Isochore Evolution in Mammals: A Human-Like Ancestral Structure

Corresponding author: Dominique Mouchiroud, Centre National de la Recherche Scientifique UMR 5558, Biométrie, Génétique et Biologie des Populations, Université Claude Bernard Lyon 1, 43, Boulevard du 11 novembre 1918, 69622 Villeurbanne Cedex, France., mouchi{at}biomserv.univ-lyon1.fr (E-mail).

Communicating editor: G. A. CHURCHILL

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Codon usage in mammals is mainly determined by the spatial arrangement of genomic G + C-content, i.e., the isochore structure. Ancestral G + C-content at third codon positions of 27 nuclear protein-coding genes of eutherian mammals was estimated by maximum-likelihood analysis on the basis of a nonhomogeneous DNA substitution model, accounting for variable base compositions among present-day sequences. Data consistently supported a human-like ancestral pattern, i.e., highly variable G + C-content among genes. The mouse genomic structure—more narrow G + C-content distribution—would be a derived state. The circumstances of isochore evolution are discussed with respect to this result. A possible relationship between G + C-content homogenization in murid genomes and high mutation rate is proposed, consistent with the negative selection hypothesis for isochore maintenance in mammals.

VERTEBRATE nuclear genomes are characterized by a peculiar structure regarding the spatial distribution of guanine and cytosine content (GC%): these genomes are mosaics of long, compositionally homogeneous DNA segments called isochores (BERNARDI et al. 1985 ; BERNARDI 1993 ). The isochore structure has been carefully studied in eutherian mammals, using both analytic ultracentrifugation experiments and DNA sequence analysis. Silent positions in protein-coding genes undergo no selection related to protein function and are therefore relevant markers of global evolutionary constraints applying at the isochore level (CLAY et al. 1996 ). Actually, codon usage in mammals appears to be uniquely determined by local GC-richness (SHARP et al. 1995 ).

A major compositional change between human and murine genomes (rat, mouse) was found from comparative genome analysis: the variance of third codon position GC% (GC3) among protein-coding genes is higher in human than in rat and mouse (MOUCHIROUD et al. 1988 ; MOUCHIROUD and GAUTIER 1990 ); GC-rich isochores are richer in human, and GC-poor isochores are poorer. Rabbit, cow, and dog mainly show a human-like isochore structure: GC3 in protein-coding genes of these species is highly correlated to GC3 in human orthologous genes (MOUCHIROUD and BERNARDI 1993 ). ROBINSON et al. 1997A showed from relevant sampling of rodent species that the mouse/rat structure is common to the whole Muridae group. Consistent results were found from analytical ultracentrifugation (SALINAS et al. 1986 ; SABEUR et al. 1993 ). Therefore, two major mammalian isochore patterns are known to date, which we call the Muridae pattern (with moderate interisochore GC% variability) and the nonrodent pattern (with high interisochore GC% variability). Taxon Muridae is a huge rodent family that includes two-thirds of rodent species and one-quarter of mammalian species (WILSON and REEDER 1993 ), among which are mouse, rat, and hamster.

Muridae are likely a monophyletic outgroup to the nonrodent eutherian orders whose isochore structure has been examined using DNA sequence data, namely Lagomorpha, Artiodactyla, Carnivora, and Primates, as supported by both mitochondrial and nuclear molecular data (GRAUR et al. 1991 , GRAUR et al. 1996 ; D'ERCHIA et al. 1996 ; CAO et al. 1997 ; JANKE et al. 1997 ; see Figure 1). Hence, the nature of the ancestral eutherian isochore structure is an open question; it may have been Muridae-like, nonrodent-like, intermediate, or different from any known present-day pattern. This question is a central one as far as mammalian molecular evolution is concerned because neither the circumstances of isochore evolution nor the evolutionary constraints maintaining this structure within mammalian genomes are well understood yet.

View larger version (22K): In this window In a new window Download PPT slide   Figure 1. Schematic phylogenetic tree of eutherian mammals. The radiation among non-Muridae eutherian lineages is considered unresolved. Mu., deeply branching Muridae; Pr., Primates; La., Lagomorpha; Ar., Artiodactyla; Pe., Perissodactyla; Ca., Carnivora. Under the molecular clock assumption, parameter is given by (c - a)/[(c - a) + (c - b)], where a, b, and c are the dates of Muridae divergence, nonrodent divergence, and Muridae/nonrodent split, respectively.

