Alu elements and hominid phylogenetics

Alu Ye5 subfamily members were identified in the draft sequence of the human genome (August 6, 2001, University of California, Santa Cruz, GoldenPath assembly) by using blast (48) queries of the draft sequence to identify exact complements to the oligonucleotide 5′-GAACCCCGGGGGGCGGAGCCTGCAG-3′, which is diagnostic for the Ye5 subfamily. We used this approach to identify 153 elements that comprise the Alu Ye5 subfamily. A complete list of all of the Alu elements identified in the searches can be found in Table 1, which is published as supporting information on the PNAS web site, www.pnas.org and at http://batzerlab.lsu.edu.

To aid in elucidating the relationship between the two chimpanzee species (Pan paniscus and Pan troglodytes) and to further clarify relationships among all of the great apes, we searched the available sequences from each of these species in GenBank by using the Alu Y subfamily consensus sequence as a blast query. Sequences containing matches of 90% or greater were submitted to blat (http://genome.ucsc.edu; April 2003 assembly) in 20,000-bp batches to identify the homologous region of the human genome and examined computationally to determine whether the Alu insertion in Pan was also present in Homo. Ifthe Alu appeared to be specific to chimpanzee, that locus was passed to the next step: primer design.

Oligonucleotide primers for PCR amplification of each of the Alu loci were designed as described (13, 15, 17). Using the primers and annealing temperatures for Alu Y lineage subfamily members shown in Table 2, which is published as supporting information on the PNAS web site and at http://batzerlab.lsu.edu, we amplified the loci from HeLa genomic DNA and from up to 12 individuals of each nonhuman primate species including Pan troglodytes (common chimpanzee), Pan paniscus (bonobo), Gorilla gorilla (lowland gorilla), Pongo pygmaeus (orangutan), Chlorocebus aethiops sabaeus (green monkey), and Aotus trivirgatus (three-striped owl monkey), which were available as a primate phylogenetic panel (PRP00001) from the Coriell Institute for Medical Research, and from Hylobates syndactylus (siamang). For the Ye5 subfamily, 117 elements identified in the draft human genomic sequence were amenable to PCR analysis. Fourteen loci obtained from the blast search of Pan spp. sequences were amenable to PCR analysis. Examination of the orthologous regions of the various nonhuman primate genomes displayed a series of different PCR bands pattern indicative of the time of insertion of each element into the primate genomes (Fig. 2). The distribution of all elements in the various primate genomes tested can be found in Table 3, which is published as supporting information on the PNAS web site and at http://batzerlab.lsu.edu.

paup 4.0b10 (49) was used to perform a parsimony analysis on the 131 loci described above and two previously identified loci (Yi6AH36 and Yd3JX167) (18, 45) that were informative with regard to the human–chimpanzee–gorilla trichotomy issue. Dollo parsimony is most appropriately applied when handling presence/absence characters for which the ancestral state can safely be assumed (50). Thus, we implemented an exhaustive search in paup using Dollo parsimony and designating a New World monkey (Aotus trivirgatus) as the outgroup taxon. If no amplification was observed for a given locus in any taxon, the character state was coded as unknown. To determine branch support for each node, 10,000 bootstrap replicates were performed. The complete data set is available in Table 3.

Gel-purified PCR products amplified from genomic DNA templates (human, bonobo, common chimpanzee, and gorilla) were cloned by using the TOPO TA cloning vector (Invitrogen). Isolates from these cloning reactions were sequenced by using chain termination sequencing (51) on an Applied Biosystems 3100 automated DNA sequencer. The sequences of the orthologous Alu insertion loci (bonobo, common chimpanzee, and gorilla) have been assigned accession numbers (AY318829–AY318854). Sequences of all of the Ye5 subfamily members were aligned by using megalign software (DNAStar version 3.1.7 for Windows 3.2). A multiple alignment of all of the Alu Ye5 subfamily members can be found in Table 1.

To estimate the age of each informative Alu insertion in the human, chimpanzee, and gorilla branch of the tree, mutations from the subfamily consensus sequence were divided into those that occur at CpG dinucleotides and those that occur at non-CpG positions without including small insertions or deletions and ages were estimated as described (13, 15, 17). The mutations are divided into these two different classes because the CpG base positions in repeated sequences mutate at a rate that is ≈10 times higher than non-CpG positions (4, 12) as a result of the spontaneous deamination of 5-methycytosine residues (52). Differences between the two age estimates based on CpG and non-CpG mutation density presumably reflect the heterogeneous level of methylation throughout the genome which influence the CpG mutation spectra of the individual Alu elements differently (53). Here we report age estimates based on CpG mutation density. The estimates were obtained by dividing the number of mutations from the consensus sequence of the subfamily by the neutral rate for CpG mutations (0.0146 per million years).

No Alu Ye5 insertions were observed in either green monkey or owl monkey. Thus, we can state that the AluYe5 subfamily began to expand after the divergence of hominids from the remainder of the primates (≈25 mya) (30). Of the 123 variable loci, 101 were parsimony-informative. Ninety-five loci support nodes unambiguously. Ambiguous loci were defined as loci where nonamplification in a particular taxon made defining the boundaries of a clade uncertain. Parsimony analysis resulted in a single most parsimonious tree (Fig. 3; 124 steps, consistency index = 0.992, RI = 0.995, HI = 0.008). With the exception of one node, bootstrap support was 100%. The node joining the subtribe Hominina (Homo and Pan) was supported with a value of 99%. The patterns observed clearly indicate a stepwise pattern of insertion reflecting the relative divergence of each group in the hominid lineage. The green monkey (Chlorocebus aethiops sabaeus) was unambiguously placed external to Hominidae (Hylobates, Pongo, Gorilla, Pan, and Homo) by 16 loci. Monophyly of Hominini (Pongo, Gorilla, Pan, and Homo) was supported by 18 unambiguous loci out of 19. Thirty-three of 35 loci unambiguously supported monophyly of the great apes (gorillas, chimpanzees, and humans).

