Evolutionary insights into the genomic organization of major ribosomal DNA in ant chromosomes

The major rDNA genes are composed of tandem repeats and are part of the nucleolus organizing regions (NORs). They are highly conserved and therefore useful in understanding the evolutionary patterns of chromosomal locations. The evolutionary dynamics of the karyotype may affect the organization of rDNA genes within chromosomes. In this study, we physically mapped 18S rDNA genes in 13 Neotropical ant species from four subfamilies using fluorescence in situ hybridization. Furthermore, a survey of published rDNA cytogenetic data for 50 additional species was performed, which allowed us to detect the evolutionary patterns of these genes in ant chromosomes. Species from the Neotropical, Palearctic, and Australian regions, comprising a total of 63 species from 19 genera within six subfamilies, were analysed. Most of the species (48 out of 63) had rDNA genes restricted to a single chromosome pair in their intrachromosomal regions. The position of rDNA genes within the chromosomes appears to hinder their dispersal throughout the genome, as translocations and ectopic recombination are uncommon in intrachromosomal regions because they can generate meiotic abnormalities. Therefore, rDNA genes restricted to a single chromosome pair seem to be a plesiomorphic feature in ants, while multiple rDNA sites, observed in distinct subfamilies, may have independent origins in different genera.


Introduction
Eukaryotic genomes have repetitive tandem sequences such as in the major ribosomal RNA genes (45S = 18S + 5.8S + 28S), herein denominated rDNA, which contain highly conserved genic sequences and are therefore useful as molecular genetic markers, allowing comparisons across distant taxa. However, intergenic spacers vary in both sequence and length (Long and Dawid, 1980;Sumner, 2003;Symonová, 2019). The 45S ribosomal genes are part of the nucleolus organizing regions (NORs) and are located in portions of the DNA that, after their condensation, usually appear as secondary constrictions on metaphase chromosomes (Sumner, 2003).
The silver nitrate impregnation technique (Ag-NOR banding) shows only transcriptionally active NORs by staining nuclear acid proteins involved in transcription (Howell and Black, 1980;reviewed by Sumner, 2003). Species with multiple rDNA clusters do not usually exhibit silver-staining in all clusters (reviewed by Sumner, 2003;Vicari et al., 2008). However, it must be assumed that if a species has only a single NOR (or single 45S rDNA gene site), then it will be transcriptionally active, as evidenced for different organisms (Dobigny et al., 2002;Barros et al., 2015;Falcione et al., 2018;Cholak et al., 2020;Malimpensa et al., 2020).
Intercellular and interindividual variations are frequently detected by the Ag-NOR banding method (Zurita et al., 1997;Cross et al., 2003;Walker et al., 2014;Schmid et al., 2017). However, the use of this technique in ant species has not produced reliable results owing to low repeatability, difficulty in obtaining good-quality markings, and the appearance of unspecified marks on heterochromatic regions of many First published online 8 March 2021.
Since the 1980s, molecular cytogenetic tools have been used to study karyotypes. For example, fluorescence in situ hybridization (FISH) has proven to be an effective and precise tool for physically mapping specific DNA sequences within chromosomes (reviewed by Levsky and Singer, 2003;Liehr, 2017). The rDNA genes can be located in single or multiple chromosome pairs (Sochorová et al., 2018). In several organisms, studies of these genes have pointed to chromosomal differences within species complexes (Mantovani et al., 2005;Barbosa et al., 2017;Dutra et al., 2020) and between species with similar karyotypes (Panzera et al., 2012;Golub et al., 2015;Gokhman et al., 2016). As a consequence, inferences can be made based on chromosomal rearrangements that shape the chromosomal evolution of a species (Roy et al., 2005;Nguyen et al., 2010;Britton-Davidian et al., 2012;Cabral-de-Mello et al., 2011;Dutrillaux and Dutrillaux, 2012;Roa and Guerra, 2012;Menezes et al., 2019;Degrandi et al., 2020).
