Fire Red Extended Evolution

[Jayme Bostic]

Centromeresare essential for accurate chromosome segregation, yet sequence conservation is low even among closely related species. Centromere drive predicts rapid turnover because some centromeric sequences may compete better than others during female meiosis. In addition to sequence composition, longer centromeres may have a transmission advantage.

We report the first observations of extremely long centromeres, covering on average 34 % of the chromosomes, in the red imported fire ant Solenopsis invicta. By comparison, cytological examination of Solenopsis geminata revealed typical small centromeric constrictions. Bioinformatics and molecular analyses identified CenSol, the major centromeric satellite DNA repeat. We found that CenSol sequences are very similar between the two species but the CenSol copy number in S. invicta is much greater than that in S. geminata. In addition, centromere expansion in S. invicta is not correlated with the duplication of CenH3. Comparative analyses revealed that several closely related fire ant species also possess long centromeres.

Our results are consistent with a model of simple runaway centromere expansion due to centromere drive. We suggest expanded centromeres may be more prevalent in hymenopteran insects, which use haplodiploid sex determination, than previously considered.

The recent discovery of metapolycentric chromosomes reveals that there is likely a continuum in centromere structures between monocentric and holocentric chromosomes [9, 10]. We first noticed unusual chromosome structures in the red imported fire ant Solenopsis invicta in a FISH experiment [18] and decided to explore further. In this article, we report the first observations of extremely long centromeres in S. invicta. We conducted cytological, bioinformatics, molecular, and comparative analyses to identify and characterize CenSol, the major centromeric satellite DNA repeat in fire ants. Our results are consistent with a model of simple runaway centromere expansion due to centromere drive for the evolution of long centromeres in fire ants.

Contrasting centromere morphologies in S. invicta and S. geminata. The centromere structure was revealed by DAPI staining of metaphase chromosomes. The elongated (dimension lines) and the small (arrowheads) primary constrictions are indicated. Scale bars, 5 μm

The most prevalent tandem repeat, or satellite, in a genome generally is assumed to be the candidate centromeric repeat [20, 21]. We followed an established bioinformatics pipeline [20] to identify high copy tandem satellites from the draft genomes of S. invicta and S. geminata (Additional files 1 and 2). The top ten satellites and their summary statistics for both species are shown in Additional file 3: Table S1. We compared the sequences by BLAST similarity searches to identify shared satellites within the two top-ten lists. We found nine repeats were shared between the ant genomes, with only the top two having identical ranks (Additional file 3: Table S1).

Centromeric positions can be used to describe the types of fire ant chromosomes. The centromeres of S. invicta previously were reported to be predominantly metacentric [27], whereas the chromosomes of S. geminata were metacentric and acrocentric [28]. Based on the CenSol signals in our FISH analysis and the cytological metaphase staining, we re-categorized S. invicta chromosomes into four metacentric, four submetacentric, seven subtelocentric, and one telocentric (or acrocentric) chromosomes (Fig. 3b). Chromosome classification was consistent between the CenSol FISH localization and our own karyotyping methods (Fig. 3c and d). On the other hand, we found six metacentric, nine submetacentric, and one subtelocentric chromosomes in S. geminata (Fig. 3b). Differences between the previous studies [27, 28] and ours likely can be explained by our inclusion of high-resolution FISH analysis and better chromosome resolution.

We compared CenSol sequences of both fire ants obtained from intact genomic arrays, which preserve the native structure, rather than the sequences from whole-genome assemblies, which may be assembled inappropriately due to their repetitive nature. Previous screening of a S. invicta BAC library by end sequencing revealed that 12 reads from six BAC clones (see methods) were composed of tandem repeats formed by the same monomer. Analysis using Tandem Repeat Finder [29] confirmed that the satellite DNA was composed of multiple 109 bp units, which corresponded to CenSol found using the bioinformatics approach. Additional screening of the S. invicta BAC genome library by PCR to survey CenSol abundance in the ant genome revealed that 66.5 % (638 of 960 clones) were positive for this satellite. This percentage was higher than the genome coverage estimated from FISH (above), possibly indicating that the BAC library is biased for centromeric DNA, that short stretches of CenSol are scattered throughout the S. invicta genome, or both.

