Comparative genomic mapping reveals mechanisms of chromosome diversification in Rhipidomys species (Rodentia, Thomasomyini) and syntenic relationship between species of Sigmodontinae

Rhipidomys (Sigmodontinae, Thomasomyini) has 25 recognized species, with a wide distribution ranging from eastern Panama to northern Argentina. Cytogenetic data has been described for 13 species with 12 of them having 2n = 44 with a high level of autosomal fundamental number (FN) variation, ranging from 46 to 80, assigned to pericentric inversions. The species are grouped in groups with low FN (46–52) and high FN (72–80). In this work the karyotypes of Rhipidomys emiliae (2n = 44, FN = 50) and Rhipidomys mastacalis (2n = 44, FN = 74), were studied by classical cytogenetics and by fluorescence in situ hybridization using telomeric and whole chromosome probes (chromosome painting) of Hylaeamys megacephalus (HME). Chromosome painting revealed homology between 36 segments of REM and 37 of RMA. We tested the hypothesis that pericentric inversions are the predominant chromosomal rearrangements responsible for karyotypic divergence between these species, as proposed in literature. Our results show that the genomic diversification between the karyotypes of the two species resulted from translocations, centromeric repositioning and pericentric inversions. The chromosomal evolution in Rhipidomys was associated with karyotypical orthoselection. The HME probes revealed that seven syntenic probably ancestral blocks for Sigmodontinae are present in Rhipidomys. An additional syntenic block described here is suggested as part of the subfamily ancestral karyotype. We also define five synapomorphies that can be used as chromosomal signatures for Rhipidomys.

The genus Rhipidomys Tschudi, 1845 is the only arboreal representative of the Thomasomyini tribe [1,5,6]. Twenty-five species are currently recognized, with a high degree of morphological similarity that comprises a taxonomically complex group [1,[5][6][7][8][9]. The species of this genus range in length from 90 mm to 210 mm, are difficult to capture due to their arboreal habits, and are among the least well-known species in the Neotropical region [5,6,10]. These rodents occur from eastern Panama and along South America up to northern Argentina [6]. However, the geographical limits of Rhipidomys species are poorly understood [11]. Eleven species occur in the Brazilian biomes [6], among them, Rhipidomys emiliae Allen 1916 is distributed in the Amazon biome and the Amazon-Cerrado ecotone, from Pará to Mato Grosso; Rhipidomys mastacalis Lund 1840 is found in the Atlantic Forest biome and also in central Brazil [6,11] (Fig 1). These two species form a clade together with the Cerrado species, R. ipukensis Rocha, Costa & Costa, 2011. This latter species is closely related to R. emiliae; together, they form a sister group with R. mastacalis [7,12].
In this genus, pericentric inversions have been identified as the main cause of the variation in the number of acrocentric versus bi-armed chromosomes, especially when species with high FN are compared with those with low FN [11,12,16,19,20]. In karyotypes with 2n other than 44 (from the R. nitela group [14]), fusions/fissions or translocations rearrangements have been suggested [14,15,22].