At eight Alu Ye5 loci and two previously identified Alu Yi and Yd loci (18, 45), amplification of filled sites was obtained in human, bonobo, and common chimpanzee. Three of these loci did not amplify filled or empty sites in Gorilla and are thus only ambiguously informative with regard to the trichotomy issue. However, five Alu Ye5 insertions as well as the Yi and Yd insertions were present in the human, bonobo, and common chimpanzee genomes but clearly absent from the gorilla genome. Therefore, a total of seven Alu insertions unequivocally support the hypothesis that the chimpanzee is the nearest living relative of humans as suggested by previous studies (22–24, 27, 28, 31–33). Six of seven Ye5 loci are unambiguously human-specific. Fourteen loci contain Alu elements that are specific to one or both chimpanzees. Thirteen of these belonged to the Yc1 subfamily (17) and one (Yx2DRPan28_2) may belong to an as yet undescribed subfamily. Of these 14 Pan-specific insertions, six were present in both P. paniscus and P. troglodytes. We are therefore able to distinguish the Homo and Pan lineages from one another and support monophyly of the two chimpanzee species with a support value of 100%.

We identified one Alu element (Ye5AH137) that was present in the genomes of human and gorilla but was absent from bonobo and common chimpanzee genomes. This locus, which supports a Homo–Gorilla clade, is interesting and may be explained by insertion homoplasy at first glance. However, sequence analysis of the locus indicated that it was the result of incomplete lineage sorting of an ancestral polymorphic locus (Fig. 4). Age estimates of this and the other seven Alu insertions at the Homo–Pan node also support this explanation. We calculated the relative ages of each Alu insertion by using CpG mutation densities. Estimates of insertion times for each of the seven loci supporting the Homo–Pan clade range from 2.74 to 5.48 mya (mean = 4.11 ± 1.12 mya). The Ye5AH137 insertion appears to be older than the others with an estimated age of ≈6.85 million years. Previously reported sequence data suggests that gorillas shared a common ancestor with humans, chimpanzees, and bonobos 6.2–8.4 mya (26), around the time of the Ye5AH137 insertion. Thus, if the ancestral population were polymorphic for the presence of an Alu insertion at this locus during this period, then fixation of one or the other character state would have occurred randomly in the daughter lineages. In this case, it appears that the insertion was fixed in both human and gorilla lineages but not in chimpanzee. Furthermore, these data lend support to the hypothesis that humans, chimpanzees, and gorillas diverged during a relatively narrow evolutionary time frame (25). It should be noted that Takahashi et al. (43), in their study of cichlid phylogeny, also found a single SINE locus that may have been influenced by lineage sorting, although interspecies hybridization could also have been invoked in that case.

In contrast, two loci that join Homo and Pan (Ye5AH31 and Ye5AH93) had an estimated age of 2.74 million years, substantially younger than the estimated time of the human chimpanzee divergence (≈4–6 mya). Although this is interesting in that these loci suggest very recent coancestry between Homo and Pan, it should be noted that age estimates were based on numbers of CpG mutations from the consensus sequence for the subfamily under examination. The stochastic nature of single nucleotide mutations and the short evolutionary times involved here produce large variances in age estimates for individual Alu elements (12). Therefore, age estimates based on the mean age of several Alu insertions are more accurate than estimates based on individual Alu repeats. Thus, the apparently young age estimates for these two insertions should not be taken as evidence for a more recent divergence between chimpanzees and humans when the mean age of all of the relevant insertions is ≈4.11 million years.

The utility of SINE insertions, and mobile elements in general, for phylogenetic analysis continues to be bolstered by studies such as this one. Here, we present the first application of SINEs to fully elucidate the phylogeny of the hominid lineage and present the strongest evidence to date for phylogenetic relationships among the hominid lineages. Of the 133 Alu insertion loci, 95 were unambiguously informative for determining the relative divergence of each of the major lineages. In addition, seven Alu insertions informative with regard to the Homo–Pan–Gorilla trichotomy unambiguously support monophyly of humans and chimpanzees. Finally, six loci discovered by using searches of the currently available chimpanzee genomic sequences unambiguously join the two Pan species to the exclusion of Homo. Only a single insertion from the entire data set showed any potential evidence of insertion homoplasy; however, the distribution of this element appears to be the result of lineage sorting of an ancestral polymorphism.

This paper was submitted directly (Track II) to the PNAS office.

Abbreviations: SINE, short interspersed elements; mya, million years ago.

Data deposition: The sequences of the orthologous Alu insertion loci (bonobo, common chimpanzee, and gorilla) have been deposited in the GenBank database (accession nos. AY318829–AY318854).

We thank Dr. Caro-Beth Stewart for providing Siamang DNA. This research was supported by Louisiana Board of Regents Millennium Trust Health Excellence Fund HEF (2000-05)-05, (2000-05)-01, and (2001-06)-02 (to M.A.B.), National Institutes of Health Grant RO1 GM59290 (to M.A.B. and L.B.J.), and National Science Foundation Grants BCS-0218338 (to M.A.B.) and BCS-0218370 (to L.B.J.).