A phylogenetic foundation for comparative mammalian genomics.

A major effort is being undertaken to sequence an array of mammalian genomes. Coincidentally, the evolutionary relationships of the 18 presently recognized orders of placental mammals are only just being resolved. In this work we construct and analyse the largest alignments of amino acid sequence data to date. Our findings allow us to set up a series of superordinal groups (clades) to act as prior hypotheses for further testing. Important findings include strong evidence for a clade of Euarchonta+Glires (=Supraprimates) comprised of primates, flying lemurs, tree shrews, lagomorphs and rodents. In addition, there is good evidence for a clade of all placental mammals except Xenarthra and Afrotheria (=Boreotheria) and for the previously recognised clades Laurasiatheria, Scrotifera, Fereuungulata, Ferae, Afrotheria, Euarchonta, Glires, and Eulipotyphla. Accordingly, a revised classification of the placental mammals is put forward. Using this and molecular divergence-time methods, the ages of the superordinal splits are estimated. While results are strongly consistent with the earliest superordinal divergences all being >65 mybp (Cretaceous period), they suffer from greater uncertainty than presently appreciated. The early primate split of tarsiers from the anthropoid lineage at ~55 mybp is seen to be an especially informative fossil calibration point. A statistical framework for testing clades using SINE data is presented and reveals significant support for the tarsier/anthropoid clade, as well as the clades Cetruminantia and Whippomorpha. Results also underline our thesis that while sequence analysis can help set up hypothesised clades, SINEs obtainable from sequencing 1-2 MB regions of placental genomes are essential to testing them. In contrast, derivations suggest that empirical Bayesian methods for sequence data may not be robust estimators of clades. Our findings, including the study of genes such as TP53, make a good case for the tree shrew as a closer relative of primates than rodents, while also showing a slower rate of evolution in key cell cycle genes. Tree shrews are consequently high value experimental animals and a strong candidate for a genome sequencing initiative.

Plastid genome phylogeny and a model of amino acid substitution for proteins encoded by chloroplast DNA.

Maximum likelihood (ML) phylogenies based on 9,957 amino acid (AA) sites of 45 proteins encoded in the plastid genomes of Cyanophora, a diatom, a rhodophyte (red algae), a euglenophyte, and five land plants are compared with respect to several properties of the data, including between-site rate variation and aberrant amino acid composition in individual species. Neighbor-joining trees from AA LogDet distances and ML analyses are seen to be congruent when site rate variability was taken into account. Four feasible trees are identified in these analyses, one of which is preferred, and one of which is almost excluded by statistical criteria. A transition probability matrix for the general reversible Markov model of amino acid substitutions is estimated from the data, assuming each of these four trees. In all cases, the tree with diatom and rhodophyte as sister taxa was clearly favored. The new transition matrix based on the best tree, called cpREV, takes into account distinct substitution patterns in plastid-encoded proteins and should be useful in future ML inferences using such data. A second rate matrix, called cpREV*, based on a weighted sum of rate matrices from different trees, is also considered.

Appropriate likelihood ratio tests and marginal distributions for evolutionary tree models with constraints on parameters.

We show how to make appropriate likelihood ratio tests for evolutionary tree models when parameters such as edge (internodes or branches) lengths have nonnegativity constraints. In such cases, under the null model of an edge length being zero, the marginal distribution of this parameter is proven to be a "half-normal", that is, 50% zero values and 50% the positive half of a normal distribution. Other constrained parameters, such as the proportion of invariant sites, give similar results. To make likelihood ratio tests between nested models, e.g., H(0): homogeneous site rates, and H(1): site rates follow a gamma distribution with variance 1/k, then asymptotically as sequence length increases, the distribution under H(0) becomes a mixture of chi distributions, in this case 50% chi(0), and 50% chi(1) (where the subscript denotes degrees of freedom, i.e. , not the usually assumed 100% chi(1); which leads to a conservative test). Such mixtures are sometimes called distributions. Simulations show that even with sequences as short as 125 sites, some parameters, including the proportion of invariant sites, fit asymptotic distributions closely.

Correspondence analysis of genes and tissue types and finding genetic links from microarray data.