The rodent genome shows great variability in diploid numbers and chromosomal morphology, both between and within species [23]. In this Order, diploid numbers range from 10 in Ctenomys steinbachi (Ctenomyidae) [24] and Akodon sp. (Sigmodontinae) [25,26] to 118 in Dactylomys boliviensis (Echimyidae) [27]. This variability in chromosome numbers can result from Robertsonian translocations (centric fusion and fission), in tandem fusions, or from a variable number of B chromosomes. The variation of chromosomal morphology can also result from pericentric inversions, reciprocal translocations or centromeric repositioning, or from variation in constitutive heterochromatin [23]. Considering the chromosomal diversity observed in rodents, any hypothesis about the origin and evolution of their chromosomes depends on the analysis of conserved syntenies between species. Classical cytogenetics combined with chromosome painting is a useful approach in comparing these karyotypes [28][29][30][31][32]. As a consequence, knowledge about the karyotypic evolution of some groups of rodents has been expanded significantly, such as in species representing the three tribes of Sigmodontinae: using Sigmodon probes in eight species of Sigmodon (tribe Sigmodontini) [33]; using Akodon probes in four species of Akodon (tribe Akodontini) [34]; using Mus musculus probes (Family Muridae) in four species of Akodon, one species of Necromys and Thaptomys (tribe Akodontini) and one species of Oligoryzomys (tribe Oryzomyini) [35,36]; using Oligoryzomys probes in seven species of Oligoryzomys (tribe Oryzomyini) [37]. In addition to these, whole chromosome probes from Hylaeamys megacephalus (HME, Oryzomyini tribe) [30] were developed for studies on comparative genomics in Sigmodontinae, and employed subsequently in six Akodontinini species belong to the genera Akodon, Thaptomys, Necromys, Oxymycterus, Blarinomys and in 13 Oryzomyini species referring to genera Neacomys, Oecomys, Cerradomys [30][31][32][38][39][40][41][42]. Studying the karyotypes of representatives of the Akodontini and Oryzomyini tribes with HME probes allowed researchers to propose characters for the putative ancestral karyotype of the subfamily [32]. This is particularly relevant since HME (Oryzomyini) is phylogenetically close to Thomasomyini and Akodontini (S1 Fig). However, we currently lack chromosomal painting data for Rhipidomys or any other genus of the Thomasomyini tribe.
In the present study, we used chromosomal painting with HME whole chromosome probes [30] and G-banding to investigate chromosomal homologies between the karyotypes of R. emiliae (2n = 44/FN = 50) and R. mastacalis (2n = 44/FN = 74), and to determine if pericentric inversions are the predominant chromosomal rearrangements responsible for  1-4). The distribution data were extracted from the literature [6,7]. Other information from the literature is shown in Table 1. The diploid number (2n) and autosomal fundamental number (FN) are based on the information in Table 1. The numbers refer to localities in Tables 1 and 2. The database was obtained from DIVA-GIS [13].
https://doi.org/10.1371/journal.pone.0258474.g001 the karyotypic divergence between these species, which were selected as representative of the groups with low and high FN, respectively. We selected HME probes because they have already been used in several taxa of the Sigmodontinae [30][31][32][38][39][40][41][42] and are phylogenetically close to the species studied here. This allowed us to compare the karyotype of the Rhipidomys with those previously hybridized with these probes, so that shared chromosomal characters could be identified. The numbers in parentheses refer to localities mentioned in

Samples
We examined six specimens of R. emiliae trapped in two localities of the Pará state, Brazil, and four specimens of R. mastacalis from two municipalities in Minas Gerais state, Brazil (Fig 1).
The Rhipidomys mastacalis specimens were identified according to their morphological features and their external and craniodental measurements, and compared with specimens deposited in the mammal collection of the Museu Nacional (UFRJ, Rio de Janeiro). Rhipidomys mastacalis can be identified by the pelage coloration-gray-brown to more redbrown in the dorsal portion [6]. Cranial characteristics, such as rostrum length, straight supraorbital ridges, not greatly inflated or rounded braincase were also considered for their identifying features as was the derived carotid circulatory pattern. Similarly, the Rhipidomys emiliae specimen was identified by the morphological features and by external and craniodental measurements [6]. This species has a dull grayish-brown to brighter orange-brown agouti dorsal pelage and cream or white ventral pelage, and has craniodental characters such as moderately developed supraorbital ridges diverging posteriorly from a point well forward, resulting in a broad interorbital region, upper toothrow length varying from 4.5 to 5.1 mm, incisive elliptical shaped foramina, and a derived carotid circulatory pattern.
Information about the samples is summarized in Table 2. The skulls and skins used to identify the species are deposited in the mammal collection of the Museu Paraense Emilio Goeldi (MPEG), Museu de Zoologia da Universidade Federal do Pará (MZUFPA), and Museu Nacional, Universidade Federal do Rio de Janeiro (MN).
