Human Molecular Genetics, 2001, Vol. 10, No. 21 2319-2328
© 2001 Oxford University Press
Understanding human disease mutations through the use of interspecific genetic variation
Department of Biology, Arizona State University, Tempe, AZ 85287-1501, USA
Received June 5, 2001; Revised and Accepted July 31, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
Data on replacement mutations in genes of disease patients exist in a variety of online resources. In addition, genome sequencing projects and individual gene sequencing efforts have led to the identification of disease gene homologs in diverse metazoan species. The availability of these two types of information provides unique opportunities to investigate factors that are important in the development of genetically based disease by contrasting long and short-term molecular evolutionary patterns. Therefore, we conducted an analysis of disease-associated human genetic variation for seven disease genes: the cystic fibrosis transmembrane conductance regulator, glucose-6-phosphate dehydrogenase, the neural cell adhesion molecule L1, phenylalanine hydroxylase, paired box 6, the X-linked retinoschisis gene and TSC2/tuberin. Our analyses indicate that disease mutations show definite patterns when examined from an evolutionary perspective. Human replacement mutations resulting in disease are overabundant at amino acid positions most conserved throughout the long-term history of metazoans. In contrast, human polymorphic replacement mutations and silent mutations are randomly distributed across sites with respect to the level of conservation of amino acid sites within genes. Furthermore, disease-causing amino acid changes are of types usually not observed among species. Using Grantham’s chemical difference matrix, we find that amino acid changes observed in disease patients are far more radical than the variation found among species and in non-diseased humans. Overall, our results demonstrate the usefulness of evolutionary analyses for understanding patterns of human disease mutations and underscore the biomedical significance of sequence data currently being generated from various model organism genome sequencing projects.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
One central purpose of genome sequencing projects is to effect a better understanding of the genetics of disease and provide assistance with the identification of disease-associated genes (1–3). However, many human mutation databases containing genetic variation found in disease patients already exist, and new databases and database entries are rapidly accumulating (4,5). Concomitant analysis of these two types of information provides unique opportunities to identify intrinsic attributes of disease-associated human genetic variation, leading to a better understanding of the relationship between mutations and the development of disease phenotypes.
Information contained in the alignments of homologous disease-associated genes has long been recognized as an important factor for understanding contemporary deleterious genetic variation in humans (4,6). For example, in a given set of homologous genes, a large fraction of amino acid sites will be conserved even among distantly related species that diverged hundreds of millions of years ago. Variations that arose at such positions throughout evolutionary history have evidently been under strong purifying selection and eliminated from populations, suggesting that the existing amino acid residues at invariant positions are critical for proper gene function. Thus, information from interspecific alignments can indicate amino acid residues in gene products that are likely to produce disease if mutated in humans. Likewise, some positions in protein sequences vary among species, and such variable sites may indicate positions that are under less severe selective constraints. These variable positions suggest sites where residue changes can be tolerated by natural selection and provide insights into the types of amino acids that can be freely exchanged without negatively impacting protein function.
Since the logic of these statements is often used by researchers to indicate the potential for an observed amino acid change to produce disease in humans (6–10), we conducted a study to directly evaluate the extent that interspecific sequence alignments reveal common attributes of the deleterious mutations observed in humans. We performed three types of analyses using disease mutation data and homologous gene sequences from seven disease-associated genes (Table 1 and Fig. 1): cystic fibrosis transmembrane conductance regulator (CFTR), glucose-6-phosphate dehydrogenase (G6PD), neural cell adhesion molecule L1 (L1CAM), phenylalanine hydroxylase (PAH), paired box 6 (PAX6), the X-linked retinoschisis gene and a gene associated with tuberous sclerosis (TSC2). First, we determined the association between the prevalence of disease mutations and the extent to which corresponding amino acid sites in other species have been conserved throughout the evolutionary history of metazoans. Secondly, we compared the frequency of a given type of amino acid change in disease patients to frequencies obtained from interspecific comparisons. Finally, we compared the chemical property differences of amino acid changes seen among species and non-diseased humans with those observed in disease patients.
|
|
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
The association of disease mutations and evolutionarily conserved amino acid residues
A null hypothesis describing the distribution of human genetic variation among amino acid sites in a gene can be generated assuming that point mutations occur randomly throughout that gene. If a set of mutations found in a population is representative of the random mutational process, then the number of mutations observed at a given type of site in a gene should be proportional to the frequency with which sites of that type appear in a sequence. Using information from interspecific comparisons, we tested the null hypothesis that disease-associated replacement mutations are randomly distributed among different classes of amino acid sites which were determined based on their variability among extant metazoans. This analysis permits a direct assessment of statements suggesting that disease mutations are more common at evolutionarily conserved residues. If we do not reject the null hypothesis of random association for a set of disease mutations, then mutations at conserved sites are no more important than those at variable sites for the development of the disease phenotype. In contrast, analyses will illustrate the importance of replacement mutations at conserved sites if the null hypothesis is rejected due to an overabundance of disease mutations at conserved positions and a deficiency at variable amino acid sites.
