Trans-Centromere Effects on Meiotic Recombination in the Zebrafish

Abstract

We report that lack of crossover along one chromosome arm is associated with high-frequency occurrence of recombination close to the opposing arm's centromere during zebrafish meiotic recombination. Our data indicate that recombination behavior on the two arms of a chromosome is linked. These results inform mapping strategies for telomeric mutants.

THE zebrafish (Danio rerio) provides the unique opportunity to recover two of the four haploid genomes produced from a single meiosis using a technique known as early pressure (EP) parthenogenesis (Streisinger et al. 1981, 1986). Recombination events can be reconstructed accurately from the genetic analysis of EP parthenogenotes (Streisinger et al. 1986; Johnson et al. 1995) (supporting information, Figure S1). Analysis of these products can reveal how the occurrence of one recombination event may affect subsequent chromosome behaviors at meiosis.

In this study we investigate recombination behavior of two zebrafish chromosomes for which telomeric recessive mutations are available. In the absence of recombination or following even numbers of crossovers along the length of the relevant chromosome arms, the early pressure (EP) half-tetrad progeny produced from heterozygous females will be homozygous for telomeric alleles. Half of such progeny will express the mutant phenotype (Figure S1). We analyzed mutant EP half-tetrad progeny in detail to identify recombination events that produced sister chromatid pairs homozygous for telomeric markers.

RESULTS AND DISCUSSION

A total of 1243 EP larvae were produced from asm^+/− heterozygous F₁ females that were generated in a mating between a unique asm^+/− P0 founder male and a unique asm^+/+ P0 female from the wild-type (WT) Tu strain (Figure 1A and Table S1). As illustrated by the representative genotyping analysis (Figure S2), EP parthenogenesis produced 5-day-postfertilization (dpf) offspring in which homologous pairs of chromosomes were derived from sister chromatid pairs. All asm mutants were genotyped using nine chromosome 18 simple sequence repeat (SSR) markers that were polymorphic between the parental asm⁻ and asm⁺ chromosomes (Figure S3). On the basis of analysis of these nine markers, 34 mutants lacked evidence of any recombination between the centromere and the asm locus. Two mutants arose from two-strand double crossovers (dco) (Figure S3C), and 3 mutants arose from four-strand dco (Figure S3D). Additionally, one exceptional individual was heterozygous for markers that flank the centromere and therefore likely was not derived only from a single sister chromatid pair (Figure S3E). In sum, 5/39 (13%) EP half-tetrad mutant progeny arose from sister chromatid pairs that experienced dcos. The products of dcos would be anticipated by established linkage maps of zebrafish, which exceed 100 cM in many cases (Johnson et al. 1995; Kauffman et al. 1995). Consistent with our genetic analyses, a recent study using the chiasma-specific MLH1 antibody indicated the occasional presence of at least two chiasmata involving a single chromosome arm during zebrafish female meiosis (Kochakpour and Moens 2008). The MLH1-staining did not distinguish between two-, three-, or four-strand dco and at what frequency they occurred. The data presented here show unambiguously that two-strand and four-strand dco occur at measurable frequency in zebrafish.