The specimens were collected following procedures recommended by the American Mammal Society. JCP and LG have permanent field licenses (numbers 13248 and 598633) from the "Chico Mendes Institute for Biodiversity Conservation". The CEABIO Cytogenetics Laboratory at UFPA has authorization from the Ministry of the Environment for the transportation of samples (number 19/2003) and the use of samples for research (number 52/2003). This research was approved by the Ethics Committee of the Federal University of Pará (Permission 68/2015). Animals were euthanized using intraperitoneal injection of barbiturates (pentobarbital, 120 mg/kg) after local anesthesia (lidocaine used topically).

Cytogenetic analysis
Chromosomal preparations were obtained from bone marrow [43] and G-banding [44], Cbanding [45] fluorescent in situ Hybridization (FISH) with human telomeric probes (All Telomere, ONCOR) and chromosome painting [30] were performed according to the described protocols. At least ten metaphases were analyzed in each sample by these techniques.

Classical cytogenetics
Rhipidomys emiliae presents 2n = 44/FN = 50, with 17 autosomal acrocentric pairs (1-17) and four bi-armed pairs (18)(19)(20)(21); the X chromosome is a medium acrocentric and the Y is a small acrocentric (Fig 2A). Constitutive heterochromatin (CH) was found to be equally distributed in the pericentromeric regions of all autosomal pairs and the X chromosome, whereas the Y chromosome is almost entirely heterochromatic, except for the short arm ( Fig 2B).
The R. mastacalis karyotype (2n = 44/FN = 74) was previously published, but the information was restricted to the 2n and NF [18]. Here, we subjected this karyotype to chromosomal painting with HME probes in addition to chromosome banding. The samples revealed five autosomal acrocentric pairs (1-5) and 16 biarmed pairs (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21); the X and Y chromosomes are medium and small acrocentrics, respectively ( Fig 3A). The distribution of CH is not homogeneous: this is evident in the pericentromeric regions of the five acrocentric pairs and pair 16 (submetacentric); in pair 14 (submetacentric) it occurs in the distal region of the short arm of one of the homologs; in the other autosomal pairs it is less evident (in the centromeric region of pair 6) or not visible (in the other pairs); in the X chromosome, there is CH in the centromeric region; and the Y chromosome is almost entirely heterochromatic, except for the distal region of the long arm ( Fig 3B).

Fluorescence in situ hybridization (FISH)
FISH with telomeric probes showed only distal markings, and there was no evidence of ITSs (interstitial telomeric sequences) in the karyotypes of R. emiliae or R. mastacalis (Fig 4).
The hybridization analysis of the 24 HME probes in the karyotypes of R. emiliae (Fig 2A  and Table 3) and R. mastacalis (Fig 3A and Table 3) revealed 36 and 37 regions of chromosomal homology, respectively.
The S2 Fig shows metaphases with hybridizations of all the whole chromosomes probes of HME.
Karyotypes with 2n = 44/FN = 50 have been described for other species of Rhipidomys (R. gardneri, R. itoan, Rhipidomys cf. macconnelli, R. macrurus, R. tribei, and Rhipidomys sp.) (See S1 Table). Considering that many chromosomes of the genus are similar in size, it is not possible to define with certainty whether chromosomes with the same morphology are or are not homoeologous in the karyotypes of the different species. Thus, we are unable to determine whether the agreement of FN in different species indicates that they have similar karyotypes or are cases of homoplasy. Studies with chromosomal painting in these species may contribute more precisely to defining the homologies of the chromosomes involved in changing the FN.
In the present study, R. mastacalis was found to have a karyotype of 2n = 44/FN = 74. This is similar to some descriptions in the literature for this species [16][17][18][19]21], while other authors [14,16,17,20] have reported different values for FN (FN = 72, 76, and 80) ( Table 1 and Fig 1). Considering the size of the chromosomes, we suggest that the difference between these karyotypes is related to chromosomes RMA 3-5 and RMA 21, which may have acrocentric or biarmed morphology.
