MBE Advance Access originally published online on September 19, 2007
Molecular Biology and Evolution 2007 24(12):2707-2715; doi:10.1093/molbev/msm202
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
Heterogeneous Rate of Protein Evolution in Serotonin Genes
Unitat de Biologia Evolutiva, Universitat Pompeu Fabra, Barcelona, Spain
E-mail: andresa{at}mail.nih.gov.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Serotonin (5-hydroxytryptamine) is a neurotransmitter crucial for cardiovascular, gastrointestinal, and brain function. It is also involved in several aspects of behavior and associated with a variety of personality disorders in humans. Its dual role as a crucial element in vital physiological functions (strictly evolutionary conserved) and in traits that differ substantially across species makes the evolution of serotonin function particularly interesting. We studied the evolution of serotonin function through the identification of the selective forces shaping the evolution of genes in its functional pathway in primates and rodents. Serotonin genes are highly conserved and show no signals of positive selection, suggesting functional constraint as the main force driving their evolution. They show, nevertheless, considerable differences in constraint between primates and rodents, with some genes showing dramatic differences between the 2 groups. These genes most likely represent cases of functional divergence between primates and rodents and point out to the relevance of using closely related species in gene-based evolutionary studies to avoid the effect of unrecognized functional differences between distant species. Within each group (rodents or primates), genes also show heterogeneity in evolution. Genes from the same gene family (with structure and function alike) tend to evolve at a similar rate, but this is not always the case. A few serotonin genes show substantial differences in constraint with the rest of members of their family, suggesting the presence of important and unrecognized functional differences among the genes, which may be involved in species-specific evolution.
Key Words: serotonin • 5-HT • conservation • evolutionary rate • functional pathway • gene family
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Serotonin (5-hydroxytryptamine, 5-HT) is a crucial neurotransmitter for normal development and behavior. Serotonin neurons innervate areas all throughout the brain and are involved in a variety of physiological processes including cardiac development and function, gastrointestinal function, and liver regeneration. It is also involved in aggression, substance abuse, addictive behavior, anxiety, depression, obsessive control, and learning. In addition, variants in serotonin genes are clearly associated with both normal and pathologic behavioral variation in humans (for a review, see Lucki 1998
). Briefly, the serotonin functional pathway can be summarized in a few steps (fig. 1). The neurotransmitter is synthesized in serotonergic neurons by a metabolic pathway whose limiting enzyme is tryptophan hydroxilase (TPH); 2 paralogous genes have been described, the nonneural (TPH1) and neural (TPH2) forms of the enzyme. Neuron activation results in serotonin release into the intersynaptic space, where it interacts with diverse serotonin receptors (HTR). Virtually all HTRs are G-protein–coupled proteins (metabotropic receptors), the best known belonging to the HTR1 and HTR2 families. HTR1 (HTR1A, B, D, E, and F) are inactivating pre- and postsynaptic receptors that reduce neurotransmitters' release. HTR2 (HTR2A, B, and C) are postsynaptic receptors, responsible for serotonin activity over a diversity of neuronal regions. The only nonmetabotropic receptors are HTR3 (HTR3A, B, and the more recently described C, D, and E), nonselective cationic channels that mediate 5-HT function by depolarization of the postsynaptic membrane. Serotonin transporter (SERT or 5-HTT) removes 5-HT from the synaptic space by reuptake to the presynaptic neuron, where it is stored in vesicles or degraded via deamination by MAOA (monoamine oxidase A).
|
The fundamental role of serotonin in basic mechanisms like development or cardiac and gastrointestinal functions is evidenced by the conservation of its function and functional pathway across highly divergent species. But serotonin function is also critical in functions that, like learning and behavior, vary widely among lineages. This duality makes the evolution of the pathway particularly intriguing. Another attractive characteristic of the pathway is its modularity. On the one side, virtually every gene family contains several members with extremely similar functions. On the other, every gene of the pathway has a specific molecular role, and they are all together necessary for the global function of serotonin. Understanding the selective forces affecting the evolution of the genes and the strength of selection they are subject to is essential to understand the evolution of complex functional pathway such as that of serotonin.
In principle, the strength of selection (positive or purifying selection) can be estimated as the proportion of variants fixed in the population by the action of positive selection or removed from the population by the action of purifying selection. In protein-coding genes the inference is possible because, under the assumption that selection affects mainly nonsynonymous variants, synonymous sites act as an "internal control" of nonselective forces. Under neutrality, dN (nonsynonymous fixed differences per nonsynonymous site) is expected to equal dS (synonymous fixed differences per synonymous site), and their ratio (dN/dS or 
) is to equal 1. dN/dS > 1 reflects the action of accelerated evolution and positive selection, and the magnitude of its departure from 1 reflects the proportion of sites fixed by selection. dN/dS < 1 denotes selective constraint, and its departure indicates the proportion of variants removed by purifying selection. Strong departures from 1 are the result of severe selective pressure, either to fast sequence change (dN/dS > 1) or to strict sequence conservation (dN/dS 
0). These 2 extremes are unusual, though, and, for example, 29% of genes have dN/dS = 0 between human and chimpanzee, and only 4.4% have dN/dS > 1 (Chimpanzee Sequence and Analysis Consortium 2005). Thus, most genes lay somewhere in between those 2 extremes, with 0 < dN/dS < 1, usually in the low range. Because positive selection is thought to affect a small portion of genes in mammals (Eyre-Walker 2006), variation in the rate of protein evolution in those genes is expected to reflect, mainly, heterogeneity in constraint, that is, the strength of purifying selection.
