TM3, P{GAL4-twi.G}2.3, P{UAS-2xEGFP}AH2.3, Sb, Ser were identified and analyzed by standard techniques. Brains and associated imaginal discs were visualized under a standard dissecting scope. Salivary gland preparations were prepared by dissection into saline solution, fixed in 45% acetic acid solution for 2 min, squashed onto Sigma-coat coverslips, and set in vapors of liquid nitrogen for 30 min. The coverslips were removed, and the slides were placed in methanol, air dried, and stained with 1.0 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) solution.

Overview of the isolation of an EMS-induced mutant in mcm5:

A genetic screen was designed to recover new meiotic mutants that did not exclude essential genes (details of screen in Page et al. 2007). Briefly, we adapted the FLP–FRT system to design a screen that required a meiotic nondisjunction event of an autosome to survive the screen. We tested each potential mutant as a germline clone for defects in meiosis. We screened 25,571 chromosome arms on 3R and isolated 10 meiotic mutants. This article describes the analysis of one mutant isolated from this screen called A7. A7 was identified as a novel meiotic mutant in that it fully complemented all known meiotic mutants on 3R, as well as all mutants isolated from the screen. After the initial isolation of mutant A7, a secondary test was done to confirm the meiotic phenotype. A7 displays a strong meiotic phenotype showing 27.5% X nondisjunction and 8% nullo fourth nondisjunction.

At the time of the screen, we were not aware that the original FRT P{ovo^D1} chromosome had an associated lethal on the 3L arm. Through the process of the FLP-induced recombination the mutant A7 stock contained this lethal, which was subsequently removed. The 3L-lethal-removed version was retested and showed similar levels of meiotic nondisjunction (data not shown), and therefore all future studies in this article were done with the lethal-removed version of A7. Upon the removal of the associated lethal, it was determined that A7 is a homozygous viable and fertile mutation.

Mapping and identification of an allele of mcm5:

The EMS-induced lesion in A7 was mapped using indel and deficiency mapping (for details see Page et al. 2007). Briefly, recombinants are made between the unmarked FRT-bearing chromosome containing A7 and a chromosome containing a distal marker element known as EP. The location of each recombination event was determined by PCR for indel markers. Selected recombinants were tested for the presence of the A7 mutation by analyzing each for a meiotic nondisjunction phenotype. Indel mapping placed the lesion between the interval 85B3–87C7. This was in agreement with mapping by deficiencies. In parallel, the lesion in A7 was mapped by testing for complementation of the meiotic nondisjunction phenotype using the 3R set of Bloomington deficiencies and additional deficiencies from the Exelixis kit. Three deficiencies [Df(3R)BSC38, Df(3R)Exel6169, and Df(3R)Exel7305] that overlap in the interval 86C6–86C7 failed to complement A7. Since the breakpoints of two of these deficiencies have been molecularly characterized, we conclude that the interval is defined by the sequence coordinates 6715093 and 6698001 on 3R (Bloomington Database). To verify that these deficiencies did not uncover a gene that was haplo-insufficient, we tested all three deficiencies that failed to complement A7 with the original FRT chromosome the mutation was induced on. All three deficiencies fully complement the original FRT chromosome, and thus no dominant effects were observed (data not shown).

Nondisjunction and recombination assays:

To measure the frequency of meiotic nondisjunction of the X and fourth chromosomes, virgin females of the listed genotype are crossed individually to attached-XY, y⁺ v f B; C(4)RM, ci ey^R males (X and fourth, Table 3) (for details see Zitron and Hawley 1989; Hawley et al. 1992; Harris et al. 2003) or y sc cv v f y⁺/B[S]Y males (X only) (Zimmering 1976; Matsubayashi and Yamamoto 2003). The frequency of recombination on the X chromosome was measured by crossing y w/y sc cv v f y⁺; FRTmcm5^A7 or y w/y sc cv v f y⁺; FRT single female virgins to y sc cv v f y⁺/B[S]Y males. Only female progeny resulting from the above cross were analyzed for the markers y sc cv v f and y⁺ (for details see Page et al. 2000; Page and Hawley 2001).

TABLE 3.

Segregation defects associated with meiotic mutant mcm5^A7

Genotype	% X ND	N
A. Germline clone assaya
Control	0	437
mcm5A7	27.5	131
B. Whole animal assayb
Control	0	1024
mcm5A7/mcm5A7	31.5	800
mcm5A7/Df(3R)BSC38	31.5	254
mcm5A7/Df(3R)Exel6159	30	479
mcm5A7/Df(3R)Exel7305	25	239
C. Transformation rescue assayc
yw; mcm5A7	31.5	800
P{mcm5+}#7; mcm5A7	0	680
P{mcm5+}#33; mcm5A7	0	450
P{mcm5+}#39; mcm5A7	0.2	1257
D. Germline specific rescue assayd
yw/pUASp-mcm5+; mcm5A7	28	536
p{nos-Gal4:VP16}/pUASp-mcm5+; mcm5A7	0.5	3083

^aGermline clones were made by crossing either control FRT82B (yw eyFLP; FRT82BEP/TM3) or mcm5^A7 (yw eyFLP; FRT82Bmcm5^A7/TM3) virgin females to males carrying an FRT82B P{ovo^D1} chromosome. At 5 days, the larval progeny of this cross are heat shocked 1 hr at 38° to induce the FLP recombinase gene expression and formation of germline clones that are homozygous for the FRT82B chromosome arm. The resulting germline clones are crossed to males with attached-XY, y⁺ v f B; C(4)RM, ci ey^R.