In the Atlantic forest, R. mastacalis presents two different karyotypes: to the south we find populations with FN = 74 (Table 1 and Fig 1, localities 3 (Table 1 and Fig 1, localities 12-15 and 23). The populations found in isolated mesic forests (mesic enclaves or brejos de altitude) within the Caatinga domain of the northeastern region also have FN = 72 ( [20] Fig 1, locality 11). If we assume a distinct evolution or biogeographic history for the brejos de altitude from northeastern Brazil [7], individuals isolated in these microenvironments provisionally identified as R. mastacalis probably represent specific entities; therefore, populations from this region should be further evaluated [6,7]. The individuals isolated in the brejos of Serra da Ibiapaba (Table 1 and Fig 1, locality  11), which were provisionally identified as R. mastacalis due to FN = 70 (which we corrected for FN = 72, see Table 1) [20], may be elevated to species status (for which the named R. cearanus is available [6]) because of this distant basal position of a clade allying R. emiliae and R. ipukensis to which R. mastacalis forms the sister clade [17]. Other populations also need to be studied in detail, such as Rhipidomys specimens from northern Goiás state provisionally referred to R. mastacallis because of their high FN (FN = 76, 80) (Table 1 and Fig 1, localities 9, 12-21) [14] without reference to the morphological data of the specimens, which may correspond to R. ipukensis [6]. If this is confirmed, it would indicate that the high FN occurs not only in R. mastacalis, but also in R. ipukensis and R. cearanus (specimens from Serra da Ibiapaba) [6,7]. The results of our C-banding analysis (Figs 2B and 3B) showed that there are differences in the amount and distribution of constitutive heterochromatin between R. emiliae and R. mastacalis. In the karyotype of specimens of R. mastacalis described [19], only the acrocentric pairs presented centromeric constitutive heterochromatin. In our samples, there was positive staining on some bi-armed chromosomes in addition to the acrocentric ones, indicating that there is intraspecific variation. These data demonstrate that the polymorphism of Rhipidomys cytotypes extends to heterochromatin and confirm that its addition/deletion is a common process in the genus [15,47].

Comparative mapping between R. emiliae and R. mastacalis
Our comparative chromosome painting analysis between R. emiliae (2n = 44/FN = 50; Fig 2A) and R. mastacalis (2n = 44/FN = 74; Fig 3A) karyotypes showed that there was no detectable difference between the species for eight autosomal pairs and the X chromosome, so these chromosomes are conserved in both species (Table 4). The divergence in FN was due to 12 pericentric inversions or centromeric repositioning and one translocation with inversion (Table 4 and Fig 6). G-banding based comparative analysis of the 12 pairs with changes in morphology (Table 4 and Figs 6 and S3) showed that there were pericentric inversions in four pairs and centromeric repositioning in eight pairs, as the latter pairs maintained the same G-banding pattern, despite their changes in chromosome morphology.
In the other genera of Sigmodontinae rodents investigated by whole-chromosome probes, such as Oligoryzomys [38], Neacomys [39,40], Oecomys [38,42] and Akodon [31,32,34,35], the diversity in 2n and FN was due to pericentric inversions, multiple fusion/fissions, and translocation events. However, Rhipidomys exhibits an unusual karyotypic evolutionary pattern. Data based on classical cytogenetics suggested that pericentric inversions are the predominant rearrangements responsible for the divergence between R. emiliae and R. mastacalis [11,12,16,19,20]. Thus, the new detection of a translocation in the present study demonstrates the great efficiency of the chromosome painting in the comparative analysis of karyotypes. It is possible that other rearrangements that are not visualized with classical methods may exist to differentiate the karyotypes of Rhipidomys species.
