Whole-Genome Scan in Thelytokous-Laying Workers of the Cape Honeybee (Apis mellifera capensis): Central Fusion, Reduced Recombination Rates and Centromere Mapping Using Half-Tetrad Analysis

DISCUSSION

This formal genetics, established in part on results from a natural laboratory (i.e., the parasitic clone), brings some light to the genetics of the honeybee but also to the more general problem of parthenogens, which until recently have been studied more at the cytological level than at the genetical level.

Central fusion: The work of Verma and Ruttner (1983) provided cytological evidence for a central fusion accounting for diploid restoration of the thelytokous parthenogens in the Cape honeybees. Genetic results expected from this mode of diploid restoration are gradients of homozygosity from the centromere toward the tip of the chromosomal arm(s). Gradients were actually observed for almost all long linkage groups. However, these gradients may also be produced by terminal fusion, i.e., fusion of the two meiotic products that were separated only at the second meiotic division. In that case, gradients of homozygosity should have the reverse direction; i.e., they should increase from the telomere to the centromere. A support for central fusion is provided by linkage group I, most probably assigned to chromosome 1. Heterozygosity is observed for all individuals at five loci located in the middle of the group, while homozygosity increases with the distance to these markers. This pattern is what is expected for a metacentric chromosome under central fusion. Less direct evidence advocating for central fusion emerges from the high level of heterozygosity that has been preserved in the clone over numerous generations. A terminal fusion cannot maintain high heterozygosity in presence of low levels of recombination.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

—Genetic distance in capensis pseudo-queens as a function of the linkage distances on the map constructed with regular meioses. All 78 pairs of adjacent markers in the data set are shown. With only one exception, all points are located well below the line of equation y = x, which indicates an important reduction of the recombination rate during thelytokous parthenogenesis.

Centromere mapping: The number of species where centromeres can be genetically mapped is relatively limited because it is necessary to have access to several products of the same meiosis. The most favorable material is obviously fungi, where tetrads give access to the four meiotic products (Perkins 1953). More common are species where only half-tetrads, i.e., two meiotic products, can be recovered. The first well-studied half-tetrads analyses were attached-X chromosomes in Drosophila (Beadle and Emerson 1935). Other examples include induced chromosome doubling at meiosis in maize (Rhoades and Dempsey 1966), gynogenesis in zebrafish (Johnsonet al. 1995), and automictic parthenogenesis in Venturia canescens and in honeybee (Verma and Ruttner 1983; Beukeboom and Pijnacker 2000). In mammals, autosomal trisomies (Mortonet al. 1990) or ovarian teratomas, where meiosis second division is suppressed (Chakravartiet al. 1989), can be used for half-tetrad analysis.

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

—Cumulative linkage distance in the capensis pseudo-queens, from centromere to telomere(s), as a function of the cumulative linkage distances of a map constructed with regular meioses. Four linkage groups are shown. (A) Linkage group I; (B) linkage group II; (C) linkage group III; (D) linkage group VIII. The group numbers refer to the current state of the linkage map of the honeybee, see Solignac et al. (2004, accompanying article). Note that the regression is not linear and that most of the chiasmata are localized at the tips of the chromosomes in the thelytokous parthenogens.

Assuming central fusion, as justified above, we have used homozygosity gradients to map the centromeric regions on the linkage groups of A. mellifera. Depending on the group, these regions are more or less extended, as a function of the number of recombinant individuals observed (see Solignacet al. 2004, accompanying article). It must be noted that in some instances the clarity of the gradient is obscured toward the tips of the chromosomes and this evanescence of homozygosity is attributable to the high abundance of multiple chiasmata and to the subterminal preferential location of crossing over.

Recombination in the Cape bee: The analyses on the Cape honeybees by Moritz and Haberl (1994) using fingerprinting (12 dominant markers) led them to conclude that recombination was absent in pseudo-queens. Our results show that recombination is not totally suppressed. This difference may be attributed to the low number of markers used by the two authors, to their dominance (only homozygotes for the absence of bands are detectable), or to a centromeric location of the loci.

Verma and Ruttner (1983) have cytologically observed an average of two chiasmata per bivalent during the meiosis of the Cape laying workers. This is equivalent to a length of 100 cM per chromosome, i.e., a total of 1600 cM for the 16 chromosomes of the genome (Beye and Moritz 1995 and references therein). The “normal” genomic length of the honeybee is ∼4500 cM (see Solignacet al. 2004, accompanying article; this is slightly longer than the randomly amplified polymorphic DNA map published by Hunt and Page 1995). By comparing linkage distances in pseudo-queens and in the linkage map, we found that linkage length is reduced on average by a factor of 12.8 in the pseudo-queens. This means a linkage length of the genome of ∼310 cM for the pseudo-queens. This value is much lower than the one inferred from the cytological distribution of chiasmata. A possible explanation could be that many chiasmata occur very close to the telomeres of chromosomes and hence are rarely detected in our data.