In an effort to determine the selective forces affecting the evolution of serotonin function we compared the evolutionary patterns, in primates and rodents, of genes responsible for the function of serotonin. Several methods were applied to measure the strength of selection affecting the evolution of the different components of the functional pathway. Serotonin genes show differences in constraint between primates and rodents, but little heterogeneity in evolution within the species of each group. We find no evidence for an important role of positive selection, but substantial variability in constraint among genes, even within families of structurally and functionally similar genes.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
The complete coding sequence (cds) of 13 genes encoding proteins involved in the serotonin pathway, including all major functions of the path, was obtained for human (Homo sapiens), bonobo (Pan paniscus), chimpanzee (Pan troglodytes), gorilla (Gorilla gorilla), orangutan (Pongo pygmaeus), mouse (Mus musculus), and rat (Rattus norvegicus). No data from genomic projects were considered to avoid the harmful effect of sequencing errors, which may bias results (Puente et al. 2006
; Taudien et al. 2006), especially if their rate varies between species. For HTR1 genes, human sequences were obtained from RefSeq database (Pruitt et al. 2007) and nonhuman primate sequences from the Silver Project Database (http://sayer.lab.nig.ac.jp/silver/index.html). MAOA coding sequences for humans and apes were previously reported (Andrés et al. 2004). The rest of genes were sequenced in 2 humans, 1 bonobo, 2 chimpanzees, 2 gorillas, and 2 orangutans. All coding exons were polymerase chain reaction amplified and sequenced (ABI PRISM sequencing) in both directions with intronic human-designed primers.
Coding sequences were combined to reconstruct the cds of the gene, and polymorphisms were discarded. The human samples are 2 Catalan (Spain) anonymous donors, and apes' samples were obtained from diverse sources (Barcelona Zoo, Coriell Cell Repositories, and European Collection of Cell Cultures). All rodent sequences were obtained from RefSeq database (Pruitt et al. 2007). Sequences used are the best reciprocal hit (with Blast and BLAT) to the human reference sequence for the gene, which ensures correct orthology assignation. Extensive BLAT searches to the chimpanzee and macaque genomes suggest the absence of additional, previously unidentified, paralog genes that could alter the specific functional relationships between components of the pathway.
For every gene, primate and rodent alignments were created (separately) with SeqMan 5.05 (DNASTAR, www.dnastar.com). Both groups were analyzed separately to avoid a strong influence of the long rodent–primate branch and to avoid problems associated with a possible saturation of synonymous substitutions (dS) in that branch. Also, a separate analysis ensures a similar function of the genes for all species considered together and allows the contrast of constraint between the 2 groups. The combined tree represents 45 MYA in primates and 70 MYA in rodents. Analyses were performed with codeml program in PAML 3.14 package (Yang 1997). The algorithm fits the data to a diversity of evolutionary models and computes the likelihood of the data under each model. Models are nested: more complex models are based on simpler ones and allow variation in additional parameters (estimated by the algorithm). The models can be compared using a likelihood ratio test, which tests the statistical fit of the data to every model (measured by the likelihood of every method) considering their different degrees of freedom. The test determines which model fits significantly better the data. The algorithm also provides estimates of the relevant parameters of the model. Relevant to this study are the following: r's, the tree length; 
, the codons equilibrium frequency; 
, the ratio of transitions to transversions; dN, the number of nonsynonymous substitutions per nonsynonymous site; dS, the number of synonymous substitutions per synonymous site; and their ratio, dN/dS or . The rate of protein evolution (dN/dS or ) reflects the strength and direction of natural selection and is the main focus of this study.
dN/dS is an implicit test of positive selection, as values >1 are considered a signal of accelerated evolution due to positive selection and values <1 reflect sequence change constraint due to purifying selection. But this is a very stringent test of positive selection that ignores site heterogeneity and has no power to detect selection in one or a few amino acid sites. Evidence of positive selection on specific sites was tested by adjusting the data to a model (M0) with a single dN/dS for all sites (estimated by the algorithm) and to 2 additional evolutionary models. The nearly neutral model (M1) allows for variation in dN/dS among sites, with all sites evolving under neutrality (dN/dS = 1) or under selective constraint (dN/dS < 1, estimated by the algorithm). The positive selection model (M2) allows for a third category of sites evolving under positive selection (dN/dS > 1, estimated by the algorithm). Given the absence of evidence for positive selection, M0 was chosen for the rest of analyses (see Results).
dN/dS reflects global selective constraints in a lineage only when the constraint is constant across time and among the species considered. Otherwise, it represents an average of diverse branch-specific rates of evolution with little biological meaning. Heterogeneity in evolutionary rates among branches was tested by the contrast of 2 models of evolution: the "one-ratio model" (a single dN/dS value for all branches in the tree) versus the "free-ratio model" (free dN/dS variation among branches). This test was not performed in rodents, with only 2 species. Given the absence of evidence for branch heterogeneity, the one-ratio model was chosen for the rest of analyses (see Results) and dN/dS estimated by this method was considered the estimate of dN/dS for that group.