Recently, GALTIER and GOUY 1998 devised a maximum-likelihood implementation of a new, nonhomogeneous model of DNA sequence evolution allowing diverging GC% among lineages. Given a phylogeny, a reasonable amount of information about past evolutionary modes can be extracted by this method from data sets of usual sizes. Especially, GC% at ancestral nodes is quite reliably recovered when data simulated under the model assumptions are used (GALTIER and GOUY 1998 ). This tool appears suitable for studying mammalian isochore evolution because unequal transition and transversion and diverging GC% among lineages—i.e., the very assumptions of Galtier and Gouy's model—are likely the major forces constraining synonymous substitution processes in mammalian genes.

In this article, we address the problems of recovering the ancestral eutherian isochore structure and locating the major genomic compositional changes in the mammalian phylogenetic tree by applying GALTIER and GOUY's (1998) method to the third-codon positions of 27 nuclear protein-coding genes.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Data:
Sequences were extracted from the HOVERGEN database (DURET et al. 1994 , release 25, July 1997). Orthologous genes were selected according to the number of represented species and the amount of GC3 change: at least six distinct eutherian species including at least two Muridae and two nonrodents were required, and the difference between human and mouse GC3 (both were available for all genes matching the above criteria) had to be >10% or <-10%, as suggested by MOUCHIROUD and GAUTIER 1990 . These genes are likely to be reliable markers of the process of compositional change. For each gene family, orthology was checked by examining the phylogenetic tree provided by the HOVERGEN interface. Putative paralogous genes were discarded. Twenty-seven protein-coding genes were finally selected (Table 1). Non-Muridae rodent sequences were removed from data sets when available because both their phylogenetic location and isochore pattern are unclear (NEDBAL et al. 1996 ; ROBINSON et al. 1997A , ROBINSON et al. 1997B ). For each gene, amino acid sequences were automatically aligned using the CLUSTALW program (THOMPSON et al. 1994 ). Nucleotidic alignments were deduced from aligned amino acid sequences. Gap-containing sites were discarded. Phylogenetic trees were reconstructed by the neighbor-joining method (SAITOU and NEI 1987 ) with the logdet distance (STEEL 1993 ) applied to all three codon positions, using program PHYLO WIN (GALTIER et al. 1996 ). Tree topologies were rooted on the branch connecting Muridae to nonrodents. First- and second-codon positions were subsequently removed from the data sets. (GenBank accession numbers of the sequences used in this study are available upon request.)

Ancestral GC-content estimation:
GALTIER and GOUY's (1998) method was used to estimate the ancestral GC% at the third-codon position of each gene. This method relies on a new model of DNA substitution that allows varying evolutionary processes among lineages: two assumptions of usual models—namely homogeneity and stationarity—are relaxed to account for variable base compositions among present-day sequences. The assumed substitution process on any branch of a given rooted tree follows TAMURA's (1992) model, with unequal equilibrium G + C contents ( in TAMURA 1992 ) among branches, so that GC% can vary with time and between lineages. The transition/transversion ratio is kept constant over the whole tree. The GC% at the root node () and the precise location of the root on its branch () are two additional parameters of the model. Parameter estimates are those values maximizing the likelihood of the model as defined by FELSENSTEIN 1981 . A reliable step-by-step optimization algorithm was designed to achieve this estimation task (GALTIER and GOUY 1998 ). Accurate estimates of the ancestral GC% were recovered from data sets simulated under the model assumptions (GALTIER and GOUY 1998 ), suggesting that a reasonable amount of information about past base compositions can be extracted from real data sets. For each gene, the above estimation algorithm was applied to third-codon position sequences, using the inferred neighbor-joining topology as a model tree.

To make inferences comparable among genes, estimates were calibrated taking into account Muridae and nonrodent mean GC3 values. A "nonrodent-likeness" index is defined,

(1)

where is the estimated ancestral GC3, GC_M is the mean GC3 among Muridae sequences, and GC_NR is the mean GC3 among nonrodent sequences. equals 0 if equals GC_M or 1 if equals GC_NR; is >1 if is more extreme than GC_NR (i.e., outside the [GC_M, GC_NR] interval on the nonrodent side), and <0 if is more median than GC_M (i.e., outside the [GC_M, GC_NR] interval on the Muridae side). GC_NR was computed taking phylogeny into account: equal weights were given to all nonrodent orders whatever the number of represented species in each order. A similar procedure was used to compute GC_M when the number of available Muridae species was >3.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Model/data adequacy:
Twenty-seven genes were analyzed (Table 1). The total number of species varied from 6 to 16 among genes. The number of Muridae species was <4 excepting gene LCAT (ROBINSON et al. 1997B , 13 Muridae species). Nonrodent sampled orders were Primates, Lagomorpha, Carnivora, Artiodactyla, and Perissodactyla. GC3 differences between nonrodents and Muridae were positive in 17 genes out of 27; these genes are expected to be located in GC-rich isochores. Two major assumptions of GALTIER and GOUY's (1998) model were empirically checked: A% = T% and C% = G% equalities, and constant substitution rate among sites.

