The role of chromosome rearrangements in speciation is a subject of discussion. The traditional underdominance model suggests that intertaxon hybrids heterozygous for chromosome rearrangements should display various irregularities in meiotic chromosome pairing, recombination and segregation, and therefore become completely or partially sterile, due to meiotic arrest or formation of unbalanced gametes [1
]. This model faces serious theoretical difficulties, because rearrangements causing sterility have a very low chance to spread in populations [2
]. A more convincing modification of this model presumes that accumulation of several different selectively neutral chromosome rearrangements in different geographically isolated populations may lead to chromosomal incompatibility in the hybrids and their sterility [3
]. Gene flow between chromosomally different populations, races or species may also be restricted, due to recombination suppression near the breakpoints of the chromosome rearrangements [6
The common shrew, Sorex araneus
, is a model species to examine various scenarios of chromosomal speciation. This species shows a high level of chromosome variation. The diploid chromosome number (2n) varies from 20 to 33, with the number of autosomal arms the same all over the species range (FNa = 40). It has been suggested that the ancestral karyotype of the common shrew consisted mostly of uniarmed (acrocentric) chromosomes [7
]. The consecutive fixation of various centric or Robertsonian fusions (Rbs) and whole-arm reciprocal translocations (WARTs) gave rise to a wide variety of combinations of biarmed (metacentric) and acrocentric chromosomes in extant chromosome races. About 70 chromosome races have been described so far [8
]. These races are distributed in a parapatric fashion, and generate many hybrid zones. The zones differ from each other in the complexity of meiotic configurations in the hybrids. In some zones, the hybrids carry only trivalents involving metacentric chromosomes paired with twin acrocentrics, while in the other zones, hybrids carrying chains of up to eleven chromosomes are observed [9
The hybrid zone between the Novosibirsk and Tomsk races is of special interest. The races differ for a series of chromosome rearrangements (Figure 1
, Supplementary Figure S1
). They form a narrow hybrid zone and generate hybrids with both simple and complex synaptic configurations [10
]. The simple configurations are trivalents, i.e., chains of three (CIII), involving metacentric and twin acrocentric chromosomes. The complex chain configurations occur due to synapsis of eight (CVIII) or nine (CIX) metacentric and acrocentric chromosomes with monobrachial homology (Figure 1
). The chromosomes involved in a CIX chain form clines of about 9 km width on average, while the cline for the CIII chromosomes is 53 km wide. Thus, the cline widths are inverse to the complexity of synaptic configurations, and the probability of meiotic errors [11
Studies on gene flow across this zone have given contradictory results. On the one hand, analysis of population structure using several microsatellite markers revealed rather weak differentiation between the races; genetic differences between the Novosibirsk and Tomsk races were less than the inter-population differences within each race [12
]. On the other hand, significant difference in morphology between the races has been demonstrated. The Novosibirsk shrews and the hybrids are significantly smaller than the Tomsk shrews [13
]. Phenotypic differentiation in the geometric shape of skulls and mandibles across this zone was found to be greater than expected under the assumption of unrestricted gene flow during the estimated time since contact [14
]. Interestingly, the morphological clines coincide with the CIX chromosome clines. The difference in the cline widths is parallel with the difference in expected level of meiotic abnormalities in CIX and CIII chromosome carriers [11
In order to determine the role of chromosomal heterozygosity in the restriction of gene flow between the races, we examined chromosome pairing and recombination in shrews of various karyotypes trapped in the hybrid zone between the Tomsk and Novosibirsk chromosome races. To visualise synaptonemal complexes (SC) and the sites of crossing over, we used fluorescently-labelled antibodies to Synaptonemal Complex Protein 3 (SYCP3), a protein of the lateral elements of the SC, and MutL Homolog 1 (MLH1), a mismatch repair protein marking about 90% of mature recombination nodules. This approach has been successfully applied for mapping the sites of recombination in several species of mammals [15
], including the common shrew [18
Here, we demonstrated a very high rate of synaptic aberration in male complex heterozygotes for chromosome rearrangements with monobrachial homology derived from the hybrid zone between the Novosibirsk and Tomsk chromosome races of the common shrew (Table 1
) and strong recombination suppression in pericentromeric regions of most chromosomes. How may these factors affect gene flow across the hybrid zone?
Partial or complete asynapsis of the chromosomes involved in the chain was the abnormality most often seen. Asynapsis is a well-known cause of impaired fertility due to meiotic arrest and germ cell loss [29
]. Completion of synapsis at pachytene is monitored by a checkpoint mechanism, which triggers apoptosis in the cells with incomplete synapsis [30
]. However, at least half of the germ cells in CVIII carriers showed normal chromosome synapsis and recombination. These cells can complete meiosis and form viable gametes. Earlier, Pavlova et al. [33
] and Matveevsky et al. [28
] detected normal pairing of the expected configurations at diakinesis–metaphase I and active sperm of normal morphology in male carriers of ring of four (RIV) and CXI from the Moscow–Neroosa and Moscow–Seliger hybrid zones, respectively. Studies in other hybrid zones of the common shrew suggest decreases in fertility in heterozygotes, particularly complex heterozygotes, but not to a massive degree [35
]. Unfortunately, we were unable to obtain direct estimates of fertility of the shrews from the Novosibirsk–Tomsk hybrid zone. Breeding shrews in captivity is extremely difficult. Trapping adult male shrews during the breeding season (April–May) in this area is impossible, due to climatic conditions. However, the indirect evidence listed above indicates that despite a high frequency of pachytene cells with aberrant chromosome pairing, the male carriers of complex meiotic configuration generally can produce viable gametes, although likely in reduced numbers.