The ratio of evolution of different regions can be compared with the "fixed-sites" models of PAML (Yang and Swanson 2002). Such models consider a group of regions jointly (with no heterogeneity among them) or with increasing parameter heterogeneity. As presented in Yang and Swanson (2002), the models are as follows:
- Model A [no G option in PAML]: all sequences are considered jointly (no parameter heterogeneity),
- Model B [G option, Mgene 0]: sequences may differ in tree length (r's),
- Model C [G option, Mgene 2]: sequences may differ in r's and codon frequencies ('s),
- Model D [G option, Mgene 3]: sequences may differ in r's and the combination of transition/transversion ratio () and protein evolutionary rate (, or dN/dS),
- Model E [G option, Mgene 4]: sequences may differ in all the above parameters (r's, 's, , and ),
- Model F [G option, Mgene 1]: sequences are analyzed separately.
To test heterogeneity in evolutionary rate within gene families, the sequence of the diverse members of every gene family was combined: HTR1 (HTR1A, HTR1B, and HTR1F), HTR2 (HTR2A, HTR2B, and HTR2C), HTR3 (HTR3A and HTR3B), and TPH (TPH1 and TPH2).
An important limitation of the fixed-sites models is that the test of variation in cannot be decoupled from the test of variation in , so a strict test of heterogeneity is not possible. In addition, a statistical comparison of is difficult because the algorithm provides point estimates of , but the calculation of the standard error assumes very long samples and normality in the likelihood curve. To avoid making assumptions about the likelihood curve, a resampling strategy (similar to the bootstrap approach widely used in phylogenetic studies) was used to estimate the dispersion of within each gene. For every gene alignment, 10,000 pseudoalignments were created by resampling with replacement over codons of the gene, and dN/dS was estimated. The 10,000 dN/dS form a distribution whose first moment estimators (mean and median) should be similar to the original dN/dS for the gene, and whose variance reflects heterogeneity among codons. Using a conservative approach, the distributions of 2 genes were considered different if their 95% confidence interval (CI) did not overlap.
The evolution of different protein domains was analyzed in a similar way. SwissProt information was used to localize the extracellular, intracellular, and transmembrane domains of the metabotropic receptors with information for all species: HTR1 (A and F) and HTR2 (A, B, and C). Domains were concatenated to obtain a single alignment for each of the domains and were analyzed as described above.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Protein Domains
Of the 13 serotonin genes analyzed, 7 belong to the group of metabotropic receptors; G-protein–coupled receptors anchored to the cellular membrane by 7 transmembrane domains with an extracellular region that binds serotonin and an intracellular one that activates the cellular response. The intrinsic differences in structure and function of domains predict some heterogeneity in their evolution. For example, transmembrane regions, under strong constraint for sequence, amino acid composition, structure, and function are expected to evolve under strict purifying selection (low dN/dS). Globular domains with higher structural flexibility and a larger set of possible amino acids are expected to evolve (on average) under a more relaxed constraint.
Actually, the ratio of protein evolution varies substantially between the 3 domains when the sequence of the metabotropic genes is combined: in rodents, dN/dS is 2.08 times lower in transmembrane than in intracellular domains and 2.7 than in extracellular ones (table 1, lower panel). In primates, the difference is even stronger, with dN/dS 3.9 times lower in transmembrane domains than in intracellular regions and 5.8 than in extracellular ones (table 1).
|
The fixed-sites models provide a comparison of evolutionary parameters between domains by allowing, in different models, increasing levels of parameter heterogeneity among domains. Results of the likelihood ratio test are presented in table 2. Focusing initially on rodents, Model B (heterogeneity in r's) fits the data significantly better than Model A (no heterogeneity), suggesting some variability in the rates of evolution between domains. Model C (r's +
) gives a better fit to the data than Model B (r's), which reflects some heterogeneity in codon usage. In addition, Model D (r's + + ) is a better fit than Model B (r's), suggesting that the transition/transversion ratio and/or the ratio of nonsynonymous/synonymous substitutions varies across domains. The significantly better fit of Model E (r's + + + ) over Model C (r's + ) further supports this result. Finally, Model F (separate analyses) does not improve the fit over Model E (r's + + + ), suggesting that few (if any) additional parameters contribute to evolutionary heterogeneity between domains.
|
The scenario in primates is more complex. The only significant contrast is that of Model C (r's +
) versus Model B (r's), reflecting the differences in codon frequencies among domains (table 2). Allowing for variation in and does not improve the fit of the models. Nevertheless, as shown above, varies greatly among domains (table 2).