^bControl (FRT82B/Df(3R)BSC38), mcm5^A7 homozygous, or mcm5^A7 hemizygous virgin females were crossed to y sc cv v f y⁺/B[S]Y males.

^cVirgin females of the above genotype were crossed to y sc cv v f y⁺/B[S]Y males.

^dFemales of the above genotype were derived by crossing yw; mcm5^A7/TM3 or p{nos-Gal4:VP16}; mcm5^A7/TM3 virgin females to pUASp-mcm5⁺/y⁺Y; mcm5^A7/TM3 males. Females were crossed to y sc cv v f y⁺/B[S]Y males.

Sequencing:

Genomic DNA was prepared from a single male fly by standard protocol (Gloor et al. 1993). mcm5 gene primers used for sequencing include primer pairs 5′-TGCCCGGTAATTTTGGCG-3′ and 5′-CACCACGCGTTCACAGAG-3′, 5′-CGACTTTGTGCCACAGGG-3′ and 5′-GCTGCCAAGACCGAACAG-3′, and 5′-GCGAGAGGATGATCGTGTG-3′ and 5′-CTCCACGTCGGATCATGG-3′.

Tilling alleles:

Additional alleles were obtained through TILLING of mcm5 at the Fred Hutchinson Cancer Research Center (FlyTILL). Mutations were verified by sequencing the mcm5 gene with the above primer pairs (data not shown). The following Zucker stocks were obtained for this study: Z3-PMM-0509 (I471I), Z3-2022 (R491Q, SIFT score 0.01), Z3-2146 (intron), Z3-3224 (M513I, SIFT score 0.15), Z3-PMM-0153 (E427E), Z3-5239 (R344R), Z3-5602 (E550E), Z3-1138 (E498K, SIFT score 0.22), Z3-0852 (P545P), Z3-1334 (S689L, SIFT score 0.00), Z3-3278 (M431I, SIFT score 0.02), Z3-4809 (F630I), and Z3-2156 (F630I).

Western blotting:

Females of the appropriate genotype were mated with males and fed yeast for 3 days. Ovaries were dissected in PBS–Triton (PBS + 0.1% Triton-X-100). Twenty ovaries were homogenized in 50 μl of 1× sample loading buffer (50 mm Tris-HCl pH 6.8, 100 mm DTT, 2% SDS, 0.1% bromophenol blue, 10% glycerol) by a small pestle. The extract was heated to 100° for 5 min and 20 μl of each genotype were loaded on a 10% SDS–PAGE gel. Protein was transferred onto PVDF membrane, blocked in PBS–Tween-20 (PBS + 0.1% Tween-20) containing 4% dry powdered milk, and probed with primary antibodies overnight at 4°. The membrane was washed in PBS–Tween-20 and probed with secondary antibody for 2 hr at room temperature. The washed membranes were reacted with alkaline phosphatase-conjugated goat anti-mouse or anti-rabbit antibody. The bound antibodies were detected by reacting with substrate solution containing 50 μg 5-bromo-4-chloro-indolyl-phosphate/ml and 100 μg 4-Nitro Blue Tetrazolium chloride/ml of rinse buffer (100 mm Tris pH 8.8, 1 mm MgCl₂). All Blue Precision Plus Protein Standards were used as a marker (Bio-Rad, Hercules, CA).

Immunostaining:

Approximately 30–40 females, which had eclosed 2–3 days previously, were mated to 10–15 males and fed on yeast for 2–3 days prior to egg chamber dissection. These females were anesthetized and the abdomens were ruptured one by one with forceps. For early egg chambers, oocytes were fixed using previously published methods (Page and Hawley 2001) with minor exceptions. Basically ovaries were dissected in PBS and immediately fixed for 20 min in 200 μl of PBS containing 2% formaldehyde (Ted Pella) and 0.5% Nonidet P-40 plus 600 μl heptane. Fixed ovaries were washed three times for 15 min each in PBS–Tween-20. Ovarioles were teased apart with forceps and blocked for 1 hr in PBS–Tween-20 containing 1% bovine serum albumin (BSA) (Calbiochem, La Jolla, CA) at room temperature. Primary antibody was incubated overnight at 4° and washed as before, and secondary antibodies were applied for 4 hr at room temperature. Ten minutes prior to washing a final concentration of 1.0 μg/ml DAPI was added. Ovarioles were washed as before and mounted in ProLong Gold (Invitrogen, Carlsbad, CA). Late-stage oocytes were fixed as previously described (Gilliland et al. 2005). Whole ovaries were dissected in 1× Robb's media (55 mm sodium acetate, 8 mm potassium acetate, 20 mm sucrose, 2 mm glucose, 0.44 mm MgCl₂, 0.1 mm CaCl₂, and 20 mm HEPES, pH 7.4) containing 1% BSA and individually ovarioles were teased apart with forceps. Ovaries were fixed in solution containing 1× fix buffer (100 mm potassium cacodylate, 100 mm sucrose, 40 mm sodium acetate,and 10 mm EGTA) and 8% formaldehyde (Ted Pella) for 5 min. Ovaries were washed in PBS–Triton three times for 15 min each, vitelline membrane was removed by rolling ovaries between frosted slides, and then ovaries were blocked 1 hr in PBS–Triton containing 5% normal goat serum. Antibody labeling was done as described above except washes were done in PBS–Triton.