Analysis of the distribution of molecular markers in the hybrid zones of the common shrew showed no evidence of reduced gene flow resulting from male hybrid sterility or reduced fertility. There was no difference in the degree of genetic differentiation within and between the Novosibirsk and Tomsk races; the markers located in common and race-specific chromosomes showed the same level of differentiation between the races [12
]. No linkage disequilibrium was detected between the race specific Y chromosome haplotypes and the race specific autosome complement in the Novosibirsk–Tomsk hybrid zone [39
]. These indirect data indicate that pairing abnormalities in hybrid CVIII and CIX carriers, detected in this study, do not lead to complete sterility of males.
Female meiosis in the hybrids has not been studied yet, due to technical difficulties. However, studies on the other mammalian species indicate that female meiosis is less vulnerable to chromosomal heterozygosity and genetic incompatibility than male meiosis. It is also known that meiotic checkpoints in female germ cells are less sensitive to incomplete synapsis than in male germ cells (see [40
] for review). For this reason, we suppose that the fertility in female hybrid shrews should be affected less than in males.
The overall recombination rate was rather similar in all shrews examined, except the male CIX carrier showing a very high frequency of pairing aberration (Table 2
). In the homozygous CIII and CVIII karyotypes, it was higher than the expected number of obligatory chiasmata sufficient for correct chromosome segregation at the first meiotic division (Table 2
). Thus, correct segregation of chromosomes is possible in the male hybrids, though direct studies are needed to estimate the frequency of meiotic non-disjunction in this hybrid zone. Certainly, the narrowness of the Novosibirsk–Tomsk hybrid zone indicates that there is a degree of infertility associated with presence of the CVIII or CIX configuration in hybrids [11
], either attributable to germ cell death (reduced reproductive lifespan of females, greater frequency of males with insufficient numbers of sperm) or non-disjunction (greater embryo death) [35
shows the balanced segregation of the CIX, when all chromosomes derived from Tomsk race move to one pole and all Novosibirsk derived chromosomes to another pole. Of course, recombination in the terminal and medial regions of the chromosomes lead to the production of recombined chromosomes. However, a strong suppression of recombination around the centromeres, spanning from a quarter to a half of the arm length in the chain forming metacentric chromosomes, enforces linkage disequilibrium between all parental alleles located in the pericentromeric regions. If in geographic isolation, chromosome races become fixed around the centromeres with alternative epistatically-interacting coadaptive genes, i.e., supergenes, then such supergenes could still be maintained after the races come into contact. Such pericentromeric supergenes have been found in several species of plants (see [41
] and references therein). In the shrew hybrids, these supergenes could be dispersed over multiple pericentromeric regions of the chromosomes involved in the long meiotic chain, and therefore, would segregate as a united race specific “super-supergene”. It has been shown that the gene flow between mouse chromosome races is more strongly restricted near the centromeres of Robertsonian chromosomes than in other regions [42
]. In the case of Novosibirsk–Tomsk hybrid zone, we would also expect strongly restricted gene flow in the proximal regions. This reduction in gene flow is probably very difficult to detect with the set of chromosome-specific microsatellite markers available at present in the common shrew [12
]. More markers with known subchromosomal localisation have to be developed.
Linkage disequilibrium, due to crossover suppression in pericentromeric regions in the complex chromosome heterozygotes occurring in the hybrid zone, might keep together race-specific alleles controlling morphological traits, and thus, contribute to the phenotypic differentiation of the parental races [13
]. Linkage disequilibrium for genes controlling local adaptations may also affect the width of clines for the chromosomes containing such genes. Polyakov at al. [45
] demonstrated altitudinal partitioning of the Novosibirsk and Tomsk races. This partitioning marks a border between two rather different biotopes: forest steppe at low altitude, and taiga at the high. It has been shown that the structure of the cline for the CIX chromosomes closely follows that predicted by a model including altitude, while the CIII cline does not fit it so well [11
In conclusion, the meiotic studies of hybrids from the Novosibirsk–Tomsk hybrid zone of the common shrew show indications that both fertility reduction and recombination suppression may contribute to reduced gene flow, in line with previous findings for the best studied house mouse chromosome hybrid zone [43
]. Together, the common shrew and house mouse have provided unusually detailed vignettes of the impacts of heterozygosity of Robertsonian chromosomes in nature, of relevance to our understanding of chromosomal rearrangements in differentiation and speciation [4
]. Given that hybrid karyotypes have very different properties in other known hybrid zones of the common shrew [12
], this species is an exceptional system to extend, even further, our understanding of the meiotic properties of a wide range of Robertsonian heterozygotes.