In this paper, we propose and use two novel procedures for the analysis of microarray gene expression data. The first is correspondence analysis which visualizes the relationship between genes and tissues as two 2 dimensional graphs, oriented so that distances between genes are preserved, distances between tissues are preserved, and so that genes which primarily distinguish certain types of tissue are spatially close to those tissues. For the inference of genetic links, partial correlations rather than correlations are the key issue. A partial correlation between i and j is the relationship between i and j after the effect of surrounding genes has been subtracted out of their pairwise correlation. This leads to the area of graphical modeling. A limitation of the graphical modeling approach is that the correlation matrix of expression profiles between genes is degenerate whenever the number of genes to be analyzed exceeds the number of distinct expression measurements. This can cause considerable problems, as calculation of partial correlations typically uses the inverse of the correlation matrix. To avoid this limitation, we propose two practical multiple regression procedures with variable selection to measure the net, screened, relationship between pairs of genes. Possible biases arising from the analysis of a subset of genes from the genome are examined in the worked examples. It seems that both these approaches are more natural ways of analyzing gene expression data than the currently popular approach of two way clustering.

Cluster inference methods and graphical models evaluated on NCI60 microarray gene expression data.

At present, there is a lack of a sound methodology to infer causal gene expression relationships on a genome wide basis. We address this first by examining the behaviour of some of the latest and fastest algorithms for tree and cluster analysis, particularly hierarchical methods popular in phylogenetics. Combined with these are two novel distances based on partial, rather than full, correlations. Theoretically, partial correlations should provide better evidence for regulatory genetic links than standard correlations. To compare the clusters obtained by many alternative methods we use tree consensus methods. To compare methods of analysis we used tree partition metrics followed by another level of clustering. These, and a tree fit metric, all suggest that the new distances give quite different trees than those usually obtained. In the second part we consider graphical modeling of the interactions of important genes of the cell cycle. Despite the models seeming to fit well on occasions, and despite the experimental error structure seeming close to multivariate normal, there are considerable problems to overcome. Latent variables, in this case important genes missing from the analysis, are inferred to have a strong effect on the partial correlations. Also, the data show clear evidence of sampling distributions conditional on the status of important cancer related genes, including TP53. Without full information on which genes are wild type the appropriate models cannot be fitted. These findings point to the need to include and distinguish not only all relevant genes but also all splice variants in the design phase of a microarray analysis. Failure to do so will induce problems similar to both latent variables and conditional distributions.

Assessing the Cretaceous superordinal divergence times within birds and placental mammals by using whole mitochondrial protein sequences and an extended statistical framework.

Using the set of all vertebrate mtDNA protein sequences published as of May 1998, plus unpublished examples for elephant and birds, we examined divergence times in Placentalia and Aves. Using a parsimony-based test, we identified a subset of slower evolutionary rate placental sequences that do not appear to violate the clock assumption. Analyzing just these sequences decreases support for Marsupionta and the carnivore + perissodactyl group but increases support for armadillo diverging earlier than rabbit (which may represent the whole Glires group). A major theme of the paper is to use more comprehensive estimates of divergence time standard error (SE). From the well-studied horse/rhino split, estimated to be 55 million years before present (mybp), the splitting time within carnivores is confidently shown to be older than 50 million years. Some of our estimates of divergence times within placentals are relatively old, at up to 169 million years, but are within 2 SE of other published estimates. The whale/cow split at 65 mybp may be older than commonly assumed. All the sampled splits between the main groups of fereuungulates (the clade of carnivores, cetartiodactyls, perissodactyls, and pholidotes) seem to be distinctly before the Cretaceous/Tertiary boundary. Analyses suggest a close relationship between elephants (representing Afrotheria) and armadillos (Xenarthra), and our timing of this splitting is coincident with the opening of the South Atlantic, a major vicariant event. Recalibrating with this event (at 100 mybp), we obtain younger estimates for the earliest splits among placentals. Divergence times within birds are also assessed by using previously unpublished sequences. We fail to reject a clock for all bird taxa available. Unfortunately, available deep calibration points for birds are questionable, so a new calibration based on the age of the Anseriform stem lineage is estimated. The divergence time of rhea and ostrich may be much more recent than commonly assumed, while that of passerines may be older. Our major concern is the rooting point of the bird subtree, as the nearest outgroup (alligator) is very distant.