Figure 1.—

(A) Pedigree of early pressure (EP) gynogenetic half-tetrad progeny that were analyzed. Fish lines: asm was induced with ethylnitrosourea and mutants were distinguished from wild-type by morphology (small, protruding eyes and small head phenotypes) or by in situ hybridization (using an antisense RNA probe for the T cell-specific kinase lck) at 5 day postfertilization (dpf) (Trede et al. 2008). Correspondence between lack of T cells and head morphology is 100% (data not shown). The heterozygous asm mutation was induced in the WIK background and a mapping hybrid cross was generated for the present study by crossing a single asm^+/− heterozygote with a single wild-type Tu individual (Figure 1A, Figure S2). asm^+/− F₁ individuals were identified by progeny test crosses. OG076 was identified in the Tübingen 2005 screen by its lack of macrophages and neutrophils (P. Herbomel and M. Redd, personal communication). It was generated on the Tu background and crossed to WIK for mapping purposes. EP parthenogenesis: Sperm collection, egg extrusion, in vitro fertilization, and timing of pressure treatment were performed as described previously (Streisinger et al. 1981). Sperm was UV-treated in a Stratagene stratalinker set to deliver 7 × 10⁴ μJ. Pressures and hydraulic press equipment used were described previously (Gestl et al. 1997). Generation of mutant offspring: EP parthenogenotes were generated from heterozygous F₁ females and phenotypically classified as either mutant (asm) or wild type (WT). For analysis of recombination affecting the asm chromosome, EP was performed on multiple occasions on eggs derived from a total of 20 asm⁺ F₁ females (Figure 1A, Figure S2). Resulting gynogenetic diploid half-tetrad larvae were raised to 5 dpf when asm mutants are viable: among >4000 offspring produced from matings between heterozygotes, ∼25% were mutant at 5 dpf (data not shown). For analysis of the OG076 chromosome, eggs from four heterozygous females were subjected to EP, larvae were fixed at 4 dpf and assayed for presence of neutrophils by staining with Sudan Black (Le Guyader et al. 2008). (B) Nonindependence of recombination on the two arms of chromosome 18. Primers and PCR: Centromere-linked microsatellite marker loci were obtained from published studies (Shimoda et al. 1999; Mohideen et al. 2000). https://wiki.zfin.org/display/prot/MGH-CVRC+Mapping+Resources. PCR products were resolved by gel electrophoresis using 3% metaphor agarose. Recombination events in 34 mutant and 23 wild-type sibling EP offspring from asm⁺ heterozygous females were analyzed using the chromosome 18 SSR markers depicted in the bar graph on the left. Half-tetrad EP offspring were divided into two groups: mutant offspring whose chromosomes 18 were nonrecombinant along the right arm (no exchange on right arm), and wild-type offspring that harbored a single chromosome 18 that was recombinant along the right arm (exchange on right arm). Because recombination affecting either of the two left arms of the sister chromatid pair could produce a heterozygous half-tetrad, we would expect that 7.4% of the EP progeny would be heterozygous for marker Z9194. Chromosomes with a right arm exchange exhibited a frequency of exchange in the left pericentric interval very close to expectations on the basis of the published map distance. In contrast, the frequency of exchange in the left pericentric interval appeared aberrantly high among larvae lacking an exchange on the right arm of chromosome 18. The two groups differed significantly with respect to the occurrence of recombination in the left pericentric interval. *These larvae constituted all asm mutant EP progeny lacking exchanges on the right arm of chromosome 18 (Figure S3B). ^†These larvae were selected at random from the phenotypically wild-type EP progeny and were genotyped to verify that a right arm exchange had occurred. (C) Nonindependence of recombination on the two arms of chromosome 14. Recombination events in 14 mutant and 47 wild-type EP offspring of OG076⁺ heterozygous females were analyzed using the markers depicted in the bar graph on the left. Half-tetrad EP offspring were divided into two groups: mutant offspring whose chromosomes 14 were nonrecombinant along the left arm and wild-type offspring that harbored a single chromosome 14 that was recombinant along the left arm. Because recombination affecting either of the two right arms of the sister chromatid pair would produce a heterozygous half-tetrad, we would expect that 8.8% of the EP progeny would be heterozygous for marker Z22094. Chromosomes with a left arm exchange exhibited a frequency of exchange in the right pericentric interval very close to expectations. In contrast, the frequency of exchange in the right pericentric interval appeared aberrantly high among larvae lacking an exchange on the left arm of chromosome 14. The two groups differed significantly with respect to the occurrence of recombination in the right pericentric interval. *These larvae were homozygous mutant EP progeny and presumably most lacked exchanges on the left arm of chromosome 14. ^†These larvae were selected at random from the phenotypically wild-type EP progeny and genotyped to verify that a left arm exchange was present.

Among mutant half-tetrad offspring, nonrecombinant along the right (asm) arm of chromosome 18, we found that the left arm pericentric marker Z9194 was heterozygous more often (20 of 34 = 59%) than would be predicted from the published female map distance of 3.7 cM for the Z9194–centromere interval (Figure 1B). As recombination affecting either of the two left arm sister chromatid strands would produce a heterozygous half-tetrad, 7.4% of the EP progeny are expected to be heterozygous. We analyzed 23 randomly chosen phenotypic wild-type half-tetrad siblings (all recombinant on the right arm of chromosome 18) and found that only 2 of 23 (8.7%) were recombinant in the left interval, very close to the expected fraction. The difference in the occurrence of recombination within the left pericentric interval was highly significant between the group of EP progeny produced from meioses that had experienced crossover along the right arm and the group that had not (P = 2.1 × 10⁻⁴, Fisher's exact test, two tailed).

To test whether effects on crossover behavior that extend beyond the centromere are a general feature in zebrafish, we examined recombination in an independent line harboring the telomeric mutation OG076. We mapped the mutant locus (W. H. Horsley and N. S. Trede, unpublished data) to the left arm of chromosome 14, adjacent (∼0.6 cM) to Z65389, located ∼93 cM from the centromere. Heterozygous OG076^+/− females were generated and their EP progeny analyzed for recombination events (Figure 1C). Fourteen OG076 half-tetrad mutants and 47 wild-type half-tetrad siblings were examined at three markers. All mutants maintained parental linkage with the centromere marker and were presumably nonrecombinant on the left arm. Twelve of the 14 mutants (86%) had an exchange in the right pericentric interval (4.4 cM between markers Z26376 and Z22094). In contrast, all 47 of the wild-type half-tetrads were recombinant on the left arm, and only 5 of these (11%) were recombinant in the right pericentric interval, close to the expected frequency of 8.8% for this interval. In sum, absence of recombination on one arm is coupled with significant differences in the probability of exchange in the pericentric region of the opposing arm on chromosome 14 (P = 2.6 × 10⁻⁷, Fisher's exact test, two tailed).