We performed a multiple sequence alignment using the same sets of cloned sequences from S. invicta and S. geminata. We used the maximum likelihood method implemented in MEGA (1000 bootstrap replicates) to construct a gene tree for these CenSol sequences. These analyses revealed that the gene sequences from each species did not cluster into species-specific groups (Additional file 4: Figure S1D). This result indicates only minor divergence between the two CenSol sequence sets of the two ant species.

Legume species with two copies of the CenH3 gene have larger centromeres than those with a single copy [9]. We searched for CenH3 paralogs in the S. invicta and S. geminata genomes to determine if either has additional CenH3 gene copies. We found only a single copy of the CenH3 gene in both genomes, and the gene sequences of both are most similar to the predicted CenH3 genes of other insect species (Additional file 4: Figure S3). The nearest similar sequence was the canonical histone H3, as predicted. Thus, there is no clear association between the CenH3 gene copy number and centromere size in fire ants.

The long centromeres in S. invicta, S. macdonaghi, and S. richteri could be due to expansion from an originally shorter centromere; alternatively long fire ant centromeres contracted in the four other fire ant species. We conducted an ancestral state reconstruction analyses to infer which condition might have been more likely. Our analysis using the linear-change parsimony model indicates that the ancestral centromere state was short or moderate in length, supporting centromere expansion in the lineage leading to S. invicta, S. macdonaghi, and S. richteri (Fig. 5d). However, the direction of centromere evolution for S. indagatrix, S. aurea, and S. geminata was unresolved. We obtained a similar ancestral state pattern using squared-change parsimony (Additional file 4: Figure S5).

Studies on ant chromosomes were led largely by Imai, Crozier and their co-workers starting in the mid-20th century [35]. As FISH was not available at that time, the vast majority of these cytological studies focused on the chromosome number and the karyotype [36, 37]. Despite knowledge of the karyotypes for >750 ant species [37], detailed examination of the centromere and identification of centromeric satellite sequences are lacking.

To the best of our knowledge, this is the first study combining bioinformatics and cytologicial examination of candidate centromere sequences in any ant. Previously, candidate centromeric repeats of four ant species were identified using computational methods to find the most dominant satellite in the genome [20]; however, no cytological evidence was provided to support centromeric localization of these satellites. We used a similar bioinformatics approach to identify the top ten satellites for two fire ant species, S. invicta and S. geminata. We then used FISH to demonstrate that the most common tandem repeat, CenSol, is localized to the centromeres on all chromosomes in both fire ants, whereas the second-most abundant satellite does not. Our results support that the CenSol satellite is a major component of fire ant centromeres, however definitive evidence will require CenH3 localization and chromatin immunoprecipitation studies (CenH3 antibodies are not yet available for fire ants).

Our cytological studies showed that one S. geminata chromosome has an extended centromeric constriction (Fig. 1). Interestingly, CenSol hybridization was localized only to the edge of this constriction (Figs. 2a, 3a, and Additional file 4: Figure S2A), possibly indicating that a new centromeric satellite has invaded this S. geminata chromosome and may represent the early stages of a centromere revolution. Alternatively, it is possible that CenSol has not yet displaced the original centromeric satellite on this chromosome. The identity of this satellite remains to be determined, but our data demonstrate it is not the second most common repeat in the genome (see Fig. 2a).

One possibility is genetic drift whereby the copy number of CenSol on each chromosome increases or decreases by mutation (e.g., replication slippage) or recombination (e.g., unequal crossover) with copy number changes in either direction being equally likely and occurring independently for each chromosome. By chance all 16 of the S. invicta centromeres may have drifted to the longer sizes. A second possibility is that centromere length evolves neutrally but cell biological processes constrain all the centromeres to be of similar size, e.g., perhaps to avoid aneuploidy during cell division. While this may explain how all centromeres are uniformly long, it cannot readily explain the initial transition from the ancestral short to current long centromeres.

3a8082e126