Both and are interrogated together in the fixed-sites models. So variation may interfere with the test of heterogeneity, and alternative strategies are necessary for a strict comparison. Such contrast can be achieved using a bootstrap approach, which provides information about dN/dS variation (in this case among codons in the domain) and allows a between-domain statistical comparison by considering the overlap between their bootstrap distributions. Consistent with the low dN/dS estimates, the bootstrap dN/dS distribution for transmembrane domains is narrow and centered at low values (fig. 2) as a result of strong and homogeneous selective pressure on transmembrane regions. The distribution for intracellular and, particularly, extracellular domains are wider and centered around the point estimate of dN/dS. According to this analysis, dN/dS differs significantly between domains in rodents (table 1), specifically between transmembrane domains (highly conserved) and intracellular–extracellular domains (more free to vary). In primates, the domains also show different distributions, but their 95% CI overlap due to the higher dispersion of the distributions (fig. 2).
|
Genes
The average dN/dS of the pathway is 0.203 in primates and 0.139 in rodents. This is a low ratio, which reflects the high conservation of the genes in the pathway. However, this average encloses considerable heterogeneity, with estimated dN/dS ranging from 0.047 to 0.378 in rodents and from 0.045 to 0.796 in primates (table 1).
None of the serotonin genes show significant heterogeneity in protein evolutionary rate between primate branches (nonsignificant contrast between the one-ratio model and the free-ratio model). Thus, the overall dN/dS can be considered as the global rate of protein evolution. No gene shows, either, signals of positive selection at specific sites (nonsignificant contrast between Model 1 and Model 2) in rodents or primates, suggesting that positive selection has not played an important role during their recent evolution in these 2 groups of species. The existence of some isolated event of positive selection cannot be completely discarded, but their influence is expected to be negligible given the very long evolutionary time considered. Thus, differences in evolutionary rate between genes likely reflect, mainly, differences in the strength of purifying selection.
In principle, the effect of heterogeneity in mutation rate between genes is accounted for by the use of dS as an internal control of the effects of mutation. A hypothetical influence of mutation rate in dN/dS can be further investigated by the relationship between dN/dS and correlates of mutation rate. Probably the best described correlates of mutation rates in mammals are GC content (Smith and Hurst 1999; Bielawski et al. 2000; Mouse Genome Sequencing Consortium et al. 2002) and recombination rate (Hellman et al. 2003). Both are uncorrelated with dN/dS in these data, suggesting a minor effect of mutational biases in the results (Spearman correlations: dN/dS vs. GC in primates –0.022, P = 0.9494; dN/dS vs. GC in rodents 0.330, P = 0.2673; dN/dS vs. recombination in primates 0.479, P = 0.1154).
A direct comparison of dN/dS between serotonin genes is complex due to intrinsic differences in sequence, structure, and function. The interacting surface area and other structural characteristics are important in evolvability (Kim et al. 2006), so the different serotonin genes are not necessarily expected to evolve at similar rates. For example, ligand-gated receptors (like HTR3) have been shown to evolve, globally, slower than G-protein–coupled receptors (like HTR1 and HTR2) between humans and mouse (Iwama and Gojobori 2002). Genes within a gene family, though, share a common ancestry and (in the case of 5-HT genes families) very similar structure and function. Their similar structure, binding molecule, protein interactions, and expression patterns predict a similar selective constraint for the different members of each serotonin gene family.
To test this, the rate of protein evolution of genes within a family was compared by different approaches: the PAML fixed-sites model (table 2), the dN/dS estimate for every gene (table 1), and the bootstrap distribution of dN/dS (table 1). The fixed-sites models allow the comparison of all genes of each family when they are tested together. Either a significant contrast between Model D (r's + + ) and Model B (r's) or between Model E (r's + + + ) and Model C (r's + ) is suggestive of (and/or ) heterogeneity within the family. Only HTR2 in rodents and HTR1 and HTR3 in primates have significant heterogeneity signals in + (table 2). This agrees with results from the direct comparison of among genes (table 1). In rodents, HTR2B rate of protein evolution (0.378) is 3 times faster than that in the other genes of the family (0.120 for HTR2A and 0.110 HTR2C). In primates, HTR1A, HTR1B, and even HTR1E (not considered in the fixed-sites model because it is not present in rodents) share low estimates (0.102, 0.047, and 0.045, respectively); HTR1F rate (0.536) is between 5 and 11 times faster. Likewise, HTR3A in primates has a typical rate of protein evolution for a serotonin gene (0.160), whereas is almost 5 times higher in HTR3B (0.796). Such differences are driven by dN rather than by dS (table 1). Extremely similar estimates of were obtained with fixed-sites Models D and E (supplementary table 1, Supplementary Material online).
The bootstrap analysis shows differences between HTR2B and the other HTR2 genes in rodents, and between HTR3B and HTR3A in primates, but not between HTR1F and the other HTR1 genes in primates. This is due to the conservative comparison between distributions and the high dispersion of the bootstrap distribution of HTR1F in primates (fig. 3) as a result of the high variation between codons. It is worth noting that in no other case a gene shows differences with other members of its family, indicative of the good correspondence between the 3 methodological approaches.
|
Lineages
The average rate of protein evolution in rodents is slightly lower than that in primates (table 1) and the bootstrap dN/dS distributions are slightly narrower (fig. 3), consistent with a more efficient action of purifying selection likely due to the larger effective population size of rodents. Nevertheless, global differences between lineages should be interpreted with caution because dN/dS estimates are based, in the 2 groups, on different number of species and total evolutionary time.