In TAMURA's (1992) substitution model (and consequently in GALTIER and GOUY's model as well), sequence base composition is described by a single parameter, namely G + C content, so that A% = T% and C% = G% are the underlying assumptions. The AT (respectively GC)-skewness of the analyzed sequences was computed:

(2)

The distributions of both statistics over 219 compared genes are shown (Figure 2). Values >0.4 or <-0.4 are infrequent.

View larger version (37K): In this window In a new window Download PPT slide   Figure 2. Distribution of AT-skewness (top) and GC-skewness (bottom) among 219 mammalian third-codon-position sequences used in the present study.

The amount of variation of substitution rates among sites was investigated by fitting to the data a substitution model involving gamma-distributed rates over sites (YANG 1994 ): the estimated shape parameter of the assumed gamma distribution ( in YANG 1994 ) gives insight about how variable rates are among sites. The mean estimated value over 27 data sets was 2.58. was >1 for 26 genes out of 27. Such values are characteristic of bell-shaped, low variance, distributions.

Simulations:
Preliminary analyses above suggest that the major part of the data essentially meets the model assumptions. Nevertheless, because the estimation method has been assessed only under conditions matching exactly the model assumptions, one may wonder about the reliability of inferences when these assumptions are (slightly) violated. We conducted a simulation study to assess the robustness of the above method to departures from the A% = T%, G% = C%, and equal-rates-among-sites assumptions. The tree of Figure 1 was used as a model tree topology, with branch lengths proportional to those of Figure 1. Two distinct processes were performed, each one aiming to simulate the evolution of the third-codon positions of a G + C-rich gene. In the first procedure ("human-like ancestor"), a 300-nucleotide-long ancestral sequence was randomly drawn with 80% average G + C-content. The diverging evolution of eight sequences was simulated according to the "HKY + gamma" model (HASEGAWA et al. 1985 ; YANG et al. 1994 ). This is a generalization of Tamura's model allowing unequal A- and T- (respectively G- and C-) pressures—resulting in sequences with nonzero expected AT (GC)-skewness—and unequal evolutionary rates among sites. Equilibrium G + C-content (i.e., GC-pressure) was set to 80% in the nonrodent lineage, leaving the expected base composition unchanged, and to 25% in the Muridae lineage, so that the G + C-content in present-day Muridae sequences was around 65%. In the second simulation process ("mouse-like ancestor"), the ancestral G + C-content was set to 65%. It was left unchanged in the Muridae lineage but was increased toward 80% in the nonrodent lineage. The transition/transversion ratio was set to 4.

In both processes, AT- and GC-skewness varied from 0 (i.e., A%/T% = 1) to 0.8 (i.e., A%/T% = 9). Equal AT- and GC-skewness was assumed. The effect of among-site rate variation was also examined: equal rates, moderately variable rates (shape parameter of the gamma distribution set to 2.5), and highly variable rates (shape parameter set to 0.5) were used. Ancestral G + C-contents were estimated by applying the above-described maximum-likelihood method to the simulated data sets. Ten replicates were performed for each combination of AT (GC)-skewness and among-site heterogeneity. The method was considered successful when the estimated ancestral G + C-content was closer to the true value than to the mean G + C-content in present-day sequences with diverged base composition (namely Muridae sequences in process 1, nonrodent sequences in process 2). The number of successes out of 10 replicates is shown (Table 2), together with the mean estimated ancestral G + C-content.

The method appeared to be biased when AT (GC)-skewness was higher than 0.6. No sensitivity to among-site rate variation was found. Therefore, no significant bias is expected in the present work because most of the analyzed sequences have observed AT (GC)-skewness <0.4.

Ancestral GC% estimation:
The main results are shown in Table 1. Ancestral GC3s estimated according to GALTIER and GOUY's (1998) algorithm () are given, as well as their calibrated version . The results clearly support the hypothesis of a nonrodent-like ancestral isochore pattern. is closer to GC_NR than to GC_M ( > 0.5) for 21 genes out of 27. The mean value over 27 genes is 0.85. Additional analyses were conducted to check the robustness of this general result.

