Rapid sequence divergence rates in the 5 prime regulatory regions of young Drosophila melanogaster duplicate gene pairs

EVOLUTIONARY GENETICS

RESEARCH ARTICLE

Rapid sequence divergence rates in the 5 prime regulatory regions of young Drosophila melanogaster duplicate gene pairs

Michael H. Kohn

Ecology and Evolutionary Biology, Rice University, Houston, Texas, United States of America

^{Send correspondence to}

ABSTRACT

While it remains a matter of some debate, rapid sequence evolution of the coding sequences of duplicate genes is characteristic for early phases past duplication, but long established duplicates generally evolve under constraint, much like the rest of the coding genome. As for coding sequences, it may be possible to infer evolutionary rate, selection, and constraint via contrasts between duplicate gene divergence in the 5 prime regions and in the corresponding synonymous site divergence in the coding regions. Finding elevated rates for the 5 prime regions of duplicated genes, in addition to the coding regions, would enable statements regarding the early processes of duplicate gene evolution. Here, 1 kb of each of the 5 prime regulatory regions of Drosophila melanogaster duplicate gene pairs were mapped onto one another to isolate shared sequence blocks. Genetic distances within shared sequence blocks (d_5’) were found to increase as a function of synonymous (d_S), and to a lesser extend, amino-acid (d_A) site divergence between duplicates. The rate d_5’/d_S was found to rapidly decay from values > 1 in young duplicate pairs (d_S < 0.3) to 0.28 or less in older duplicates (d_S > 0.8). Such rapid rates of 5 prime evolution exceeding 1 (~neutral) predominantly were found to occur in duplicate pairs with low amino-acid site divergence and that tended to be co-regulated when assayed on microarrays. Conceivably, functional redundancy and relaxation of selective constraint facilitates subsequent positive selection on the 5 prime regions of young duplicate genes. This might promote the evolution of new functions (neofunctionalization) or division of labor among duplicate genes (subfunctionalization). In contrast, similar to the vast portion of the non-coding genome, the 5 prime regions of long-established gene duplicates appear to evolve under selective constraint, indicating that these long-established gene duplicates have assumed critical functions.

Key words: gene duplication, gene expression, selection, promoter evolution.

Introduction

The alignment of orthologous sequences sampled from two or more related species can reveal evolutionarily conserved sequence blocks, an approach referred to as ‘phylogenetic footprinting’ (e.g. Fickett and Wasserman, 2000). The approach relies on the assumption that sequence blocks that contain functionally important motifs evolve under functional constraint (purifying selection), and thus, remain similar in their sequence over long periods of time (e.g. Koop, 1995). In contrast, alignments of non-functional sequences that evolve free of such constraint usually are less clear or not significant (Bergman and Kreitman, 2001). Overall, the footprint of varying degrees of selective constraint along alignments of orthologous, or homologous, sequences is manifest as a mosaic pattern of aligned and non-aligned sequence blocks (Bergman and Kreitman, 2001; Shabalina et al., 2001; Bergman et al., 2002; Castresana, 2002; Webb et al., 2002). In non-coding sequences, such as enhancers and promoters, sequence blocks conserved between orthologs may be enriched for potential transcription factor binding sites (Fickett and Wasserman, 2000; Berman et al., 2002). As more whole genome sequences begin to accumulate in the databases, comparative genomic approaches have become widely applied to aid with the annotation and evolutionary study of non-coding DNA (de Meaux, 2006; Haberer et al., 2006; Li and Stephan, 2006; Hahn, 2007; Thomas et al., 2007).

A wide range of evolutionary divergence times is captured within a single genome through the duplication of genes and their subsequent divergence (e.g. Ohno, 1970; Lynch and Conery, 2000; Conery and Lynch, 2001). Presumably, the extent to which gene duplicates, or paralagous genes, occur in the genome reflects their potential to provide a source for biological adaptation and diversification (e.g. Zhang et al., 1998; Lynch and Conery, 2000; Conant and Wagner, 2002; Gu et al., 2002a; Hughes, 2002; Zhang, 2003). In recognition of the pivotal role gene duplication may play in evolution the mechanisms driving their origins and preservation have been a vibrant field of study that is experiencing a renaissance owing to the ever-growing number of genome sequencing projects (e.g. Ohno, 1970; Ohta, 1987; Clark, 1994; Hughes, 1994; Ohta, 1994; Walsh, 1995; King, 1998; Force et al., 1999; Lynch and Force, 2000; Wagner, 2001; Hughes, 2002; Wagner, 2002a; Zhang, 2003; Taylor and Raes, 2004).

Whereas the origin and subsequent silencing of duplicate genes both appear to be frequent events, the evolutionary trajectories conducive to duplicate gene preservation may be restrictive (Force et al., 1999; Lynch and Force, 2000). Importantly, the complement of functional duplicate genes that is sampled by genome sequencing projects and that can be studied for their molecular evolution should be comprised predominantly of those that have passed the ‘selective sieve’. In other words, gene duplications detrimental to fitness have been removed by purifying selection and gene duplications free of selective constraint may have undergone mutations that rendered them non-functional pseudogenes whose evolution is governed by drift. Functional diversification of duplicates leading to the evolution of novel functions (neo-functionalization), or the partitioning of labor between them (sub-functionalization) could provide avenues for escape from non-functionalization and loss, because purifying selection would remove detrimental mutations from the functional duplicate genes once these have become indispensable (Ohta, 1988; Basten and Ohta, 1992; Hughes, 1994; Walsh, 1995; Force et al., 1999; Lynch and Force, 2000; Wagner, 2002a,b).

Mutations in the 5 prime cis-regulatory regions of gene duplicates may promote functional diversification of duplicate genes (Wagner, 2000; Gu et al., 2002b; Makova and Li, 2003; Papp et al., 2003). To examine this possibility the 5 prime regulatory regions of gene duplicates could be searched for the footprint regulatory diversification, be it through positive selection or the loss of constraint (i.e. neutral processes), may have left. One such approach would be to compare the rate of divergence in the 5 prime regions relative to that at synonymous sites (Bird et al., 2006; Eyre-Walker, 2006; Hahn 2007), as long as it is assumed synonymous sites follow neutral dynamics (see Akashi, 1995).