In ants, the physical mapping of rDNA genes using the FISH technique was first described in Australian ants by Hirai et al. (1994Hirai et al. ( , 1996. Since then, the number of studies mapping these genes has increased (Mariano et al., 2008;Santos et al., 2016;Micolino et al., 2019a;Teixeira et al., 2020) and other repetitive sequences, such as telomeres (Meyne et al., 1995;Pereira et al., 2018;Micolino et al., 2020;Castro et al., 2020), satellite DNA (Lorite et al., 2004;Huang et al., 2016), 5S ribosomal genes , and microsatellites (Barros et al., 2018;Micolino et al., 2019b), have been mapped in the chromosomes using the FISH technique. To date, molecular cytogenetic studies on rDNA genes in ants have improved understanding of chromosomal evolution and phylogeny and provided taxonomic resolutions for different ant groups (Hirai et al., 1994(Hirai et al., , 1996Santos et al., 2010Santos et al., , 2016. In this study, we physically mapped 18S rDNA clusters using the FISH technique and verified if they were GC-rich in 13 Neotropical ant species from four subfamilies (Ectatomminae, Formicinae, Myrmicinae, and Ponerinae). In addition, we reviewed previous molecular cytogenetic data related to rDNA gene clusters (45S, 18S, or 28S) in ants. Using these data, we investigated whether the number and location of the ribosomal gene clusters followed a specific pattern or were randomly distributed in order to understand the genomic organization and evolutionary dynamics of these genes in ants.
The remaining species showed GC-rich 18S rDNA clusters across the entire chromosome arm, occupying either the long arm, as in Gnamptogenys tortuolosa (Smith, 1858) (Fig. 3A, S3A), or residing in the short arm, as in Strumigenys diabola Bolton, 2000 (Fig. 2G, S2E), Camponotus atriceps (Smith, 1858) (Fig. 3B, S3B), and Gigantiops destructor (Fabricius, 1804) ( Fig. 3C, S3C, D). Heteromorphism of 18S rDNA clusters was detected in all of the analysed C. atriceps and G. tortuolosa individuals. In the latter species, the heteromorphism of the NOR resulted in differences in total size between homologous chromosomes, which changed their morphology such that one was submetacentric while the other was subtelocentric. In G. destructor, additional GC-rich bands were located in the interstitial region of the long arm of the largest subtelocentric chromosome pair (Fig. S3C, D).

Chromosome mapping review of rDNA clusters in ants
Cytogenetic data available in the literature related to the rDNA genes of 50 ant species from 12 genera and six subfamilies were reviewed (Table 1; Fig. 4). Most data were concentrated on Neotropic ants, with information on 33 species, while the Palearctic and Australian regions had data on only one and 16 species, respectively. A single rDNA site localized in the intrachromosomal region was observed in most species (Fig. 4A, B). However, Camponotus renggeri Emery, 1894, Dinoponera gigantea (Perty, 1833), and most of the studied Myrmecia species presented multiple rDNA sites over the entire short chromosome arm. The subfamily Myrmicinae possessed most of the rDNA data, and the Myrmeciinae subfamily showed  a different pattern in relation to other subfamilies with multiple rDNA sites observed in the majority of species (Fig. 4C). Dolichoderus voraginosus Mackay, 1993 did not show any co-localization of 18S rDNA clusters and GC-rich bands. Fig. 5 summarizes the available data on the number and position of NORs in ant species and the phylogenetic relationships among these species and is based on the published molecular phylogenies.

Patterns of rDNA clusters in the karyotypes of specific ant groups
Specific rDNA patterns can be observed in the karyotypes of some ant groups where several species have been studied. In species of the genus Dolichoderus, the chromosome numbers range from 2n = 10 to 58, and a single rDNA site in the interstitial region has been observed in the majority of the investigated species. However, there are two exceptions: D. voraginosus and Dolichoderus attelaboides (Fabricius, 1775), where the rDNA clusters are located in the terminal region of the long arm and in the short arm, respectively (Santos et al., 2016). Despite the chromosomal variation observed in Gnamptogenys spp. (2n = 24-44), they all have a single rDNA site in the intrachromosomal region (Teixeira et al., 2020). Even Gnamptogenys moelleri (Forel, 1912), a species with differing chromosome numbers between two populations (2n = 34 and 44), has a single intrachromosomal rDNA site. Gnamptogenys tortuolosa is an exception, with rDNA clusters occurring over the entire long arm (this study). In the genus Anochetus, the chromosome number ranges from 2n = 30 to 46, and a single pericentromeric rDNA site has been observed (Santos et al., 2010).