Sensitivity to the location of the root:
The actual location of the root on its branch (i.e., the fraction of the root branch length lying on a given side of the root, say the Muridae side) was not reliably estimated when simulated data sets were used in GALTIER and GOUY 1998 . This has little effect on the estimation of ancestral GC% if the root branch is short. However, if a significant amount of the evolutionary change occurs along the root branch, inaccurate estimates of may mislead estimation. Because the branch connecting Muridae to nonrodents was generally long for the present data sets, we checked the sensitivity of the estimate to the location of the root: three additional estimation procedures were performed assuming fixed values for parameter , namely 0.2, 0.5, and 0.8. A 0.8 value means that the branch connecting the root to the ancestral Muridae node is four times longer than the branch connecting the root to the ancestral nonrodent node (Figure 1). The resulting estimates are called _0.2, _0.5, and _0.8, respectively (Table 1). -Dependent nonrodent-likeness _0.2, _0.5, and _0.8 can be defined by replacing by _0.2 (respectively _0.5, _0.8) in Equation 1.

Genes were classified into three groups depending on the sensitivity of the estimate to the value. For 19 genes out of 27, the highest difference between _0.2, _0.5, and _0.8 values was <5%. These genes are called root insensitive (Table 1, top). The remaining 8 genes were again split into two groups. For 3 of them, the estimated ancestral GC3 was closer to GC_NR than to GC_M whatever the assumed value: _0.2, _0.5, and _0.8 are >0.5 (Table 1, middle). For the last 5 genes, the estimate appears highly sensitive to the location of the root: was either Muridae-like or nonrodent-like depending on (Table 1, bottom). Seventeen root-insensitive genes out of 19 supported the nonrodent ancestral pattern hypothesis; the mean value over 19 root-insensitive genes was 0.96. Inversely, genes supporting the Muridae ancestral pattern hypothesis generally belong to the root-sensitive group (4 out of 6): inferences vary depending on . When is set to 0.8, the estimated ancestral GC3 is closer to GC_NR than to GC_M for all 27 genes. These results provide additional support for the nonrodent ancestral isochore pattern hypothesis because genes supporting an alternative scheme appear less reliable regarding estimation and may have been misled by wrong estimates.

Obtaining a priori estimates for would help in interpreting root-sensitive results. This is not an easy task because it depends on species sampling, speciation times, and evolutionary rates. However, let us consider those genes in the data set (18 out of 27) where Muridae are represented by mouse and rat. If constant evolutionary rates among lineages are assumed (the molecular clock hypothesis), accepting 13, 80, and 100 mya as rough date estimates for, respectively, mouse/rat divergence (JACOBS and DOWNS 1994 ), nonrodent radiation, and murid/nonrodent divergence would result in = 0.81 (Figure 1). Using the estimated date of Muridae radiation (20–25 mya, HUGUENEY and MEIN 1993 ) rather than mouse/rat split gives = 0.79–0.80. Accounting for the higher substitution rate in Muridae (LI et al. 1996 ) would increase these values. These are imprecise estimates. However, they suggest that values >0.5 are more likely than low values. This reinforces the nonrodent ancestral pattern hypothesis since estimates for root-sensitive genes are closer to GC_NR when is high.

Sensitivity to species sampling:
The above results could be a consequence of unbalanced species sampling: nonrodent species are generally more numerous than Muridae in our data set, which might bias estimation toward the GC_NR value. Three genes, namely albumin (9 represented species), interleukin 6 (14 species), and LCAT (16 species), were carefully studied to check this putative methodological artifact. Incomplete data sets were set up by removing species, and the above analyses were performed again. Results are shown in Table 3. Both value and root-sensitivity remain almost unchanged when the number of species decreases, suggesting that asymmetric species sampling is unlikely to have misled the above findings.

