Integrating cytogenetics and genomics in comparative evolutionary studies of cichlid fish

BAC-FISH mapping

The chromosome diploid number and morphology were analyzed in Giemsa-stained metaphases (Table 1). Such analysis confirmed the previous karyotype data available for cichlids composed of meta and submetacentric (m/sm), and telo/acrocentric (t/a) chromosomes [17, 47]. Metaphases of males and females of O. niloticus (the species source of the BAC clones used for FISH) were used to set up the BAC-FISH mapping experiments (Figure 1). After that, a comparative mapping, including eight other Pseudocrenilabrinae species, was carried out to elucidate the evolutionary history of chromosomes carrying LG1, 3, 5, and 7 (Figure 2, Additional file 1).

Subfamily	Groups or tribes	Species	2n	Sex	Karyotype	Origin
Etroplinae		Etroplus maculatus	46	1M/1F/1ns	18m/sm+18t/a+10 micro	Petshop, Botucatu, Brazil
Pseudocrenilabrinae	Tilapiine	Oreochromis niloticus	44	2M/3F/1ns	2m/sm+42t/a	TAF-UMD
		Oreochromis mossambicus	44	1M/1F	4m/sm+40t/a	TAF-UMD
		Oreochromis aureus	44	1F	2m/sm+42t/a	TAF-UMD
		Tilapia mariae	40	1M/1F	8m/sm+32t/a	TAF-UMD
	Haplochromine	Astatotilapia latifasciata	44	2M/3F/1ns	12m/sm+32t/a	Petshop, Botucatu, Brazil
		Astatotilapia burtoni	40	1M/2F	14m/sm+26t/a	TAF-UMD
		Metriaclima lombardoi	44	1M/3F/1ns	14m/sm+30t/a	TAF-UMD
		Labeotropheus trewavasae	44	2M/3F	14m/sm+30t/a	TAF-UMD
	Hemichromine	Hemichromis bimaculatus	44	2 ns	4m/sm+40t/a	Petshop, Botucatu, Brazil
Cichlinae	Cichlini	Cichla kelberi	48	2M	48 t/a	Araguaia River, Brazil
	Astronotini	Astronotus ocellatus	48	2M	12m/sm+36t/a	Tietê River, Brazil
	Heroini	Symphysodon aequifasciatus	60	3ns	46m/sm+4t/a+10micro	Petshop, Botucatu, Brazil
	Geophagini	Geophagus brasiliensis	48	1M/1F	2m/sm+46t/a	São Paulo State rivers, Brazil

2n, diploid chromosome number; m/sm, metacentric/submetacentric; t/a, telocentric/acrocentric; micro, microchromosomes; M/F, males/females; ns, not identified sex.

Four BACs from LG1 mapped on the long arm of a small t/a chromosome pair of O. niloticus (Figure 1A). The largest chromosome of O. niloticus evidenced labeling signals of eight BACs of LG3 (Figure 1B) and the LG3 mapping confirmed the presence of a lot of repetitive DNA in the end of the largest chromosome of all analyzed species (Figure 2).

Three BACs from LG5 mapped on a medium t/a chromosome different from that containing the LG1 (Figure 1C), and three BACs from LG7 mapped on the second largest pair of O. niloticus (Figures 1D). No differences were found between males and females for any of the hybridized BAC-probes.

At least two individuals of each species were used in BAC-FISH experiments and the signal patterns observed for LG1, 3, 5, and 7 were conserved in number and position when compared to the reference species O. niloticus (Figure 2; See Additional file 1). No results were obtained for BACs of LG7 in Labeotropheus trewavasae due to the low quality chromosome preparations of this species for FISH procedures (Figure 2). In South American cichlids (Cichlinae), none of the BAC probes produced identifiable chromosomal signals. Furthermore, no hybridization signal was observed in the Asian Etroplinae species, Etroplus maculatus.

Besides the conservation of the studied LGs, some differences were observed in the chromosomes’ morphology of the species (Figure 2): Astatotilapia burtoni showed a different pattern for LG5 which is located on the short arm of a large m/sm chromosome instead of a smal t/a chromosome as observed in the other cichlids. The single BAC signal observed in the short arms of the chromosomes was related to repeated sequences as observed for the UNH115 marker in LG3 of Metriaclima lombardoi (Figure 2; Additional file 1). Additionally, small variations in the chromosomal position of the markers were also observed (See Additional file 1), but these could be just a consequence of variation in the chromosome condensation of the analyzed individuals.