Pericentric inversions play an important role in the reorganization of rodent genomes, since they can act in reproductive isolation [48]. Comparative studies of chromosomes in primates, other mammals, and birds have shown that centromeres can change their position throughout evolution without any change in the order of DNA markers around the new centromeric location (centromeric repositioning) [49]. The most parsimonious way for this to occur would be through the inactivation of the original centromere and formation of a new centromere in another location [49,50]. Several evolutionary studies have indicated that centromeric repositioning is not rare in karyotype evolution, and that it should be considered on equal terms with traditional chromosomal rearrangements when examining the evolution of chromosomal structure [50,51]. As seen for inversions, the repositioning of the centromere on a chromosome provides an effective mechanism for reproductive isolation and, therefore, speciation [49]. Inversions can create linkage groups that cause sterility between hybridizing taxa, and natural selection will have a greater opportunity to decrease the frequency of interspecies mattings [52]. They also can reduce gene flow by suppressing recombination and extending the effects of linked isolation genes [53].
In the present study, we observed four pericentric inversions and eight examples of centromeric repositioning (S3 Fig). We speculate that the constancy of this type of rearrangement may be related to the process of reproductive isolation between these species.
The chromosomal evolution process that occurs in Rhipidomys can be classified as "karyotype orthoselection", wherein certain strains acquire a series of rearrangements of a particular type [54,55]. There is evidence that orthoselection may be associated with specific adaptive values that have certain evolutionary connotations [56]. Some rearrangements would be selectively advantageous in mammals due to their effect on gene recombination, where the elevation of FN would increase the amount of recombination [55]. The 13 events of change in the centromeric position between REM and RMA, six of which occurred within conserved blocks and seven between syntenic blocks (Fig 6), can be understood as karyotypic orthoselection.
The phylogenetic analysis of the genus Rhipidomys carried out using cytochrome b showed that R. macconnelli may be one of the first species within the genus to diverge [7]. The fact that this species has 2n = 44/FN = 50 suggests that the ancestral karyotype of the genus would have 2n = 44 and a low FN, from which the high FN would be derived. This evidence is reinforced Table 4. Chromosomal rearrangements that differentiate the R. emiliae and R. mastacalis cytotypes, as identified by HME whole chromosome probes.

Characters shared among species of Sigmodontinae subfamily
Representatives of two tribes of the Sigmodontinae subfamily (Oryzomyini and Akodontini) have been studied with probes from Hylaeamys megacephalus (HME) [30][31][32][40][41][42]. In the present study, we extended this analysis to two species of the Thomasomyini tribe. We found , showing chromosomes involved in their karyotypic divergence, as assessed using HME probes [31]. HME probes are shown beside the idiograms, while each chromosomal pair is identified below. Idiograms within the box correspond to the H. megacephalus karyotype elaborated by Oliveira da Silva et al. [40].
The HME 8, 6/21, 7/[9,10], 11/(16,17), 19/14/19, and 24 (acrocentric) syntenies remain as independent blocks in the R. emiliae and R. mastacalis karyotypes, and are not associated with other probes (Figs 2A and 3A). The HME 20/ (13,22) association is fused to one of the HME 18 segments, originating the HME 18/20/ (13,22) association. In R. emiliae, the HME 20 probe is found as a single block, as has been seen in most species analyzed with the same probe (S2 Table). In R. mastacalis, by contrast, it is fragmented into two blocks of different sizes and the smallest segment resulting from this fission is associated with one of the two HME 1 signals, originating the HME 20/1/ � /1 (RMA 10) association. Therefore, it is possible that, in Rhipidomys, a single signal for HME 20 (REM 5 interstitial) is the original chromosomal condition (Fig 2A), while a dual signal is the derivative form, arising from translocation, and is probably an autapomorphic characteristic of R. mastacalis (Fig 3A).
https://doi.org/10.1371/journal.pone.0258474.g007 (CLA) + Akodon montensis (AMO) + Akodon sp. (ASP); in Rhipidomys it suffered a fission that generated two independent units HME (16,17) (REM 17 and RMA 4) and HME 5. It is therefore difficult to define what region would be homologous to this region of HME 5 in Rhipidomys, since this pair is the most fragmented in Rhipidomys and other species.