Here the evolution of 5 prime regulatory sequences of duplicate gene pairs in the D. melanogaster genome was studied. Specifically, (i) 1 kb of each of the 5 prime regions of the two members of a duplicate gene pairs identified previously (Lynch and Conery, 2000; Conery and Lynch 2001) were aligned. It was assumed that blocks of aligned sequence indicate regions of homology preserved owing to their recent divergence and/or by purifying selection. In analogy to phylogenetic footprinting this approach has been dubbed ‘intragenomic footprinting’ (Haberer et al. 2004; Haberer et al. 2006). (ii) Divergence of the 5 prime regions of duplicate gene pairs (d_5’) was expressed relative to divergence at synonymous sites (d_S) and amino-acid replacement sites (d_A) in these gene pairs. This is analogous to studies considering rates of coding sequence evolution of duplicate genes (e.g. Ohta, 1994; Lynch and Conery, 2000; Barrier et al., 2001; Conery and Lynch, 2001; Thornton and Long, 2002; Kondrashov, 2005; Kondrashov and Kondrashov, 2006). (iii) Gene expression data from microarray experiments was compiled and related to Drosophila duplicate gene divergence (c.f. Wagner, 2000; Gu et al., 2002b; Makova and Li, 2003; Castillo-Davis et al., 2004; Haberer et al., 2004; Casneuf et al., 2006; Wang et al., 2006; Tirosh and Barkai, 2007).

Methods

Collection and analysis of sequence data: The identification numbers for a set of 456 D. melanogaster duplicate gene pairs (Lynch and Conery, 2000, Conery and Lynch, 2001) were retrieved from (http://www.csi.uoregon.edu/ projects/genetics/duplications/D.melanogaster.txt) and 1 kilobase (kb) of the nucleotide sequences annotated as the upstream 5 prime flanking regions and 5 prime untranslated regions (5’ UTR) were retrieved for each gene via the Berkley Drosophila Genome Project (BDGP, Release 2) (http://www.fruitfly.org). Estimates of synonymous site divergence (d_S) and amino acid replacement site divergence (d_A) for the protein coding sequences of each duplicate gene pair were adopted from Lynch and Conery (2000) and Conery and Lynch (2001), who deduced them using PAML (Yang, 1997).

Duplicate gene pairs were grouped into divergence bins: d_S < 0.1 (N = 19), 0.1 < d_S < 0.25 (N = 20), 0.25 < d_S < 0.5 (N = 27), 0.5 < d_S < 0.75 (N = 15), 0.75 < d_S < 1.0 (N = 14), 1.0 < d_S < 1.25 (N = 14), 1.25 < d_S < 1.5 (N = 17) and d_S > 1.5 (N = 274). Young duplicated genes (e.g. d_S < 1.0) were comparatively scarce (N = 95 or ~22.5%) in this dataset, and in the Drosophila genome as a whole (Lynch and Conery, 2000; Conery and Lynch, 2001; Conant and Wagner, 2002; Gu et al., 2002b; Thornton and Long, 2002). Similarly, d_A values were grouped into bins: d_A < 0.1 (N = 76), 0.1 < d_A < 0.2 (N = 97), 0.2 < d_A < 0.3 (N = 65), 0.3 < d_A < 0.4 (57), 0.4 < d_A < 0.5 (N = 43) and d_A > 0.5 (N = 86). It was assumed that gene conversion has not affected the estimation of genetic distances between gene duplicates.

A non-redundant set of 5 prime regions of D. melanogaster genes (set of single-copy genes) retrieved from http://www.fruitfly.org/seq_tools/datasets/ Drosophila/promoter/ (Ohler et al., 2002) was analyzed for comparison. The 5 prime regions of the duplicate genes and of the set of single-copy genes had similar GC contents (40.3 and 41.2%). Both datasets were screened for the presence of sequence elements known to occur in the Drosophila genome using the RepeatMasker software using the settings for insect genomes (http://repeatmasker.genome.washington.edu/cgi-bin/RepeatMasker) (Thompson et al., 1994), masked with "N", and excluded prior to alignment.

As done by Bergman and Kreitman (2001) the alignments of the 5 prime regions of each duplicate gene pair were done using the Dialign software (setting T = 1) (Morgenstern, 1999). For comparison, 5,000 alignments of randomly paired 5 prime regions drawn from the set of single-copy genes were done. Even if the Dialign alignment procedure may have its biases, as most procedures do, the comparison between the alignments of the 5 prime regions of duplicate genes and the alignments of randomly paired single copy genes should enable qualitative and quantitative statements regarding the significance of the sequence similarities observed in the 5 prime regions of the duplicate genes. Regions in the 5 prime regions that were aligned were converted as capital letters in the Fasta-formatted Dialign output. Aligned regions at least 10 nucleotides long were extracted and concatenated. The percentages of nucleotides that fell within such aligned regions was noted and referred to as 5 prime similarities (Table S1). Subsequently, for each alignment the number of perfectly matched base pairs within each aligned region was computed, leading to an estimate of sequence similarity within them (5 prime block similarity, Table S1). 5 prime block similarity values were transformed into a genetic distance (d_5’) using the HKY method (Hasegawa et al., 1985) as implemented in PAML.

Distance estimation at high divergence levels can be associated with errors. Therefore, no attempt was made to resolve divergence times of d_S > 1.5. The estimation of very low synonymous divergence levels also can be associated with errors, in particular when the examined genes are short in length. To address this issue all 95 duplicate gene pairs with d_S up to 1 were re-analyzed to obtain estimates of d_5’ and d_5’/ d_S that should be less likely to be affected by stochastic sampling. Specifically, first, sequences were extracted and aligned using the Dialign software. Second, Kimura’s 2 parameter method was used to estimate d_5’, d_S, and d_A for each gene separately (Figure S1 and Table S2). Third, divergence times d_5’, d_S, and d_A were deduced from the concatenated sequences, the latter allows to obtain a weighted (by gene length) average of divergence times that should be less prone to stochastic sampling. For the concatenation process duplicate genes were grouped into the divergence bins d_S < 0.1 (N = 17), 0.1 < d_S < 0.2 (N = 25), 0.2 < d_S < 0.3 (N = 16), 0.3 < d_S < 0.4 (N = 13), 0.4 < d_S < 0.5 (N = 2), 0.5 < d_S < 0.6 (N = 5), 0.6 < d_S < 0.7 (N = 4), 0.7 < d_S < 0.8 (N = 3), 0.8 < d_S < 0.9 (N = 3), 0.9 < d_S < 1.0 (N = 4).