Sensitivity to phylogeny:
estimates may also be sensitive to the assumed phylogenetic tree. For three genes, we reconducted the above analyses after modifying the assumed tree topology. In each case, five distinct trees were randomly drawn by modifying those branching orders not considered firmly established. For genes albumin and interleukin 6, branching orders among nonrodent orders were randomly shifted. For gene LCAT, internal branches supported by bootstrap percentages >80 when all three codon positions are analyzed by the neighbor-joining method (logdet distance) were kept, but less supported branching orders were randomly shifted. Results are given in Table 4: the first line in each part of the table recalls the initial estimate, and the next five lines are for modified trees. Again, estimates are remarkably stable when the assumed phylogenetic tree varies.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Locating the compositional change in the mammalian tree:
When applied to 27 protein-coding genes showing variable GC3 in Muridae and nonrodent mammals, GALTIER and GOUY's (1998) method unambiguously recovers a nonrodent-like ancestral pattern. Data are highly self-consistent with respect to this result: all 27 examined genes support the nonrodent-like ancestor hypothesis when a plausible value is assumed, which is quite unlikely to occur by chance. This result is not sensitive to species sampling, nor to assumed phylogenetic trees. Our method accurately recovered ancestral GC-contents from data sets simulated under the underlying model (GALTIER and GOUY 1998 ). It also performed well when the model assumptions were slightly violated (this study). Therefore, we think that the above inferences deserve reasonable confidence.

Within-genome GC-content heterogeneity is far lower in fishes and amphibians than in mammals or birds (BERNARDI and BERNARDI 1990 ; BERNARDI et al. 1997 ). Actually, the cold-blooded vs. warm-blooded discrepancy is the most striking pattern of isochore structure variability among vertebrates. This result, together with the deep phylogenetic location of Muridae among eutherian mammals, raised an attractive hypothesis: Muridae may have kept an "intermediate" state between poorly structured cold-blooded and highly structured nonrodent mammals (for example, SACCONE et al. 1997 ). The present results suggest that this hypothesis should be dismissed. We propose that a highly structured genomic GC% compartmentation is ancestral in eutherian mammals. This structure has remained unchanged in many eutherian orders, while Muridae have been (are) undergoing a "reversal" toward a less-structured state, i.e., GC-homogenization. A brief look at the eutherian phylogenetic tree (Figure 1) shows that the latter hypothesis, although less parsimonious than the alternative, is not less likely on general grounds: assuming a Muridae-like ancestral pattern would require that the nonrodent-like pattern either evolved during a short period of time—which is unlikely as far as the whole genomic structure is concerned—or evolved independently in several lineages—which is less parsimonious than the human-like ancestor hypothesis.

Several peculiar characteristics of the genomic evolution of rodents (and especially Muridae) have been reported, including a high rate of chromosomic rearrangements (VIEGAS-PEQUINOT et al. 1986 ), more numerous and longer microsatellites (DURET et al. 1995 ), and erosion of CpG islands (MATSUO et al. 1993 ; CROSS et al. 1997 ). Most importantly, relative rate tests showed that both synonymous and nonsynonymous substitution rates in nuclear genes are significantly (possibly 10 times) higher in the rat/mouse lineage than in primates (LI et al. 1996 ). This discrepancy is likely a consequence of higher mutation rate; rodents show less efficient repair of DNA lesions (HART and SETLOW 1974 ) and lower generation time than most mammals. Therefore, the GC-homogenization we report occurred in the eutherian lineage undergoing the highest mutation rate. Both phenomena may be related, as we discuss below.

Isochore evolution in mammals:
Debates rage on about the questions of isochore evolution and maintenance in mammalian genomes. Two main hypotheses compete. Bernardi has consistently argued that a negative selection pressure was acting to maintain this structure (see BERNARDI 1993 ; ZOUBAK et al. 1995 ; ALVAREZ-VALIN et al. 1998 ); a possible relationship between isochore evolution and endothermy has been suggested. Sharp and colleagues criticized this view and proposed that the isochore structure is a consequence of varying mutational pattern within genomes (see SHARP et al. 1995 ). Several neutral, putatively isochore-forming molecular mechanisms were put forward, including varying free deoxyribonucleotide concentration through the cell cycle (WOLFE et al. 1989 ; but see EYRE-WALKER 1992 ) and unequal repair pattern across chromosomes (HOLMQUIST and FILIPSKI 1994 ). Locating the compositional change on the Muridae branch in the mammalian tree provides a new argument. If isochores were the consequence of varying neutral mutation pattern across genomes, increasing mutation rate should result in increasing GC-heterogeneity. The opposite pattern is found: the isochore structure gets weaker in fast-evolving Muridae. This result, however, appears consistent with the alternative hypothesis. If negative selection against slightly deleterious variants is contributing to isochore maintenance, less structure is expected when DNA repair is less efficient because more deleterious mutations per generation accumulate.