This analysis was performed for the sets of mutations obtained from each of the seven disease mutation databases examined. Type I and types II, III and IV mutations found in G6PD were analyzed in separate data sets. For comparison purposes, we also tested the sets of polymorphic replacement mutations and silent mutations obtained from the CFTR and TSC2 databases. Our analyses revealed that the mutations observed in disease patients are not randomly distributed across sites in genes. Instead, with the exception of the analysis of less severe types II, III and IV mutations to G6PD, significantly more mutations were observed at invariant amino acid positions than expected by chance alone (Fig. 2). All other sites in the genes examined, even those that have experienced a single amino acid substitution over evolutionary history, showed reduced numbers of disease mutations. Polymorphic human replacement mutations and silent mutations in CFTR and TSC2 were distributed at random with respect to the level of conservation of amino acid sites in the gene (Fig. 3), as were the types II, III and IV mutations to G6PD (Fig. 2).
|
|
This overall pattern illustrates that disease-associated mutations are only a subset of the full complement of random mutations that may occur in a gene. These subsets mainly encompass mutations at conserved residues and reflect the evolutionary constraints placed on coding regions of genes, as the proper function of a given gene product likely requires specific amino acids at certain positions of the protein. Mutations to those residues have severe phenotypic consequences, thus bringing the patients in question to the attention of clinicians. Our analyses also revealed in some instances (for example, CFTR, L1CAM, RS1, and TSC2) relatively large numbers of disease mutations at sites that vary among species (Fig. 2). However, the vast majority of the disease mutations found at variable alignment positions occur at sites that have undergone single substitutions in the phylogenetic lineages least related to humans. For example, of the six type I G6PD disease mutations observed at sites that have changed once among the species analyzed, five occur at positions that have mutated in fish, Drosophila or Caenorhabditis elegans. Likewise, we observed 79% of the CFTR disease mutations, 93% of the L1CAM mutations, and all of the RS1 mutations in once-changed sites at alignment positions that differ from humans only in fish lineages. Similarly, of the TSC2, PAH and PAX6 disease mutations found at once-changed sites, 91, 99 and 100%, respectively, occur at positions that vary only in the invertebrate sequences analyzed. Therefore, the majority of the amino acids that can produce disease mutations are conserved, at least among mammals. While the inclusion of distantly related species in analyses may seem to provide counterintuitive results (and result in the observation of disease mutations at variable sites), it is important to note that from a statistical perspective, the amino acid variation introduced through the inclusion of distantly related species is critical if one is to address the influence of evolutionary history on contemporary patterns of disease mutations. Without the genetic variation observed from the inclusion of non-mammalian species, insufficient information is available to generate counts of disease mutations at non-conserved sites for analysis purposes. Thus, we recommend using as many orthologous gene sequences as possible from both closely and distantly related metazoan species to better understand the degree with which gene sites have been conserved throughout evolutionary history.
We did not reject the null hypothesis in our analyses of polymorphic replacement and silent mutations in CFTR and TSC2, indicating that these sets of mutations were randomly distributed with respect to the level of conservation of amino acid sites (Fig. 3). However, it should be noted that the biological relevance of this pattern is not equivalent for each mutation type. In the case of silent mutations, the random association of mutations with sites is a reflection of their selectively neutral nature. Silent mutations may occur anywhere in a coding region without affecting the protein product and, barring any significant codon usage bias (11,12), should show minimal effects on an organism’s phenotype. However, the random distribution of polymorphic replacement mutations may indicate the presence of slightly deleterious amino acid changes in these genes. Since invariant positions in a sequence alignment suggest positions that are critical for proper protein function, only variable sites indicate positions where residue changes are likely tolerated due to relaxed selection constraints. Recent studies of CFTR have in fact revealed mutations that, while not responsible for the development of cystic fibrosis, are instead putatively associated with less severe disease phenotypes such as asthma, disseminated brochiectasis or chronic obstructive pulmonary disease (9,10,13,14). Several of the mutations reported in patients of these diseases were found in our list of polymorphic amino acid changes obtained from the cystic fibrosis mutation database, and many of these mutations occur at fully conserved alignment positions. Thus, the random distribution of sets of CFTR and TSC2 polymorphic replacement mutations may suggest the presence of recently evolved random replacement mutations that have relatively mild, albeit deleterious, phenotypic consequences. Perhaps only those mutations found at variable positions are truly selectively neutral. The remainder, specifically those found at conserved sites, may be eliminated by natural selection over long-term evolutionary history.
Based on the calculation of amino acid sequence alignment profile scores, among other criteria, Sunyaev et al. (15) predicted that 
20% of the non-synonymous single nucleotide polymorphisms (SNPs) in humans are to some extent deleteriou. In their study, alignment scores quantified the agreement of amino acids found at each site within a set of homologous sequences. Any observed human amino acid that changed the site-specific score beyond a set threshold was interpreted to be damaging to the protein. Likewise, Notaro et al (16). compiled all available sequence data from bacteria to vertebrates and plants for G6PD. Based on the calculation of amino acid similarities of other extant organisms with humans, they concluded that more mutations than expected are found at moderate to highly conserved sites; however, fewer mutations were seen at fully conserved residues among the 52 species analyzed. The authors suggested that mutations at fully conserved sites may be lethal, thus accounting for their relative absence in data sets, while those at moderately conserved sites are more likely to produce the observed disease phenotype in patients. If such an interpretation is correct, then we overestimated the number of fully conserved sites among species by restricting our analyses to metazoans. However, including such a wide variety of species in analyses can be problematic, as amino acid variation observed among ancient phylogenetic lineages may be correlated with functional differences in gene products. Thus, the inclusion of such distantly related species in analyses may inappropriately indicate amino acid variation at sites that are critical to the function of the gene in humans and result in an underestimate of the number of fully conserved sites within metazoan or mammalian lineages.