Using novel phylogenetic methods to evaluate mammalian mtDNA, including amino acid-invariant sites-LogDet plus site stripping, to detect internal conflicts in the data, with special reference to the positions of hedgehog, armadillo, and elephant.

We look at the higher-order phylogeny of mammals, analyzing in detail the complete mtDNA sequences of more than 40 species. We test the support for several proposed superordinal relationships. To this end, we apply a number of recently programmed methods and approaches, plus better-established methods. New pairwise tests show highly significant evidence that amino acid frequencies are changing among nearly all the genomes studied when unvaried sites are ignored. LogDet amino acid distances, with modifications to take into account invariant sites, are combined with bootstrapping and the Neighbor Joining algorithm to account for these violations of standard models. To weight the more slowly evolving sites, we exclude the more rapidly evolving sites from the data by using "site stripping". This leads to changing optimal trees with nearly all methods. The bootstrap support for many hypotheses varies widely between methods, and few hypotheses can claim unanimous support from these data. Rather, we uncover good evidence that many of the earlier branching patterns in the placental subtree could be incorrect, including the placement of the root. The tRNA genes, for example, favor a split between the group hedgehog, rodents, and primates versus all other sequenced placentals. Such a grouping is not ruled out by the amino acid sequence data. A grouping of all rodents plus rabbit, the old Glires hypothesis, is also feasible with stripped amino acid data, and rodent monophyly is also common. The elephant sequence allows confident rejection of the older taxon Ferungulata (Simpson, 1945). In its place, the new taxa Scrotifera and Fereuungulata are defined. A new likelihood ratio test is used to detect differences between the optimal tree for tRNA versus that for amino acids. While not clearly significant as made, some results indicate the test is tending towards significance with more general models of evolution. Individual placement tests suggest alternative positions for hedgehog and elephant. Congruence arguments to support elephant and armadillo together are striking, suggesting a superordinal group composed of Xenarthra and African endemic mammals, which in turn may be near the root of the placental subtree. Thus, while casting doubt on some recent conclusions, the analyses are also unveiling some interesting new possibilities.

Mammalian evolution: timing and implications from using the LogDeterminant transform for proteins of differing amino acid composition.

We explore the tree of mammalian mtDNA sequences, using particularly the LogDet transform on amino acid sequences, the distance Hadamard transform, and the Closest Tree selection criterion. The amino acid composition of different species show significant differences, even within mammals. After compensating for these differences, nearest-neighbor bootstrap results suggest that the tree is locally stable, though a few groups show slightly greater rearrangements when a large proportion of the constant sites are removed. Many parts of the trees we obtain agree with those on published protein ML trees. Interesting results include a preference for rodent monophyly. The detection of a few alternative signals to those on the optimal tree were obtained using the distance Hadamard transform (with results expressed as a Lento plot). One rearrangement suggested was the interchange of the position of primates and rodents on the optimal tree. The basic stability of the tree, combined with two calibration points (whale/cow and horse/rhinoceros), together with a distant secondary calibration from the mammal/bird divergence, allows inferences of the times of divergence of putative clades. Allowing for sampling variances due to finite sequence length, most major divergences amongst lineages leading to modern orders, appear to occur well before the Cretaceous/Tertiary (K/T) boundary. Implications arising from these early divergences are discussed, particularly the possibility of competition between the small dinosaurs and the new mammal clades.

The complete mitochondrial DNA sequence of the shark Mustelus manazo: evaluating rooting contradictions to living bony vertebrates.