Microscopy was conducted using a DeltaVision microscopy system (Applied Precision, Issaquah, WA) equipped with an Olympus 1× 70 inverted microscope and high-resolution CCD camera. Images were deconvolved using the SoftWoRx v.25 software (Applied Precision). Images are shown as maximum-intensity projections of the complete germarium, or as a subset of sections, or as a single deconvolved section (noted in each figure legend). His2Av foci were counted by manually examining all optical sections of anti-His2Av staining nuclei.

Antibodies:

Mouse anti-Orb antibodies (4H8 and 6H4) obtained from the Iowa Hybridoma Bank (Lantz and Schedl 1994) were used together at a 1:30 dilution. Mouse anti-C(3)G antibody (Anderson et al. 2005) was used at a dilution of 1:500. Guinea pig anti-C(3)G antibody was used at 1:500 dilution (Page and Hawley 2001). Rabbit anti-His2Av (gift from Kim McKim; Mehrotra and McKim 2006) was used at 1:500. Rabbit anti-Mcm5 (gift from Paul O'Farrell; Su et al. 1996) was used at a dilution of 1:666. Rat anti-tubulin antibodies, MAS078P (Harlan Ser-Lab) and MAS1864 (Chemicon), were used together at a dilution of 1:250. Mouse anti-α tubulin antibody (Sigma, St. Louis) was used for Western blotting at 1:1500. Secondary goat anti-mouse or rabbit Alexa-488 and Alexa-555 conjugated antibodies (Molecular Probes, Eugene, OR) were used at 1:500, and donkey anti-rat Cy3 conjugated antibody was used at 1:200 (Jackson ImmunoReseach). Anti-rabbit and anti-mouse alkaline phosphatase-conjugated antibodies (Sigma) were used at 1:5000.

MMS and X-ray treatment:

Sensitivity to different DNA-damaging agents was determined for two developmental stages of Drosophila. Two- to 3-day-old virgin yw; mcm5^A7/TM3 Sb¹ females were mated to yw/y⁺Y; mcm5^A7/TM3 Sb¹ males. For chemical treatment, flies were set up in standard food vials (5 females and 5 males per vial testing 5 vials per concentration), and for X-ray treatment flies were set up in cages with grape plates (100 females and 30 males), and each allowed to lay eggs for 24 hr. After an additional 24 or 48 hr, the embryos or larvae were treated with various concentrations of 0.25 ml methyl methanesulfonate (MMS) (Sigma) in water or X-rayed with an ∼1000-R dose (shelf 6, 115 kV, and 5 mA for 2 min) (Faxitron Cabinet X-ray Systems). Untreated larvae of the same parents were used as controls. After X-ray treatment, embryos or larvae from grape plates were transferred to bottles containing standard food. The offspring were analyzed starting at day 10 and continued to be analyzed until day 18. Zero, 0.02%, and 0.04% MMS-treated embryos and larvae began to hatch at day 10; 0.08% on day 11; and 0.16% on day 12. In both treatments, the eclosed adult flies were counted to determine viability. In controls, the relative viability of homozygotes is 33% of total viable progeny.

RESULTS

Isolation and characterization of a null allele of mcm5:

In all organisms for which loss-of-function mutants are available for members of the Mcm2-7 complex, studies indicate that each of these members is essential for survival due to its role in DNA replication. The fact that we uncovered a mutant in mcm5 from a meiosis-specific screen that is both viable and fertile suggests either that we have uncovered a separation-of-function allele of mcm5 that is primarily meiosis-specific or that the mcm5 gene product is not essential in Drosophila. To distinguish between these two possibilities, we created a null allele of mcm5 (mcm5^exc222) by imprecise excision of a neighboring P element (Figure 1). The mcm5^exc222 excision event removes the 5′ regulatory elements and 650 bp of the gene and is therefore a null allele of mcm5. The null allele was lethal over three deficiencies that uncover mcm5 (Table 1). In addition, the lethality of the null allele could be rescued when a transgene carrying a wild-type copy of mcm5 was present (Figure 1 and Table 1). This indicates that mcm5 is an essential gene, presumably due to the role it plays in mitotic DNA replication.

