Three structure-selective endonucleases are essential in the absence of BLM helicase in Drosophila

Abstract

DNA repair mechanisms in mitotically proliferating cells avoid generating crossovers, which can contribute to genome instability. Most models for the production of crossovers involve an intermediate with one or more four-stranded Holliday junctions (HJs), which are resolved into duplex molecules through cleavage by specialized endonucleases. In vitro studies have implicated three nuclear enzymes in HJ resolution: MUS81-EME1/Mms4, GEN1/Yen1, and SLX4-SLX1. The Bloom syndrome helicase, BLM, plays key roles in preventing mitotic crossover, either by blocking the formation of HJ intermediates or by removing HJs without cleavage. Saccharomyces cerevisiae mutants that lack Sgs1 (the BLM ortholog) and either Mus81-Mms4 or Slx4-Slx1 are inviable, but mutants that lack Sgs1 and Yen1 are viable. The current view is that Yen1 serves primarily as a backup to Mus81-Mms4. Previous studies with Drosophila melanogaster showed that, as in yeast, loss of both DmBLM and MUS81 or MUS312 (the ortholog of SLX4) is lethal. We have now recovered and analyzed mutations in Drosophila Gen. As in yeast, there is some redundancy between Gen and mus81; however, in contrast to the case in yeast, GEN plays a more predominant role in responding to DNA damage than MUS81-MMS4. Furthermore, loss of DmBLM and GEN leads to lethality early in development. We present a comparison of phenotypes occurring in double mutants that lack DmBLM and either MUS81, GEN, or MUS312, including chromosome instability and deficiencies in cell proliferation. Our studies of synthetic lethality provide insights into the multiple functions of DmBLM and how various endonucleases may function when DmBLM is absent.

Figure 1. Apoptosis in larval imaginal discs.

Apoptosis levels are expressed as the average number of cells per imaginal wing disc that stain with antibody to cleaved caspase-3. n = number of discs scored. mus81 and Gen mutant larvae had apoptosis levels indistinguishable from the wild-type control (y w), but mus81 Gen double mutants had significantly increased apoptosis compared to either wild-type or either of the single mutants. *** p<0.0001 (Fisher's exact test).

Figure 2. Lethal stages of various mutants.

The Drosophila life cycle is illustrated, with the lethal stages of different genotypes indicated. The arrows at the bottom left are intended to signify diminishing contribution of maternally-deposited protein. In our crosses, there is no maternal MUS81, but there is half the normal amount of maternal DmBLM, MUS312, and GEN; the developmental stage to which this maternal protein perdures is unknown, and may be different in different tissues.

Figure 3. Nuclear defects in mutants.

(A-B) DAPI-stained metaphase neuroblast nuclei. Normal wild-type nuclei (A) contain one pair of sex chromosomes, two pairs of large autosomes, and one pair of small autosomes. Most nuclei of single mutants for mus81, mus309, Gen, or mus312 are normal (see Figure 4). (B) An example of a cell from a mus312 mus309 mutant, illustrated polyploidy and broken chromosomes (arrowhead). (C-F) DAPI-stained larval salivary gland nuclei. Arrows point to polyploid nuclei; arrowheads point to diploid imaginal nuclei. Compared to wild-type (C) and mus309 (D) and mus312 (E) single mutants, diploid imaginal ring cell nuclei are reduced in number and enlarged in size in mus312 mus309 double mutants (F) and in Gen mus309 spn-A (not shown). Endocycling polytene cells are similar in all genotypes shown.

Figure 4. Chromosome breaks and polyploidy in mutants.

A. Fraction of chromosomes in metaphase neuroblast nuclei with breaks. B. Fraction of metaphase neuroblast nuclei with polyploidy in different genotypes. * p<0.05, ** p<0.01, *** p<0.0001 (Fisher's exact test). Number of nuclei scored (left to right): 46, 27, 44, 54, 24, 54, 38, 18, 33, 26, 59, 53, 43.

Figure 5. Models for roles of DmBLM and endonucleases in replication fork repair.

A. The first structure (i) represents a replication fork with a block (diamond) on the leading strand. Arrowheads on dark lines indicate the 3′ ends of the template strands; arrows on light lines indicate 3′ ends of the nascent leading (blue) and lagging (red) strands. It is possible that blocked forks can be cleaved on the lagging strand template by GEN or on the leading strand template by MUS81–MMS4. More typically, however, the fork is regressed (ii), possibly with template switching (iii). After removal of the block, DmBLM catalyzes reversal of the regressed structure to re-establish the replication fork. In the absence of DmBLM, regressed forks without or with template switching can be cut by MUS81–MMS4 or GEN, respectively (iv and v). Blocked forks can also spontaneously break (dotted line), especially if not protected by SPN-A. Collapsed forks resemble one-ended DSBs, but replication from a fork to the right converts these into DSBs (vi and vii), which are repaired by standard DSB repair pathways (see Figure S1). B. Converging replication forks (viii) sometimes experience problems that are solved through a DmBLM-dependent migration/decatenation process (ix). In the absence of DmBLM, MUS312–SLX1 cuts a fork, generating a DSB (x). It is also possible that both forks are cut, leading to DSBs on both chromatids (not shown). These could both be repaired using the homologous chromosome, except in the case of the male X or Y chromosome.