Analysis of co-regulation of gene duplicates: Gene expression data from 267 Affymetrix GeneChips representing six independent investigations on D. melanogaster were retrieved from http://jbiol.com/content/supplementary/1475-4924-1-5-S1.txt (Spellman and Rubin, 2002). These dealt with embryo development, aging, DNA damage, immune response, and DDT resistance in adult flies and embryos subjected to 88 distinct conditions or experimental manipulations. For the description of the gene expression data and their analysis see Spellman and Rubin (2002). Here, Pearson’s correlation coefficient (R) was computed across the expression levels provided by Spellman and Rubin (2002) to quantify the degree of co-regulation of duplicate genes. R was transformed using the expression ln((R+1)/R-1)) (Gu et al., 2002b; Gu and Su 2007) and referred to as ln(R). The transformation of R into ln(R) enabled the analysis of sequence divergence and gene expression using linear regression (Gu et al., 2002b). The expectation was that co-regulated duplicate genes would display high ln(R)-values when calculated over a series of conditions, because more similar regulatory regions should mediate more similar responses. For comparison, sampling with replacement from the expression profiles of the duplicate genes was done to yield 5,000 ln(R)-values computed between 10,000 randomly paired genes (Figure S2).

Results

Levels of 5 prime sequence similarities between duplicate genes: Alignments of the 5 prime non-coding regions of duplicate gene pairs resulted in a mosaic of aligned and non-aligned stretches of sequence. Only a percentage of sites in the 5 prime regions of duplicate genes fell within aligned stretches of sequence. Specifically, 5 prime similarities, a number that summarizes the percentage of nucleotide sites that fell within aligned stretches of sequence, were between 2 and 60% (median, 20.0%, mean 21.6%, 95% CI of mean, 20.4-22.1%) (Figure S3). 5 prime similarities were weakly correlated with synonymous and amino acid replacement site divergence between duplicate gene pairs (ANOVA, F_ratio 13.2, R² = 0.06, p < 0.001 and F_ratio 6.7, R² = 0.03, p = 0.0014, respectively).

The distribution of 5 prime block similarity values derived from the alignments of duplicate genes was compared to the distribution derived from 5,000 alignments of randomly paired genes (Figure 1). The expectation was that the 5 prime regions of randomly paired single-copy genes should reflect the degree to which DiAlign generated alignments between unrelated 5 prime regions of genes. For ~26% of random alignments no regions of any similarity were found that were 10 bp or longer. For a lower percentage 38/456 (~9%) of the duplicate gene pair dataset DiAlign could not identify such sequence blocks. These were excluded because they cannot be analyzed within a framework that considers per nucleotide site divergence rates. Their omission should have introduced a bias towards higher average levels of 5 prime block similarities among duplicate genes.

Levels of sequence similarity between the 5 prime regions of Drosophila duplicate genes exceeded those that were obtained from random alignments (Figure 1). Specifically, whereas the distribution of 5 prime block similarity scores that was based on alignments of randomly paired genes had a mean 5 prime block similarity of 0.553 (95% CI of mean: 0.552-0.555), mean 5 prime block similarity in alignments of the 5 prime regions of duplicate genes was 0.723 (95% CI: 0.717-0.729). The distribution of 5 prime block similarities between randomly paired single-copy genes was normally distributed, and was used to deduce the probability P to observe 5 prime block similarity values that were observed in the duplicate gene pairs after correction for multiple testing with the Bonferoni method. Forty-four duplicate gene pairs had average 5 prime block similarities that were not significant (< 0.67, n.s.), but 373 had average block similarities that exceeded random levels (= 0.67, p = 0.01). In total, average 5 prime block similarity between the 5 prime regions of duplicate genes was between 0.6 and 0.7 for 158 (34.6%) duplicate pairs, 0.7-0.8 for 216 (47.3%) pairs, 0.8-0.9 for 32 (7.0%) pairs, and 0.9-1.0 (2.4%) for 11 pairs. Thus, while DiAlign tended to find short sequence blocks even between 5 prime regions of random pairs of single copy genes, sequence similarity in the set of single-copy genes generally remained below those deduced from alignments of duplicated genes. For rate calculations below it was assumed that sequence similarities among the 5 prime regions of duplicated genes reflect sequence homology.

A contrast between d_5’ and d_S should enable inferences concerning the role of drift and selection on the evolution of the 5 prime regions of duplicated genes. Here it was found that d_5’ significantly increased with d_S (Figure 2A, p < 0.0001, F_Ratio = 79.5, R² = 29%, ANOVA). This was less pronounced when d_5’ was related to d_A (Figure 2B, p < 0.0001, F_Ratio = 24.6, R² = 11%, ANOVA). In addition, a decay of d_5’/d_S as a function of d_S (Figure 2C, p < 0.001, F_Ratio = 109.8, R² = 36, ANOVA), and to a less systematic degree d_A (not shown, p < 0.001, F_Ratio = 38.9, R² = 16%, ANOVA), was observed. Values for d_5’/d_S larger than 1 were observed for nearly all, ~50%, and ~10% of duplicate pairs with d_S < 0.1, 0.1-0.25, and > 0.25-0.5, respectively. Duplicate gene pairs with d_S < 0.25 had mean and median d_5’/d_S values exceeding 1. Thus, rapid rates of 5 prime block evolution close to 1, or exceeding 1, predominantly occurred in young duplicated genes, and these high rates were suggestive of relaxed constraint and/or positive selection. In contrast, the rate of 5 prime evolution observed slowed relative to that at synonymous sites, a pattern consistent with purifying selection and functional constraint. However, other homogenizing forces, such as gene conversion, should be considered as well.