Maybe the most surprising result is the lack of correlation between the ratio of protein evolution in primates and rodents (Spearman correlation –0.103, P = 0.785). Such disjoint is not expected given the good correlation of gene evolutionary rates across species at the genomic level (Lindblad-Toh et al. 2005; supplementary fig. S11, Supplementary Material online) and suggests some differences in selective constraints of serotonin pathway elements in rodents and primates.
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
As early as 1977, Wilson et al. (1977)
predicted that essential proteins, evolving under strong selective constraints, would show higher conservation than less essential or nonessential ones. This prediction was subsequently supported by the recurrent finding, in diverse organisms, of a correlation between sequence evolutionary rate and dispensability (Hirsh and Fraser 2001) or associated traits (Pal et al. 2001; Fraser et al. 2002; Jordan et al. 2002; Fraser et al. 2003; Pal et al. 2003; Hahn and Kern 2005; Liao et al. 2006). The relationship between "essentiality" and evolutionary rate is controversial (e.g., Jordan et al. 2003; Yang et al. 2003), but the principle that sequence conservation reflects functional constraint is very much in place. It is used, for example, to help predict genes and their structure (Guigo et al. 2003; Parra et al. 2003), understand the functional relevance of protein domains (Madsen et al. 2002), or localize noncoding functional segments of the genome (Dermitzakis et al. 2002; Margulies et al. 2003; Bejerano et al. 2005; Siepel et al. 2005). In coding regions, in the absence (or nearly absence) of positive selection and with sufficiently large branch lengths, dN/dS measures the strength of purifying selection in preventing the fixation of deleterious replacement variants, that is, dN/dS reflects selective constraint, here understood as the limitation that selection poses on gene variation between species. The relationship between constraint and dN/dS is exemplified by the differences in protein evolutionary rate between transmembrane and globular domains in metabotropic receptors (due to their intrinsic differences in structural and functional requirements) also reported for another G-protein–coupled receptor, A2AB (Madsen et al. 2002). Similarly, evolutionary patterns can help understand the evolutionary constraints affecting individual genes or the combined evolution of groups of proteins functionally related, for example, by jointly creating a given functional pathway. But the comparison of individual genes requires consideration of possible nonconstraint factors whose genomic heterogeneity may interfere in the estimates of constraint, like positive selection, heterogeneity between species, and differences in mutation and recombination rates. None of such factors seem to play an important role in this case. First, the raise in dN/dS observed in some genes is driven in all cases by an increase in dN (mainly affected by selection) rather than a reduction of dS (mostly affected by mutation). Second, dN/dS does not correlate with GC content or recombination rate in these data, suggesting that the effect of mutational load, interference, and positive selection are minimal. Finally, dN/dS is consistently low, and we find no evidence of positive selection or rate branch heterogeneity. In summary, all the characteristics of the data suggest that dN/dS reflects mainly selective constraint.
Serotonin genes have evolved under strong purifying selection, as shown by their low dN/dS both in primates and rodents, close to 0 for most genes. In general, gene families evolve under relatively homogeneous constraint, but there are interesting exceptions. The gene with the highest rate of protein evolution of the path is HTR3B in primates. HTR3 are pentameric inotropic receptors responsible for fast responses to 5-HT release; they are involved in pain sensitivity, emetic reflex, and anxiety. HTR3A subunits are necessary for the channel function, and they may or may not be combined with (nonessential) HTR3B subunits to produce heteromeric receptors (Davies et al. 1999). All HTR3 subunits, which include also the recently discovered C, D, and E subunits, are involved in the formation of HTR3 receptors. They all share a common ancestor and sequence and structural similarity (Niesler et al. 2003). Homomeric and heteromeric channels have different conductance properties (Niesler et al. 2007) suggesting that non–A subunits may play a role in the creation of a collection of HTR3 receptors with subtle physicochemical and functional differences. HTR3A, the essential subunit, is highly conserved both in primates and rodents. HTR3B, in contrast, has evolved under more relaxed constraint in primates. This is a surprising difference because the 2 proteins must interact physically and functionally to create functional receptors. It may reflect certain freedom to change for a subunit mainly responsible for variability in the properties of different types of HT3 receptors.
HTR1E and HTR1F are a particularly interesting case. They are close paralogs with similar sequence and structure, so much so that antibodies do not specifically bind one but not the other (Barnes and Sharp 1999; Bai et al. 2004). HTR1E is present in primates but absent in mouse and rat, where it was probably lost because it is present in guinea pig (Bai et al. 2004) and a variety of mammals. In primates the gene is highly conserved, with no protein change in the human–chimp–gorilla lineages (representing about 20 MYA of evolution) and high conservation in human populations (Shimron-Abarbanell et al. 1995). HTR1F represents the opposite case. In primates, the rate of HTR1F is almost 12 times higher than the rate of HTR1E, with dN 3 times higher. The gene is, though, highly conserved in rodents. It is certainly possible that the function of HTR1E is completely carried on by HTR1F in rodents, but this seems unlikely. Their sequence divergence (60% at the protein level in humans) and their conservation in divergent species suggest that their function is not identical. An obvious explanation for their evolutionary asymmetry involves differences in the influence in fitness of their function, high for HTR1E in primates and for HTR1F in rodents. Other possible scenarios, though, exist. For example, the 2 genes may still partially overlap in function, and the loss of HTR1E in rodents (by drift or reduced constraint) may have increased constraint for HTR1F in that lineage. Actually, even distantly related genes may provide backup function after deletion (Ihmels et al. 2007), so HTR1F may be at least partially compensating for the loss of HTR1E in rodents. This would require a considerable overlap in expression pattern and cellular localization of the 2 genes, and such overlap is still unclear. Also, the 2 genes may differ in the interaction spectra, with different partners and/or number of partners, and those differences may strongly influence the sequence and structural constraints of the 2 receptors. If such interactions vary between primates and rodents (implying there are important differences in the pathways responsible for the transmission of the signal in the 2 lineages), they would explain the large differences in constraint observed. Whatever the mechanism that maintains both genes under differential selective pressures in the 2 lineages, our data suggest that primates and rodents probably differ, at least partially, in the functions carried out by these 2 genes. A deep analysis of other elements involved in signal transmission, like proteins interacting with HTR1E and HTR1F, may help better understand the functional differences between the 2 genes in these lineages.