All analysed fungus-farming ants (Attina) had a single rDNA site (Table 1). In the genus Mycetophylax, although chromosome number differs among species (2n = 26-36), a single rDNA site is located in the pericentromeric or terminal region in all the species of the genus (Micolino et al., 2019a). The leaf-cutting ants are considered the most derived among the Attina species (Schultz and Brady, 2008). Atta spp. (2n = 22) have a single rDNA site in the interstitial region and Acromyrmex spp. (2n = 38) have a single rDNA site in the terminal region (Barros et al., 2015(Barros et al., , 2016Teixeira et al., 2017). Acromyrmex striatus (Roger, 1863), a sister group of the leaf-cutting ants, has 2n = 22 and pericentromeric rDNA clusters that are not located on the same chromosome pair relative to that in Atta spp. (Cristiano et al., 2013;Teixeira et al., 2017). In A. echinatior (2n = 38), rDNA clusters are located in the interstitial region of the same pair as observed in other Acromyrmex spp. (this study).
The chromosome number of the Australian bulldog ants is highly variable, ranging from 2n = 2 to 76 (Imai et al., 1994). In addition, Myrmecia spp. present remarkable patterns of multiple 28S rDNA clusters that are highly dispersed throughout their genomes. The number of rDNA sites increases with the chromosome number of the species. This suggests several ribosomal gene amplification events have occurred in the different species of Myrmecia and that they have accumulated in karyotypes throughout the evolution of the genus. Only four out of 16 species from this monophyletic genus have the entire arm or intrachromosomal rDNA clusters restricted to a single pair of chromosomes (Hirai et al., 1994(Hirai et al., , 1996Hirai, 2020). This pattern is observed in species with small chromosome numbers, which suggests that a single NOR is plesiomorphic among Myrmecia (Hirai, 2020).
Carpenter ants (Camponotus) from the subgenus Myrmothrix have 2n = 40 chromosomes, with its studied species having a single rDNA site in the terminal region. Camponotus renggeri is the only exception, having an additional rDNA cluster at the terminal region of a medium-sized subtelocentric pair .
In the giant ants of the genus Dinoponera, two contrasting patterns have been observed: D. gigantea (2n = 82) has multiple rDNA sites located on its short chromosome arms (Aguiar et al., 2011), whereas Dinoponera lucida Emery, 1901 has a higher chromosome number (2n = 120), but only a single rDNA site restricted to the intrachromosomal region of its largest chromosome pair (Mariano et al., 2008).
Homeology patterns among chromosomal pairs bearing ribosomal genes can be detected in a few ant genera, such as Gnamptogenys (striatula group), Camponotus (Myrmothrix), Atta, and Acromyrmex (Table 1). However, it is speculative to infer such homeology patterns for the entire Formicidae family. Ants constitute an ultra-diverse monophyletic group with more than 13 800 described species (Bolton, 2020). They show a wide range of karyotype variation both in number (2n = 2-120) and chromosomal morphology (reviewed by Lorite and Palomeque, 2010;Mariano et al., 2019).

Insights concerning the organizational patterns of ribosomal gene clusters in the ant genome
The mapping of the ribosomal gene clusters of 63 species distributed in 19 genera and six subfamilies, together with information on their phylogenetic relationships, demonstrated that a single pair of chromosomes bearing the GC-rich rDNA clusters is the most frequent trait among the studied species, regardless of the chromosome number (Table 1, Fig. 4A). Genomes carrying rDNA clusters in more than a single chromosome pair have been observed in non-related taxa, such as D. gigantea (Aguiar et al., 2011), C. renggeri , and Myrmecia spp. (Hirai et al., 1994(Hirai et al., , 1996. We hypothesize that having a single rDNA site should be considered a plesiomorphic trait because multiple rDNA sites were observed in different nonrelated lineages that do not share exclusive common ancestry and appear de novo throughout the Formicidae family.