Kauhausen (1978), using five morphological characters on the thelytokous progeny of pseudo-queens of capensis ancestry, observed a high number of recombinant individuals homozygous for recessive alleles. The linkage distances to the centromeres deduced from her values are 36.0, 25.7, 18.2, 3.6, and 1.6 cM for chartreuse, cream, cordovan, bayer, and an eye color mutation, respectively. The position of these loci is unknown but their average distance to the centromere is relatively high compared to the “thelytokous” distances (see Figure 4). This might be due to a preferential telomeric location of these morphological markers but can also result from the fact that the pseudo-queens were not pure Cape bees but were produced by two or three recurrent backcrosses of capensis and carnica hybrid drones with capensis queens.

Even if from these results the exact linkage length of the map of the capensis pseudo-queens cannot be determined, it is clear that there is a strong reduction of recombination during their meiosis. This raises the problem of its origin. Ruttner (1988) has concluded that thelytoky in A. m. capensis is under the control of a single Mendelian gene with incomplete penetrance. It is highly improbable that the reduction of recombination is a pleiotropic effect of the factor responsible for thelytoky. More probably several genes are involved, as suggested by the intermediate recombination rates observed in interracial hybrids between carnica and capensis (Kauhausen 1978).

Thelytoky and recombination: Thelytokous parthenogenesis (the production of diploid females from unfertilized eggs) is relatively rare in the animal kingdom with hardly >1500 thelytokous species (White 1984). In Hymenoptera, it also does not appear to be a frequent phenomenon (reviewed in Slobodchikoff and Daly 1971; Schilder 1999), although, on the basis of the sporadic appearance of thelytoky in several parasitoid families (Crozier 1975), it has been argued that Hymenoptera might quite easily make the transition from arrhenotoky to thelytoky (Cornell 1988). The various cytological mechanisms of thelytoky are traditionally divided in two groups. When offspring are produced without meiosis, the parthenogenesis is called apomixis. When reproduction involves chromosomal reduction and restoration of the original ploidy (by fusion of two meiotic products or through gamete duplication), the parthenogenesis is called automixis. The majority of Hymenopteran parthenogens for which the cytological mechanism has been investigated are automictic (Cook 1993). Automixis leads to complete homozygosity or at least to an increase in the homozygosity of the offspring compared to the mother. In A. m. capensis, the increase of homozygosity is much reduced because of the low recombination rate during thelytoky. It is currently not known whether the other automictic Hymenopteran species also have mechanisms to control the homozygosity increase associated with automixis. In a few automictic insect species, it has been shown that most individuals are heterozygous for chromosomal inversion (Stalker 1956; Carson 1962; Suomalainenet al. 1987), which may be a way to preserve heterozygosity.

This reduction, limited to the capensis pseudo-queens, appears as a specific genome-subspecies-caste interaction. It would be interesting to investigate crossover rate in laying workers in other subspecies where thelytoky is observed at low frequency (Tucker 1958).

Maintenance of heterozygosity: The reduction of recombination in laying workers is a way to preserve most of the heterozygosity present in the parthenogens. A rate similar to the one observed in the queens would produce in a few generations a progeny that is mostly homozygote (autozygote). Even in a single generation, the reduction of heterozygosity would cause damage because the honeybee is very sensitive to inbreeding for many characters (Brueckner 1976a,b).

Reduced recombination is probably a compromise between the necessity of chiasmata to ensure a faithful chromosome segregation (aneuploids are produced when crossing over is absent; Bascom-Slacket al. 1997; Moore and Orr-Weaver 1998) and the requirement to reduce or suppress recombination to maintain heterozygosity. This maintenance is enhanced by the location of the “residual” chiasmata: as long as the linkage map is approximately the reflection of the physical distance on the chromosomes, our results indicate that crossing over is preferentially located at the tips of the chromosomes in the Cape bees. This means that fewer loci become homozygous per recombination event.

The sex locus is located on one of the groups we have studied. Suomalainen et al. (1987) have predicted that, because Cape bee pseudo-queens are able to produce heterozygous females parthenogenetically, this locus must be localized near a centromere. This prediction has not been verified by subsequent analysis: Beye et al. (1996) have located the sex locus in subtelomeric position on the large arm of chromosome 8. In the Cape honeybee clone, the chromosomal region that encompasses this locus has regularly conserved heterozygosity thanks to double crossing over.

In spite of the high number of generations elapsed since the origin of the clone, as testified by the observation of mutations at 10 loci, the reduction of heterozygosity by a central fusion process was rather moderate. Considering the 161 loci analyzed, which were heterozygous in the foundress, the average reduction of heterozygosity per individual is only 19.1% (including 7 loci that were homozygous in all individuals for two alternate alleles and the few mutations being taken into account).

Several researchers took advantage of the thelytokous pseudo-queens in the Cape bee to study the heritability or the phenotypic variance of quantitative traits (Moritz and Hillesheim 1985; Moritz and Klepsch 1985; Brandes 1988, 1991; Radloffet al. 2002). The situation (assuming the absence of recombination) seemed favorable because the parthenogen and its daughters were thought to have exactly the same genotype. In fact, this is only approximately true, every meiosis generating daughters that are homozygous for a fraction of the genes that are heterozygous in the mother. It remains that the daughters of a laying worker are highly related (far more than true sisters that are sexually produced). However, a calculation of relatedness in thelytokous families has to take into account the position of genes (centromeric or telomeric) on the chromosome.