While the interspecific comparison approaches of Notaro et al. (16) and Sunyaev et al. (15) address the prevalence of deleterious mutations at evolutionary conserved sites, we note that both suffer technical problems relative to our approach of estimating numbers of substitution events via phylogenetic analyses. First, the similarity based approach (16) will overestimate the variability of a given site because an identical residue may appear in multiple species of a phylogenetic lineage, indicating that it also was present in the common ancestor. This means that a single substitution at a given site may be counted multiple times depending on the species sampled for analyses. It has long been recognized that species cannot be treated as independent observations for use in statistical analyses (17), as groups of species within phylogenetic lineages share common attributes (such as amino acid sequences) due to their common ancestry. However, our approach to quantifying interspecific variation makes use of known phylogenetic relationships to naturally account for the correlated amino acid changes that appear among related organisms. In addition, the quantification of amino acid similarities tends to result in large ranges of values that need to be arbitrarily clustered to form groups for analysis purposes. Such grouping is not necessary when using tree-based statistics, as the actual count of the number of substitutions provides a natural classification (Figs 2 and 3). Secondly, when considering the alignment profile score approach (15), researchers need to establish an arbitrary criterion to eliminate closely related sequences [set at 95% similar or more in the study by Sunyaev et al. (15)] from analyses to avoid overly influencing the site-specific alignment score. The inclusion of excessive closely related sequences would make the site-specific scores indicate highly conserved residues when in fact the relative lack of variation is due to the inclusion of recently diverged lineages that have not had sufficient time to accumulate amino acid changes. In contrast, our phylogenetic approach permits the use of all available sequence data, even those from species closely related to humans (Materials and Methods).
Frequencies and chemical differences of amino acid changes
It is well known that the relative rates of substitution are not uniform for different pairs of amino acids. This is clearly illustrated by matrices such as PAM (18) and BLOSUM (19) that provide information on the frequency with which a given amino acid is substituted for another over the long term evolutionary history of a large number of proteins. Our analyses showed that both the type and frequency of a given amino acid change differs considerably between those observed in disease patients and those found among species. Scatterplots of the frequencies of each type of change revealed no correlation between mutations observed in disease patients and those derived from the PAM matrix or directly observed among species in our main set of seven disease genes (Fig. 4A and B). However, estimated frequencies of changes obtained from the PAM matrix and interspecific comparisons of the seven disease genes were highly correlated (r = 0.87; Fig. 4C), illustrating that both sources of data provide similar information. Figure 5 shows in more detail the differences between the amino acid substitutions observed among species and changes commonly found in disease patients. For example, in the set of amino acid changes involving glutamic acid (E), the most common substitution among species is an aspartic acid (D). The disease-associated mutation data set, in contrast, shows changes that most often involve lysine (K) (Fig. 5). Likewise, phenylalanine
tyrosine changes (FY) are commonly observed among species. However, this change was never seen among the 1004 disease-associated mutations analyzed. We note that these types of trends are not due to the limited sampling of disease mutations from only seven genes. To illustrate this, we computed amino acid change frequencies from an extensive collection of 10 262 disease-associated replacement mutations observed in a wide variety of disease genes. These frequencies, which were calculated from data obtained at the Human Gene Mutation Database (HGMD; 20,21), were highly correlated with disease associated amino acid change frequencies observed in our main set of seven genes (r = 0.86; Fig. 4D). Overall, this finding illustrates that there is a general pattern to the types of amino acid changes seen in disease patients and that they are not of the types accepted by natural selection over long term evolutionary history.
|
|
A logical explanation for the lack of correlation between interspecific and disease-associated amino acid substitution frequencies (Fig. 4A and B) is that differences in the chemical properties of disease-associated amino acid changes are larger than those generally tolerated by natural selection. Many researchers have noted the similar physicochemical properties of the amino acids commonly substituted among species. This has been demonstrated through the use of multiple different measures of the chemical differences of amino acids (22–24) and through analyses based on the assignment of residues to categories based on common shared characteristics (25). Studies such as these illustrate that variation in the rate of substitution of different pairs of amino acids over the long-term history of species is in part due to negative selection against amino acid changes to residues with dramatically different chemical properties. While there are many coarse classifications of amino acid residues, the single most informative residue grouping is unknown. In addition, while it is seemingly logical to label a set of possible amino acid changes as ‘conservative’, the alternative label of ‘radical’ has no quantitative component and does not accurately reflect the severity of the amino acid change. Therefore, we used the chemical difference matrix of Grantham (26), a multivariate combination of residue side chain composition, polarity and volume differences, to quantify the severity of amino acid changes.
Our analyses revealed that amino acid changes associated with the development of a disease phenotype are on average more radical (larger chemical difference) than the types of amino acid changes that are observed in interspecific comparisons of species (Table 2). In all cases, chemical differences for disease-associated mutations in humans were significantly greater than those observed for amino acid changes among species. Scores for human polymorphic amino acid changes in CFTR and TSC2 were likewise significantly less than scores observed for disease-associated amino acid changes; in both cases, values for the polymorphic human variation were statistically indistinguishable from the amino acid substitutions observed among species (Table 2). In the case of G6PD, we found no significant differences between type I and types II, III and IV disease mutations. However, the average for the less severe mutation types was noticeably lower than that seen for the more severe type I mutations. Mean values for both disease mutation classes of G6PD were significantly greater than those observed for amino acid changes among species. Thus, interspecific amino acid substitutions are to some extent constrained by selection against mutated residues with dramatically different properties, and the presence of a dramatically different mutated amino acid in humans is therefore more likely to be associated with disease. Grantham’s chemical difference matrix itself is negatively correlated with the relative amino acid substitution frequency matrix of McLachlan (27), further indicating that changes to dissimilar residues are rarely tolerated by natural selection. Likewise, Krawczak et al. (28) showed that Grantham’s measure is correlated with the clinical observation likelihood of disease patients, indicating that individuals with mutant amino acid residues quite different from the wild-type are more likely to come to the attention of clinicians.
|
Although we observed no significant overabundance of less severe types II, III and IV mutations in G6PD at conserved sites (Fig. 2), chemical difference scores for these mutations were on average greater than those observed among species, but less severe than those associated with type I mutations. In this case, it appears that the mild phenotypic consequences of these changes are mainly due to the mutated residue’s radical chemical properties relative to the wild-type rather than a tendency towards the disruption of a presumably critical amino acid residue. Miyata et al. (24) have also demonstrated that the chemical differences of human hemoglobin mutations resulting in severe and mild hemolytic anemia were greater than those expected by chance, and average differences for mutations producing severe phenotypes were likewise greater than those that produce milder forms.