The chromosomal positions of all selected BACs were correlated with their corresponding order on their linkage maps. This enabled the association of LGs with chromosomes and to orientate their position according to the short/long arm of the corresponding chromosomes. Using the convention that the short arm is north oriented to and the long arm is south oriented, the north/south orientation of the current Oreochromis LG1 matches with the short/long arm (Figure 3); however, the orientation of LG3, 5, and 7 should be inverted in relation to the order of previously published molecular markers (Figures 4, 5 and 6).

In silico comparative analyses

LG3 proved to be conserved with at least two correspondent markers between the fish species medaka (chromosome 18) and stickleback (chromosome 7). In the pufferfish genome, markers of LG3 mapped at a non identified region (Table 2). Despite the synteny of LG3 markers identified in medaka and stickleback, an inversion of the markers UNH115 and UNH180, separeted by 8 Mb, was detected in medaka in relation to Oreochromis (Figure 4).

LG3 contains few known genes that could be used for comparative analyses in other vertebrates. CLCN5 was the only gene identified in other vertebrates. Using CLCN5 in a comparative analysis against the zebrafish genome, this gene appears duplicated and located at chromosomes 7 and 21. On the other hand, in all mammal species, CLCN5 has been conserved in X chromosome (Figure 4).

The majority of LG5 markers corresponds to chromosome 5 in medaka, chromosome 17 in stickleback, and chromosome 11 in pufferfish. Although other LG5 markers (Green OPSIN and c-Ski) were found in stickleback and medaka genome, it was not possible to categorize their chromosome position (Table 2). The position of the MME gene differs, as it can appear in chromosome 13, 1, or 16, respectively, in the three mentioned species (Table 2) (Figure 5). With the addition of zebrafish in Ensembl datasheet, an extremely conserved pattern of LG5 in chromosome 11 was evidenced. Once more, a duplicated marker was found in zebrafish (see c-Ski, for instance, in Figure 5). Genes observed in LG5 (AKR, ATP and c-Ski) have their positions conserved in chromosome 1 of primates. Additionally, this conservation is also related to horse chromosome 2, mouse chromosome 4, chicken chromosome 21, and turkey chromosome 23 (Figure 5). Despite located in the same chromosome, the position of these genes are inconsistent - in O. niloticus these genes appear tight together at 17 cM; however, in other species they appear separated, sometimes around 70Mb apart (data not show). The OPN gene (Blue opsin), present in the end of O. niloticus LG5, matched in the X chromosome for all analyzed mammals (Figure 5).

Analysis of LG7 markers in BouillaBase showed a block of synteny between Oreochromis, stickleback chromosome 19 and pufferfish chromosome 13, even taking into account that the distances between these markers are similar among the species. In medaka genome, only the UNH179 marker was identified in chromosome 6 (Figure 6). It was impossible to identify the position of the IGFBP2 marker in medaka and stickleback genomes, but in pufferfish it is located together with DUSP6 (Table 2), as also observed for O. niloticus (Table 2). In Ensembl comparative studies, the analysis of a small region (about 3cM) of LG7 that contains the genes RERG, KCNE1L, DUSP6, and IGFBP2, revealed a conserved position of RERG and DUSP6 at the same chromosome in vertebrates: Homo sapiens chromosome 12, Macaca mulatta chromosome 11, Bos taurus chromosome 5, Gallus gallus and Meleagris gallopavo chromosome 1, and indeed in the reptile Anolis carolinensis chromosome 5. However, these genes are separated by 10 to 75 Mb (data not shown) in non-fish species (Figure 6). Moreover, the gene KCNE1L was conserved in the X chromosome of all mammals (Figure 6).

Cytogenetic mapping and chromosome stability in Pseudocrenilabrinae

The most important find in our study is the extreme chromosome conservation observed within African cichlids. It is already known that African cichlids (Pseudocrenilabrinae) comprise about 1.400 species [48] and that their modal chromosome number is 2n = 44 [17]. Extensive comparative mapping has demonstrated that the genetic maps for tilapia and Malawi cichlids are almost perfectly collinear [41]. This conservation could be evidenced through comparative physical chromosomal mapping of linkage groups as shown in the present study. On the other hand, variability at the cytogenetic level among cichlids was mostly observed based in repetitive DNA chromosome mapping [49, 50] and seems to reflect the evolutionary dynamics of the repetitive genomic fraction and not broad processes acting in the whole karyotype.