Figure 1.—

P-element excision to create a null allele of mcm5. The null allele was generated by imprecise excision of P{KG05409}. The P element is located between art1 and mcm5. mcm5^exc222 is deleted for 982 bases (332 bases upstream of mcm5 and 650 bases of mcm5), creating a null allele. The region used to generate the genomic rescue construct P{mcm5⁺}.

TABLE 1.

mcm5 is an essential gene

Genotype	Presence of Mcm5	Viability
Lethal phenotype of the null allele
mcm5exc222/Df(3R)BSC38	−	−
mcm5exc222/Df(3R)Exel6159	−	−
mcm5exc222/Df(3R)Exel7305	−	−
Transformation rescue of the null allele
P{mcm5+}#7; mcm5exc222	+	+
P{mcm5+}#33; mcm5exc222	+	+
P{mcm5+}#39; mcm5exc222	+	+

To determine whether mcm5 plays a role in mitotic DNA replication and/or endo-reduplication, we further analyzed the null allele. Similar to what has been shown in Drosophila mcm2, mcm4, and mcm6 mutants, the null allele of mcm5 displays a larval phenotype consistent with a role in mitotic DNA replication (Table 2). Specifically, homozygotes survive to third instar larval stage and are of wild-type size but never finish pupariation and develop into adults. These larvae have small brains and no identifiable imaginal discs, indicating that the mitotic DNA replication cycles required for embryonic development are impaired. We did not observe any of the defects in the ability of homozygotes to complete endo-reduplication of the salivary gland polytene chromosomes, which has been identified as a function for Mcm6 (Table 2 and Figure 2). Homozygotes for the null mutation developed normal size salivary glands that contained banded polytene chromosomes. These data indicate that mcm5 in Drosophila is an essential gene required for mitotic DNA replication, but dispensable for endo-reduplication.

TABLE 2.

Developmental defects in mcm5 mutants

	Adults		Third instar larval phenotype
Genotype	Viability	Fertility	CNS	Imaginal discs	Salivary glands/ polytene chromosomes
mcm5exc222	Lethal	ND	Small	Absent	Normal
mcm5A7	Viable	Fertile	Normal	Normal	Normal

Figure 2.—

Polytene chromosomes from salivary glands. (A) control, (B) mcm5^exc222, and (C) mcm5^A7. All parts are a single deconvolved optical section. Bar, 20 μm.

mcm5^A7, a viable and fertile allele of mcm5 that exhibits high levels of meiotic nondisjunction:

As noted in the Introduction we first identified an allele of mcm5 in a screen for meiotic mutants. This screen is described in detail in Page et al. (2007), but briefly the right arms of third chromosomes mutagenized with EMS were made homozygous in the female germline using FLP/FRT-mediated recombination. Females carrying such germline clones were then mated to appropriate tester males to screen for those mutants that caused high levels of autosomal nondisjunction. As shown in Table 3A, females bearing germline clones homozygous for one such mutagenized third chromosome generated ∼30% X chromosome nondisjunction. Mapping of the mutant, denoted mcm5^A7, indicates the gene responsible for this elevated level of nondisjunction is mcm5. Indeed, the mcm5^A7 allele results from a single A → T transversion (A2081T) in the first codon of the last exon that corresponds to an aspartic acid to valine change (D694V) (Figure 3A). This acidic residue, which is located in the C-terminal region outside of the conserved MCM box, is highly conserved from yeast to humans (Figure 3B).

Figure 3.—

(A) Schematic diagram of the mcm5 gene on 3R and the location of the A7 lesion. Solid boxes depict the four exons. Asterisks denote the position of the mutation and underlines denote the first codon of exon 4 and the corresponding D → V codon change in A7. Intron sequence is in italics. (B) Alignment of mcm5 genes from various organisms in the region mutated in A7 showing the conservation of the aspartic acid residue (*). (C) Schematic diagram of the Mcm5 protein showing point mutations generated from the Fly-TILLING project. The solid box depicts the ATPase domain containing the Walker A (WA), Walker B (WB), and arginine finger (RF) motifs. The asterisk denotes the A7 allele (D694V).

Males and females homozygous for the mcm5^A7 mutant are viable and fertile and show no developmental defects that are observed in the null allele (see Table 2). mcm5^A7 homozygotes or mcm5^A7/Df heterozygotes demonstrate levels of X chromosome nondisjunction similar to those observed in females bearing germline clones homozygous for mcm5^A7 (Table 3B). To verify that the disjunction defects observed were indeed the consequence of this mutation, we rescued the disjunction defect using a transgene construct carrying a wild-type allele of mcm5 (Table 3C and see Figure 1). Three independent transgenes that express Mcm5 from the genomic promoter fully suppressed the phenotype of mcm5^A7. In addition, the germline expression construct was able to rescue the nondisjunction phenotype only when the germline driver was present (Table 3D).