Polymorphic amino acid changes in CFTR and TSC2 were commonly observed, albeit not in excess, at invariant sites in the sequence alignments (Fig. 3). While information from the alignments suggests that such mutations are deleterious, the chemical difference scores for polymorphic mutations in these data sets are on average no different from those observed among species (Table 2), indicating that the types of amino acid changes seen in these non-diseased individuals should be functionally compatible with the structure of the protein. Indeed, 14 of the polymorphic CFTR mutations and eight of the polymorphic TSC2 mutations observed at invariant or once-changed sites have chemical difference scores <50, well below the average score for amino acid substitutions observed among species. Thus, in some instances, the ‘conserved’ sites where human polymorphism is observed may indicate less critical residues that, by chance alone, appear to be invariant based on the reference sequences used in analyses. These positions could likewise vary among unanalyzed individuals of the other species or vary among species that were not analyzed. In comparison, a total of seven polymorphic CFTR and two polymorphic TSC2 mutations from our data sets have chemical difference scores >100. Perhaps, of the polymorphic variation analyzed, it is the mutations with the most radical amino acid changes that are most likely to be removed from populations over long term evolutionary history.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
We have shown that disease-associated amino acid changes in humans differ from human polymorphic changes or variation seen among species in several ways. First, disease-associated amino acid changes are overabundant at conserved residues, thus illustrating the critical importance of such sites for proper gene function. Secondly, the types of amino acid changes producing disease are not likely to be similar to the types of amino acid substitutions commonly observed among species. Chemical differences of these uncommon disease-associated amino acid changes are more radical than the commonly encountered polymorphic amino acid variation found in humans or permitted by natural selection throughout evolutionary history. In some respects, the analyses conducted here are analogous to the many commonly employed procedures that use patterns of within and among species variation to elucidate the relative importance of neutral versus selective (purifying or positive) processes in the evolution of genes (29–31). In our case, the comparison of intraspecific disease-associated genetic variation and variation observed among species reveals several common trends and shows that comparative molecular analyses are directly applicable to the deleterious genetic variation found in humans. Thus, researchers should strive to obtain homologous gene sequences from both closely and distantly related species to obtain an evolutionary perspective towards the level of conservation of amino acid sites in disease associated genes and the types of amino acid changes that have been permitted in a given gene over time. The identification of disease gene homologues from the annotated genomes of organisms such as Drosophila, C.elegans, Takifugu, mouse and rat, will likely provide this evolutionary context for a wide variety of human genetic disorders.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
Data acquisition
Using available information from online databases, we obtained data for single base pair replacement mutations observed in seven disease-associated genes (Table 1). Each entry in the databases contained information on a nucleotide position, a reference nucleotide typically observed in humans at that position, and the nucleotide residue observed at that position in a disease patient. Based on nucleotide position information, we excluded all point mutations that occurred in non-coding regions, termination codons or those mutations from the database whose reference nucleotide could not be reconciled with a reference human gene sequence obtained from GenBank. Not all of the databases examined contained information on the observation frequencies of a given mutation. Therefore we included each mutation only once in our analyses to ensure that comparable data were present for each disease gene. In addition, the elimination of frequency information prevented bias in our results towards the properties of commonly observed mutations over those less frequently reported for a specific genetic disease. CFTR and TSC2 databases also contained information on ‘polymorphic’ replacement (presumably non-disease associated) and silent mutations in sufficient quantities for analysis. In the case of G6PD, information on the severity of the disease phenotype was also available (32), which permitted us to analyze severe (type I) mutations resulting in chronic non-spherocytic hemolytic anemia and less severe (types II, III and IV) mutations resulting only in enzyme deficiencies separately. In total, we examined 1004 disease-associated replacement mutations, 50 polymorphic replacement mutations and 94 silent mutations (Table 1). We further obtained complete GenBank reference cDNA sequences for human and other metazoan species for each of the disease-associated genes (Fig. 1). cDNA sequences were aligned at the amino acid level using Clustal X (33).
Determining the association between disease mutations and evolutionarily conserved sites
We performed a test to determine whether human disease mutations are more common at evolutionarily conserved amino acids than expected by chance. Interspecific variability (i.e. degree of conservation) was quantified by calculating the minimum number of amino acid substitutions that have occurred at each amino acid site throughout the evolutionary history of the given gene. This process was performed for each position using the algorithm of Fitch (34) assuming the tree topologies given in Figure 1. Alignment gaps in the human reference sequences were excluded from analyses since they provide no information about mutations that may occur in disease patients. However, human disease mutations were in a few instances observed at alignment positions containing gaps in some of the other species analyzed. In these cases, alignment gaps were treated as 21st amino acid states. Overall, this parsimony procedure provides information on the relative extent that amino acid variation has been tolerated at different sites in each gene while simultaneously accounting for shared amino acids observed within phylogenetic lineages. Ideally, parsimony is best suited for the analysis of slow evolving genes where there is no need to account for multiple hits. In other instances, we recommend the use of maximum likelihood (ML) methods in conjunction with the known phylogenetic tree to obtain counts of numbers of substitutions at each amino acid site. In this case we estimate rj, the relative evolutionary rate of site j, under a gamma distribution of rate heterogeneity (35) and a Poisson (or PAM) model of amino acid substitution, and rescale each rj such that 
rj =1. Given the total number of substitutions per site of the tree (S) and the sequence length (L), the estimated number of substitutions at site j is Sj = rj x S x L (35). For the genes examined here, estimated counts derived from the parsimony procedure described above and from ML-based analyses as implemented in PAML (36) were highly correlated (r 
0.987 for all genes). Thus, we rely on the parsimony-based results in the analyses that follow, as the estimates are discrete integer values that provide a convenient means of classifying individual sites in genes based on their variability.