We note there is conflicting data whether recombination behavior on one chromosome arm is linked with that on the opposing arm (Mather 1936; Colombo and Jones 1997; Broman and Weber 2000; Falque et al. 2007). The different findings may reflect differences in experimental designs to detect trans-centromere associations or species-specific differences in the occurrence of such associations.

We considered the possibility that the evidence for pericentric recombination reflected the unexpected acquisition of markers from paternal DNA from the “UV-treated” sperm used to activate parthenogenetic development, creating aneuploid half-tetrad progeny carrying multiple pericentric alleles. This model appeared unlikely in that the 5-dpf-EP progeny homozygous for the telomeric mutations appeared uniform and distinctive in phenotype, as expected for normal diploid mutants. We directly examined this possibility by using genotyped parents and our standard EP method to produce gynogenetic offspring. Embryos were tested at 3 dpf for the presence of detectable levels of a paternal allele at a marker locus on chromosome 18. Among 132 half-tetrad progeny generated with two independent batches of UV-treated sperm, we failed to detect any paternal contribution (Figure S4). Similarly, Streisinger et al. (1981) reported no genetic contribution from UV-treated sperm in >600 gynogenetic embryos. We conclude that the vast majority of cases in which embryos were heterozygous for pericentromeric markers arose from bona fide recombination events.

A second possibility is that recombination events affecting each arm of a chromosome occurred independently, but that some combinations of crossovers were not recovered in the 5-dpf-EP progeny we analyzed, leading to the enriched recovery of half-tetrads with absence of recombination on one arm and pericentric recombination on the other arm of a chromosome. Selective recovery of half-tetrad progeny could have occurred in two ways to produce the results we observed. First, it is possible that meiotic products arising from a combination of pericentric recombination on one arm and a crossover anywhere on the opposing arm were selectively lost. The combination of crossovers might have increased the likelihood of aberrant chromosome segregation during meiosis, yielding aneuploid EP progeny that failed to develop and were lost from analysis (Buonomo et al. 2000; Rockmill et al. 2006; Subramanian and Bickel 2008). In this scenario, lack of crossover along one entire arm is the permissive condition, and analysis of the EP offspring that lacked crossover on the asm or OG076 arms (Figure 1, B and C) would indicate the true frequencies with which recombination occurred on the opposing arms. This interpretation would require that well over half of the half-tetrads in which an asm or OG076 arm experienced a crossover were lost prior to analysis. By extrapolation, pericentric recombination on any chromosome arm would frequently contribute to the production of aneuploid gametes, a prediction that is incompatible with the very high viability (often >95%) that accompanies most fertilization events with zebrafish in the laboratory.

A second selective loss model that could account for our observed results involves loss of half-tetrad progeny that failed to experience recombination on one arm and failed to experience recombination in the pericentric interval on the opposing arm. Such selective loss would produce an apparent inflation of the proportion of half-tetrads with no crossover on one arm and pericentric recombination on the other arm. Applying this interpretation to the results we observed for chromosomes 18 and 14 (Figure 1, B and C), in which pericentric recombination was recovered in 8- to 10-fold excess above the expected frequency for the pericentric interval, would require loss of 85–90% of the half-tetrads in which an asm or OG076 arm failed to experience a crossover. Once again, if we extrapolate the selective loss of products to meioses in which any of the 50 chromosome arms failed to experience a crossover, we would predict a very high fraction of nonviable gametes to be produced under normal conditions, a prediction incompatible with the observed high viability of zygotes in the laboratory.

A third possibility is that our analyses have uncovered a bona fide linkage between crossover behaviors on opposing chromosome arms. The mechanism underlying this association is not revealed by our experiments. It is possible that pericentromeric exchanges exert an inhibitory effect that propagates across the centromere and blocks the generation of additional crossovers. Alternatively, there may be mechanisms that work to ensure the occurrence of a crossover so that absence of crossing over along a chromosome arm is causally linked with the generation of a nearby pericentromeric event. Whether crossovers arise independently but then are subsequently subject to selection, or mechanisms that operate across centromeres coordinate the generation of multiple chiasmata, the net result is the transmission in the zebrafish of meiotic products with a highly nonrandom distribution of crossovers.

Our data have implications for using EP for mapping purposes. First, given the high frequency of crossovers close to the centromere opposite nonrecombinant chromosome arms, markers on both sides of the centromere need to be used. Second, the occurrence of four-strand dco implies that a homozygous mutant with the opposing centromere marker allele does not exclude linkage to that chromosome. Third, telomere markers should be used to confirm linkage.