HTR2B shows a relatively low conservation pressure in rodents. Its proposed role in brain and heart development (Nebigil et al. 2000, 2001) and in lung regeneration (Lesurtel et al. 2006) contrasts with a relatively high protein evolutionary rate in rodents. Its evolution in rodents differs as well with its high conservation in primates and the high conservation of the rest of genes of the family (HTR2A and HTR2B) in rodents.
Other genes have evolved under more strict conservation pressures both in primates and rodents. TPH genes (responsible for the synthesis of serotonin) and SERT (responsible for its reuptake) show very low evolutionary rates. The constraint is slightly lower in MAOA both in rodents and primates. This protein has been proposed to have experienced a very recent episode of positive selection in humans (Gilad et al. 2002) but not in other species, probably due to the fixation of a single amino acid change (Andrés et al. 2004). Selection was detected, in that case, using intraspecific variability data in humans and apes. The disparity between those results and the ones obtained from interspecific comparisons in this study illustrates the limitations of phylogenetic methods to detect possible isolated events of positive selection. On the other hand, the patterns of MAOA in the analyses presented in this study demonstrate the robustness of the methods to the presence of isolated positively selected substitutions in a single species.
Globally, the observed differences between members of a single gene family suggest that they perform unique functions, either because of differences in location and expression, response to serotonin levels, or protein interactions. Still, these particular functions are not static but may vary between lineages, changing the constraint under which every member of the family evolves. As seen with HTR1E and HTR1F, even the set of paralogs present in different species may differ, by the acquisition or loss of members of the family (Hahn et al. 2005). Recognizing and understanding these differences is extremely important, particularly for the translation of biological lessons learned from model organisms to humans.
Iwama and Gojobori (2002) provided an extensive comparison of the levels of constraint of neurotransmitter receptors, measured as the dN/dS of genes between human–mouse, human–rat, and mouse–rat. They detected differences between different types of receptors (G-protein–coupled receptors vs. ligand-gated receptors) and certain heterogeneity in constraint within some receptor families. Our approach is similar, but differs in at least 2 important aspects. First, the 2 studies differ on the genes analyzed, as that is a large-scale study of neurotransmitter receptors, and ours focuses on the complete functional network of a single neurotransmitter; we find that receptors evolve, on average, under similar constraints as the rest of genes of the pathway, but some of them show drastic differences in given groups of species. The second main difference lays in the species considered, in their case humans–rodents and in our case closely related species. Iwama and Gojobori (2002) detected a small number of receptors showing significantly relaxed constraint in the long human–rodent branch, whereas we detect a number of genes showing lower constraint in one specific species group (primates or rodents). Genes showing reduced constraint in the 2 studies do not necessarily overlap, and in fact they do not. Actually, the differences between both studies point out the relevance of analyzing closely related species when comparing levels of constraint between genes, because the rate of protein evolution may vary drastically between lineages, even within mammals. This may be particularly important in functional pathways involved in traits that vary between species, like behavior or cognition.
Serotonin represents a very special case, similar to that of other neurotransmitters (like dopamine) with a critical function, on the one side, for basic activities strictly conserved among species and, on the other, for processes (like behavior) that vary widely across species. Accordingly, most genes of the pathway evolve under strong constraint, and only a few of them show some freedom to vary between species. Interestingly, there is absence of correlation in the level of constraint between primates and rodents, probably because different genes have experienced some relaxed constraint in the 2 lineages. We speculate that genes showing the most divergent rates of protein evolution between groups may be good candidates to explain fine serotonin functional differences between them. Likewise, genes showing the fastest evolution in every group may be the best candidates to explain phenotype variability associated with serotonin function among the species of every group. At the very end, comparative analysis is a powerful tool for the understanding of functional constraints and differences (among species, among gene family members, and among domains) within a functional pathway framework; it may thus be a fundamental element in the growing systems biology effort for understanding function in biology.
|
Supplementary Material |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
The supplementary table 1 is available at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).
|
Acknowledgements |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
The authors would like to thank Monica Vallès for technical support, David Comas and Anna Pérez-Lez
un for their assistance in the design and development of the sequencing scheme, and Marta Soldevila for comments and suggestions. The manuscript, and in particular the discussion section, was improved thanks to the suggestions of anonymous reviewers. Some primate samples were supplied by the Barcelona Zoo, under agreement with the Universitat Pompeu Fabra; the authors are especially grateful to Jesús Fernández and Maria Teresa Abelló. This work was supported by MEC (BFU 2004-02002/BMC) and Generalitat de Catalunya (Grup de Recerca Consolidat 2005SGR00608 and Distinció per a la Recerca Universitària). A.M.A. was financially supported by a fellowship from the Generalitat de Catalunya, 2000FI 00686. |
Footnotes |
|---|
1 Present address: Genome Technology Branch, National Human Genome Research Institute, National Institutes of Health, Bethesda, MD.