To obtain d_5’/d_S rates less likely affected by stochastic sampling of sites from individual gene pairs with low d_S, the sequences of duplicate gene pairs with d_S < 1 were concatenated in bins (c.f. Table 1). Bins with a weighted average of d_S < 0.3 displayed d_5’/d_S ratios > 1 (Table 1). The corresponding average of d_A was 0.173 (Table 1). Thus, high rates of 5 prime sequence block evolution between young duplicate genes were not caused by the inclusion of a few genes with particularly high d_5’/d_S. The d_5’/d_S ratios of more divergent duplicate gene pairs remained smaller than 1 in the concatenated data sets. The use of the concatenated sequences should provide conservative, i.e. lower, estimates for the rate d_5’/d_S. This was most notable in the divergence bin d_S < 0.1, where the average rate d_5’/d_S computed as the mean over individual duplicate gene pairs yielded a value close to 7 (c.f. Table S2 for divergence estimates and rates derived from individual duplicate gene pairs). In contrast, when estimated from the concatenated sequences, a d_5’/d_S value of 2.75 was obtained. Similarly, the rates d_5’/d_S obtained for the remaining divergence bins were lower than the corresponding d_5’/d_S values calculated as the mean over individual duplicate gene pairs. Taken together, the decay of d_5’ and d_5’/d_S as function of d_S was suggestive of a phase of accelerated evolution in the 5 prime regions of young duplicated genes, i.e. those with d_S < 25%-30%. This was also true, qualitatively, when each individual duplicate gene pair was examined (Table S2).

Masked sequences in the 5 prime regions of duplicate genes: Besides nucleotide substitution, a range of possible other mutational events following gene duplication may alter the functionality of 5 prime regulatory sequences. These involve the insertion or deletion of various types of sequence elements (retro-elements and low-complexity/repeat sequences), or the insertion of the duplicate gene copies into regions that already were densely occupied by such sequence elements. As a proxy for the frequency of such events, the percentage of 5 prime sequence occupied by sequence elements that was recognized and masked by the RepeatMasker software was tabulated (Table 2). On average, only 4% of the total sequence data covering the 5 prime regions of duplicate genes were masked. A similar percentage (3.8%) was masked in the single-copy 5 prime regions, indicating that the majority of the duplicated 5 prime regions were not atypical with respect to such sequence elements when compared to 5 prime regions of single copy genes.

For about 10% of duplicate pairs masked sequences occupied as much as 18%-77% of the 5 prime region, indicating that larger-scale insertions or deletions of elements could affect the function of the 5 prime region. Simple repeats and low complexity-type sequences occupied the largest percentage of the masked sequence (Table 2). There was a trend towards higher percentages of masked sequence (median > 10%) in comparatively young (d_S~25% or less) duplicate gene pairs when compared to the usually less than 5% masked sequence in duplicate gene pairs separated by d_S values > 25%. Perhaps, some of these repeats or low-complexity-type sequences are deleterious or form the basis for the evolution of motifs not recognized by RepeatMasker. However, the percentage of 5 prime sequence masked by RepeatMasker was not significantly related to d_S (not shown).

Evolution and expression of 5 prime sequences of duplicate genes: In yeast, young duplicate gene pairs tend to be more similar in their expression than are old duplicate pairs (Gu et al., 2002a; Papp et al., 2003; but see Wagner, 2000). Correlation of gene expression between duplicated genes might be a useful proxy for functional equivalence (Gu et al., 2002c). Here, an analysis of the co-regulation of gene duplicates, as inferred from ln(R), showed that about 40% of D. melanogaster duplicate pairs were above the 99% range of randomly generated ln(R)-values (-0.79 and +0.99) and 10% were below that (c.f. Figure S1). Thus, half of the examined duplicate gene pairs conformed to random expectations. Co-regulation of duplicate genes may be 4 times more common than extreme divergence in regulation.

The correlation of expression of duplicate genes, expressed as ln(R), was found to decay as duplicate genes diverged at synonymous sites and at amino-acid replacement sites, but the relationships were weak. Specifically, a reduction of ln(R) between d_S < 0.25% (mean ln(R) = 1.8 ± 0.2, median 1.49) and d_S > 0.25 (mean ln(R) = 0.79 ± 0.1, median 0.64) was observed, but no further systematic trend was observed at higher divergence levels. Moreover, at high d_S the median and mean of ln(R) remained compatible with random expectations (Figure S1). However, expression was assayed over the whole fly, larvae, and embryos (Spellman and Rubin, 2002), such that only limited power would be expected to detect diversification of expression between gene duplicates e.g. at the level of tissues (c.f. Makova and Li, 2003). Overall, the Drosophila data fell in between the previously observed strong correlation between ln(R) and d_S (Gu et al., 2002a) and a much weaker such relationship (Wagner, 2000), both observed in yeast. However, in this study emphasis was placed on the expression divergence after only 25 percent synonymous site divergence was observed, i.e. in young pairs of duplicated genes.

Duplicate gene pairs that diverged in their expression patterns displayed rapid rates of 5 prime sequence evolution. Specifically, gene pairs with d_5’/d_S > 1 displayed higher levels of correlation in gene expression than duplicate pairs with d_5’/d_S < 1 (Figure 3C, median ln(R) = 1.80, mean 1.64 ± 0.23 vs. median ln(R) = 0.69, mean 0.83 ± 0.07, respectively). More than 60% of the duplicate pairs with d_5’/d_S > 1 had ln(R) values that fell outside the random distribution of ln(R) values (Figure 3C). In contrast, none had ln(R) values that were below random levels.

5 prime block similarities between duplicate genes were not a good indicator for the co-regulation of expression. When d_5’ and ln(R) were grouped into those that were compatible with random expectations and those that were not, then one would have expected that random values of 5 prime block similarity values predominantly coincide with random ln(R) values (or vice versa). This was not the case. Only duplicate pairs with d_5’ higher than 0.8 differed from the remaining duplicate pairs in their correlation of expression (ln(R) > 1.4 vs. ln(R) < 0.9). However, for duplicate gene pairs with d_5’ exceeding 0.8 ln(R) values as low as -0.61 (c.f. Figure S1) were not uncommon, e.g. they were found in ~10% of the cases. Conversely, ln(R) values as high as 2.7 were found in ~10% of the gene pairs with d_5’ less than 0.8. Thus, while there was weak indication that d_5’ and ln(R) were dependent variables, the statistical resolution to document such a relationship was either limited or obscured by biological factors or the functional regulatory elements are located in regions that could not be aligned, and thus, d_5’ more closely approximates non-functional rates of evolution.