We also obtained 12 additional point mutations in both the MCM box and C-terminal region in mcm5 from the Seattle Fly-TILLING project (Figure 3C). Of these, 5 were noncoding, 1 was in an intron, and 6 resulted in amino acid substitutions. Of these 6, 3 were predicted to cause deleterious changes on the basis of SIFT scores of <0.05. However, none of these additional alleles was lethal over the deficiency Df(3R)BSC38, which uncovers the mcm5 gene or showed a meiotic phenotype when analyzed over this deficiency (data not shown).

Defects observed in mcm5^A7 are not due to reduced levels of Mcm5 protein:

One possible explanation for the lack of an effect of the mutant on viability might be that the process of oogenesis requires higher levels of Mcm5 protein than do the mitotic divisions required for growth and viability, and that the mcm5^A7 reduced the level of Mcm5 protein in the ovaries below some threshold required for meiotic function, but not to a sufficient degree to be lethal. To verify that the ovaries homozygous for the mcm5^A7 mutant expressed a Mcm5 protein of predicted size and at wild-type levels, we compared the expression of Mcm5 from mcm5^A7 and control ovaries. By Western blotting, it was confirmed that ovaries from females homozygous for mcm5^A7 produced levels of Mcm5 protein that were similar or identical to wild-type protein (Figure 4).

Figure 4.—

Expression of Mcm5 in mcm5^A7 and wild-type ovaries. Ovarian protein extracts from mcm5^A7 or wild-type females were prepared and analyzed by Western blotting. Equal amounts of each extract were loaded on a 10% SDS–PAGE gel and transferred to PVDF membrane. The membrane was cut horizontally under the 75-kDa protein marker (M). The upper half was probed with antibody to Mcm5 and the lower half was probed with antibody to α-Tubulin (Tub). Predicted molecular weight for Mcm5 is 82 kDa and 50 kDa for α-Tubulin.

The high levels of meiotic nondisjunction observed in mcm5^A7 females are a consequence of a 10-fold decrease in the frequency of meiotic recombination:

Most meiotic mutants that are recovered on the basis of their ability to induce high levels of nondisjunction in fact define gene products that are required for normal levels of meiotic recombination (Sandler et al. 1968; Baker and Carpenter 1972; Sekelsky et al. 1999). This reflects the fact that normal levels of meiotic recombination are required for proper chromosome segregation in Drosophila (for review see Baker and Hall 1976). To test whether or not the mcm5^A7 mutant was in fact recombination deficient, we measured recombination along the length of the X chromosome in both mcm5^A7 mutant and control females. As shown in Table 4, the mcm5^A7 mutant reduces the level of meiotic recombination as demonstrated by both a 10-fold decrease in the absolute frequency of recombination along the entire X chromosome (total map length), when compared to the wild-type control, and by a >10-fold increase in the fraction of oocytes in which the X chromosomes failed to undergo exchange (the E_o frequency).

TABLE 4.

Frequency of crossing over on the X chromosome

	sc-cv	cv-v	v-f	f-y+	Total map length(cM)	Frequency of oocytes that did not undergo X exchange (Eo)	N
Wild type	14	23	20	12	69	0.066	649
mcm5A7	1	1.3	1.5	2.5	6.9	0.883	875
% wild type	7.1	5.6	7.5	20.8	10	—	—

In addition to reducing the absolute frequency of meiotic exchange, the mcm5^A7 mutant also alters the distribution of exchange as is evident by looking at the residual exchanges. The reduction in exchange is most severe in the two regions (cv-v and v-f) that show the highest levels of exchange in control females. Similar patterns of exchange reduction are also observed in females homozygous for either of two other meiotic mutants, rec/mcm8 and mei-218, both of which define genes that encode MCM or MCM domain-containing proteins (Matsubayashi and Yamamoto 2003; Blanton et al. 2005; J. Sekelsky and K. McKim, personal communication). The level of exchange reduction observed is fully sufficient to account for the levels of X chromosome nondisjunction observed in our assays (Baker and Hall 1976).

Oocytes homozygous for the mcm5^A7 mutant are proficient in synapsis:

In Drosophila, formation of the synaptonemal complex (SC), a structure that is required for synapsis and meiotic recombination, occurs only after premeiotic S-phase is complete (Grell 1984). The transverse filament protein, C(3)G (the yeast Zip1 homolog), is a marker often used for visualization of SC structure. Previous localization studies by Page and Hawley (2001) show that C(3)G is first present within multiple cells of the germline cyst in region 2A. This staining is reduced to two adjacent cells in region 2B and is lost by region 3 in all cells but the oocyte, which can be detected by cytoplasmic Orb staining (Lantz and Schedl 1994). As the oocyte enters the vitellarium stage (stage 2) the C(3)G thread-like staining persists early, gradually breaking down and most is lost by stage 6. Comparison of the localization of C(3)G in mcm5^A7 to previous studies and to FRT controls (data not shown) indicates that there is no observable defect in the timing of C(3)G expression, the ability to form SC, or in the SC structure by immunocytology (Figure 5), which indicates the mutant is able to complete synapsis.

