Extraordinary centromeres: differences in the meiotic chromosomes of two rock lizards species Darevskia portschinskii and Darevskia raddei

Early stages of meiotic prophase I in D. portschinskii and D. raddei: leptotene and zygotene

In both species D. portschinskii and D. raddei, leptotene begins with the formation of axial structures, which resemble dotted lines (Figs. 1A and 1B) until the early zygotene (Figs. 1C and 1D). Assembly of chromosome axial structures from the fragments was often observed together with the beginning of homologous synapsis (early “bouquet” formation) in different zones of one nucleus (Figs. 1C and 1D). Immunostaining of kinetochore proteins revealed progressive clustering of the centromeres of all 38 acrocentric chromosomes from the leptotene to early zygotene stages in both species (Figs. 1C and 1D).

Our detailed study of early prophase I stages (preceding the pachytene) revealed previously unknown phenomenon specific for D. raddei—additional foci of anti-kinetochore proteins antibodies ACA (Figs. 1B and 1D). Additional ACA foci were detected on the still unsynapsed axial elements of D. raddei homologous chromosomes at zygotene stage (Figs. 2A and 2D). Thus, four ACA foci were visible on several SC bivalents during the synapsis, with two on each axial element (Figs. 2B, 2C and 2E). Totally in 26 of 30 leptotene–zygotene nuclei studied one to four additional centromere signals were detected in both D. raddei individuals. Fragmented axial cores of chromosomes at the presynaptic stages (before the pachytene) did not allow us to study actual number and distribution of additional ACA foci among the meiotic chromosomes. The next meiotic prophase I stage, pachytene, was analyzed in details due to complete SCs assembly and applicability for the distinct chromosomes identification.

Late stages of meiotic prophase I in D. portschinskii and D. raddei: pachytene and diplotene

The pachytene stage of both species studied (D. portschinskii and D. raddei) is characterized by complete synapsis of all 19 acrocentric SC bivalents (18 autosomal bivalents and the ZZ sex bivalent) (Figs. 3A and 3B).

In D. portschinskii pachytene nuclei single ACA focus was detected at one end of each of the 19 SCs (78 nuclei studied) (Fig. 3A). However, in D. raddei apart from to the usual 19 ACA foci on the SC ends we detected additional ACA foci on the SC bivalents (Fig. 3B). These signals were located closely in the SC structure at a distance of 0.27–1.39 μm, on average 0.62 ± 0.21 (mean ± SD, 54 nuclei studied) and demonstrated similar or slightly different signal intensities.

Pachytene stage was the most representative for the precise chromosome length measurements, other prophase I stages are inapplicable for this study. Nevertheless SC karyotyping revealed minor differences in the length of medium-sized SC bivalents in both species (Fig. 4A) making the identification of distinct dicentric (Fig. 4C) chromosomes more challenging.

We performed the detailed analysis of 54 pachytene nuclei of the six preparations from two D. raddei animals. We selected immunostained SC-karyotypes without bivalent overlapping and used relative chromosomes length in order to minimize possible influence of the different spreading conditions between nuclei studied. Additional ACA foci were detected at chromosomes 1–17 with enrichment on chromosomes 1–4 (1—90.7%; 2—79.6%; 3—88.9%; 4—63.0% of all 54 nuclei studied), medium occurrence on chromosomes 5–14 (5—29.6%; 6—24.1%; 7—29.6%; 8—16.7%; 9—20.4%; 10—11.1%; 11—20.4%; 12—25.9%; 13—24.1%; 14—29.6% of 54 nuclei studied), sporadic occurrence on chromosomes 15–17 (15—3.7%; 16—1.9%; 17—5.6% of 54 nuclei studied) and not been detected on the chromosomes 18 and 19 (Figs. 4D and 4E).

In total, additional ACA foci were detected in 198 out of the 211 D. raddei primary spermatocyte nuclei studied (leptotene–diplotene stages).

