Molecular Phylogenetics of Mastodon and Tyrannosaurus rex
Chris L. Organ,1,2 Mary H. Schweitzer,3,4 Wenxia Zheng,3 Lisa M. Freimark,5 Lewis C. Cantley,5,6 John M. Asara5,6*
We report a molecular phylogeny for a nonavian dinosaur, extending our knowledge of trait evolution within nonavian dinosaurs into the macromolecular level of biological organization. Fragments of collagen {alpha}1(I) and {alpha}2(I) proteins extracted from fossil bones of Tyrannosaurus rex and Mammut americanum (mastodon) were analyzed with a variety of phylogenetic methods. Despite missing sequence data, the mastodon groups with elephant and the T. rex groups with birds, consistent with predictions based on genetic and morphological data for mastodon and on morphological data for T. rex. Our findings suggest that molecular data from long-extinct organisms may have the potential for resolving relationships at critical areas in the vertebrate evolutionary tree that have, so far, been phylogenetically intractable.
1 Harvard University, Cambridge, MA 02138, USA.
2 Museum of Comparative Zoology, Cambridge, MA 02138, USA.
3 North Carolina State University, Raleigh, NC 27695, USA.
4 North Carolina Museum of Natural Sciences, Raleigh, NC 27601, USA.
5 Beth Israel Deaconess Medical Center, Boston, MA 02115, USA.
6 Harvard Medical School, Boston, MA 02115, USA.
* To whom correspondence should be addressed. E-mail: jasara@bidmc.harvard.edu
Protein sequences from bone-derived collagen as old as 68 million years have been detected by using mass spectrometry (1, 2). A BLAST search of the resulting peptide sequences found the highest similarities between the collagen peptides obtained from extracts of Tyrannosaurus rex fossil bone and those of birds (Gallus gallus) and between mastodon (Mammut americanum) and other mammals, including elephant (Loxodonta africana). We performed phylogenetic analyses to infer the evolutionary relationships of the T. rex [Museum of the Rockies (MOR) 1125] and mastodon (MOR 605). The results extend our knowledge of trait evolution within nonavian dinosaurs into the macromolecular level of biological organization.
We used 21 extant organisms in the analyses. Collagen {alpha}1(I) and {alpha}2(I) protein sequences from 19 extant organisms (3) were obtained from the National Center for Biotechnology Information (NCBI) and ENSEMBL databases. Collagen {alpha}1(I) and {alpha}2(I) peptide sequences from a metacarpal of Alligator mississippiensis were obtained by microcapillary liquid chromatography tandem mass spectrometry (LC/MS/MS) with an ion trap mass spectrometer using the Sequest algorithm (1, 2) and the Paragon algorithm (4). Ostrich collagen sequences were determined elsewhere (1, 2). Sequences from alligator (Crocodilia) and ostrich (Aves) represent 63% and 43% of the full-length collagen {alpha}1(I) protein available for other organisms, respectively. The collagen {alpha}1(I) and {alpha}2(I) sequences for elephant (L. africana), tenrec (Echinops telfairi), and green anole (Anolis carolinensis) were obtained by using gene translations from online databases (5).
Bayesian, likelihood, parsimony, and distance methods were used to generate evolutionary trees (6). In the Bayesian analysis, the posterior distribution of trees reconstructed all extant groups in generally agreed-upon relationships (the posterior probability of clades ranged from 0.80 to 1.00), with the exception of green anole (A. carolinensis), which is inferred here to lie at the base of amniotes instead of grouping as the sister taxon to alligator and birds (archosaurs) (Fig. 1). LC/MS/MS from tryptic digests produces fragmentary protein sequence data; however, we found unequivocal support (posterior probability of 1.00) uniting mastodon with elephant as members of Elephantinae, which together group with tenrec (E. telfairi) as members of the mammalian group Afrotheria (7). Maximum likelihood produces the same groupings, although with less support (approximate likelihood ratio test; aLRT = 0.855 for Elephantinae and 0.872 for Afrotheria). Maximum parsimony analysis also groups mastodon, elephant, and tenrec together (fig. S1, B to D). For the T. rex sample, we used five peptide sequences from collagen {alpha}1(I) and one from collagen {alpha}2(I) (1, 2, 8) for a total of 89 amino acids (Fig. 1). The T. rex clusters within the Archosauria (posterior probability of 0.92), more closely related to birds (chicken and ostrich, 0.9) than alligator, although a lack of informative sites in the ostrich and T. rex leaves Dinosauria unresolved. The likelihood tree is identical to the Bayesian tree, except for higher support at these locations in the tree (aLRT = 0.969 for Archosauria and 0.907 for Dinosauria). Branch lengths (expected rates of change per site) indicate a relatively stable and uniform rate of evolution, lacking evidence for a deviation from a molecular clock. Maximum parsimony analysis also groups the T. rex with the chicken and ostrich, although bootstrap support is low (fig. S1, B to D). Neighbor joining groups the T. rex with the birds, but miscalculates the branching order and misplaces alligator, mastodon, and several extant organisms (fig. S1, B to D).
Figure 1 Fig. 1. Inferred evolutionary relationships of major vertebrate groups hypothesized from collagen {alpha}1(I) and {alpha}2 (I) protein data by using a Bayesian approach. The node (bifurcation) labels are measures of support, which indicate the proportion of trees in the posterior distribution to containing the node. Branch lengths are in expected changes per site. [View Larger Version of this Image (48K GIF file)]
The slight disagreement between the distance results compared with the Bayesian, likelihood, and parsimony results (all three of which are concordant) are predictable given that distance methods perform poorly for taxa with large amounts of missing data (9). Nevertheless, there is congruence between three out of the four methods for the mastodon and four out of four methods for the T. rex despite the problem of missing sequence data. The recovered sequences contain informative phylogenetic signal consistent with predictions based on genetic and morphological data for mastodon (10, 11) and on morphological data for T. rex (12, 13).
These results support the endogenous origin of the preserved collagen molecules and confirm the prediction based on morphology that, if biomolecules could be retrieved from a nonavian dinosaur, they would share a higher degree of similarity with birds than with other extant vertebrates. Our findings suggest that molecular data from long-extinct organisms may have the potential for resolving relationships at critical areas of the vertebrate evolutionary tree that have, so far, been intractable. The findings presented here also bolster the use of morphology in phylogenetics because our results are consistent with studies on the evolutionary relationships of fossil forms that rely on morphology (12, 13).