Figure 5.—

Structure and timing of synaptonemal complex formation in mcm5^A7. (A) Maximum-intensity projection of all optical sections through a mcm5^A7 germarium stained with DAPI (blue) and antibodies to C(3)G (red) and Orb (green). C(3)G is expressed in multiple cells in region 2A, reduced to two cells in region 2B, concentrated in the pro-oocyte (identified by cytoplasmic Orb staining) in region 3, and is thread-like as it enters the vitellarium stage 2 (S2). Bar, 30 μm. (B) Maximum-intensity projection of all optical sections through the nurse cells and oocyte in a cyst in region 2B stained DAPI (blue) and antibodies to C(3)G (red) and Orb (green). Bar, 10 μm. (C) Same cysts as in B showing only the C(3)G staining. Two adjacent cells within the cyst show the predicted thread-like pattern of SC.

Oocytes homozygous for the mcm5^A7 mutant are proficient in the production of meiotic DSBs but cannot resolve those DSBs into functional chiasmata:

To determine whether the reduction in recombination observed in mcm5^A7 mutant oocytes was due to the inability to form and/or repair double-strand breaks (DSBs), which are formed prior to meiotic recombination, we analyzed DSB formation in mcm5^A7. In Drosophila, DSBs can be detected with an antibody that recognizes a phosphorylated histone variant γHIS2Av (gift from Kim McKim; Mehrotra and McKim 2006). Using this antibody, McKim and colleagues show that DSBs first appear after SC formation in region 2A where the highest numbers of foci are detected. Most His2Av foci disappear by region 2B where only a few foci are occasionally detected, and all foci are gone by region 3, presumably as repair is being completed. On the basis of His2Av staining pattern, we found no significant difference in the number of DSBs formed in mcm5^A7 compared to control (in region 2A, average number DSBs in mcm5^A7 is 10.1, N = 22 and 9.7, N = 13 in controls) or in the ability to repair breaks by region 3 that had formed in region 2A (Figure 6).

Figure 6.—

Formation and repair of meiotic DSBs in mcm5^A7. Maximum-intensity projection of optical sections through the C(3)G staining of cells labeled a, b, and c in a germarium from mcm5^A7 stained with DAPI (blue) and antibodies to C(3)G (red) and His2Av (green). Within the germarium maximum-intensity projections of all optical sections through three cells labeled a (region 2A), b (region 2B), and c (region 3) are shown at higher magnification at the right. C(3)G and His2Av are shown in gray scale at the higher magnification. His2Av foci are abundant in region 2A (a), most disappear by region 2B (b), and none are detected in region 3 (c). Bar, 15 μm.

Consistent with our observation that DSBs are being repaired, mcm5^A7 females do not display the phenotypic characteristics of other mutants that are defective in repair of meiotic DSBs. Mutants in genes required for DSB repair are sterile and lay eggs with specific patterning defects, have abnormal chromatin condensation, and arrest in prophase I (Ghabrial et al. 1998). None of these phenotypes are observed in mcm5^A7. This argues strongly that the defect induced by the mcm5^A7 mutant cannot be ascribed to a simple failure to repair the DSBs, which suggests that the DSBs are likely repaired by either gene conversion or sister chromatid exchange (see discussion).

Although the DSBs created in mcm5^A7 mutant oocytes are not left unrepaired, they produce significantly reduced levels of crossovers and chiasmata (the physical manifestation of exchange). Homozygotes for mutants that greatly reduce the level of meiotic recombination usually also display defects in the ability to arrest meiotic progression at metaphase I (McKim et al. 1993). This failure to arrest reflects a requirement for at least one chiasmata, the physical manifestation of an exchange event, to hold at least a pair of homologs together at the midspindle and thus create a metaphase plate. Accordingly, we analyzed meiotic spindles from both mcm5^A7 and wild-type ovaries to determine if there was a failure to arrest at metaphase I. Although we did not observe any defects in prometaphase figures (Figure 7, compare A and B to E and F), we did observe a failure to arrest at metaphase I in 42% (n = 54) of mcm5^A7 oocytes (Figure 7, compare C and D to G and H). Indeed, the predicted frequency of cells without a chiasmata on all five chromosome arms in mcm5^A7 (E₀ frequency of 0.88) is 52%, and thus we would expect this high frequency of a failure to arrest at metaphase I (McKim et al. 1993). In addition to the anaphase I-like figures seen in Figure 7, we also less commonly observed (11%) figures that contained greater than two spindles.

Figure 7.—

Prometaphase and metaphase figures from wild-type (A–D) and mcm5^A7 (E–H) oocytes. Images are maximum-intensity projections of the optical sections through the tubulin-staining region. Meiotic figures are stained with DAPI (blue) and tubulin (red) in A, C, E, and G. Corresponding DNA is shown in gray scale to the right. A, B, E, and F represent prometaphase figures, and C, D, G, and H represent metaphase I figures. Bar, 10 μm.

