The structure of an endogenous Drosophila centromere reveals the prevalence of tandemly repeated sequences able to form i-motifs

Abstract

Centromeres are the chromosomal loci at which spindle microtubules attach to mediate chromosome segregation during mitosis and meiosis. In most eukaryotes, centromeres are made up of highly repetitive DNA sequences (satellite DNA) interspersed with middle repetitive DNA sequences (transposable elements). Despite the efforts to establish complete genomic sequences of eukaryotic organisms, the so-called 'finished' genomes are not actually complete because the centromeres have not been assembled due to the intrinsic difficulties in constructing both physical maps and complete sequence assemblies of long stretches of tandemly repetitive DNA. Here we show the first molecular structure of an endogenous Drosophila centromere and the ability of the C-rich dodeca satellite strand to form dimeric i-motifs. The finding of i-motif structures in simple and complex centromeric satellite DNAs leads us to suggest that these centromeric sequences may have been selected not by their primary sequence but by their ability to form noncanonical secondary structures.

Figure 1. Structure of the centromere of the third chromosome of D. melanogaster.

(a) 2.5 Mb physical map across the centromere of chromosome 3. The regions containing the 10 bp satellite repeats and the dodeca satellite repeats appear in green and red, respectively. The segmental duplications from the NOR and from region 2L-29C are indicated in purple and blue, respectively. The position and GenBank number of the eight centromeric scaffolds are also indicated. Abbreviations are: B, BamHI; H, BssHII; E, BstEII, R, EcoRI; N, NaeI; Hp, HpaI; P, PmeI, S, SwaI. (b) Extended chromatin fibers from S2 cells were processed for immunofluorescence with an anti-CID antibody followed by FISH with the dodeca satellite oligo probe. A representative image showing CID immunostaining (green) overlapping with dodeca (red) is shown. Of a total of 43 chromatin fibers stained with the anti-CID antibody, 14 showed co-localization with dodeca, a proportion in agreement with the karyotype of the polyploid S2 cells. CID signals do not encompass all dodeca satellite repeats. A minor number of fibers containing dodeca satellites are not stained with the anti-CID antibody. Scale bar is 5 μm.

Figure 2. The 10 bp satellite DNA localizes on the third chromosome at h52p instead of h48.

(a) Metaphase chromosomes counterstained with DAPI. (b) Hybridization signals from a dodeca satellite probe (in red). (c) Hybridization signals from a 10 bp satellite probe (in green). (d) Hybridization signals superimposed with DAPI-stained chromosomes. The Scale bar is 2 μm. (e) Diagram representing the heterochromatic regions of chromosomes 2 (regions 35–46) and 3 (regions 47–58) showing the localization of the 10 bp (in green) and dodeca (in red) satellites. The position of the centromeres (C) is indicated. (f) High molecular weight DNA from red e embryos was digested with BssHII, electrophoresed through a 1% (w/v) agarose gel in a “Waltzer” apparatus at 150 V for 24 h with a 130 s pulse time, blotted onto a nylon filter and hybridized successively with the dodeca satellite probe pBK6E218 at 68 °C and with the 10 bp satellite probe 5′-AATAACATAGAATAACATAGAATAACATAGAATAACATAGAATAACATAG-3′ at 50 °C. The asterisks indicate a 1.2 Mb fragment that hybridizes with both probes. (g) DNA sequence showing a junction between 10 bp satellite repeats (in green) and dodeca satellite repeats (in red).

Figure 3. The centromeric dodeca satellite DNA is able to form dimeric i-motif structures.

Imino region of the NMR spectra of the C-rich strands of the undeca (a) and dodeca repeats (b). Experimental conditions: Oligo concentration = 0.8 mM, 25 mM sodium phosphate, 100 mM NaCl, T = 5 °C, pH 4. CD spectra of the C-rich strands of undeca (c) and dodeca repeats (d) at different temperatures. Oligo concentration = 100 μM, same buffer as the NMR experiments. Mass spectrometry data showing the peaks of the single stranded [1] and dimeric [2] species formed by C-rich strands of undeca (e) and dodeca (f). Buffer conditions: 100 mM NH₄OAc, spectra at pH 4. See Supplementary Fig. S7 legend for details.

Figure 4. The dimeric i-motif structure of the centromeric dodeca satellite DNA is a head-to-head association of two hairpins.

(a) Exchangeable proton region of the NOESY spectra. Each of the six cytosine imino signals exhibit two cross-peaks with cytosine amino protons, indicating the C:C⁺ base pairs occur between magnetically equivalent cytosines. (b) Region of the NOESY spectra of the C-rich strand of the dodeca repeat, showing characteristic H1’-H1’ cross-peaks (same experimental conditions as in Fig. 3b). (c) Scheme of a hemiprotonated C:C⁺ base pair, indicating the observable NOE cross-peaks between the cytosine imino and amino protons. (d) Schematic representation of the dimeric structure of C-rich strand of the dodeca repeat.