Discussion

The principle onto which ‘phylogenetic footprinting’ is based is that conservation between orthologous coding sequences reflects functional constraint (Fickett and Wasserman, 2000). Conservation between orthologous non-coding sequences also has been viewed as evidence for functional constraint (Tautz and Nigro, 1998; Bergman and Kreitman, 2001; Wasserman et al., 2000; Bergman et al., 2002; Webb et al., 2002; Dermitzakis et al., 2003; Haberer et al., 2006, Thomas et al., 2007). The possibility that negative selection on the 5 prime regions of genes may indeed be prevalent has been raised (Tautz and Nigro, 1998; Stone and Wray, 2001; Dermitzakis et al., 2003; Hahn et al., 2003; Kohn et al., 2004; Andolfatto, 2005; Eyre-Walker, 2006; Hahn, 2007). More rapid rates of substitution take place in regions free of functional constraint (Andolfatto, 2005; Shapiro et al., 2007). In the case of non-coding sequences rapid rates may be driven by nucleotide substitution, but also by mutational events (insertions, deletions, replication slippage) whose dynamics are not well understood (e.g. Comeron, 2001; Eyre-Walker, 2006). The dynamics of selective constraint on the 5 prime regions of D. melanogaster duplicate genes over time was manifest in the rate d_5’/d_S (Figure 2A, and Table 1). Initially, for duplicate pairs separated by d_S < 0.25-0.3 d_5’/d_S was larger than one. If it is assumed that d_S represents neutral divergence (Akashi, 1999), then d_5’/d_S = 1 indicate selective neutrality and d_5’/d_S > 1 positive selection. The majority of genes used here had low levels of codon usage bias (ENC 35 or more, Gu et al., 2002b) and only 2% of genes had ENC levels between 32 and 35, suggesting that synonymous sites in this dataset should conform to neutrality reasonably well. Thus, as has been assumed by others here it was assumed that synonymous site divergence is useful measure for the relative ages of gene duplicate pairs (Kim and Yi, 2006; Wang et al., 2006; Gu and Su, 2007; Guan et al., 2007; Ha et al., 2007; Jiang et al., 2007; Johnston et al., 2007; Roth et al., 2007).

Duplicate gene pairs separated by d_S > 0.25-0.3 displayed lower d_5’ than d_S values (Figure 2, Table 1), i.e. d_A/d_S 1 (Table 2). Thus, levels of constraint on the 5 prime regions of duplicate genes were found to be comparable to those at amino-acid replacement sites once substantial coding sequence divergence levels have been reached. In contrast, young duplicate pairs may experience reduced levels of constraint on their amino-acid changes (Figure 2B and Table 1) (Clark, 1994; Lynch and Conery, 2000; Kondrashov et al., 2002). The degree to which the 5 prime regions of ancient duplicate pairs, which are fully saturated at synonymous sites, still can be aligned is remarkable. In the absence of constraint neutral sites should be entirely diverged after a few million years, or at d_S ~ 1.

However, the constraint imposed on 5 prime regions that can directly be attributed to transcription control may be less than intuition would suggest. During a previous study this conclusion was based on the similar levels of sequence similarity that can be detected from alignments of 5 prime regions of orthologous Drosophila genes as well as alignments of introns of orthologous genes (Bergman and Kreitman, 2001). Here, the weak relationship between 5 prime block similarities between duplicates and their weak correlations with expression (Figure 3) indicated that the constraint detected here at best was in part a direct result of transcription requirements. This could reflect a limited resolution of this study. However, biological implications of this finding are plausible, as much remains to be learned about regulatory non-coding sequences (e.g. Comeron, 2001; Fessele et al., 2002; Ludwig, 2002; Hahn et al., 2003; Bird et al., 2006). Additional forces, such as gene conversion tracts spanning regions that are not involved in regulation can maintain sequence similarity in the 5 prime regions of duplicate genes (Ohta, 1985; Basten and Ohta, 1992; King, 1998; Maside et al., 2003).

It is noteworthy that various other types of sequence elements (retro-elements and low-complexity/repeat sequences) located in the 5 prime regions of D. melanogaster duplicates became increasingly rare as duplicate genes diverged. Even though this was not further investigated here, the pattern pointed to their reduction over time. In human, repeat sequences occasionally have been linked to deleterious effects when located in the regulatory region of genes (Usdin and Grabczyk, 2000). Many types of low-complexity/repeat sequences may act as spurious transcription factor binding sites that are slightly deleterious (Stone and Wray, 2001).

The important assertion made in this report refers to the accelerated evolution in the 5 prime regions of young duplicates. The interpretation of the d_5’/d_S rates relies on the premise that d_5’ and d_S of duplicate genes may be directly compared to one another, which may be questioned on a number of grounds. Most importantly, while it is quite certain that homologous sites in the coding regions of duplicate genes were compared, the possibility remains that non-homologous sites in the 5 prime regions of duplicate genes were compared. However, both the alignment and divergence estimation generally should be less problematic in young duplicate pairs compared to the alignment of old duplicate gene pairs. In fact, accelerated evolution in the 5 prime regions of young duplicate gene pairs was deduced from generally longer and more reliable alignments than those alignments of ancient duplicate pairs from which constraint was inferred.

The rapid divergence in the 5 prime regions of young D. melanogaster duplicates was found to coincide with their divergence at amino-acid replacement sites and low correlations of expression, as was expressed as ln(R) (Figure 3). To some degree this may reflect the functional diversification of duplicates with time. Data from yeast indicate that d_A and correlation of gene expression reflect functional equivalency of duplicates in vivo (Gu et al., 2002b; Gu et al., 2002c). In humans (Makova and Li, 2003) and in yeast duplicate gene expression patterns diverge rapidly (Gu et al., 2002a; Papp et al., 2003 but see Wagner, 2000). Data from orthologous genes now available from the newly released multiple Drosophila genome projects could be used next to assess whether the 5 prime regions of one or both copies of duplicate genes display accelerated evolution. This could help distinguishing between neo-functionalization (one copy accelerated) and sub-functionalization models (both copies accelerated), and to polarize the direction of change.

