Molecular Organization of the 5S rDNA in Engystomops
This study verified the occurrence of two types of 5S rDNA in the genomes of the Amazonian species of Engystomops, a feature widely documented in many vertebrates, including fish [10, 11, 36–42], the anurans Xenopus laevis and Xenopus borealis [6, 16, 21–24, 43, 44] chickens [45–47], and mammals [12, 48, 49]. Similarly to the findings reported in these studies, the two types of 5S rDNA sequences found in E. freibergi and E. petersi varied slightly in their corresponding coding regions, with the main difference between them found in the NTS region, which varied in length (84 bp for the type I 5S rDNA and approximately 650 bp for the type II 5S rDNA) and nucleotide composition. These two types of 5S rDNA do not, however, appear to be related to the dual system observed in Xenopus (i.e., the oocyte and somatic types) [6, 13, 16, 20–22, 24] because we found no similarities between the NTSs of the oocyte- or somatic-type sequences to those in the present study.
Since Ohno's publication , the origin of genic variants has been attributed to events of sequence duplication followed by processes that result in the divergence of the duplicated sequences. This hypothesis has been corroborated by many studies (reviewed in references [51–55]) and may also explain the presence of two types of 5S rDNA. Gene duplication may result from unequal crossing over, retropositioning, or chromosomal (or genomic) duplication , and the outcomes of these events are quite different, including neofunctionalization, pseudogene origin or simple preservation of gene duplicates [50, 56, 57].
Although our data for the Amazonian Engystomops do not allow us to elucidate the events involved in the origin of either type of the 5S rDNA, those events may have preceded the divergence of Engystomops and Physalaemus. Such an inference follows from the observation of higher nucleotide divergence between the sequences of the type I and type II 5S rDNAs of each species in these genera than among the sequences of the same type found in distinct species.
With respect to the functionality of the 5S rDNA sequences found in Engystomops, the analysis of the secondary structure of the presumed rRNAs shows that both types of sequences are consistent with the general eukaryotic 5S rRNA structure [58, 59], suggesting that the type I and type II 5S rDNA sequences may have transcriptional potential. The functionality of the type I 5S rDNA sequences of Engystomops is corroborated by the recognition, in this type of sequences, of elements that are similar to those considered to be important for the transcriptional activity of 5S rRNA genes. Those type I 5S rDNA elements are: (i) sequences quite similar to the ICR elements; (ii) a T-rich region downstream from the presumed coding region; (iii) a TATA-box located 25 nucleotides upstream from the coding region; and (iv) the nucleotide C at position -1. In addition, in the type I 5S rDNA of Engystomops, the presence of a GAACAAA segment was noted, which is very similar to the sequence GAAACAA suggested to act as a terminal region of the 5S gene transcription in fish . A region tentatively named the TATA-like region was also found in the Engystomops type I 5S rDNA, located 12 nucleotides upstream from the TATA-box. Campo et al.  reported an additional TATA-like region in the NTS regions of the 5S rDNA sequences of the fishes Merluccius merluccius, Merluccius senegalensis, and Merluccius capensis, and suggested that this TATA-box may serve as a "backup". The same hypothesis may be considered for the similar sequence found in the NTS of the type I 5S rDNA of Engystomops.
In contrast, some doubts remain about the transcriptional potential of the type II 5S rDNA sequences of Engystomops. In the type II repeats, the nucleotide at position -1 is a T and not a C; the ICR segment differed more from the ICRs of the other vertebrates used for comparison than the ICR of the type I repeats; and no TATA-box was found in the NTS. The only segment that resembles a TATA-like motif in the type II repeats of Engystomops was observed very distant from the region considered to be the coding region, approximately at position -420. However, a T-rich region downstream from the presumed coding region is also present in the Engystomops type II 5S rDNA.
It is also interesting to note that the results of previous experiments with Xenopus suggest that the oligonucleotides AGAAGC and AAAAGT, located at positions -28 to -23 and -18 to -13, respectively, may be involved in the initiation of 5S rDNA transcription instead of a TATA-box . In the type II 5S rDNA sequences of Engystomops, the hexanucleotides AGAAGC and GCAAGT were found at positions -53 to -48 and -21 to -16, respectively. The similarity of these oligonucleotides to those described for Xenopus, despite the low similarity of the remaining NTS sequence, is an interesting issue to be considered in further analyses of the functionality of the Engystomops type II 5S rDNA.
Physical Mapping of the 5S rDNA in the Engystomops karyotypes X evolutionary diversification in 5S rDNA
The FISH assays suggest the existence of two sites of sequence accumulation of the 5S rDNA in the karyotypes of E. freibergi and E. petersi, one on 3p (E. freibergi and E. petersi from Puyo) or 5p (E. petersi from Yasuní) and another on 6q. The results of these assays also suggest that the former site is exclusive to or preferentially constitutes type I sequences, whereas the latter, on 6q, is associated with type II sequences.
Ribosomal DNA repeating units are evolutionarily dynamic and appear to be able to spread throughout the genome, creating new rDNA loci [3, 60–62]. The presence of two distinct 5S rDNA sequence types organized in different chromosomal regions or even on different chromosomes has been described for several fish, e.g., Salmo solar , Oncorhynchus mykiss , Coregonus artedi, C. zenithicus , and Oreochromis niloticus .
In anurans, the first chromosome mapping experiments for the 5S rRNA genes were conducted in Xenopus laevis. Using specific probes, Harper et al.  revealed a differential distribution of the two types of 5S rDNA in the Xenopus karyotypes, mapping the somatic-type 5S rDNA to the distal end of the long arm of Chromosome 9 in Xenopus laevis and X. borealis, and the oocyte-type to the distal ends of the majority of Xenopus laevis chromosomes. The authors also mapped a trace oocyte-type 5S rDNA in X. laevis, which is a minor class of the oocyte type, to the distal end of the long arm of Chromosome 13.