A remarkable example of a misleading mitochondrial protein tree is presented, involving ray-finned fishes, coelacanths, lungfishes, and tetrapods, with sea lampreys as an outgroup. In previous molecular phylogenetic studies on the origin of tetrapods, ray-finned fishes have been assumed as an outgroup to the tetrapod/lungfish/coelacanth clade, an assumption supported by morphological evidence. Standard methods of molecular phylogenetics applied to the protein-encoding genes of mitochondria, however, give a bizarre tree in which lamprey groups with lungfish and, therefore, ray-finned fishes are not the outgroup to a tetrapod/lungfish/coelacanth clade. All of the dozens of published phylogenetic methods, including every possible modification to maximum likelihood known to us (such as inclusion of site heterogeneity and exclusion of potentially misleading hydrophobic amino acids), fail to place the ray-finned fishes in a biologically acceptable position. A likely cause of this failure may be the use of an inappropriate outgroup. Accordingly, we have determined the complete mitochondrial DNA sequence from the shark, Mustelus manazo, which we have used as an alternative and more proximal outgroup than the lamprey. Using sharks as the outgroup, lungfish appear to be the closest living relative of tetrapods, although the possibility of a lungfish/coelacanth clade being the sister group of tetrapods cannot be excluded.

Conflict among individual mitochondrial proteins in resolving the phylogeny of eutherian orders.

The phylogenetic relationship among primates, ferungulates (artiodactyls + cetaceans + perissodactyls + carnivores), and rodents was examined using proteins encoded by the H strand of mtDNA, with marsupials and monotremes as the outgroup. Trees estimated from individual proteins were compared in detail with the tree estimated from all 12 proteins (either concatenated or summing up log-likelihood scores for each gene). Although the overall evidence strongly suggests ((primates, ferungulates), rodents), the ND1 data clearly support another tree, ((primates, rodents), ferungulates). To clarify whether this contradiction is due to (1) a stochastic (sampling) error; (2) minor model-based errors (e.g., ignoring site rate variability), or (3) convergent and parallel evolution (specifically between either primates and rodents or ferungulates and the outgroup), the ND1 genes from many additional species of primates, rodents, other eutherian orders, and the outgroup (marsupials + monotremes) were sequenced. The phylogenetic analyses were extensive and aimed to eliminate the following artifacts as possible causes of the aberrant result: base composition biases, unequal site substitution rates, or the cumulative effects of both. Neither more sophisticated evolutionary analyses nor the addition of species changed the previous conclusion. That is, the statistical support for grouping rodents and primates to the exclusion of all other taxa fluctuates upward or downward in quite a tight range centered near 95% confidence. These results and a site-by-site examination of the sequences clearly suggest that convergent or parallel evolution has occurred in ND1 between primates and rodents and/or between ferungulates and the outgroup. While the primate/rodent grouping is strange, ND1 also throws some interesting light on the relationships of some eutherian orders, marsupials, and montremes. In these parts of the tree, ND1 shows no apparent tendency for unexplained convergences.

Hadamard conjugations and modeling sequence evolution with unequal rates across sites.

This paper considers the many different distributions that may approximate the distribution of site rates in DNA sequences and shows how the Hadamard conjugation may be modified to take these into account. This is done for both 2-state and 4-state data. Distributions which give simple closed forms include the gamma (gamma) distribution, the inverse Gaussian distribution (which is similar to the lognormal), and a mixture of either of these with a proportion of sites which cannot change (invariant sites). It is seen that the tail of a distribution can have major effects upon the coefficient of variation of site rates. Because the Hadamard conjugation can be used to either correct data or predict the data given the model (i.e., the likelihood of site patterns), light is shed on properties of maximum likelihood tree selection with unequal site rates. Analysis of rRNA shows how unequal rates across sites can change the optimal tree. Maximum likelihood analysis also shows that distinct distributions fit each data set, with the gamma often not being the best. Analyzing both these data and a long stretch of primate mtDNA reveals evidence of many "hidden" multiple substitutions, while signals not corresponding to the preferred biological tree generally decrease an unequal rates are allowed for. Last, we discuss the expected behavior of sequences evolving by models where stabilizing selection alone explains unequal site rates. Such models do not explain "synapomorphies" or informative changes in ancient molecules, because while stabilizing selection can vastly decrease change at a site, it will also vastly accelerate back-substitution (leaving only a covarion model to explain old synapomorphies). When and why models allowing a continuous distribution of site rates (e.g., gamma) will approximate covarion evolution requires further study.