The GC-content at third-codon positions of eutherian nuclear protein-coding genes is highly variable within genomes (23 to 98% in human) and highly correlated between genomes: the isochore structure appears well conserved among eutherian orders (MOUCHIROUD and BERNARDI 1993 ). In one lineage, incidentally the ones evolving the fastest, GC% homogenized. GC3 homogenization in Muridae may be due to the loss of evolutionary constraints maintaining isochores in mammals, either mutational or selective. However, we still need to understand why such GC-pressure applying to most eutherian orders would slow down in murids. A less ad hoc hypothesis involves a relationship between GC-homogenization and high mutation rate, two peculiar features of Muridae genomic evolution. This scheme is consistent with selected vs. strictly neutral isochore maintenance in mammalian genomes.

The selective hypothesis leaves two points unexplained. First, we do not know which selective advantage may arise from highly structured isochores. Especially, is isochore evolution related to endothermy? Genome analysis of additional cold-blooded vertebrate species, e.g., crocodilians, should help address this question. Second, the way selection may act within mammalian populations is unclear. SHARP et al. 1995 pointed out that a major role of natural selection is unlikely because the selective advantage of point mutations must be very low and effective population sizes small, so that random genetic drift should overcome selection. This criticism remains unanswered. Thus, the mechanisms of isochore evolution are far from being elucidated. This question is a challenging one for both molecular biologists and population geneticists.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALVAREZ-VALIN, F., K. JABBARI, and G. BERNARDI, 1998  Synonymous and nonsynonymous substitutions in mammalian genes: intragenic correlations. J. Mol. Evol. 46:37-53[Medline].

BERNARDI, G., 1993  The vertebrate genome: isochores and evolution. Mol. Biol. Evol. 10:186-204[Abstract].

BERNARDI, G. and G. BERNARDI, 1990  Compositional patterns in the nuclear genome of cold-blooded vertebrates. J. Mol. Evol. 31:265-281[Medline].

BERNARDI, G., B. OLOFSSON, J. FILIPSKI, M. ZERIAL, and J. SALINAS et al., 1985  The mosaic genome of warm-blooded vertebrates. Science 228:953-958[Abstract/Free Full Text].

BERNARDI, G., S. HUGHES, and D. MOUCHIROUD, 1997  The major compositional transition in the vertebrate genome. J. Mol. Evol. 44:44-51.

CAO, Y., N. OKADA, and M. HASEGAWA, 1997  Phylogenetic position of Guinea pigs revisited. Mol. Biol. Evol. 14:461-464[Medline].

CLAY, O., S. CACCIO, S. ZOUBAK, D. MOUCHIROUD, and G. BERNARDI, 1996  Human coding and non-coding DNA: compositional correlations. Mol. Phylogenet. Evol. 5:2-12[Medline].

CROSS, S. H., M. LEE, V. H. CLARK, J. M. CRAIG, and A. P. BIRD et al., 1997  The chromosomal distribution of CpG islands in the mouse: evidence for genome scrambling in the rodent lineage. Genomics 40:454-461[Medline].

D'ERCHIA, A. M., C. GISSI, G. PESOLE, C. SACCONE, and U. ARNASON, 1996  The guinea-pig is not a rodent. Nature 381:597-600[Medline].

DURET, L., D. MOUCHIROUD, and M. GOUY, 1994  HOVERGEN: a database of homologous vertebrate genes. Nucleic Acids Res. 22:2360-2365[Abstract/Free Full Text].

DURET, L., D. MOUCHIROUD, and C. GAUTIER, 1995  Statistical analysis of vertebrate sequences reveals that long genes are scarce in GC-rich isochores. J. Mol. Evol. 40:308-317[Medline].

EYRE-WALKER, A., 1992  Evidence that both G + C rich and G + C poor isochores are replicated early and late in the cell cycle. Nucleic Acids Res. 20:1497-1501[Abstract/Free Full Text].

FELSENSTEIN, J., 1981  Evolutionary trees from DNA sequences: a maximum likelihood approach. J. Mol. Evol. 17:368-376[Medline].

GALTIER, N. and M. GOUY, 1998  Inferring pattern and process: maximum-likelihood implementation of a new, non-homogeneous model of DNA sequence evolution for phylogenetic analysis. Mol. Biol. Evol. 15:871-879[Abstract].

GALTIER, N., M. GOUY, and C. GAUTIER, 1996  SEAVIEW and PHYLO WIN: two graphic tools for sequence alignment and molecular phylogeny. Comp. Appl. Biosci. 12:543-548[Abstract/Free Full Text].

GRAUR, D., W. A. HIDE, and W.-H. LI, 1991  Is the guinea-pig a rodent? Nature 351:649-651[Medline].

GRAUR, D., L. DURET, and M. GOUY, 1996  Phylogenetic position of the order Lagomorpha (rabbits, hares and allies). Nature 379:333-335[Medline].

