Introduction

Speciation is the evolutionary process by which populations evolve to become distinct species. Several models and theories have been proposed for this highly complicated process, including gene regulatory networks, community ecology, and mating preferences (for a review see [1]). Natural selection may be considered a major outcome associated with, and linking the above propositions. With an exceptionally high degree of polymorphism and plasticity, short tandem repeats (STRs) (also known as microsatellites/simple sequence repeats) may be a spectacular source of variation required for speciation and evolution [2,3,4,5,6]. The impact of STRs on speciation is supported by their various functional implications in gene expression, alternative splicing, and translation [4, 7,8,9,10,11,12,13].

STRs are a source of rapid and continuous morphological evolution[14], for example, in the evolution of facial length in mammals[15]. These highly evolving genetic elements may also be ideal responsive elements to fluctuating selective pressures. A role in evolutionary selection and adaptation is consistent with deep evolutionary conservation of some STRs, as “tuning knobs”, including several in genes with neurological and neurodevelopmental function[16].

While a limited number of studies indicate that purifying selection and drift can shape the structure of STRs at the inter- and intra-species levels [17,18,19,20,21,22], the global abundance of STRs at the crossroads of speciation remains largely unknown.

Mononucleotide and dinucleotide STRs are the most common categories of STRs in the vertebrate genomes[23, 24]. In addition to their association with frameshifts in coding sequences and pathological [25] and possibly evolutionary consequences, recent evidence indicates surprising functions for the mononucleotide STRs, such as their proposed role in translation initiation site selection[12, 26]. Several groups have found evidence on the involvement of a number of dinucleotide STRs in gene regulation, speciation, and evolution[4, 23, 27,28,29,30]. Trinucleotide STRs are frequently linked to human neurological disorders, most of which are specific to this species[31, 32].

Here, we analyzed the global hierarchical clustering of all types of mono-, di-, and trinucleotide STRs in nine mammalian species, encompassing primates and rodents, Those species belong to the superordinal group of Euarchontoglires [33], and form three distinct and unambiguous phylogenetic < clusters>. The aim of this analysis was to examine whether the global abundance of STRs in the selected species conforms to the phylogenetic < clusters > of the selected species, or not.

Materials and methods

Species and whole-genome sequences

The UCSC genome browser (https://hgdownload.soe.ucsc.edu) was used to download and analyze the latest genome assemblies of nine species as follows (genome sizes are indicated following each species): rat (Rattus norvegicus): 2,647,915,728, mouse (Mus musculus): 2,728,222,451, gelada (Theropithecus gelada): 2,889,630,685, olive baboon (Papio anubis): 2,869,821,163, macaque (Macaca mulatta): 2,946,843,737, gorilla (Gorilla gorilla gorilla): 3,063,362,754, chimpanzee (Pan troglodytes): 3,050,398,082, bonobo (Pan paniscus): 3,203,531,224, and human (Homo sapiens): 3,099,706,404. Those species encompassed rodents: rat and mouse, Old World monkeys: gelada, olive baboon, macaque, and great apes: gorilla, bonobo, chimpanzee, human.

Extraction of STRs from genomic sequences

The whole-genome abundance of mononucleotide STRs of ≥ 10-repeats, dinucleotide STRs of ≥ 6-repeats, and trinucleotide STRs of ≥ 4-repeats were studied in the nine selected species. To that end, we designed a software package in Java (https://github.com/arabfard/Java_STR_Finder). All possibilities of mononucleotide motifs, consisting of A, C, T, and G, all possibilities of dinucleotide motifs, consisting of AC, AG, AT, CA, CG, CT, GA, GC, GT, TA, TC, and TG, and all possibilities of trinucleotide motifs, consisting of AAC, AAT, AAG, ACA, ACC, ACT, ACG, ATA, ATC, ATT, ATG, AGA, AGC, AGT, AGG, CAA, CAC, CAT, CAG, CCA, CCT, CCG, CTA, CTC, CTT, CTG, CGA, CGC, CGT, CGG, TAA, TAC, TAT, TAG, TCA, TCC, TCT, TCG, TTA, TTC, TTG, TGA, TGC, TGT, TGG, GAA, GAC, GAT, GAG, GCA, GCC, GCT, GCG, GTA, GTC, GTT, GTG, GGA, GGC, and GGT were analyzed.

The written program calculated based on perfect (pure) STRs. The algorithm started from an initial point, which was the first nucleotide of each genome, and iteratively repeated a series of steps during walking on the genome, nucleotide by nucleotide. In the first step, it investigated a window frame of 2*N, where 2 was the definition of tandem repeats i.e., two identical continuous sequences, and N was the length of the STR core. If the first half of the sequence inside the window was not equal to the second half, the algorithm moved one nucleotide forward. If equal, the algorithm checked the nucleotides, and this process continued until all identical continuous nucleotides, which were the same as the core were found. The final selected sequence- M*N- was introduced as a new STR, which had a core with a length of N and M repeats. All steps were repeated to find new STRs from the end of the previous STR. We repeated the algorithm for different values of N (N was between 1 and 3 in each genome to detected mono, di, and trinucleotide STRs).