FISH with telomere probes on pachytene chromosomes of D. portschinskii and D. raddei

Fluorescent in situ hybridization with telomere probes (Figs. 4B and 4C) revealed a standard distribution of telomere repeats in the SC karyotypes of D. portschinskii and D. raddei, with no interstitial signals (Rovatsos et al., 2015) detected in the pachytene SC bivalents (25 nuclei for each species studied). In particular, we studied the SC regions of additional ACA foci in D. raddei SC bivalents, and no one telomere FISH-signal was detected at any of these locations in all nuclei studied (Fig. 4C).

Immunodetection of crossing over sites in D. portschinskii and D. raddei spermatocytes I

Immunolocalization of the MLH1 protein (late recombination nodules marker, prospective chiasmata) on pachytene stage in the SC preparations was performed for both species (Figs. 5A and 5B). The average number of crossing over sites (MLH1 foci) was 28.43 ± 2.11 (mean ± SD) in D. portschinskii (Fig. 5C) and 28.64 ± 2.07 (mean ± SD) in D. raddei (Fig. 5C). We did not detect MLH1 foci in the SC regions between the two ACA foci in all 55 immunostained D. raddei spermatocyte I nuclei (Fig. 5B).

Synaptonemal complex processing and number of crossing over sites during prophase I in D. portschinskii and D. raddei

In general, the characteristics of the stages of meiotic prophase I were similar between species (Figs. 1 and 3), and were comparable to the results of our previous study of two other bisexual species, D. raddei nairensis and D. valentini (Spangenberg et al., 2017). For example, the absence of a classical leptotene stage with long threads (presynaptic stage with completely formed chromosome axial elements) is a common feature among all four rock lizards species studied by us to date. Furthermore, early zygotene, organization of “bouquet” formation, progression of axial element synapsis during zygotene and SC assembly at pachytene were similar in both species (Figs. 1C, 1D, 3A and 3B). No significant difference in numbers of crossing over sites was detected between D. portschinskii and D. raddei males (Fig. 5C) in the studied population.

Extraordinary difference of distribution of ACA foci in the SC bivalents in D. portschinskii and D. raddei

As visible in our results, the SC karyotypes of the two species display a striking difference in relation to the immunostaining of kinetochore proteins. Two ACA foci can be observed in the structure of SC bivalents 1–17 in primary spermatocyte nuclei of D. raddei (Figs. 4D and 4E), one near the telomere, similar to the acrocentric organization of SCs in D. portschinskii, and the second located at some distance along the SC axis (Fig. 4C). Spreading technique allow us to study meiotic chromosomes in detail (Fig. 4C) and to detect additional ACA foci. This phenomenon has not been previously described in mitotic chromosomes due to high levels of chromatin compaction as well as in previous studies of the meiotic chromosome structure of rock lizards using light microscopy (Darevsky & Kupriyanova, 1982; Darevsky, Kupriyanova & Danielyan, 1986).

Immunodetection of additional ACA foci in D. raddei during the leptotene and zygotene stages indicates the presence of two ACA foci in both homologous chromosomes prior to synapsis (Figs. 1B, 1D, 2B, 2C and 2E).

The differences in additional ACA foci number and distribution were detected in the nuclei from one sample preparation as well as between preparation slides derived from different D. raddei animals. We suppose that this result can be explained by the fact that closely located double ACA foci often cannot be distinguished from enlarged (or elliptical) single focus due to limitations of fluorescence microscopy as well as because the nearest chromosome lengths are very close in this species (Fig. 4A). On the other hand, the different chromosome numbers with additional ACA foci from the one sample could indicate a high level of instability of centromere protein distribution in the pericentromeric regions in D. raddei.

Additional ACA foci we detected in SC-karyotype of D. raddei males leads us to assume that the formation of the additional foci occurred not via inversions or duplication but via an epigenetic mechanism in these species. This is supported by the fact that we did not observe any disruption of synapsis or inversion loops in the pericentomeric regions of SC bivalents with additional ACA foci in all 198 nuclei studied.