The mcm5^A7 mutant is proficient in somatic DNA repair:

Many genes involved in meiotic recombination are also required for the repair of mitotically damaged DNA (Baker et al. 1978). To determine whether mcm5^A7 is required for somatic DNA repair, we analyzed the ability to repair DNA lesions at two different time points in development that were induced by two classes of mutagens. X rays were used to induce DSBs and methyl methanesulfonate (MMS) was used to create lesions that are repaired by the base excision repair pathway (Lindahl and Wood 1999). As shown in Table 5, there was no significant difference in the relative rates of viability of mcm5^A7 homozygotes compared to heterozygous siblings at either stage of development. A similar effect was seen when lesions were produced by X rays (Table 6). To verify that the ability to repair breaks resulting from DNA damaging agents was not due to the maternal loading of Mcm5 from the heterozygous mothers, MMS treatment was performed on embryos produced from homozygous mcm5^A7 females and males at both time points. There was no difference in number of progeny produced from mock and 0.8% MMS-treated embryos at either time point (data not shown), indicating that the inability to repair DSBs into crossover products is due to a defect in the meiotic recombination pathway and not to a general inability to repair DNA lesions. Therefore, it appears that all three MCM-containing genes (mei-218, rec/mcm8, and mcm5) that show reduced levels of meiotic exchange are all proficient in somatic DNA repair (Baker et al. 1978; Matsubayashi and Yamamoto 2003; Blanton et al. 2005).

TABLE 5.

MMS sensitivity in mcm5^A7

	No. homozygotes/total no. progeny		% relative viabilitya
% MMS	24–48 hr old	48–72 hr old	24–48 hr old	48–72 hr old
0	69/193	176/549	36	32
0.02	74/220	211/641	33	33
0.04	61/166	196/560	36	35
0.08	69/229	171/548	30	31
0.16	15/47	26/435	32	29

^aThe expected relative viability of homozygotes in the control is 33% of total number of progeny.

TABLE 6.

X-ray sensitivity in mcm5^A7

	No. homozygotes/total no. progeny		% relative viabilitya
X-ray dose	24–48 hr old	48–72 hr old	24–48 hr old	48–72 hr old
1000 R	98/370	195/719	26.5	27

^aThe relative viability of homozygotes in the control is 33% of total number of progeny.

DISCUSSION

Mcm5 is a component of the Mcm2-7 complex that has been shown to be required for initiating replication origins during S-phase of the mitotic cell cycle in most organisms; however, until now, no detailed studies have been done on the Drosophila mcm5 ortholog. The genetic analysis of the mcm5 gene in this study allowed us to determine that the mcm5 gene product has two distinguishable functions in Drosophila: an essential function in mitosis and a specialized function in the meiotic recombination pathway. We have shown that like other MCM family members in Drosophila, mcm5 has a phenotype consistent with a role in mitotic DNA replication. However, it is not required for the processive DNA replication cycles that occur during endo-reduplication in the salivary glands of the developing embryo.

We have also identified a function of Mcm5 in the maturation of DSBs into crossovers and shown that this meiotic function is separable from the role of Mcm5 in mitotic replication. We have identified a residue or domain in Mcm5 that is specifically required for this meiotic function. The observation that mcm5^A7 homozygotes and mcm5^A7/Df heterozygotes demonstrate similar levels of X chromosome nondisjunction argues that the mutant constitutes a null allele with respect to the role of this protein in meiotic recombination, while the fact that this mutant so strongly affects meiosis without affecting viability demonstrates that the mcm5^A7 mutant is a clear separation-of-function allele in terms of the role of the Mcm5 protein in mitosis and meiosis.

Recombination-deficient mutants in Drosophila can coarsely be grouped into four classes: those like mei-W68 (which encodes the fly homolog of SPO11) that block the formation of DSBs, those like mei-9 and mus312 that are involved in the resolution of recombination intermediates, those like spn-A and spn-B that are involved in the repair of DSBs, and a class of mutants, often referred to as precondition mutants, that appear to simply alter the probability that DSBs will be processed into crossover events (Carpenter and Sandler 1974; Lindsley and Sandler 1977; Bhagat et al. 2004). Precondition mutants are characterized by the fact that they not only decrease the total number of exchange events, but also alter the mechanisms that normally control the distribution of exchanges, such that exchanges occur more commonly in proximal regions than in distal regions (Baker and Hall 1976; Lindsley and Sandler 1977). They also usually ablate crossover interference, the process that serves to distribute crossover events along the arms of chromosomes. The mcm5^A7 mutant, as well as mutants in the rec/mcm8 and mei-218 genes, are all precondition mutants. As shown in Figure 8, the three genes defined by these mutants encode either bona fide MCM proteins (Mcm5 or Mcm8) or a protein with a MCM domain (Mei-218) (J. Sekelsky and K. McKim, personal communications).