In eukaryotes, it is common to observe variations in the number of rDNA clusters and the location of these genes in the chromosomes within genera ( . The possession of terminal rDNA clusters seems to be a common trait among mammals, fish, and molluscs, but less so in arthropods (Sochorová et al., 2018). Within the ultra-diverse insect group, the location of rDNA clusters may follow distinct patterns in its two largest orders; terminal rDNA sites are more abundant in Coleoptera, whereas pericentromeric rDNA clusters are more frequent in Orthoptera (Sochorová et al., 2018). In Formicidae, terminal rDNA sites are a less common feature and all species with multiple rDNA clusters show these genes in the entire short chromosome arms including terminal/subterminal regions, such as C. renggeri, D. gigantea, and Myrmecia spp. (Aguiar et al., 2011Hirai et al., 1994Hirai et al., , 1996. Different hypotheses have been proposed to explain the cytogenetic pattern (conservative or variable) of these rDNA clusters in the genome of several organisms. Such hypotheses are linked to the specific locations of these rDNA clusters on the chromosomes. Rearrangements, such as translocations, unequal exchange, and ectopic recombination mechanisms (i.e., between non-homologous chromosomes), which can lead to gene dispersion or increases in number in the genome, are more likely in the terminal/subterminal regions of chromosomes and are uncommon in the intrachromosomal regions (Martins and Wasko, 2004;Mantovani et al., 2005;Nguyen et al., 2010;Roa and Guerra, 2012;Hirai, 2020).
Based on rDNA mapping data in fish karyotypes, Martins and Wasko (2004) proposed that translocations are more likely to occur in telomeric regions because of their proximity within the interphase nucleus, which originates from the ordering of chromosomes according to Rabl's model. Effects due to the location of ribosomal genes in relation to their dispersion in the karyotype were also observed in Coleoptera (Dutrillaux et al., 2016) and primates (Gerbault-Seureau et al., 2017). These authors argued that translocations in the interstitial position could result in abnormal meiosis and, therefore, unbalanced gametes. In contrast, translocations in terminal positions may increase the number of rDNA genes in the genome. This would lead to fewer meiotic abnormalities and highlights the selection for interstitial rDNA site stability (Dutrillaux et al., 2016;Gerbault-Seureau et al., 2017).
Ectopic recombination is another mechanism suggested to explain the rDNA patterning in moths and butterflies (Nguyen et al., 2010) and plants (Roa and Guerra, 2012). It is also included in the recent model proposed by Hirai (2020). In this model, two mechanisms are important: the "site effect" and the "molecular effect." The former allows terminal region associations due to the proximity of these regions in a meiotic bouquet. The "site effect" is a precondition for the "molecular effect," which refers to systems of affinity/non-affinity due to the similarity between rDNA sequences with other repetitive sequences. Thus, rDNA clusters in the terminal regions tend to associate with other repetitive sequences of non-homologous chromosomes more easily, facilitating the occurrence of ectopic recombination and dispersion of these genes in the genome (Hirai, 2020).
There are reports of species with multiple rDNA clusters associated with the centromeres of acrocentric chromosomes (Cazaux et al., 2011). In the recent model proposed by Hirai (2020), the centromeric region of acrocentric chromosomes (chromosomes with a short and heterochromatic arm) that have rDNA genes associated with the centromere may behave as subterminal regions. Therefore, such an arrangement would also facilitate eventual associations of rDNA genes with other repetitive sequences and the occurrence of ectopic recombination, which leads to their dispersal in the genome (for details, see Hirai, 2020).
The ant rDNA chromosome evolution seems to be in accordance with the above-mentioned hypothesis about dispersal and NOR location because the single rDNA clusters of most studied species are interstitial or pericentromeric (Fig. 4B). In ant species, terminal rDNA clusters are  prone to rearrangements that lead to their dispersal. Camponotus renggeri , D. gigantea (Aguiar et al., 2011), and Myrmecia spp. present multiple NORs in the entire short chromosome arms including terminal/subterminal regions, which facilitates the association of these genes with the heterochromatic sequences of other nonhomologous acrocentric chromosomes during meiosis and the subsequent occurrence of ectopic recombination (Hirai, 2020). This pattern may be applicable to different ant groups. In addition, inversions have been shown to change the position of rDNA genes in A. echinatior (this study), Dolichoderus spp. (Santos et al., 2016), and Myrmecia spp. (Hirai et al., 1996).