If ni is the observed number of amino acid sites in the alignment that have undergone i substitutions throughout evolutionary history and N is the total number of amino acid sites in the gene, then under the null hypothesis of random association we expect to observe the fraction ni/N of the human mutations in a sample at sites that have undergone i substitutions. Thus, the expected number of human mutations at sites of type i is
where D is the total number of mutations in the data set, given by
2 tests with i degrees of freedom where
Frequencies of different amino acid changes in disease-associated mutations and interspecific comparisons
We generated four data sets to explore the null hypothesis that frequencies of different types of amino acid changes associated with disease are comparable to those observed among species. We first tabulated the frequencies of all 1004 disease-associated amino acid changes seen in the seven databases analyzed. Based on the observed counts Nij of mutations from residues i to j, we calculated the relative frequency of mutations from residues i to j as
where the summation involves only the set of k amino acid changes from residue i that are possible as a result of a single nucleotide mutation. Secondly, we calculated Mij’s for amino acid substitutions observed among species based on the combined analysis of the interspecific alignments for each of the seven disease genes. To keep frequency data comparable to those found in the disease mutation data set, only amino acid changes that could arise via single nucleotide substitutions were considered in analyses. Furthermore, each amino acid change from residue i to j at a given site in the alignment was counted only once to account for the common ancestry of residues within phylogenetic lineages. For the third data set, we modified the PAM-1 matrix (18) to obtain information on accepted point mutation frequencies from more generalized sets of homologous proteins. The PAM matrix provides instantaneous transition probabilities for all possible pairs of amino acid substitutions. Thus, we treated the probability of change from amino acid i to amino acid j as a frequency and recalculated new relative frequencies as described above, again considering only the set of amino acid changes that can result from a single base pair substitution. Finally, we used equation 3 to compute Mij’s for an extensive collection of 10 262 disease-associated replacement mutations obtained from the HGMD (20,21). This information permitted us to evaluate the extent to which disease mutations observed in a large set of disease genes are comparable to those observed in our main set of seven genes. To better understand differences in the types of amino acid changes seen among disease patients and among species, we generated simple visual depictions of the frequencies of each type of amino acid change to compare disease data to interspecific derived data. Likewise, we constructed scatterplots to evaluate the correlation of the frequencies of each type of amino acid change among data sets.
Characterizing amino acid property differences
Using Grantham’s chemical difference matrix (26), we tested the null hypothesis of no difference in chemical properties among disease-associated amino acid changes, human polymorphic amino acid variation and amino acid variation observed among species for each disease gene. Because our human mutation data sets contained information on single nucleotide mutations, only a subset of the 380 possible amino acid changes could have been observed. Thus, as with our analysis of the frequencies of different amino acid changes, we only analyzed amino acid changes among species that could have been the result of a single nucleotide mutation and scored each type of amino acid change seen at a site once to account for the residue’s common ancestry within a phylogenetic lineage. In the cases of CFTR and TSC2, we further analyzed the observed polymorphic amino acid changes in humans. From the chemical distance scores, we used a non-parametric Kruskal–Wallace test to determine if significant overall differences existed among the sets of human and interspecific amino acid changes. Scores for CFTR and TSC2 were further subjected to Mann–Whitney U tests post hoc to determine where the significant differences lie among the disease associated, human polymorphic, and interspecific scores. This same procedure was used for comparisons of type I mutations, types II, III and IV mutations and interspecific mutation scores in G6PD. These analyses were conducted using SPSS 10.0 (SPSS Inc, 1999).
|
ACKNOWLEDGEMENTS |
|---|
We thank Drs Georgia Floyd-Smith, Chris Lorson, Michael Rosenberg, Sudhindra Gadagkar, David Cutler and two anonymous reviewers for comments on this manuscript. This research was supported by grants from the National Science Foundation (DBI-9983133), National Institute of Health (HG-02096) and Burroughs Wellcome Fund (BWF-1001311) to S.K.
|
FOOTNOTES |
|---|
+ To whom correspondence should be addressed. Tel: +1 480 727 6949; Fax: +1 480 965 2519; Email: s.kumar@asu.eduPresent address:Mark P. Miller, Department of Fisheries and Wildlife, Utah State University, Logan, UT 84322-5210, USA
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTS AND DISCUSSIONCONCLUSIONSMATERIALS AND METHODSREFERENCES |
|---|
1 Green, E.D. (2000) The human genome project and its impact on the study of human disease. In Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D. (eds), The Metabolic and Molecular Bases of Inherited Disease. McGraw-Hill, New York, Vol. 1, pp. 259–298.
2 Lander, E.S., Linton, L.M., Birren, B., Nusbaum, C., Zody, M.C., Baldwin, J., Devon, K., Dewar, K., Doyle, M., FitzHugh, W. et al. (2001) Initial sequencing and analysis of the human genome. Nature, 409, 860–921.[Medline]
3 Venter, J.C., Adams, M.D., Myers, E.W., Li, P.W., Mural, R.J., Sutton, G.G., Smith, H.O., Yandell, M., Evans, C.A., Holt, R.A. et al. (2001) The sequence of the human genome. Science, 291, 1304–1351.