Whole-genome STR data aggregation, abundance, and hierarchical cluster analysis across species

Whole-genome chromosome-by-chromosome data were aggregated and analyzed in the nine species. STR abundances across the selected species were obtained and depicted by boxplot diagrams and hierarchical clustering, using boxplot and hclust packages[34] in R, respectively. Boxplots illustrate abundance differences among segments across the selected species, and hierarchical clustering plots demonstrate the level of similarity and differences across the obtained abundances. The input data to these packages were numerical arrays . Each array consisted of a number of columns, each column corresponding to the STR abundance in different chromosomes. It should be noted that the focus of our analysis was to evaluate the global abundance of STRs across those species, regardless of the homologous regions.

Statistical analysis

The STR abundances across the nine selected species were compared by repeated measurements analysis, using one and two-way ANOVA tests. These analyses were confirmed by nonparametric tests.

Results

Global abundance of mono, di, and trinucleotide STRs coincides with the phylogenetic distance of the nine selected species

Whole-genome data was collected on the abundance of mononucleotide STRs across the nine species (Table 1). We found massive expansion of the mononucleotide STR compartment in all primate species versus rat and mouse. Hierarchical clustering yielded three < clusters > as follows: <rat, mouse>, <gelada, olive baboon, macaque>, and < gorilla, chimpanzee, bonobo, human>, which coincided with the phylogenetic distance of the nine selected species (P = 6.3E-09) (Fig. 1) namely < rodents>, <Old World monkeys>, and < great apes>.

Table 1 Mononucleotide STR abundance across the nine selected species
Full size table
Fig. 1
figure 1

Whole-genome mononucleotide STR abundance in the nine selected species. Global incremented pattern was observed in the primate species versus rodents (left graph). The overall hierarchical clustering yielded three <clusters>, which conformed to <rodents>, <Old World monkeys>, and <great apes> (right graph).

Full size image

The whole-genome STR abundances from aggregated chromosome-by-chromosome analysis in the dinucleotide category (Table 2) was decremented in primates versus rodents. Similar to the mononucleotide STR compartment, the dinucleotide STR compartment conformed to the genetic distance among the three < clusters > of species (P = 7.1E-08) (Fig. 2).

Table 2 Dinucleotide STR abundance across the nine selected species
Full size table
Fig. 2
figure 2

Whole-genome dinucleotide STR abundance in the nine selected species. Global decremented patterns were observed in all primate species versus mouse and rat (left gragh). The global pattern conformed to the three <clusters> across the nine species and their phylogenetic distance (right graph)

Full size image

There was global shrinkage of the trinucleotide STR compartment in primates versus rodents (P = 3.8E-05) (Table 3; Fig. 3). Remarkably, human stood out among all other species in the trinucleotide STR compartment.

Table 3 Trinucleotide STR abundance across the nine selected species
Full size table
Fig. 3
figure 3

Whole-genome trinucleotide STR abundance in the nine selected species. While global decremented patterns were observed in primates versus rodents (left graph), human stood out in this category, in comparison to all other species (right graph)

Full size image

Differential abundance patterns of various STRs and STR lengths across rodents and primates

Numerous STRs and STR lengths across the mono, di, and trinucleotide STR categories conformed to the phylogenetic distances of the nine selected species, for example, in the instance of T/A mononucleotides of 10, 11, and 12 repeats, which were the most abundant STRs across all nine species (Fig. 4). In another example, (ct)6 and (taa)4 conformed to the phylogeny of the studied species in the di and trinucleotide STR categories, respectively.

Fig. 4
figure 4

Example of STRs and STR lengths, abundance of which coincided with the phylogeny of the nine selected species. Three STRs are depicted as examples for each of mono, di, and trinucleotide categories. Data from all studied STRs are available at: https://figshare.com/articles/figure/STR_Clustering/17054972

Full size image

On the other hand, numerous STRs did not follow perfect phylogenetic patterns, such as (C)10, (at)8, and (ttg)4 (Fig. 5). Hierarchical clusters of all studied STRs across the three categories are available at: https://figshare.com/articles/figure/STR_Clustering/17054972.

Fig. 5
figure 5

Example of STRs and STR lengths, abundance of which appeared to be predominantly random across the nine selected species. Three STRs are depicted as examples for each of mono, di, and trinucleotide categories. Data from all studied STRs are available at: https://figshare.com/articles/figure/STR_Clustering/17054972

Full size image

Discussion

While the mechanisms underlying speciation are extremely complicated and largely based on theories and models, the impact of genetics seems to be significant in respect of adaptation, gene flow, and natural selection. In fact, natural selection may be a central converging point of the evolutionary propositions for speciation. However, the various mechanisms involved in speciation have different impact on natural selection, and it is the net effect which may ultimately result in the emergence of a new species.

As one of the most abundant genetic elements in various animal genomes, it is largely unknown whether at the crossroads of speciation, STRs evolved as a result of purifying selection, genetic drift, and/or in a directional manner.