The possibility that advantageous mutations occur and positive selection acts on duplicate gene promoters has been raised before (Papp et al., 2003; Seoighe et al., 2003; Castillo-Davis et al., 2004; Huminiecki and Wolfe, 2004; Jordan et al., 2004; Lynch and Katju, 2004; He and Zhang, 2005; Crow and Wagner, 2006; Kim and Yi, 2006; Kondrashov and Kondrashov, 2006; Gu and Su, 2007; Jiang et al., 2007; Johnston et al., 2007). Here a pattern consistent with selection in Drosophila was observed. Complex selection patterns (Ohta, 1988; Basten and Ohta, 1992; Force et al., 1999; Ludwig et al., 2000; Lynch and Force, 2000; Tautz, 2000; Ludwig, 2002; Wagner, 2002a) and the diffuse link between sequence context and regulatory function (Carroll et al., 2001; Fessele et al., 2002) pose considerable challenges to the conclusive documentation of selection. However, the results presented here suggest that rapid evolution in the 5 prime regulatory regions of young duplicate genes, which tend to be rather equivalent in their function, appears to be a part of the footprint left by functional diversification. That positive selection is driving rates d_5’/d_S in excess of 1 is conceivable when assuming that single nucleotide substitutions within 5 prime blocks are their major mode of change.

In sum, the 5 prime regulatory regions of very young Drosophila duplicate gene pairs diverge at rates faster than at synonymous sites. If the latter are viewed as a proxy for neutral divergence rates, then we can infer that the evolution of 5 prime sequences in young duplicate genes is driven by positive selection. Conceivably the process is facilitated by initial relaxation of selective constraint due to the overlapping functions of young duplicate pairs. Low levels of nonsynonymous site divergence and an analysis of Drosophila duplicate gene expression data presented supported functional redundancy of young gene duplicates. In contrast, as duplicate genes diverge over time in their coding sequences and expression patterns the 5 prime regulatory regions of them were found to display divergence rates as low as those at amino-acid replacement sites, suggesting that they evolve under selective constraint. An important next step in the analysis of duplicated gene evolution in Drosophila would be concerned with the symmetric, or asymmetric divergence of duplicate genes, which appears to be commonly seen in other organisms (Casneuf et al., 2006; Chung et al., 2006; Kim and Yi,2006; Tirosh and Barkai, 2007).

Acknowledgments

I thank Chung-I Wu for stimulating discussions and financial support, and Kevin Thornton, Casey Bergman, Matthias Gerberding, Sebastian Zöllner, Bettina Harr, and Jian Lu and two anonymous reviewers for important feedback.

Internet Resources

Received: October 23, 2007; Accepted: March 13, 2008.

Associate Editor: Louis Bernard Klaczko

Supplementary Material

The following online material is available for this article:

Table S1
- Click to enlarge

Table S2
- Click to enlarge

This material is available as part of the online article from http://www.scielo.br/gmb.