HART, R. W. and R. B. SETLOW, 1974  Correlation between deoxyribonucleic acid excision-repair and life-span in a number of mammalian species. Science 71:2169-2173.

HASEGAWA, M., H. KISHINO, and T. YANO, 1985  Dating the human-ape splitting by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22:160-174[Medline].

HOLMQUIST, G. P. and J. FILIPSKI, 1994  Organization of mutations along the genome: a prime determinant of genome evolution. TREE 9:65-68.

HUGUENEY, M. and P. MEIN, 1993  A comment on the earliest Spalacinae (Rodentia, Muroidea). J. Mammal. Evol. 1:215-223.

JACOBS, L. L., and W. R. DOWNS, 1994 The evolution of murine rodents in Asia, pp. 149–156 in Rodent and Lagomorph Families of Asian Origin and Diversification, edited by Y. TOMIDA, C. LI and T. SETOGUCHI. National Science Museum, Tokyo.

JANKE, A., X. XU, and U. ARNASON, 1997  The complete mitochondrial genome of Wallaroo (Macropus robustus) and the phylogenetic relationship among Monotrema, Marsupialia and Eutheria. Proc. Natl. Acad. Sci. USA 94:1276-1281[Abstract/Free Full Text].

LI, W.-H., D. L. ELLSWORTH, J. KRUSHKAL, B. J. H. CHANG, and D. HEWETT-EMETT, 1996  Rates of nucleotide substitution in primates and rodents and the generation time hypothesis. Mol. Phylogenet. Evol. 5:182-187[Medline].

MATSUO, K., O. CLAY, T. TAKAHASHI, J. SILKE, and W. SCHAFFNER, 1993  Evidence for erosion of mouse CpG island during mammalian evolution. Somatic Cell. Mol. Genet. 19:543-555[Medline].

MOUCHIROUD, D. and G. BERNARDI, 1993  Compositional properties of coding sequences and Mammalian phylogeny. J. Mol. Evol. 37:109-116[Medline].

MOUCHIROUD, D. and C. GAUTIER, 1990  Codon usage changes and sequence dissimilarity between human and rat. J. Mol. Evol. 31:81-91[Medline].

MOUCHIROUD, D., C. GAUTIER, and G. BERNARDI, 1988  The compositional distribution of coding sequences and DNA molecules in humans and murids. J. Mol. Evol. 27:311-320[Medline].

NEDBAL, M. A., R. L. HONEYCUTT, and D. A. SCHLITTER, 1996  Higher-level systematics of rodents (Mammalia:Rodentia): evidence from the mitochondrial 12S rRNA gene. J. Mammal. Evol. 3:201-237.

ROBINSON, M., C. GAUTIER, and D. MOUCHIROUD, 1997a  Evolution of isochores in rodents. Mol. Biol. Evol. 14:823-828[Abstract].

ROBINSON, M., F. CATZEFLIS, J. BRIOLAY, and D. MOUCHIROUD, 1997b  Molecular phylogeny of rodents, with special emphasis on murids: evidence from nuclear gene LCAT. Mol. Phylogenet. Evol. 8:423-434[Medline].

SABEUR, G., G. MACAYA, F. KADI, and G. BERNARDI, 1993  The isochore pattern of mammalian genomes and their phylogenetic implications. J. Mol. Evol. 37:93-108[Medline].

SACCONE, C., S. CACCIO, P. PERANI, L. ANDREOZZI, and A. RAPISARDA et al., 1997  Compositional mapping of mouse chromosomes and the identification of the gene-rich regions. Chromosome Res. 5:293-300[Medline].

SAITOU, N. and M. NEI, 1987  The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425[Abstract].

SALINAS, J., M. ZERIAL, J. FILIPSKI, and G. BERNARDI, 1986  Gene distribution and nucleotide sequence organization in the mouse genome. Eur. J. Biochem. 160:469-478[Medline].

SHARP, P. M., M. AVEROF, A. T. LLOYD, G. MATASSI, and J. F. PEDEN, 1995  DNA sequence evolution: the sounds of silence. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 349:241-247[Medline].

STEEL, M. A., 1993  Recovering a tree from the leaf colorations it generates under a Markov model. Appl. Math. Lett. 7:19-23.

TAMURA, K., 1992  Estimation of the number of nucleotide substitutions when there are strong transition-transversion and G + C-content biases. Mol. Biol. Evol. 9:678-687[Abstract].