Within Pseudocrenilabrinae, the major karyotype feature in the tilapiines is the presence of one large t/a chromosome pair (LG3), which is significantly larger than all other elements of the karyotype [17, 51]. On the other hand, the haplochromine and hemichromine cichlids have shown two outstanding chromosome pairs [17, 52] - the first and the second largest elements that correspond to LG3 and LG7, respectively (Figure 2). The chromosome mapping of LG3 and LG7 markers (present report) in Pseudocrenilabrinae species confirms the proposed homology between the two largest chromosome pairs of tilapiines and haplochromines/hemichromines [52]. However, the hypothesis concerning an independent chromosome fusion that originated the largest chromosomes in tilapiines and haplochromines [52] is not in agreement with the results observed for the cytogenetic mapping of LG3 and LG7 markers.

It is known that the largest chromosome pair of O. niloticus originated by a centric fusion event of three other pairs of the ancestral cichlid karyotype composed of 48 acrocentric chromosomes [53]. A previous study [52] proposed that a first chromosome fusion took place before the divergence of the main East African cichlid groups. The second chromosome fusion occurred independently in the tilapiines and non-tilapiines. In the tilapiines, a new chromosome was fused to the largest pair, and in the non-tilapiines, the second fusion did not involve the largest chromosome, but two other chromosomes, which gave rise to the chromosome pair 2. However, when we compared the distribution of BAC signals (LG3 and LG7) through the long arm of the two largest chromosomes in tilapiines and non-tilapiines species, we detected that the second chromosome fusion seems to be identical in these two groups, producing a chromosome with the same genomic content that remained conserved in all Pseudocrenilabrinae species so far analyzed (Figure 2).

Despite displaying a highly conserved karyotype structure, many events of duplication, inversion, and fusion occurred during the diversification of African cichlids and resulted in diploid chromosome numbers divergent from the pattern of 44 chromosomes, as observed in Tilapia mariae with 2n = 40 [52, 60], T. sparrmanii with 2n = 42 [60], Oreochromis alcalicus with 2n = 48 [17], O. karongae with 2n = 38 [10], and Astatotilapia burtoni with 2n = 40 [17], among others.

A. burtoni has two metacentric chromosome pairs, which are probably the result of centric fusions of two small t/a chromosome pairs. According to our results, the ancestral t/a chromosome that contains the LG5 (observed in other haplochromines) might be involved in one of these events resulting in a m/sm chromosome. Such assumption could not be inferred for T. mariae (tilapiine), in which the diploid chromosome number reduction to 2n = 40 must be the result of rearrangements that did not involve the linkage groups investigated here (Figure 2).

Additionally, the reduction of chromosome number observed in O. karongae (2n = 38) and the presence of three pairs of medium- sized chromosomes, that are not found in the typical Oreochromis species, were originated by chromosome fusion events involving LG1 [10, 61] and could represent recent chromosomal rearrangements that have occurred independently in tilapiine and haplochromine groups. The additional signal observed in the 1p arm (LG3 – UNH115) of the largest chromosome of males and females of M. lombardoi could be a consequence of specific rearrangements or, even more, the presence of chimeric BAC inserts.

The absence of BAC-FISH signals in South American cichlids (Cichlinae) belonging to different tribes (Table 1) could be associated to genomic rearrangements that have disrupted in a small-scale level the genomic blocks carried by the BAC clones of O. niloticus in relation to these cichlines. Instead of being a monophyletic group [13], Neotropical cichlids harbor significantly higher levels of genetic variation compared to the African Pseudocrenilabrinae group [62]. Although genomic rearrangements seem to have occurred differentiating South American and African cichlids, it is plausible that large genomic blocks are still conserved between them as well as in relation to other fish groups. Unfortunately, deep analyses integrating cytogenetics and genomic data were not possible for Cichlinae species since there is no available large-scale genome information for this subfamily so far.