• Akashi H (1995) Inferring weak selection from patterns of polymorphism and divergence at silent sites in Drosophila DNA. Genetics 139:1067-1076.
• Akashi H (1999) Within- and between-species DNA sequence variation and the footprint of natural selection. Gene 238:39-51.
• Andolfatto P (2005) Adaptive evolution of non-coding DNA in Drosophila Nature 437:1149-1152.
• Barrier M, Robichaux RH and Purugganan MD (2001) Accelerated regulatory gene evolution in an adaptive radiation. Proc Natl Acad Sci USA 98:10208-10213.
• Basten CJ and Ohta T (1992) Simulation study of a multigene family, with special reference to the evolution of compensatory advantageous mutations. Genetics 132:247-252.
• Bergman CM and Kreitman M (2001) Analysis of conserved noncoding DNA in Drosophila reveals similar constraints in intergenic and intronic sequences. Genome Res 11:1335-1345.
• Bergman CM, Pfeiffer BD, Rincon-Limas DE, Hoskins RA, Gnirke A, Mungall CJ, Wang AM, Kronmiller B, Pacleb J, Park S, et al. (2002) Assessing the impact of comparative genomic sequence data on the functional annotation of the Drosophila genome. Genome Biol 3:0086.1-0086.20.
• Berman BP, Nibu Y, Pfeiffer BD, Tomancak P, Celniker SE, Levine M, Rubin GM and Eisen, MB (2002) Exploiting transcription factor binding site clustering to identify cis-regulatory modules involved in pattern formation in the Drosophila genome. Proc Natl Acad Sci USA 99:757-762.
• Bird CP, Stranger BE and Dermitzakis ET (2006) Functional variation and evolution of non-coding DNA. Curr Opin Genet Dev 16:559-564.
• Carroll SB, Grenier JK and Weatherbee SD (2001) From DNA to Diversity: Molecular Genetics and the Evolution of Animal Design. Blackwell Science, Malden, 192 pp.
• Casneuf T, De Bodt S, Raes J, Maere S and Van de Peer Y (2006) Nonrandom divergence of gene expression following gene and genome duplications in the flowering plant Arabidopsis thaliana Genome Biol 7:R13.
• Castillo-Davis CI, Hartl DL and Achaz G (2004) cis-Regulatory and protein evolution in orthologous and duplicate genes. Genome Res 14:1530-1536.
• Castresana J (2002) Estimation of genetic distances from human and mouse introns. Genome Biol 3:0028.1-0028.7.
• Chung WY, Albert R, Albert I, Nekrutenko A and Makova KD (2006) Rapid and asymmetric divergence of duplicate genes in the human gene coexpression network. Bmc Bioinfo 7:46.
• Clark AG (1994) Invasion and maintenance of a gene duplication. Proc Natl Acad Sci USA 91:2950-2954.
• Comeron JM (2001) What controls the length of noncoding DNA. Curr Opin Genet Dev 11:652-659.
• Conant GC and Wagner A (2002) Genome History - A software tool and its application to fully sequenced genomes. Nucleic Acids Res 30:3378-3386.
• Conery JS and Lynch M (2001) Nucleotide substitutions and the evolution of duplicate genes. Pacific Symp Biocomp 6:167-178.
• Crow KD and Wagner GP (2006) What is the role of genome duplication in the evolution of complexity and diversity? Mol Biol Evol 23:887-892.
• de Meaux J (2006) An adaptive path through jungle DNA. Nat Genet 38:506-507.
• Dermitzakis ET, Bergman CM and Clark AG (2003) Tracing the evolutionary history of Drosophila regulatory regions with models that identify transcription factor binding sites. Mol Biol Evol 20:703-714.
• Eyre-Walker A (2006) The genomic rate of adaptive evolution. Trends Ecol Evol 21:569-575.
• Fessele S, Maier H, Zischek C, Nelson PJ and Werner T (2002) Regulatory context is crucial part of gene function. Trends Genet 18:60-63.
• Fickett JW and Wasserman WW (2000) Discovery and modeling of transcriptional regulatory regions. Curr Opin Biotechnol 11:19-24.
• Force A, Lynch M, Pickett FB, Amores A, Yan Y-L and Postlethwait J (1999) Preservation of duplicate genes by complementary degenerative mutations. Genetics 151:1531-1545.
• Gu Z, Nicolae D, Henry H-S and Li W-H (2002a) Rapid divergence in expression between duplicate genes inferred from microarray data. Trends Genet 18:609-613.
• Gu Z, Cavalcanti A, Chen F-C, Bouman P and Li W-H (2002b) Extend of gene duplication in the genomes of Drosophila, nematode and yeast. Mol Biol Evol 19:256-262.
• Gu Z, Steinmetz LM, Gu X, Scharfe C, Davis RW and Li W-H (2002c) Role of duplicate genes in genetic robustness against null mutations. Nature 421:63-66.
• Gu X and Su ZX (2007) Tissue-driven hypothesis of genomic evolution and sequence-expression correlations. Proc Natl Acad Sci USA 104:2779-2784.
• Guan YF, Dunham MJ and Troyanskaya OG (2007) Functional analysis of gene duplications in Saccharomyces cerevisiae Genetics 175:933-943.
• Ha M, Li WH and Chen ZJ (2007) External factors accelerate expression divergence between duplicate genes. Trends Genet 23:162-166.
• Haberer G, Hindemitt T, Meyers BC and Mayer KFX (2004) Transcriptional similarities, dissimilarities, and conservation of cis-elements in duplicated genes of arabidopsis. Plant Physiol 136:3009-3022.
• Haberer G, Mader MT, Kosarev P, Spannagl M, Yang L and Mayer KFX (2006) Large-scale cis-element detection by analysis of correlated expression and sequence conservation between arabidopsis and Brassica oleracea Plant Physiol 142:1589-1602.
• Hahn MW, Stajich JE and Wray GA (2003) The effects of selection against spurious transcription factor binding sites. Mol Biol Evol 20:901-906.
• Hahn MW (2007) Detecting natural selection on cis-regulatory DNA. Genetica 129:7-18.
• Hasegawa M, Kishino H and Yano T (1985) Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J Mol Evol 22:160-174.
• He XL and Zhang JZ (2005) Rapid subfunctionalization accompanied by prolonged and substantial neofunctionalization in duplicate gene evolution. Genetics 169:1157-1164.
• Hughes AL (1994) The evolution of functionally novel proteins after gene duplication. Proc R Soc London Soc Ser B 256:119-124.
• Hughes AL (2002) Adaptive evolution after gene duplication. Trends Ecol Evol 18:433-434.
• Huminiecki L and Wolfe KH (2004) Divergence of spatial gene expression profiles following species-specific gene duplications in human and mouse. Genome Res 14:1870-1879.
• Jiang, HF, Liu DY, Gu ZL and Wang W (2007) Rapid evolution in a pair of recent duplicate segments of rice. J Exp Zool 308B:50-57.
• Johnston CR, ODushlaine C, Fitzpatrick DA, Edwards RJ and Shields DC (2007) Evaluation of whether accelerated protein evolution in chordates has occurred before, after, or simultaneously with gene duplication. Mol Biol Evol 24:315-323.
• Jordan IK, Marino-Ramirez L and Koonin EV (2005) Evolutionary significance of gene expression divergence. Gene 345:119-126.
• Kim SH and Yi SV (2006) Correlated asymmetry of sequence and functional divergence between duplicate proteins of Saccharomyces cerevisiae Mol Biol Evol 23:1068-1075.
• King LM (1998) The role of gene conversion in determining sequence variation and divergence in the Est-5 gene family in Drosophila pseudoobscura Genetics 148:305-316.
• Kohn MH, Fang S and Wu C-I (2004) Inference of positive and negative selection on the 5 regulatory regions of Drosophila genes. Mol Biol Evol 21:374-383.