In addition to these data for the Xenopus karyotypes, only the 5S rDNA chromosomal sites were detected in the karyotypes of Physalaemus ephippifer  and Physalaemus cuvieri . A probe containing the entire repeat of the type I 5S rDNA of P. cuvieri detected a pericentromeric region of the short arm of Chromosome 3 in both Physalaemus species karyotypes [18, 19] whereas a probe with the entire repeat of the type II 5S rDNA of P. cuvieri preferentially detected a distal region of the long arm of Chromosome 6 . Similarly to the above-mentioned cases, a differential localization of the two types of 5S rDNA found in the Engystomops species was observed in this study.
As mentioned above, taking into account the similarity of the sequences, probably the origin of the two types of 5S rDNA found in Engystomops and Physalaemus species preceded the divergence of these genera. Apparently, the origin of this dual-system involved translocation or transposition events that lead to the separation of two groups of 5S rDNA sequences, favoring the dominance of divergence forces over homogenization processes between these groups. On the other hand, the homogeneity of the 5S rDNA repeats clustered in the same chromosomal site was maintained, what may be explained by concerted evolution. As a result of these processes, two distinct types of 5S rDNA sequences, occupying different chromosomal sites, have arisen. It is worth mentioning that purifying selection may also have been involved in this scenario. In addition to concerted evolution, purifying selection has been invoked to justify the homogenization in a gene family [64, 65]. Since the comparison between both types of 5S rDNA sequences of Engystomops showed a higher variation between the presumed coding-regions than between their NTSs, it is likely that purifying selection has been acting over these coding-regions, avoiding high level of divergence.
Another intriguing finding of this study was the hybridization of the probe that corresponds to the type I NTS to the centromeric region of various chromosomes, sites which were not detected in this analysis by the probes that potentially contain the transcribed region of the 5S rDNA. A possible explanation for this result is that the sequences associated with the centromeric regions are segments of satellite DNA derived from the 5S rDNA, a phenomenon previously reported for the fish Hoplias malabaricus  and the frog Physalaemus cuvieri .
The 5S rDNA cluster has been reported to be linked to the major rDNA sequences [4, 36, 37, 67–69] but, in some cases, is localized to different chromosomes [18, 63, 70–73]. The differential chromosomal localization of the 5S and 45S rDNAs is the prevalent condition, not only in the cited examples but also in other groups, including plants [74–77].
Several authors have previously discussed the prevalence of different chromosomal sites for the 45S and 5S rDNAs over their linkage in other organisms, and a probable explanation is intrinsically related to the repetitive nature of these sequences. Others have suggested that because tandem repeated sequences are frequently involved in events of unequal crossing-over and gene conversion, the separation of the two great families of ribosomal DNA at different chromosomal sites would avoid disruptive interference in its organization such as undesired rearrangements between the 45S and 5S arrays [3, 11].
The 5S rDNA chromosomal sites unlinked to the NORs of the Engystomops species represent new cytogenetic markers to be considered for their karyotypic comparison. Based on classic cytogenetic techniques, CMA3 and DAPI staining, and in situ localization of nucleolar rDNA, Targueta et al.  described the three Engystomops karyotypes of the present study and noted that recognizing chromosomal homeology was difficult, especially between the Yasuní karyotype and the other two. In addition, difficulties in differentiating the three very morphologically similar chromosome pairs (pairs 3, 6, and 8) in the Yasuní karyotype have been reported. In the present study, we were able to suggest a homeology between Chromosome 6 of the E. petersi karyotype from Yasuní and Chromosome 6 of E. freibergi and E. petersi from Puyo based on the mapping of the type II 5S rDNA sequences. Additionally, this chromosome site also constitutes a distinctive marker for Chromosome 6 in the karyotype of E. petersi from Yasuní, distinguishing it from Chromosome 8 and the NOR-bearing Chromosome 3. Finally, the mapping of the type I 5S rDNA sequence to Chromosome 5 of the specimens from Yasuní and Chromosome 3 of E. freibergi and of E. petersi from Puyo, which are similar chromosomes in size and morphology, may suggests that these chromosomes are homeologous.
In addition to the recognition of chromosomal homeologies among the Engystomops species, the 5S rDNA mapping performed here allows for a better cytogenetic comparison of Engystomops with its sister genus, Physalaemus. Chromosomes 3 of P. cuvieri  and P. ephippifer , bearing the type I 5S rDNA sequence, are morphologically similar to Chromosome 5 of E. petersi from Yasuní and Chromosome 3 of E. freibergi and E. petersi from Puyo. Therefore, the homeology of all these chromosomes can be strongly inferred. Similarly, we can deduce homeology among the metacentric Chromosome 6 of the Engystomops species and Chromosome 5 of P. cuvieri, which all carry the type II 5S rDNA sequences. Chromosome 5 or 6 of P. ephippifer likely also bears this type of sequence; however, the presence of this sequence has not been verified with a specific probe for type II 5S rDNA sequences . These cytogenetic data suggest that the chromosomal sites of the 5S rDNA may be conserved in these leiuperid genera, which has not been observed for the NORs [18, 33, 78–84]. Therefore, the two 5S rDNA arrays appear to be independent units of evolution in Engystomops species, and further studies of their functionality and their relation to a possible centromeric DNA satellite sequence are necessary to provide a better understanding of the evolution of these sequences.