A single rDNA site located in the terminal region or entire chromosomal arm was observed in some ant species. The repetitive sequences in the subterminal/terminal chromosome regions probably do not form affinity systems with ribosomal genes (the so-called molecular effect; for details, see Hirai, 2020) in these species. Therefore, rDNA clusters are restricted to a single chromosomal pair. Future studies focusing on the characterization of repetitive sequences that make up the heterochromatin of these species will help clarify this hypothesis.
Size heteromorphisms are frequent in karyotypes where the rDNA clusters have terminal positions in the chromosomes, as reported in this study (G. tortuolosa and C. atriceps) as well as in other ants  and insects in general (Cabral-de-Mello et al., 2011;Marya nska-Nadachowska et al., 2016;Andrade-Souza et al., 2018). Subtle variations in the size of the rDNA clusters between homologous chromosomes can be observed as a result of late condensation during cell division (Sumner, 2003). However, large variations, such as those mentioned above, at the terminal region on the chromosome are usually related to duplications/deletions as a result of unequal exchange (Schubert and Lysack, 2011). It is believed that exchanges are less common in intrachromosomal regions (Hirai, 2020).
Size variations in the rDNA clusters can be observed when these genes are located in the interstitial/pericentromeric region of the chromosomes, as seen in Gnamptogenys regularis Mayr, 1870 (Teixeira et al., 2020) and in O. bauri (this study). A different path seems to be involved in the evolution of these karyotypes compared to the rearrangements involved in terminal rDNA heteromorphisms. In these cases, the mechanism may be associated with the formation of extrachromosomal circular DNA (eccDNA), which is likely to form tandem repetitive sequences similar to the rDNA genes (Cohen and Segal, 2009). These eccDNAs may be lost, leading to deletions in the original rDNA sequences, or they may be replicated via a rolling circle mechanism and reintegrated into the original chromosome, producing duplications of these repetitive sequences (Cohen and Segal, 2009).
In ants as well as in other eukaryotes, rDNA clusters are located in GC-rich regions (Symonová, 2019) and, therefore, usually coincide with CMA 3 + bands, possibly as a result of GC-biased gene conversion (gBGC) over the course of evolutionary time (Escobar et al., 2011). This co-localization of GC-rich regions and rDNA was observed for all the ants studied to date (Table 1), with the exception of D. voraginosus (Santos et al., 2016). However, GC-rich chromatin is not always an indication of ribosomal genes, as seen here in G. destructor, P. gilberti, and in some Dolichoderus spp. and fungus-farming ants (Table 1). In insects, a compilation of previous data concerning rDNA genes has been conducted, for example, in moths and butterflies (Lepidoptera; Nguyen et al., 2010), beetles (Coleoptera;Dutrillaux and Dutrillaux, 2012), and kissing bugs (Heteroptera; Panzera et al., 2012). However, this is the first survey of Hymenoptera species. We have compiled available information and new data on 13 Neotropical ant species. In different organisms, including ants, the number and location of chromosomes bearing rDNA clusters within the genome follow general patterns that govern the modes of evolution for these genes (Martins and Wasko, 2004;Nguyen et al., 2010;Dutrillaux et al., 2016;Gerbault-Seureau et al., 2017;Hirai, 2020;this study). We can conclude that having only a single pair of chromosomes bearing rDNA clusters is more common in the ant genome because of the pericentromeric/interstitial location of these genes on the chromosomes. Intrachromosomal regions are sites with low frequencies of rearrangements, such as non-Robertsonian translocations and ectopic recombination, and are therefore less prone to meiotic abnormalities. It should be assumed that the chromosomal location of rDNA clusters influences the dispersion of these genes within the karyotype. Future studies will allow the mapping of rDNA genes in more ant taxa, including the other remaining subfamilies. Other repetitive sequences, such as 5S rDNA and histone genes, may also be mapped in ant species as a tool to investigate further patterns that reflect the relationship between chromosomal location and dispersion in the genome. Finally, a solid understanding of the evolutionary patterns of ribosomal gene dispersal in ant chromosomes may provide a comparative model for other insects.