2 Present address: Laboratoire de Neurophysiologie, Université catholique de Louvain, Brussels, Belgium.
Richard Thomas, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSupplementary MaterialAcknowledgementsReferences |
|---|
Andrés AM, Soldevila M, Navarro A, Kidd KK, Oliva B, Bertranpetit J. Positive selection in MAOA gene is human exclusive: determination of the putative amino acid change selected in the human lineage. Hum Genet (2004) 115:377–386.[CrossRef][Web of Science][Medline]
Bai F, Yin T, Johnstone EM, Su C, Varga G, Little SP, Nelson DL. Molecular cloning and pharmacological characterization of the guinea pig 5-HT1E receptor. Eur J Pharmacol (2004) 484:127–139.[CrossRef][Web of Science][Medline]
Barnes NM, Sharp T. A review of central 5-HT receptors and their function. Neuropharmacology (1999) 38:1083–1152.[CrossRef][Web of Science][Medline]
Bejerano G, Siepel AC, Kent WJ, Haussler D. Computational screening of conserved genomic DNA in search of functional noncoding elements. Nat Methods (2005) 2:535–545.[CrossRef][Web of Science][Medline]
Bielawski JP, Dunn KA, Yang Z. Rates of nucleotide substitution and mammalian nuclear gene evolution. Approximate and maximum-likelihood methods lead to different conclusions. Genetics (2000) 156:1299–1308.
Chimpanzee Sequencing and Analysis Consortium. Initial sequence of the chimpanzee genome and comparison with the human genome. Nature (2005) 437:69–87.[CrossRef][Medline]
Davies PA, Pistis M, Hanna MC, Peters JA, Lambert JJ, Hales TG, Kirkness EF. The 5-HT3B subunit is a major determinant of serotonin-receptor function. Nature (1999) 397:359–363.[CrossRef][Medline]
Dermitzakis ET, Reymond A, Lyle R, et al, (11 co-authors). Numerous potentially functional but non-genic conserved sequences on human chromosome 21. Nature (2002) 420:578–582.[CrossRef][Medline]
Eyre-Walker A. The genomic rate of adaptive evolution. Trends Ecol Evol (2006) 21:569–575.[CrossRef][Medline]
Fraser HB, Hirsh AE, Steinmetz LM, Scharfe C, Feldman MW. Evolutionary rate in the protein interaction network. Science (2002) 296:750–752.
Fraser HB, Wall DP, Hirsh AE. A simple dependence between protein evolution rate and the number of protein-protein interactions. BMC Evol Biol (2003) 3:11.[CrossRef][Medline]
Gilad Y, Rosenberg S, Przeworski M, Lancet D, Skorecki K. Evidence for positive selection and population structure at the human MAO-A gene. Proc Natl Acad Sci USA (2002) 99:862–867.
Guigo R, Dermitzakis ET, Agarwal P, et al, (12 co-authors). Comparison of mouse and human genomes followed by experimental verification yields an estimated 1,019 additional genes. Proc Natl Acad Sci USA (2003) 100:1140–1145.
Hahn MW, De Bie T, Stajich JE, Nguyen C, Cristianini N. Estimating the tempo and mode of gene family evolution from comparative genomic data. Genome Res (2005) 15:1153–1160.
Hahn MW, Kern AD. Comparative genomics of centrality and essentiality in three eukaryotic protein-interaction networks. Mol Biol Evol (2005) 22:803–806.
Hellmann I, Ebersberger I, Ptak SE, Paabo S, Przeworski M. A neutral explanation for the correlation of diversity with recombination rates in humans. Am J Hum Genet (2003) 72:1527–1535.[CrossRef][Web of Science][Medline]
Hirsh AE, Fraser HB. Protein dispensability and rate of evolution. Nature (2001) 411:1046–1049.[CrossRef][Medline]
Ihmels J, Collins SR, Schuldiner M, Krogan NJ, Weissman JS. Backup without redundancy: genetic interactions reveal the cost of duplicate gene loss. Mol Syst Biol (2007) 3:86.[Medline]
Iwama H, Gojobori T. Identification of neurotransmitter receptor genes under significantly relaxed selective constraint by orthologous gene comparisons between humans and rodents. Mol Biol Evol (2002) 19:1891–1901.
Jordan IK, Rogozin IB, Wolf YI, Koonin EV. Essential genes are more evolutionarily conserved than are nonessential genes in bacteria. Genome Res (2002) 12:962–968.