4 Horaitis, R., Scriver, C.R. and Cotton, R.G.H. (2000) Mutation databases: overview and catalogues. In Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D. (eds), The Metabolic and Molecular Bases of Inherited Disease. McGraw-Hill, New York, Vol. 1, pp. 113–125.
5 Cotton, R.G. (2000) Progress of the HUGO mutation database initiative: a brief introduction to the human mutation MDI special issue. Hum. Mutat., 15, 4–6.[Web of Science][Medline]
6 Welsh, M.J., Ramsey, B.W., Accurso, F. and Cutting, G.R. (2000) Cystic Fibrosis. In Scriver, C.R., Beaudet, A.L., Sly, W.S. and Valle, D. (eds), The Metabolic and Molecular Bases of Inherited Disease. McGraw-Hill, New York, Vol. 1, pp. 5121–5188.
7 Poyau, A., Buchet, K., Bouzidi, M.F., Zabot, M.T., Echenne, B., Yao, J., Shoubridge, E.A. and Godinot, C. (2000) Missense mutations in SURF1 associated with deficient cytochrome c oxidase assembly in Leigh syndrome patients. Hum. Genet., 106, 194–205.[Web of Science][Medline]
8 Girodon, E., Cazeneuve, C., Lebargy, F., Chinet, T., Costes, B., Ghanem, N., Martin, J., Lemay, S., Scheid, P., Housset, B. et al. (1997) CFTR gene mutations in adults with disseminated bronchiectasis. Eur. J. Hum. Genet., 5, 149–155.[Web of Science][Medline]
9 Lazaro, C., de Cid, R., Sunyer, J., Soriano, J., Gimenez, J., Alvarez, M., Casals, T., Anto, J.M. and Estivill, X. (1999) Missense mutations in the cystic fibrosis gene in adult patients with asthma. Hum. Mutat., 14, 510–519.[Web of Science][Medline]
10 Tzetis, M., Efthymiadou, A., Strofalis, S., Psychou, P., Dimakou, A., Pouliou, E., Doudounakis, S. and Kanavakis, E. (2001) CFTR gene mutations—including three novel nucleotide substitutions – and haplotype background in patients with asthma, disseminated bronchiectasis and chronic obstructive pulmonary disease. Hum. Genet., 108, 216–221.[Web of Science][Medline]
11 Akashi, H. (1994) Synonymous codon usage in Drosophila melanogaster: natural selection and translational accuracy. Genetics, 136, 927–935.[Abstract]
12 Akashi, H. (1995) Inferring weak selection from patterns of polymorphism and divergence at ‘silent’ sites in Drosophila DNA. Genetics, 139, 1067–1076.[Abstract]
13 Pignatti, P.F., Bombieri, C., Marigo, C., Benetazzo, M. and Luisetti, M. (1995) Increased incidence of cystic fibrosis gene mutations in adults with disseminated bronchiectasis. Hum. Mol. Genet., 4, 635–639.
14 Pignatti, P.F., Bombieri, C., Benetazzo, M., Casartelli, A., Trabetti, E., Gile, L.S., Martinati, L.C., Boner, A.L. and Luisetti, M. (1996) CFTR gene variant IVS8-5T in disseminated bronchiectasis. Am. J. Hum. Genet., 58, 889–892.[Web of Science][Medline]
15 Sunyaev, S., Ramensky, V.V., Koch, I.I., Lathe, I.W., Kondrashov, A.S. and Bork, P. (2001) Prediction of deleterious human alleles. Hum. Mol. Genet., 10, 591–597.
16 Notaro, R., Afolayan, A. and Luzzatto, L. (2000) Human mutations in glucose 6-phosphate dehydrogenase reflect evolutionary history. FASEB J., 14, 485–494.
17 Felsenstein, J. (1985) Phylogenies and the Comparative Method. Am. Naturalist, 125, 1–15.
18 Dayhoff, M.O., Schwartz, R.M. and Orcutt, B.C. (1978) A model of evolutionary change in proteins. In Dayhoff, M.O. (ed.), Atlas of Protein Sequence and Structure. National Biomedical Research Foundation, Silver Spring, pp. 345–352.
19 Henikoff, S. and Henikoff, J.G. (1992) Amino acid substitution matrices from protein blocks. Proc. Natl Acad. Sci. USA, 89, 10915–10919.
20 Krawczak, M. and Cooper, D.N. (1997) The human gene mutation database. Trends Genet., 13, 121–122.[Web of Science][Medline]
21 Krawczak, M., Ball, E.V., Fenton, I., Stenson, P.D., Abeysinghe, S., Thomas, N. and Cooper, D.N. (2000) Human gene mutation database-a biomedical information and research resource. Hum. Mutat., 15, 45–51.[Web of Science][Medline]
22 Epstein, C.J. (1967) Non-randomness of amino-acid changes in the evolution of homologous proteins. Nature, 215, 355–359.[Medline]
23 Clarke, B. (1970) Selective constraints on amino-acid substitutions during the evolution of proteins. Nature, 228, 159–160.[Medline]
24 Miyata, T., Miyazawa, S. and Yasunaga, T. (1979) Two types of amino acid substitutions in protein evolution. J. Mol. Evol., 12, 219–236.[Web of Science][Medline]
25 Zhang, J. (2000) Rates of conservative and radical nonsynonymous nucleotide substitutions in mammalian nuclear genes. J. Mol. Evol., 50, 56–68.[Web of Science][Medline]
26 Grantham, R. (1974) Amino acid difference formula to help explain protein evolution. Science, 185, 862–864.
27 McLachlan, A.D. (1971) Tests for comparing related amino-acid sequences. Cytochrome c and cytochrome c 551. J. Mol. Biol., 61, 409–424.[Web of Science][Medline]
28 Krawczak, M., Ball, E.V. and Cooper, D.N. (1998) Neighboring-nucleotide effects on the rates of germ-line single base-pair substitutions in human genes. Am. J. Hum. Genet., 63, 474–488.[Web of Science][Medline]
29 Hudson, R.R., Kreitman, M. and Aguade, M. (1987) A test of neutral molecular evolution based on nucleotide data. Genetics, 116, 153–159.