Comparative analyses of vertebrates

Using BAC-end sequence data available at BouillaBase, it was possible to perform comparative analyses and detect regions of synteny among African cichlids and model fish species, and also vertebrates. In this work, conserved (syntenic) chromosome segments have been successfully identified by means of comparative cytogenetics between cichlid markers (mainly LG5 and 7) and mammals and birds chromosomes. The marker distances are very similar among the analyzed teleosts. However, considering non-fish groups, the markers’ distances are, in most cases, very divergent, evidencing that genomes are suffering rearrangements, despite the maintenance of conserved large genomic blocks as part of the same LG. Although large genomic blocks are conserved among vertebrates, variability in the genes/DNA sequences harbored in these regions is expected.

Comparative BAC mapping has already shown that some stickleback chromosomes have a nearly complete synteny with those of cichlids. Although stickleback does not belong to the order Perciformes, it is considered to be the closest related genome to cichlids among the currently available sequenced teleost genomes [63] and is currently the best reference sequence for assembling comparative maps of tilapia [64].

A range of studies using comparative cytogenetics has demonstrated that karyotypes are conserved in a macro scale throughout vertebrates. In Canidae, for example, despite the extensive variation in chromosome numbers and morphology, the majority of conserved chromosome segments appear to have remained largely intact in the karyotypes of extant canid species, although the relative orientation and distances are not always conserved [65]. Conserved karyotypes are also observed in birds, which show a slow rate of interchromosomal rearrangements [9], and within the reptile family Scincidae, a character that was shown by cross-species chromosome painting [66]. Similarly, at different taxonomic levels, comparative gene mapping has revealed a highly conserved linkage homology between an agamid lizard (Leiolepis reevesii) and a snake (Elaphe quadrivirgata) [67].

Conservation of large genomic blocks were also identified among different species of the Salmonidae family based on BAC-FISH blocks [8], and on SOX genes regions in cichlids that revealed a large genomic block that was conserved through vertebrates [12]. These conservations of gene orders at scales of several Mb in diverse vertebrate groups permit the use of relatively complete sequences of model fish species to accelerate gene discovery and positional cloning of non-model species [63, 68]. The synteny observed in different vertebrates could be due to intrinsic chromosomal properties that confer selective pressure on large genomic blocks that should be preserved from major changes [69].

There is a classical idea that fish genomes have high rates of chromosomal rearrangements compared to other vertebrates and then, probably, evidence low rates of synteny [70–73]. However, our data associated to previous genomic studies are revealing that when we compare the genomic synteny among teleosts and other vertebrates in a macro scale level, large syntenic blocks can be clearly identified.

Comparative analysis of sex chromosome regions

Three genes from different LGs of cichlids (CLCN5-LG3, OPSIN-LG5, and KCNE1L-LG7) are located on the X chromosome of human and other mammals, highlighting the conservation of this sex chromosome through mammals [74]. However, in bird clade these genes do not correspond to the Z or W chromosomes and, instead of this, are located in autosomes. This is explained by the fact that the ZW and XY chromosomes have no homology and both sexual chromosome pair systems were derived from different autosomes from their common ancestors [75].

In contrast to mammals and birds, the pufferfish, like most fish, does not possess heteromorphic sex chromosomes, and the genetic mechanism(s) of sex determination is still unclear. However, the human X is a mosaic of orthologs from three chromosomes of the pufferfish: most human Xp and Xq genes are syntenic on pufferfish chromosomes 1, 2, and 7 [69]. In addition, zebrafish LGs 9 and 23 are also related with both human Xp and Xq orthologs [18, 76].

The largest chromosome of Nile tilapia that contains the LG3 was previously thought to be the sex chromosome of this species [77–80]. However, the major sex-determining region in the Nile tilapia was mapped on LG1 [81] located in a small t/a chromosome [present work, 7]. LG1 also contains the WT1 and CYP19a genes involved in mammalian sex differentiation [42], although they are not considered anymore to be candidate genes for sex determination in African cichlids [38]. Even though our results are still limited, the integrated comparative analysis approach seems to be promising in the clarification of the complex evolutionary dynamics of sex chromosomes among fishes.