Jordan IK, Wolf YI, Koonin EV. No simple dependence between protein evolution rate and the number of protein-protein interactions: only the most prolific interactors tend to evolve slowly. BMC Evol Biol (2003) 3:1.[CrossRef][Medline]
Kim PM, Lu LJ, Xia Y, Gerstein MB. Relating three-dimensional structures to protein networks provides evolutionary insights. Science (2006) 314:1938–1941.
Lesurtel M, Graf R, Aleil B, Walther DJ, Tian Y, Jochum W, Gachet C, Bader M, Clavien PA. Platelet-derived serotonin mediates liver regeneration. Science (2006) 312:104–107.
Liao BY, Scott NM, Zhang J. Impacts of gene essentiality, expression pattern, and gene compactness on the evolutionary rate of mammalian proteins. Mol Biol Evol (2006) 23:2072–2080.
Lindblad-Toh K, Wade CM, Mikkelsen TS, et al, (236 co-authors). sequence, comparative analysis and haplotype structure of the domestic dog. Nature (2005) 438:803–819.[CrossRef][Medline]
Lucki I. The spectrum of behaviors influenced by serotonin. Biol Psychiatry (1998) 44:151–162.[CrossRef][Web of Science][Medline]
Madsen O, Willemsen D, Ursing BM, Arnason U, de Jong WW. Molecular evolution of the mammalian alpha 2B adrenergic receptor. Mol Biol Evol (2002) 19:2150–2160.
Margulies EH, Blanchette M, Haussler D, Green ED. Identification and characterization of multi-species conserved sequences. Genome Res (2003) 13:2507–2518.
Mouse Genome Sequencing Consortium. Waterston RH, Lindblad-Toh K, Birney E, et al, (223 co-authors). sequencing and comparative analysis of the mouse genome. Nature (2002) 420:520–562.[CrossRef][Medline]
Nebigil CG, Choi DS, Dierich A, Hickel P, Le Meur M, Messaddeq N, Launay JM, Maroteaux L. Serotonin 2B receptor is required for heart development. Proc Natl Acad Sci USA (2000) 97:9508–9513.
Nebigil CG, Etienne N, Schaerlinger B, Hickel P, Launay JM, Maroteaux L. Developmentally regulated serotonin 5-HT2B receptors. Int J Dev Neurosci (2001) 19:365–372.[CrossRef][Web of Science][Medline]
Niesler B, Frank B, Kapeller J, Rappold GA. Cloning, physical mapping and expression analysis of the human 5-HT3 serotonin receptor-like genes HTR3C, HTR3D and HTR3E. Gene (2003) 310:101–111.[CrossRef][Web of Science][Medline]
Niesler B, Walstab J, Combrink S, Moeller D, Kapeller J, Rietdorf J, Boenisch H, Goethert M, Rappold G, Bruess M. 5-HT3E. Mol Pharmacol (2007) 72:8–17.
Pal C, Papp B, Hurst LD. Highly expressed genes in yeast evolve slowly. Genetics (2001) 158:927–931.
Pal C, Papp B, Hurst LD. Genomic function: rate of evolution and gene dispensability. Nature (2003) 421:496–497.[Medline]
Parra G, Agarwal P, Abril JF, Wiehe T, Fickett JW, Guigo R. Comparative gene prediction in human and mouse. Genome Res (2003) 13:108–117.
Pruitt KD, Tatusova T, Maglott DR. NCBI reference sequence (RefSeq): a curated non-redundant sequence database of genomes, transcripts and proteins. Nucleic Acids Res (2007) 35. Database issue: D26–D31.
Puente XS, Velasco G, Gutierrez-Fernandez A, Bertranpetit J, King MC, Lopez-Otin C. Comparative analysis of cancer genes in the human and chimpanzee genomes. BMC Genomics (2006) 7:15.[CrossRef][Medline]
Shimron-Abarbanell D, Nothen MM, Erdmann J, Propping P. Lack of genetically determined structural variants of the human serotonin-1E (5-HT1E) receptor protein points to its evolutionary conservation. Brain Res Mol Brain Res (1995) 29:387–390.[Medline]
Siepel A, Bejerano G, Pedersen JS, et al, (16 co-authors). Evolutionarily conserved elements in vertebrate, insect, worm, and yeast genomes. Genome Res (2005) 15:1034–1050.
Smith NG, Hurst LD. The effect of tandem substitutions on the correlation between synonymous and nonsynonymous rates in rodents. Genetics (1999) 153:1395–1402.
Taudien S, Ebersberger I, Glockner G, Platzer M. Should the draft chimpanzee sequence be finished? Trends Genet (2006) 22:122–125.[CrossRef][Web of Science][Medline]
Wilson AC, Carlson SS, White TJ. Biochemical evolution. Annu Rev Biochem (1977) 46:573–639.[CrossRef][Web of Science][Medline]
Yang Z. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci (1997) 13:555–556.
Yang J, Gu Z, Li WH. Rate of protein evolution versus fitness effect of gene deletion. Mol Biol Evol (2003) 20:772–774.
Yang Z, Swanson WJ. Codon-substitution models to detect adaptive evolution that account for heterogeneous selective pressures among site classes. Mol Biol Evol (2002) 19:49–57.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||