General time-reversible distances with unequal rates across sites: mixing gamma and inverse Gaussian distributions with invariant sites.

A series of new results useful to the study of DNA sequences using Markov models of substitution are presented with proofs. General time-reversible distances can be extended to accommodate any fixed distribution of rates across sites by replacing the logarithmic function of a matrix with the inverse of a moment generating function. Estimators are presented assuming a gamma distribution, the inverse Gaussian distribution, or a mixture of either of these with invariant sites. Also considered are the different ways invariant sites may be removed and how these differences may affect estimated distances. Through collaboration, we implemented these distances into PAUP in 1994. The variance of these new distances is approximated via the delta method. It is also shown how to predict the divergence expected for a pair of sequences given a rate matrix and a distribution of rates across sites, allowing iterated ML estimates of distances under any reversible model. A simple test of whether a rate matrix is time reversible is also presented. These new methods are used to estimate the divergence time of humans and chimps from mtDNA sequence data. These analyses support suggestions that the human lineage has an enhanced transition rate relative to other hominoids. These studies also show that transversion distances differ substantially from the overall distances which are dominated by transitions. Transversions alone apparently suggest a very recent divergence time for humans versus chimps and/or a very old (> 16 myr) divergence time for humans versus orangutans. This work illustrates graphically ways to interpret the reliability of distance-based transformations, using the corrected transition to transversion ratio returned for pairs of sequences which are successively more diverged.

Evolution of chlorophyll and bacteriochlorophyll: the problem of invariant sites in sequence analysis.

Competing hypotheses seek to explain the evolution of oxygenic and anoxygenic processes of photosynthesis. Since chlorophyll is less reduced and precedes bacteriochlorophyll on the modern biosynthetic pathway, it has been proposed that chlorophyll preceded bacteriochlorophyll in its evolution. However, recent analyses of nucleotide sequences that encode chlorophyll and bacteriochlorophyll biosynthetic enzymes appear to provide support for an alternative hypothesis. This is that the evolution of bacteriochlorophyll occurred earlier than the evolution of chlorophyll. Here we demonstrate that the presence of invariant sites in sequence datasets leads to inconsistency in tree building (including maximum-likelihood methods). Homologous sequences with different biological functions often share invariant sites at the same nucleotide positions. However, different constraints can also result in additional invariant sites unique to the genes, which have specific and different biological functions. Consequently, the distribution of these sites can be uneven between the different types of homologous genes. The presence of invariant sites, shared by related biosynthetic genes as well as those unique to only some of these genes, has misled the recent evolutionary analysis of oxygenic and anoxygenic photosynthetic pigments. We evaluate an alternative scheme for the evolution of chlorophyll and bacteriochlorophyll.

Improved analyses of human mtDNA sequences support a recent African origin for Homo sapiens.

New quantitative methods are applied to the 135 human mitochondrial sequences from the Vigilant et al. data set. General problems in analyzing large numbers of short sequences are discussed, and an improved strategy is suggested. A key feature is to focus not on individual trees but on the general "landscape" of trees. Over 1,000 searches were made from random starting trees with only one tree (a local optimum) being retained each time, thereby ensuring optima were found independently. A new tree comparison metric was developed that is unaffected by rearrangements of trees around many very short internal edges. Use of this metric showed that downweighting hypervariable sites revealed more evolutionary structure than studies that weighted all sites equally. Our results are consistent with convergence toward a global optimum. Crucial features are that the best optima show very strong regional differentiation, a common group of 49 African sequences is found in all the best optima, and the best optima contain the 16 !Kung sequences in a separate group of San people. The other 86 sequences form a heterogeneous mixture of Africans, Europeans, Australopapuans, and Asians. Thus all major human lineages occur in Africa, but only a subset occurs in the rest of the world. The existence of these African-only groups strongly contradicts multiregional theories for the origin of Homo sapiens that require widespread migration and interbreeding over the entire range of H. erectus. Only when the multiregional model is rejected is it appropriate to consider the root, based on a single locus, to be the center of origin of a population (otherwise different loci could give alternative geographic positions for the root). For this data, several methods locate the root within the group of 49 African sequences and are thus consistent with the recent African origin of H. sapiens. We demonstrate that the time of the last common ancestor cannot be the time of major expansion in human numbers, and our results are thus also consistent with recent models that differentiate between the last common ancestor, expansion out of Africa, and the major expansion in human populations. Such a two-phase model is consistent with a wide range of molecular and archeological evidence.