THOMPSON, J. D., D. G. HIGGINS, and T. J. GIBSON, 1994  CLUSTAL W: improving the sensitivity of progressive multiple alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680[Abstract/Free Full Text].

VIEGAS-PEQUINOT, E., D. PETIT, T. BENAZZOU, M. PROD'HOMME, and M. LOMBARD et al., 1986  Phylogénie chromosomique chez les Sciuridés, Gerbillidae et Muridae, et étude d'espèces appartenant à d'autres familles de Rongeurs. Mammalia 50:164-202.

WILSON, D. E., and D. M. REEDER, 1993 Mammal Species of the World. A Taxonomic and Geographic Reference. Smithsonian Institution Press, Washington, DC and London.

WOLFE, K. W., P. M. SHARP, and W.-H. LI, 1989  Mutation rates differ among regions of the mammalian genome. Nature 337:283-285[Medline].

YANG, Z., 1994  Maximum-likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39:105-111[Medline].

YANG, Z., N. GOLDMAN, and A. FRIDAY, 1994  Comparison of models for nucleotide substitution used in maximum-likelihood phylogenetic estimation. Mol. Biol. Evol. 12:451-458[Abstract].

ZOUBAK, S., G. D'ONOFRIO, S. CACCIO, and G. BERNARDI, G. BERNARDI, 1995  Specific compositional pattern of synonymous positions in homologous mammalian genes. J. Mol. Evol. 40:293-307[Medline].

This article has been cited by other articles:

				J. L. Chojnowski and E. L. Braun Turtle isochore structure is intermediate between amphibians and other amniotes Integr. Comp. Biol., October 1, 2008; 48(4): 454 - 462. [Abstract] [Full Text] [PDF]

				C. Webber and C. P. Ponting Hotspots of mutation and breakage in dog and human chromosomes Genome Res., December 1, 2005; 15(12): 1787 - 1797. [Abstract] [Full Text] [PDF]

				M. M. Alba and R. Guigo Comparative Analysis of Amino Acid Repeats in Rodents and Humans Genome Res., April 1, 2004; 14(4): 549 - 554. [Abstract] [Full Text] [PDF]

				P. F. Arndt, D. A. Petrov, and T. Hwa Distinct Changes of Genomic Biases in Nucleotide Substitution at the Time of Mammalian Radiation Mol. Biol. Evol., November 1, 2003; 20(11): 1887 - 1896. [Abstract] [Full Text] [PDF]

				A. E. Vinogradov Isochores and tissue-specificity Nucleic Acids Res., September 1, 2003; 31(17): 5212 - 5220. [Abstract] [Full Text] [PDF]

				A. E. Vinogradov DNA helix: the importance of being GC-rich Nucleic Acids Res., April 1, 2003; 31(7): 1838 - 1844. [Abstract] [Full Text] [PDF]

				M. J. Lercher, N. G. C. Smith, A. Eyre-Walker, and L. D. Hurst The Evolution of Isochores: Evidence From SNP Frequency Distributions Genetics, December 1, 2002; 162(4): 1805 - 1810. [Abstract] [Full Text] [PDF]

				L. Duret, M. Semon, G. Piganeau, D. Mouchiroud, and N. Galtier Vanishing GC-Rich Isochores in Mammalian Genomes Genetics, December 1, 2002; 162(4): 1837 - 1847. [Abstract] [Full Text] [PDF]

				A. Eyre-Walker, P. D. Keightley, N. G. C. Smith, and D. Gaffney Quantifying the Slightly Deleterious Mutation Model of Molecular Evolution Mol. Biol. Evol., December 1, 2002; 19(12): 2142 - 2149. [Abstract] [Full Text] [PDF]

				J. Castresana Genes on human chromosome 19 show extreme divergence from the mouse orthologs and a high GC content Nucleic Acids Res., April 15, 2002; 30(8): 1751 - 1756. [Abstract] [Full Text] [PDF]

				R. Tarrio, F. Rodriguez-Trelles, and F. J. Ayala Shared Nucleotide Composition Biases Among Species and Their Impact on Phylogenetic Reconstructions of the Drosophilidae Mol. Biol. Evol., August 1, 2001; 18(8): 1464 - 1473. [Abstract] [Full Text] [PDF]

				J. P. Bielawski, K. A. Dunn, and Z. Yang Rates of Nucleotide Substitution and Mammalian Nuclear Gene Evolution: Approximate and Maximum-Likelihood Methods Lead to Different Conclusions Genetics, November 1, 2000; 156(3): 1299 - 1308. [Abstract] [Full Text]