In recent years, studies in mammals have reported that neocentromeres can be formed in intact chromosomes (without rearrangements), which can functionally replace the native centromeres (Rocchi et al., 2012). De novo centromere formation occurs by an epigenetic sequence-independent mechanism involving the deposition of a centromere-specific histone H3 variant, CENP-A (Allshire & Karpen, 2008; Black & Cleveland, 2011; Perpelescu & Fukagawa, 2011). A study of artificial neocentromere formation in chickens demonstrated that while they can be formed in any region of the chromosome, the most likely location of neocentromere formation is close to the native centromere. The authors suggested that this is due to the potential enrichment of epigenetic marks in the zone of the native centromere (Shang et al., 2013; Zhao et al., 2017). According to studies of mitosis in different species, the formation of dicentric chromosomes (chromosomes with two active centromeres) may lead to breakage or loss of such chromosomes during the process of cell division and, consequently, cell death (Stimpson, Matheny & Sullivan, 2012; Lopez et al., 2015). However, dicentric chromosomes can be inherited if one of the centromeres is inactivated without altering the DNA sequence (Earnshaw & Migeon, 1985; Valente, Silva & Jansen, 2012). Furthermore, immunocytochemical markers allow the identification of which of the two centromeres is active (Voullaire et al., 1993; Warburton et al., 1997). However, the molecular basis for centromere inactivation is not well understood (Stimpson, Matheny & Sullivan, 2012).

Our results do not allow the status of additional ACA foci to be determined, specifically whether they are associated with active or inactivated centromeres. However, the number of nuclei with clear additional ACA foci, identified in two D. raddei males, indicates that this phenomenon is not random. Moreover, numerous mature elongated spermatids with normal morphology were found in preparations from both D. raddei males in the current study. The successful formation of mature germ cells by D. raddei males indicates that the peculiarities of centromere organization described above does not affect their fertility.

The formation of double centromeres in the structure of SCs has been previously reported for Danio rerio fish (Moens, 2006). In this study, three to four double centromeres were detected in the nuclei of primary spermatocytes, which the author characterized as being misaligned centromeres. The author also pointed out that such double centromeres are not necessarily found on the same chromosomes, judging by the length and centromere position (Moens, 2006). This pattern correlates well with the distribution of the double ACA foci observed for D. raddei in the current study. However, in the early zygotene nuclei of D. raddei males, we were able to detect two ACA foci on the yet unpaired axial elements of chromosomes. Furthermore, four ACA foci were visible by the zygotene stage, with two on each axial element (Figs. 2B, 2C and 2E). This indicates the presence of two ACA antibody binding sites on each axial element but not incomplete alignment of homologous chromosomes.

It should be noted that we used similar antibodies (ACA) to those used by other studies (ACA, CREST) (Brenner et al., 1981; Oppedisano et al., 2002; Chambon et al., 2013). The applicability of these antibodies for the detection of centromeres in a wide range of vertebrate species has long been established. In addition, numerous studies in the literature have discussed the specific characteristics of immunostaining using these antibodies during different phases of the cell life cycle (Brenner et al., 1981; Chambon et al., 2013). Our comparative studies of CENP-A and CENP-C proteins by alignment of amino acids sequences confirmed our negative immunostaining experiments with the commercially available CENP-antibodies. Poor similarity was detected between reptilian proteins and the human immunogen polypeptides used in commercial antibodies.

Further studies using monoclonal antibodies against specific proteins are needed in order to distinguish active and inactivated centromeres, in addition to FISH with pericentromeric satellite DNA. An interesting finding was the detection of additional ACA foci in SC preparations of the most closely related species, D. raddei nairensis (in supplemental data: see ‘Data Availability’ section). D. raddei and D. raddei nairensis are considered recently divergent species, and are actively investigated as a model system for the study of speciation mechanisms (Omelchenko et al., 2016).