Soma neurofilament immunoreactivity is related to cell size and fibre conduction velocity in rat primary sensory neurons.

1. Intracellular recordings were made in dorsal root ganglia in vitro at 37 degrees C. The L4, L5 and L6 ganglia from 46- to 51-day-old female Wistar rats were used. In each neuron conduction velocity (CV) was measured and fluorescent dye was injected. Later the intensity of the immunoreactivity to RT97 (a monoclonal antibody to the phosphorylated 200 kDa neurofilament subunit) as well as the cell size (cross-sectional area at the nuclear level) were measured in the dye-injected neurons. RT97 was used to distinguish between the L (light, neurofilament-rich) and the SD (small dark, neurofilament-poor) neuronal somata. 2. Neurons were classified as C neurons (CV less than 1.3 m/s), C/A delta neurons (1.3-2 m/s), A delta neurons (2-12 m/s) or A alpha/beta neurons (greater than 12 m/s). 3. All A-fibre somata were RT97 positive (L) and all C-fibre somata were RT97 negative (SD), although in the C/A delta group both positive and negative neurons were seen. Thus, RT97-negative somata had C (unmyelinated) or C/A delta fibres, while RT97-positive somata had A (myelinated) or C/A delta fibres. 4. The size distributions of A neurons and C neurons were consistent with their classification as L- and SD-cell neurons respectively. The size distribution of A delta cells was skewed with a peak of small cells and a tail of medium-sized cells. 5. There was a loose positive correlation between cell size and fibre CV. 6. RT97 intensity was positively correlated with CV if all neurons were considered together, but no correlation was seen within the C, A delta or A alpha/beta CV groups. 7. RT97 intensity was positively correlated with cell size when all neurons were considered together. Although no correlation was seen within the C or the A delta CV groups, a clear positive correlation was seen for A alpha/beta neurons. 8. The relationship of RT97 intensity to cell size was not demonstrably altered by axotomy, time in vitro or the presence of intracellular dye in control experiments. 9. RT97-negative and -positive neurons could be seen in neonatal rat ganglia. Their size distributions resembled, respectively, the SD- and L-neuron populations at this age. RT97 immunoreactivity may therefore be a useful predictor of the cell type and myelinated state which a sensory cell is destined to reach in the adult rat.

Electrophysiological properties of subpopulations of rat dorsal root ganglion neurons in vitro.

Intracellular recordings were made from rat dorsal root ganglion neurons in vitro. Action potentials with an inflection on the falling phase occurred in all cells conducting up to 5.2 m/s and in a proportion of faster conducting cells which decreased with increasing conduction velocity, until no cells conducting faster than 31 m/s had an inflection. Overall, all C-cells (less than 1.3 m/s), 61% of A delta-cells (1.3-12 m/s) and 23% of A alpha/beta-cells (greater than 12 m/s) had inflections. A-cells with inflections were found to be electrophysiologically distinct from those without as they differed in the mean and distribution of every action potential and afterhyperpolarization parameter measured. C-cells differed from all A-cell groups, but the means and distributions of most parameters were closer to those of A-cells with inflections than of A-cells without. In addition, all A- and C-cell action potentials with inflections were tetrodotoxin resistant, while all those without were sensitive. The only parameters whose means differed between A alpha/beta- and A delta-neurons were ones which correlated with conduction velocity (action potential duration, overshoot and maximum rate of rise and fall). The response pattern to prolonged current injection did not correlate with conduction velocity, but slightly more A-cells with inflections were single firing. A-cells with long afterhyperpolarizations always fired singly, while those with shorter durations fired singly or multiply. Somatic following frequency was most strongly limited by long afterhyperpolarization duration; it was also slightly lower in A delta- than in A alpha/beta-cells, and lower in A-cells with inflections than in those without. Fibre following frequencies were highest in the fastest conduction neurons.