Here, we selected multiple species across rodents and primates, and investigated the clustering patterns of all possible types and lengths of mononucleotides, dinucleotide, and trinucleotide STRs on the whole-genome scale in those species. Hierarchical clustering yielded clusters that predominantly conformed to the phylogenetic distances of the selected species. Hierarchical clustering is an unsupervised clustering method that is used to group data. This algorithm is unsupervised because it uses random, unlabeled datasets. As the number of clusters increases, the accuracy of the hierarchical clustering algorithm improves.

Our findings may be of significance in a number of aspects. Firstly, there were significant differential abundances separating rodents from primates, for example, massive decremented abundance of dinucleotide and trinucleotide STRs in primates versus the rodent species, and massive incremented abundance of mononucleotide STRs in primates versus rodents. Secondly, the three major < clusters > obtained from global hierarchical cluster analysis matched the phylogeny of the three < clusters > of species, i.e., <rodents>, <Old World monkeys>, and < great apes>. It is possible that there are mathematical channels/thresholds required for the abundance of STRs in various orders. This is in line with the hypothesis that STRs function as scaffolds for biological computers[35]. In addition, our data indicate that various STRs and STR lengths behave differently with respect to their colossal abundance. Not all the studied STRs conformed to the phylogenetic distances of the nine selected species. We hypothesize that those which did, had a link with the speciation of those species, whereas those which did not, apparently followed random patterns for the most part. The potential effect of STRs in non-genic regions is largely unknown. However, when located at genic regions, various STRs and repeat lengths can potentially recruit transcription factors (TFs), which differ in qualitative and quantitative terms (http://alggen.lsi.upc.es/cgi-bin/promo_v3/promo/promoinit.cgi?dirDB=TF_8.3) [36]. Those various TF sets may differentially regulate expression of the relevant genes during the process of evolution. For example, T-blocks of 10, 12, and 14-repeats recruit various combinations of FOXD3, HNF-3, and Hb (Fig. 6). Interestingly, (T)10 and (T)12 were among the mononucleotide STRs, which conformed to the phylogenetic distance of the nine species (Fig. 4), and (t)14 did not (https://figshare.com/articles/figure/STR_Clustering/17054972). The concept of various TF sets stands for other STRs as well. For example, (ct)6 conforms to the phylogenetic clusters, and recruits a number of TFs, whereas (ct)7, which does not conform to those clusters, recruits quantitatively different set of those TFs (Fig. 7).

Fig. 6
figure 6

Potential recruitment of qualitatively and quantitatively different TFs to various lengths of (T)-repeats. (T)10 (A) and (T)12 (B) conformed to the phylogenetic < clusters>, whereas (T)14 (C) did not. Differential recruitment of TFs may differentially regulate the relevant genes in evolutionary processes

Full size image
Fig. 7
figure 7

Potential differential TF recruitments to various lengths of (ct)6 A) and (ct)7 B). Those two lengths result in alternative quantitative binding of three TFs. (ct)6 conformed and (ct)7 did not conform to the phylogenetic < clusters>

Full size image

Mononucleotide STRs impact various processes, such as gene expression, translation alterations, and frameshifts of various proteins, which may have evolutionary and pathological consequences[12, 25]. They can overlap with G4 structures, many of which associate with evolutionary consequences[37].

In a number of instances, dinucleotide STRs located in the protein-coding gene core promoters have been subject to contraction in the process of human and non-human primate evolution[38]. A number of those STRs are identical in formula in primates versus non-primates, and the genes linked to those STRs are involved in characteristics that have diverged primates from other mammals, such as craniofacial development, neurogenesis, and spine morphogenesis. Structural variants are enriched near genes that diverged in expression across great apes[39], and genes with STRs in their regulatory regions are more divergent in expression than genes with fixed or no STRs[40]. STR variants are likely to have epistatic interactions, which can have significant consequences in complex traits, in human as well as model organisms[6, 41].

Trinucleotide STRs are predominantly focused on in human because of their link with several neurological disorders[42,43,44,45]. We found an exceptional global hierarchical distance between human and all other species in that compartment. In view of the fact that most of the phenotypes attributed to trinucleotide STRs are human-specific in nature, it is conceivable that their evolution is also significantly distant from all other species studied.

The observed abundances were independent of the genome sizes of the selected species. For example in the instances of di- and trinucleotide STRs, we observed higher abundances in rodents versus primates despite the smaller genome sizes of the former. These findings are in line with the previous reports of lack of relationship between genome size and abundance of STRs[46, 47].

It should be noted that this is a pilot study based on hierarchical clustering, and future studies are warranted to further examine our hypothesis, using phylogenetic platforms and additional orders and species. Functional studies are also warranted to examine the biological impact of the relevant STRs.

Conclusion

We propose that the global abundance of STRs is non-random across rodents and primates. We also propose the STRs and STR lengths, which predominantly conformed to the phylogenetic distances of those species, such as (t)10, (ct)6, and (taa4). Additional species encompassing other orders and phylogenetic platforms are warranted to further examine this proposition.

Limitations

This research was a pilot study based on hierarchical clustering of the collected data in a number of mammalian species. Phylogenetic platforms and additional orders of species are warranted to further examine our hypothesis.