• Kondrashov FA, Rogozin IB, Wolf Y I and Koonin EV (2002) Selection in the evolution of gene duplications. Genome Biol 3:0008.1-0008.9.
• Kondrashov AS (2005) Evolutionary biology - Fruitfly genome is not junk. Nature 437:1106.
• Kondrashov FA and Kondrashov AS (2006) Role of selection in fixation of gene duplications. J Theor Biol 239:141-151.
• Koop BF (1995) Human and rodent DNA sequence comparisons: A mosaic model of genomic evolution. Trends Genet 11:367-371.
• Li HP and Stephan W (2006) Inferring the demographic history and rate of adaptive substitution in Drosophila PLoS Genetics 2:1580-1589.
• Ludwig MZ, Bergman C, Patel NH and Kreitman M (2000) Evidence for stabilizing selection in a eukaryotic enhancer element. Nature 403:564-567.
• Ludwig MZ (2002) Functional evolution of noncoding DNA. Curr Opin Genet Dev 12:634-639.
• Lynch M and Conery JS (2000) The evolutionary fate and consequences of duplicate genes. Science 290:1151-1155.
• Lynch M and Force A (2000) The probability of duplicate gene preservation by subfunctionalization. Genetics 154:459-473.
• Lynch M and Katju V (2004) The altered evolutionary trajectories of gene duplicates. Trends Genet 20:544-549.
• Makova KD and Li W-H (2003) Divergence in the spatial pattern of gene expression between human duplicate genes. Genome Res 13:1638-1645.
• Maside X, Bartolome C and Charlesworth B (2003) Inferences on the evolutionary history of the S-element family of Drosophila melanogaster Mol Biol Evol 20:1183-1187.
• Morgenstern B (1999) DIALIGN 2: Improvement of the segment-to-segment approach to multiple sequence alignment. Bioinformatics 15:211-218.
• Ohler U, Liao G-C, Niemann H and Rubin GM (2002) Computational analysis of core promoters in the Drosophila genome. Genome Biol 3:0087.1-0087.12.
• Ohno S (1970) Evolution by Gene Duplication. George Allen and Unwin, London, 160 pp.
• Ohta T (1985) A model of duplicative transposition and gene conversion for repetitive DNA families. Genetics 110:513-524.
• Ohta T (1987) Simulating the evolution of gene duplication. Genetics 115:207-213.
• Ohta T (1988) Evolution by gene duplication and compensatory advantegous mutations. Genetics 120:841-847.
• Ohta T (1994) Further examples of evolution by gene duplication revealed through DNA sequence comparisons. Genetics 138:1331-1337.
• Papp B, Pal C and Hurst LD (2003) Evolution of cis-regulatory elements in duplicated genes of yeast. Trends Genet 19:417-422.
• Roth C, Rastogi S, Arvestad L, Dittmar K, Light S, Ekman D and Liberles DA (2007) Evolution after gene duplication: Models, mechanisms, sequences, systems, and organisms. J Exp Zool Part 308B:58-73.
• Seoighe C, Johnston CR and Shields DC (2003) Significantly different patterns of amino acid replacement after gene duplication as compared to after speciation. Mol Biol Evol 20:484-490.
• Shabalina SA, Ogurtsov AY, Kondrashov VA an Kondrashov AS (2001) Selective constraint in intergenic regions of human and mouse genomes. Trends Genet 17:373-376.
• Shapiro JA, Huang W, Zhang CH, Hubisz MJ, Lu J, Turissini DA, Fang S, Wang HY, Hudson RR, Nielsen R, et al. (2007) Adaptive genic evolution in the Drosophila genomes. Proc Natl Acad Sci USA 104:2271-2276.
• Spellman PT and Rubin GM (2002) Evidence for large domains of similarly expressed genes in the Drosophila genome. J Biol 1:5.
• Stone JR and Wray GA (2001) Rapid evolution of cis-regulatory sequences via local point mutations. Mol Biol Evol 18:1764-1770.
• Tautz D (2000) Evolution of transcriptional regulation. Curr Opin Genet Dev 10:575-579.
• Tautz D and Nigro L (1998) Microevolutionary divergence pattern of the segmentation gene hunchback in Drosophila Mol Biol Evol 15:1403-1411.
• Taylor JS and Raes J (2004) Duplication and divergence: The evolution of new genes and old ideas. Annu Rev Genet 38:615-643.
• Thomas BC, Rapaka L, Lyons E, Pedersen B and Freeling M (2007) Arabidopsis intragenomic conserved noncoding sequences. Proc Natl Acad Sci USA 104:3348-3353.
• Thompson JD, Higgins DG and Gibson TJ (1994) CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22:4673-4680.
• Thornton K and Long M (2002) Rapid divergence of gene duplicates on the Drosophila melanogaster X chromosome. Mol Biol Evol 19:918-925.
• Usdin K and Grabczyk E (2000) DNA repeat expansions and human disease. Cell Mol Life Sci 57:914-931.
• Wagner A (2000) Decoupled evolution of coding region and mRNA expression patterns after gene duplication: Implications for the neutralist-selectionist debate. Proc Natl Acad Sci USA 97:6579-6584.
• Wagner A (2001) Birth and death of duplicated genes in completely sequenced eukaryotes. Trends Genet 17:237-239.
• Wagner A (2002a) Selection and gene duplication: A view from the genome. Genome Biol 3:1012.1-1012.3.
• Wagner A (2002b) Asymmetric functional divergence of duplicate genes in yeast. Mol Biol Evol 19:1760-1768.
• Walsh JB (1995) How often do duplicated genes evolve new functions? Genetics 139:421-428.
• Wang R, Chong K and Wang T (2006) Divergence in spatial expression patterns and in response to stimuli of tandem-repeat paralogues encoding a novel class of proline-rich proteins in Oryza sativa J Exp Bot 57:2887-2897.
• Wasserman WW, Palumbo M, Thompson W and Fickett JW (2000) Human-mouse genome comparisons to locate regulatory sites. Nat Genet 26:225-228.
• Webb CT, Shabalina SA, Ogurtsov AY and Kondrashov AS (2002) Analysis of similarity within 142 pairs of orthologous intergenic regions of Caenorhabditis elegans and Caenorhabditis briggsae Nucleic Acids Res 30:1223-1229.
• Yang Z (1997) PAML: A program package for phylogenetic analysis by maximum likelihood. Comput Appl Biosci 13:555-556.
• Zhang J, Rosenberg HF and Nei M (1998) Positive Darwinian selection after gene duplication in primate ribonuclease genes. Proc Natl Acad Sci USA 95:3708-3713.
• Zhang J (2003) Evolution by gene duplication: An update. Trends Ecol Evol 18:292-298.
• Web page that contains the information and analysis methods used for the paper on the evolution of duplicate genes by Lynch and Conery (2000), http://www.csi.uoregon.edu/projects/genetics/duplications/D.melanogaster.txt (December 2002).
• Berkley Drosophila Genome Project (BDGP, Release 2) (http://www.fruitfly.org) (December 2002).
• A non-redundant set of the 5 prime regions of D. melanogaster genes, http://www.fruitfly.org/seq_tools/datasets/Drosophila/promoter/ (December 2002) (Ohler et al., (2002).
• RepeatMasker software, (http://repeatmasker.genome.washington.edu/cgi-bin/RepeatMasker) (Thompson et al., 1994).
• Gene expression data from 267 Affymetrix GeneChips representing six independent investigations on D. melanogaster compiled by Spellman and Rubin (2002), http://jbiol.com/content/supplementary/1475-4924-1-5-S1.txt (December 2002).

Publication Dates

• Publication in this collection
  24 June 2008
• Date of issue
  2008

History

• Accepted
  13 Mar 2008
• Received
  23 Oct 2007