Animals and sampling

Cichlids from Lake Malawi were collected from the wild from 2005–2008 and maintained in the Tropical Aquaculture Facility (TAF) of the University of Maryland (UMD), College Park, MD, USA. Additional African and Asian species of uncertain origin were obtained from commercial sources in Botucatu, SP, Brazil, and South American species were collected from the wild in Brazilian rivers (Table 1) and maintained in the fish room of the Laboratory of Genômica Integrativa (FR-LGI) at Sao Paulo State University (UNESP), Botucatu, SP, Brazil. All the examined specimens were fixed in formaldehyde and then stored in alcohol in the fish collections of TAF-UMD and FR-LGI.

BAC clones and probes labeling

BAC clones containing specific markers of LG1, 3, 5, and 7 (Table 3) were obtained from a BAC library of the Nile tilapia, O. niloticus, which was previously developed [32], and were used as probes for FISH. BAC extraction was conducted using the PhasePrep®^TM BAC DNA Kit (Sigma-Aldrich, St Louis, MO, USA) according to supplier’s protocol. The BAC clones were labeled with biotin, digoxigenin coupled nucleotides (Roche Applied Sciences, Indianapolis, IN, USA), CY3- and CY5-avidin (GE-Healthcare, UK) using whole genome amplification (WGA2 & 3 kits) (Sigma-Aldrich), according to the supplier’s protocol. After that, DNase-I (Sigma-Aldrich) concentration was titrated to yield labeled DNA fragments ranging from 100 to 500 base pair products. For multicolor FISH, we used 16 μl of probe mixture containing: 10 μl of hybridization mixture (62.5% deionized formamide, 12.5% of 20XSSC, 12.5% of dextran sulfate 50%), 4 μg of blocking DNA (salmon/herring sperm DNA or Cot-1 DNA resuspended in the hybridization mixture) and 100 nanograms of each probe. The probe mixture was denatured for 10 min at 65°C and immediately cooled on ice.

Chromosome preparation and FISH procedure

Mitotic chromosomes of cichlid species belonging to the Pseudocrenilabrinae (including representatives of tilapiine, haplochromine, and hemichromine groups), Cichlinae and Etroplinae subfamilies (Table 1) were prepared from anterior kidney cells with in vivo colchicine treatment [82]. The slides with chromosomes were air-dried, treated with pepsin (0.01% in 10 mM HCl) and dehydrated in an ethanol series one day before use. The slides were denatured in 70% formamide/2xSSC, pH 7.0 for 40 s, and dehydrated in an ice-cold ethanol series. The probe mixture was hybridized under a 24 x 50 mm cover slip in a 37°C moist chamber for 48 h. Slides were washed two times for 5 min each in 50% formamide/2xSSC, pH 7.0 at 43°C under agitation, then 10 min in 2xSSC, pH 7.0 at 42°C under continuous agitation. For undirected labeled probes, the hybridization signals were detected with avidin-FITC and rhodamine-anti-DIG (Roche Applied Sciences, Indianapolis, IN, USA), according to the supplier’s protocol. After three washes of 2 min in phosphate buffer detergent (4xSSC/1% Tween-20), slides were mounted with antifade solution containing DAPI (Vectashield mounting medium). Results were recorded with an Olympus BX61 microscope equipped with an Olympus digital camera DP71 and the software Image-Pro MC 6.0.

The selected BACs were checked in order to identify if the FISH mapping results were in agreement to the anchoring marker positions on LGs. Each group of BACs from a particular linkage group was checked to ensure if it hybridizes to the same chromosome using at least two markers from each LG.

Comparative genomic database analyses

The comparative analyses among O. niloticus BAC-end sequences and Oryzias latipes (medaka), Tetraodon nigroviridis (pufferfish) and Gasterosteus aculeatus (stickleback) were done based on the BAC ID (Table 3) of the clones used for FISH against the updated version of the comparative genome browsers of the BouillaBase database [46]. The BAC ID was the starting point to detect a landmark or region of similarity in the databank source. The identification of genomic positions of genes located in LG1, 3, 5, and 7 of O. niloticus in other vertebrates was determined using the currently available public genomic databases Sanger Institute Ensembl Database [83], Genomicus genome browser [84], and NCBI Map Viewer [85]. Once a query gene name has been entered in the search box, the position of the gene in all species that have their genome available and annotated was retrieved. The accession numbers of all genes used in the comparative analyze are available as Additional file 2.