30 McDonald, J.H. and Kreitman, M. (1991) Adaptive protein evolution at the Adh locus in Drosophila. Nature, 351, 652–654.[Medline]
31 Nei, M. and Kumar, S. (2000) Molecular Evolution and Phylogenetics. Oxford University Press, Oxford.
32 Vulliamy, T., Luzzatto, L., Hirono, A. and Beutler, E. (1997) Hematologically important mutations: glucose-6-phosphate dehydrogenase. Blood Cells Mol. Dis., 23, 302–313.[Web of Science][Medline]
33 Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. and Higgins, D.G. (1997) The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res., 25, 4876–4882.
34 Fitch, W.M. (1971) Toward defining the course of evolution: minimum change for a specific tree topology. Syst. Zool., 20, 406–416.
35 Yang, Z. (1993) Maximum-likelihood estimation of phylogeny from DNA sequences when substitution rates differ over sites. Mol. Biol. Evol., 10, 1396–1401.[Abstract]
36 Yang, Z. (1997) PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci., 13, 555–556.
37 Van Camp, G., Fransen, E., Vits, L., Raes, G. and Willems, P.J. (1996) A locus-specific mutation database for the neural cell adhesion molecule L1CAM (Xq28). Hum. Mutat., 8, 391.[Web of Science][Medline]
38 Scriver, C.R., Waters, P.J., Sarkissian, C., Ryan, S., Prevost, L., Cote, D., Novak, J., Teebi, S. and Nowacki, P.M. (2000) PAHdb: a locus-specific knowledge base. Hum. Mutat., 15, 99–104.[Web of Science][Medline]
39 Brown, A., McKie, M., van Heyningen, V. and Prosser, J. (1998) The human PAX6 mutation database. Nucleic Acids Res., 26, 259–264.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. G. Kann Advances in translational bioinformatics: computational approaches for the hunting of disease genes Brief Bioinform, January 1, 2010; 11(1): 96 - 110. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Kumar, M. P. Suleski, G. J. Markov, S. Lawrence, A. Marco, and A. J. Filipski Positional conservation and amino acids shape the correct diagnosis and population frequencies of benign and damaging personal amino acid mutations Genome Res., September 1, 2009; 19(9): 1562 - 1569. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Saito, Q. Chi, H. Zhuang, H. Matsunami, and J. D. Mainland Odor Coding by a Mammalian Receptor Repertoire Sci. Signal., March 3, 2009; 2(60): ra9 - ra9. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Domazet-Loso and D. Tautz An Ancient Evolutionary Origin of Genes Associated with Human Genetic Diseases Mol. Biol. Evol., December 1, 2008; 25(12): 2699 - 2707. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Errasti-Murugarren, P. Cano-Soldado, M. Pastor-Anglada, and F. J. Casado Functional Characterization of a Nucleoside-Derived Drug Transporter Variant (hCNT3C602R) Showing Altered Sodium-Binding Capacity Mol. Pharmacol., February 1, 2008; 73(2): 379 - 386. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Sandelin, A. Nordlund, P. M. Andersen, S. S. L. Marklund, and M. Oliveberg Amyotrophic Lateral Sclerosis-associated Copper/Zinc Superoxide Dismutase Mutations Preferentially Reduce the Repulsive Charge of the Proteins J. Biol. Chem., July 20, 2007; 282(29): 21230 - 21236. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Thusberg and M. Vihinen The structural basis of hyper IgM deficiency - CD40L mutations Protein Eng. Des. Sel., March 1, 2007; 20(3): 133 - 141. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. M. Goldstein, M. Chan, M. Harland, E. M. Gillanders, N. K. Hayward, M.-F. Avril, E. Azizi, G. Bianchi-Scarra, D. T. Bishop, B. Bressac-de Paillerets, et al. High-risk Melanoma Susceptibility Genes and Pancreatic Cancer, Neural System Tumors, and Uveal Melanoma across GenoMEL. Cancer Res., October 15, 2006; 66(20): 9818 - 9828. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Lang, M. Haberl, D. Jung, A. Drescher, R. Schlagenhaufer, A. Keil, E. Mornhinweg, B. Stieger, G. A. Kullak-Ublick, and R. Kerb Genetic Variability, Haplotype Structures, and Ethnic Diversity of Hepatic Transporters MDR3 (ABCB4) and Bile Salt Export Pump (ABCB11) Drug Metab. Dispos., September 1, 2006; 34(9): 1582 - 1599. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. G. Randles, I. Lappalainen, S. B. Fowler, B. Moore, S. J. Hamill, and J. Clarke Using Model Proteins to Quantify the Effects of Pathogenic Mutations in Ig-like Proteins J. Biol. Chem., August 25, 2006; 281(34): 24216 - 24226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. S. Sullivan, S. J. Bowne, D. G. Birch, D. Hughbanks-Wheaton, J. R. Heckenlively, R. A. Lewis, C. A. Garcia, R. S. Ruiz, S. H. Blanton, H. Northrup, et al. Prevalence of disease-causing mutations in families with autosomal dominant retinitis pigmentosa: a screen of known genes in 200 families. Invest. Ophthalmol. Vis. Sci., July 1, 2006; 47(7): 3052 - 3064. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Conde, J. M. Vaquerizas, H. Dopazo, L. Arbiza, J. Reumers, F. Rousseau, J. Schymkowitz, and J. Dopazo PupaSuite: finding functional single nucleotide polymorphisms for large-scale genotyping purposes. Nucleic Acids Res., July 1, 2006; 34(Web Server issue): W621 - W625. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Mathe, M. Olivier, S. Kato, C. Ishioka, P. Hainaut, and S. V. Tavtigian Computational approaches for predicting the biological effect of p53 missense mutations: a comparison of three sequence analysis based methods Nucleic Acids Res., March 6, 2006; 34(5): 1317 - 1325. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. J. Urban, R. Sebro, E. H. Hurowitz, M. K. Leabman, I. Badagnani, L. L. Lagpacan, N. Risch, and K. M. Giacomini Functional genomics of membrane transporters in human populations Genome Res., February 1, 2006; 16(2): 223 - 230. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Lopez-Bigas, B. J. Blencowe, and C. A. Ouzounis Highly consistent patterns for inherited human diseases at the molecular level Bioinformatics, February 1, 2006; 22(3): 269 - 277. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. F. Rudd, R. D. Williams, E. L. Webb, S. Schmidt, G. S. Sellick, and R. S. Houlston The Predicted Impact of Coding Single Nucleotide Polymorphisms Database Cancer Epidemiol. Biomarkers Prev., November 1, 2005; 14(11): 2598 - 2604. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. D. Jones and D. J. Begun Parallel evolution of chimeric fusion genes PNAS, August 9, 2005; 102(32): 11373 - 11378. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. A. Stone and A. Sidow Physicochemical constraint violation by missense substitutions mediates impairment of protein function and disease severity Genome Res., July 1, 2005; 15(7): 978 - 986. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Conde, J. M. Vaquerizas, C. Ferrer-Costa, X. de la Cruz, M. Orozco, and J. Dopazo PupasView: a visual tool for selecting suitable SNPs, with putative pathological effect in genes, for genotyping purposes Nucleic Acids Res., July 1, 2005; 33(suppl_2): W501 - W505. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L M Messiaen and K Wimmer Pitfalls of automated comparative sequence analysis as a single platform for routine clinical testing for NF1 J. Med. Genet., May 1, 2005; 42(5): e25 - e25. [Full Text] [PDF]
|
|
|
|
|
|
S. Balasubramanian, Y. Xia, E. Freinkman, and M. Gerstein Sequence variation in G-protein-coupled receptors: analysis of single nucleotide polymorphisms Nucleic Acids Res., March 22, 2005; 33(5): 1710 - 1721. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Wu, S. R. Hinson, A. Ohashi, D. Farrugia, P. Wendt, S. V. Tavtigian, A. Deffenbaugh, D. Goldgar, and F. J. Couch Functional Evaluation and Cancer Risk Assessment of BRCA2 Unclassified Variants Cancer Res., January 15, 2005; 65(2): 417 - 426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Sivendran, D. Patterson, E. Spiegel, I. McGown, D. Cowley, and R. F. Colman Two Novel Mutant Human Adenylosuccinate Lyases (ASLs) Associated with Autism and Characterization of the Equivalent Mutant Bacillus subtilis ASL J. Biol. Chem., December 17, 2004; 279(51): 53789 - 53797. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. M. Sinilnikova, S. M. Ginolhac, C. Magnard, M. Leone, O. Anczukow, D. Hughes, K. Moreau, D. Thompson, C. Coutanson, J. Hall, et al. Acetyl-CoA carboxylase {alpha} gene and breast cancer susceptibility Carcinogenesis, December 1, 2004; 25(12): 2417 - 2424. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. D. Thomas and A. Kejariwal Coding single-nucleotide polymorphisms associated with complex vs. Mendelian disease: Evolutionary evidence for differences in molecular effects PNAS, October 26, 2004; 101(43): 15398 - 15403. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
V Abkevich, A Zharkikh, A M Deffenbaugh, D Frank, Y Chen, D Shattuck, M H Skolnick, A Gutin, and S V Tavtigian Analysis of missense variation in human BRCA1 in the context of interspecific sequence variation J. Med. Genet., July 1, 2004; 41(7): 492 - 507. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Conde, J. M. Vaquerizas, J. Santoyo, F. Al-Shahrour, S. Ruiz-Llorente, M. Robledo, and J. Dopazo PupaSNP Finder: a web tool for finding SNPs with putative effect at transcriptional level Nucleic Acids Res., July 1, 2004; 32(suppl_2): W242 - W248. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Lopez-Bigas and C. A. Ouzounis Genome-wide identification of genes likely to be involved in human genetic disease Nucleic Acids Res., June 4, 2004; 32(10): 3108 - 3114. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Gray, L. M. Mangravite, R. P. Owen, T. J. Urban, W. Chan, E. J. Carlson, C. C. Huang, M. Kawamoto, S. J. Johns, D. Stryke, et al. Functional and Genetic Diversity in the Concentrative Nucleoside Transporter, CNT1, in Human Populations Mol. Pharmacol., March 1, 2004; 65(3): 512 - 519. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. K. Leabman, C. C. Huang, J. DeYoung, E. J. Carlson, T. R. Taylor, M. de la Cruz, S. J. Johns, D. Stryke, M. Kawamoto, T. J. Urban, et al. Natural variation in human membrane transporter genes reveals evolutionary and functional constraints PNAS, May 13, 2003; 100(10): 5896 - 5901. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. M. Cooper, M. Brudno, N. C. S. Program, E. D. Green, S. Batzoglou, and A. Sidow Quantitative Estimates of Sequence Divergence for Comparative Analyses of Mammalian Genomes Genome Res., May 1, 2003; 13(5): 813 - 820. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||