Experimental procedures
Obtaining samples for analysis Field surveys to collect ant colonies were performed in French Guiana, Brazil, and Panama ( In this study, previous molecular cytogenetic data related to rDNA genes (45S, 18S, or 28S) mapped by FISH from 50 ant species were used. For comparative analysis, the following traits were considered for each species: chromosome number, number of rDNA-bearing chromosomes, location of rDNA clusters in the karyotype, and co-localization of CMA 3 fluorochrome and rDNA clusters. Ag-NOR data were disregarded due to the highly variable patterns and unreliable results (for details, see Introduction).

Chromosome preparation
Mitotic metaphase chromosomes were obtained from the cerebral ganglia of larvae after meconium elimination, using colchicine hypotonic solution (0.005%) and fixatives according to the methods described by Imai et al. (1988).
Staining with fluorochromes chromomycin A 3 (CMA 3 ) and 4 0 6-diamidino-2-phenylindole (DAPI) Metaphase chromosomes of all the species, except A. echinatior, were stained with the fluorochromes CMA 3 and DAPI for the detection of GC and AT-rich regions, respectively, based on the technique proposed by Schweizer (1980). The A. echinatior samples studied here correspond to the same colonies studied by Barros et al. (2016) and the CMA 3 /DAPI staining in this species was performed by these authors.
The rDNA 18S genes were mapped by FISH, following the protocol of Pinkel et al. (1986). The slides were treated with RNase A (100 μg/ml) and kept in a moist chamber at 37 C for 1 h. After that, they were washed in 2 × SSC for 5 min, incubated in 5 μg/ml pepsin in 0.01 N HCl for 10 min, washed in 1 × PBS for 5 min, and dehydrated in 50%, 70% and 100% alcohol series for 2 min each. After this pretreatment, metaphase chromosomes were denatured in 70% formamide/2 × SSC at 75 C for 3 min, and 20 μl of hybridization mix including 200 ng of labelled probe, 2 × SSC, 50% formamide, and 10% dextran sulfate was denatured for 10 min at 85 C and added on preparations. The slides were kept in a moist chamber at 37 C overnight. Then, the slides were washed in 2 × SSC for 5 min; the detection solution including anti-digoxigeninrhodamine was added on slides that were kept in a moist chamber at 37 C for 1 h. The slides were washed in 4 × SSC/Tween and dehydrated in an alcohol series. Finally, counterstaining with DAPI (DAPI Fluoroshield, Sigma Aldrich) was performed.

Chromosomal analysis
Chromosomes were arranged in order of decreasing size and based on the ratio of the chromosomes arm lengths (r = long arm/short arm), according to the classification proposed by Levan et al. (1964). The chromosomes were classified as m = metacentric (r = 1-1.7), sm = submetacentric (r = 1.7-3), st = subtelocentric (r = 3-7), and a = acrocentric (r > 7); they were organized using Adobe Photoshop® 21.1.1 and measured using Image Pro Plus®. Ideograms of the NOR-bearing chromosome/ chromosomes (i.e., graphical representation of the chromosomes concerning the rDNA clusters) of the ant species were then designed with the Easy Idio software (Diniz and Xavier, 2006).
For the fluorochrome staining and FISH 18S rDNA technique, 30 metaphases from at least three individuals of each species were analysed. In the case of O. bauri, which presented a chromosomal polymorphism involving rDNA clusters, seven individuals were analysed (six females and one male). The metaphases were analysed and photographed using a fluorescence microscope, Olympus BX60, attached to an image system, QColor Olympus®, with the filters WB (450-480 nm), WU (330-385 nm), and WG (510-550 nm) for the fluorochromes CMA 3 , DAPI, and rhodamine, respectively.

Phylogenetic relationships
The phylogenetic relationship among ant species was determined by associating with previously published molecular phylogenies. The resultant cladogram topology at the subfamily level was determined following Moreau and Bell (2013). The Poneroid clade topology was determined according to Schmidt (2013) and Larabee et al. (2016); the clade topology for the subfamily Dolichoderinae was determined according to Santos et al. (2016), and that for the subfamily Myrmicinae was determined according to Bacci et al. (2009), Ward et al. (2015, Queiroz EC (2015, unpublished data), Solomon et al. (2019), and Micolino et al. (2019a). The topology of the species groups within the subfamily Myrmeciinae was determined according to Hasegawa and Crozier (2006).