The HME 15 (not associated) has been described as a symplesiomorphic character (Table 5). This configuration is present in Oryzomyini and, among the Akodontini, it is observed only in Thaptomys nigrita (TNI) and Oxymycterus amazonicus (OAM). In the other Akodontini and Thomasomyini, it is associated with another HME probe. In R. emiliae and R. mastacalis, this block is associated with one of the HME 5 segments, giving rise to the association HME 5/15 (REM 8 and RMA 14).
The HME 1/12 association suffered a fission to produce two segments, one corresponding to HME 12 (REM 12 and RMA 17) and the other to HME 1. The latter associated with one of the HME 14 fragments, giving rise to the HME 14/1 (REM 7 and RMA 12). Pereira et al. [32] compared their results with the hybridization data obtained using Mus musculus (MMU) probes [33], and found that the association HME 1/12 (MMU 3/18) is absent in Oligoryzomys. Later studies found that this association is also absent in most species of tribe Oryzomyini, occurring only in Neacomys sp. B (NSP-B) and C. langguthi. This association is also absent in Rhipidomys, suggesting it is an exclusive character of the Akodontini tribe. Its presence in Neacomys sp. B and C. langguthi would then be a homoplasy. We observed that in all of these analyzed species, the HME 1 probe shows at least two signals (three in the Oecomys species already analyzed to date), which reinforces the proposal that in H. megacephalus this chromosome is the result of fusion [31].
The association HME 25/3 that has been suggested as an exclusive ancestral character of Akodontini is also present in Rhipidomys. Fissioned HME 18 (18a and 18b, S2 Table) is a character that is also shared among these tribes. Thus, we speculate that these characters were present in the ancestral karyotype of Sigmodontinae and appear in a derived condition in Oryzomyini. Alternatively, it could be a synapomorphy (or even a homoplasy) only between Akodontini and Thomasomyini. Two other associations of Rhipidomys are shared with some Akodontini species: HME 5/11 (REM 16 Table 5).
The associations HME (13,22)/26/25/3, 18/20/(13,22), (13,22)/4, 14/1, and 5/15 were found exclusively in Rhipidomys, and can be considered synapomorphies for the genus. The future inclusion of other representatives of the Thomasomyini tribe in comparative analyses with HME probes may shed further light on whether these characters are exclusive to Rhipidomys, or if they can be considered as signatures for the tribe.

Conclusions
The results of this work demonstrate that the rearrangements responsible for genomic diversification between the karyotypes of R. emiliae and R. mastacalis and, possibly, of the other species of Rhipidomys, involve a combination of translocations, centromeric repositioning and pericentric inversions. The translocation found herein could not be easily detected through classical cytogenetics, and thus our work demonstrates the usefulness of chromosomal painting for such analyses. Comparative analysis with other species of Sigmodontinae shows that Rhipidomys (Thomasomyini) shares the synthetic blocks HME 8, 6/21, 20/ [13,22], 5/11, 7/ [9,10] and 19/14/19 and 24 with species from the Akodontini and Oryzomyini tribes. We also suggest that HME 25/3 association and HME 18a and 18b may be a synapomorphies between Akodontini and Thomasomyini; and fissioned HME 1 may be a symplesiomorphic character for the Sigmodontinae subfamily. It will be interesting to expand the use of HME probes to other species of Rhipidomys, in order to define chromosomal signatures that may be used to elucidate the taxonomic and phylogenetic relationships between species of this genus and enable a better reconstruction of the ancestral karyotype for the Sigmodontinae subfamily. Each probe refers to a chromosome pair, with the exception of HME [9,10], [16,17], [13,22] which are equivalent to 2 pairs of chromosomes each. Avidin-Cy3 (red) and avidin-FITC (green).  Table. FISH signals detected for Sigmodontinae species based on hybridization with Hylaeamys megacephalus (HME) whole-chromosome probes [30]. (DOCX)