Figure 8.—

Drosophila meiotic mutants that contain MCM domains. MCM domains were identified using the Conserved Domain Search service (CD-Search) (Marchler-Bauer and Bryant 2004).

We have shown that DSBs are both created and disappear with normal kinetics in Drosophila mcm5^A7 oocytes. The fertility of mcm5^A7 homozygotes, the absence of the so-called “spn” phenotype, which is exhibited by mutants that are defective in the repair of DSBs, and the absence of chromosome fragments during prometaphase (see Figure 7) all argue strongly that these breaks are repaired, but not in a fashion that generates crossover. It is tempting to suggest, as Blanton et al. (2005) have done for mutants in the rec/mcm8 gene, that MCM proteins are required for the processive DNA synthesis that is necessary for the formation of a recombination intermediate (Szostak et al. 1983; Cromie et al. 2006), and thus their absence prevents the maturation of a DSB into a mature crossover. Such speculation is bolstered by the observations by Carpenter (1982, 1989) and by Blanton et al. (2005) that while conversion events were at least as frequent in mei-218 and rec/mcm8 oocytes as they were in wild type, the conversion tracts themselves were shorter than observed in wild type. Thus, as suggested by Blanton et al. (2005), perhaps the ability of the oocyte to extend a 3′ end in the process of initiating recombination is not sufficient to stabilize a recombination intermediate, but rather “falls back” to creating a conversion event by a synthesis-dependent strand annealing (SDSA)-like mechanism. This would be plausible if the mcm5^A7 mutation, which is in the C-terminal conserved domain, is required for an interaction with machinery involved specifically in processive DNA replication in meiosis. There are, however, at least two problems with this explanation. First, a further analysis by Curtis and Bender (1991) of the mei-218 conversion events recovered by Carpenter failed to confirm an alteration in tract length, and second, at least for mcm5, the ability of mcm5 null alleles to still properly perform endo-reduplicative DNA synthesis, producing normally appearing polytene salivary gland chromosomes, argues against a requirement for at least this protein in general processive DNA synthesis.

So then how might these mutants suppress crossing over by a mechanism that is unrelated to their traditional roles in replication? One possibility is that at least Mcm5 is known to regulate transcription via a physical interaction with Stat1 (Zhang et al. 1998; Snyder et al. 2005). Thus it is at least possible, even though the mcm5^A7 mutation lies well outside the putative Stat1 interacting domain of Mcm5 (DaFonseca et al. 2001), that the change created by this mutation impairs the interaction of Mcm5 with Stat1 or some other transcriptional regulator, and in doing so prevents the expression of one or more genes that function in the maturation of DSBs to crossovers.

However, on the basis of a recent finding by Gasser and her collaborators (Shimada and Gasser 2007) that origin recognition complex proteins in yeast function in the process of establishing and maintaining sister chromatid cohesion, in a fashion that is independent of their role in replication initiation, we propose that MCM proteins in flies might play a similar role in meiosis. We imagine that like the fly Ord and C(2)M proteins, which are thought to be involved in conversion of the cohesion complex into the lateral elements of the synaptonemal complex (for review of this process see Page and Hawley 2004), the Mcm5, Rec, and Mei-218 proteins also play a role in the function of axial and/or lateral elements and that it is this defect, rather than a problem in replication per se, that underlies their meiotic defects.

We should note that this adaptation of the Mcm5 protein for a meiotic function may not be universal. Forsburg and her collaborators have created the corresponding mutation to that found in mcm5^A7 in S. pombe (see Figure 2B) and failed to observe any defect in meiotic recombination (S. Forsburg, personal communication). This may reflect the rather unusual constellation of repair and recombination proteins found in Drosophila (Sekelsky et al. 2000). Notably lacking from the fly genome are obvious homologs of the Dmc1 protein, which at least in other organisms is required to promote interhomolog exchange events and suppress sister chromatid exchange events. Although Abdu et al. (2003) have proposed that the missing Dmc1 function might be provided by fly Rad51 homologs, Spn-B and Spn-D, one could imagine that the MCM proteins also play such a role in flies, and thus measurement of meiotic sister chromatid exchange in these mutants would be of real interest. Alternatively, it has recently been demonstrated that the mechanism of recombination in S. pombe is fundamentally different from the double Holliday junction mechanism that prevails in S. cerevisiae (Bishop 2006; Cromie et al. 2006), leaving the possibility open either that flies are more like the budding yeast in their mechanism of recombination (in a fashion that makes Mcm5 nonessential for recombination in S. pombe) or that perhaps there are even more than two variations on a theme with respect to the process of meiotic recombination, such that flies have their own unique set patterns of nucleic acid needlework with which to perform crossing over.

Finally, it is worth noting that Shima et al. (2007) have recently identified a hypomorphic viable allele of mcm4 that causes chromosome instability and identified a unique function of this protein in tumor suppression in mice. Thus, it is obvious that core MCM proteins play roles outside of DNA replication and that the identification of separation-of-function mutants is going to be essential in elucidating the multiple roles of MCM proteins.