Nucleotide Variability at the Acetyl Coenzyme A Carboxylase Gene and the Signature of Herbicide Selection in the Grass Weed Alopecurus myosu
http://www.100md.com
分子生物学进展 2004年第5期
Institut National de la Recherche Agronomique, UMR Biologie et Gestion des Adventices, Dijon Cédex, France
E-mail: delye@dijon.inra.fr.
Abstract
Acetyl coenzyme A carboxylase (ACCase) is the target of highly effective herbicides. We investigated the nucleotide variability of the ACCase gene in a sample of 18 black-grass (Alopecurus myosuroides [Huds.]) populations to search for the signature of herbicide selection. Sequencing 3,396 bp encompassing ACCase herbicide-binding domain in 86 individuals revealed 92 polymorphisms, which formed 72 haplotypes. The ratio of nonsynonymous versus synonymous substitutions was very low, in agreement with ACCase being a vital metabolic enzyme. Within black grass, most nonsynonymous substitutions were related to resistance to ACCase-inhibiting herbicides. Differentiation between populations was strong, in contrast to expectations for an allogamous, annual plant. Significant H tests revealed recent hitchhiking events within populations. These results were consistent with recent and local positive selection. We propose that, although they have only been used since at most 15 black-grass generations, ACCase-inhibiting herbicides have exerted a positive selection targeting resistant haplotypes that has been strong enough to have a marked effect upon ACCase nucleotide diversity. A minimum-spanning network of nonrecombinant haplotypes revealed multiple, independent apparitions of resistance-associated mutations. This study provides the first evidence for the signature of ongoing, recent, pesticide selection upon variation at the gene encoding the targeted enzyme in natural plant populations.
Key Words: acetyl coenzyme A carboxylase ? Alopecurus myosuroides ? herbicide ? local selection ? population structure
Introduction
Genetic variation is the basis for adaptation to environmental variation. Environment-driven, positive selection leads to the fixation of adaptive mutations in populations. However, the overall picture that emerges from studies of molecular variation is that most mutations are either neutral, or slightly deleterious. Their frequencies, thus, depend on stochastic effects (genetic drift). So far, relatively few studies of local positive selection within species have been carried out. Although there has recently been increasing interest in characterizing patterns of genetic variation within and between populations in human (Bamshad and Wooding 2003), only few studies concerning plants have considered sequence variation at both the between-population and within-population levels. Even when there are well-characterized phenotypic differences between populations, the underlying genes are often not clearly identified. Furthermore, selection pressures generally studied are ancient, and their action in shaping molecular variation cannot easily be discriminated from that of demography, (e.g., postglacial recolonization [see Dvornyk et al. {2002}, Kuittinen, Salguero, and Aguadé {2002}, and J?rvinen et al. {2003}]). In contrast, highly managed environments such as agricultural fields are of interest to study adaptation of short-lived plant populations. In agricultural fields, herbicides targeting well-identified genes are used to kill weed populations. Herbicides generally represent the most powerful selective pressure exerted upon weed populations. Such populations, thus, provide the basis for studying recent but drastic selection pressures targeting one or a few metabolic genes.
We considered the selection pressure exerted by herbicides in the grass weed Alopecurus myosuroides Huds. (black grass). Black grass is a very widespread, annual, allogamous (Chauvel and Gasquez 1994) weed of cereal crops in northwestern Europe. To control this weed, herbicides targeting the chloroplastic acetyl coenzyme A carboxylase (ACCase, EC 6.1.4.2) have been broadly used since the end of the 1980s. Chloroplastic ACCase is a vital enzyme that catalyzes the first committed step in fatty acid biosynthesis (Harwood 1988). In Poaceae (grasses), the chloroplastic ACCase is a multidomain enzyme (Konishi et al. 1996) encoded by a nuclear gene (Gornicki et al. 1994; Podkowinski et al. 1996) of about 12,500 bp, with a complex intron-exon structure (Podkowinski et al. 1996). It encodes a protein of about 250 kDa, containing three distinct functional domains, biotin-carboxylase, biotin-carboxyl carrier protein, and carboxyl-transferase (CT) (Gornicki et al. 1994). Herbicides targeting ACCase are known to specifically bind to the CT domain of grass chloroplastic ACCase, thus blocking fatty acid biosynthesis and causing plant death (Gronwald 1991). Recently, mutant ACCase alleles with reduced sensitivity to ACCase-inhibiting herbicides were identified in black-grass populations (Délye, Calmès, and Matéjicek 2002; Délye et al. 2003), suggesting that, in agricultural fields, strong selection pressure can drive fast adaptation of weed populations. Hence, the ACCase gene in black grass is an excellent candidate for use in the exploration of the selective effects of herbicides in a weed.
In this work, we examined nucleotide variation in ACCase CT domain within and between 18 black-grass populations to find out whether a signature of the selective pressure exerted by herbicides specifically targeting ACCase could be detected. We found that ACCase CT domain was highly conserved both between and within grass species. The distribution of DNA polymorphisms found within ACCase sequences revealed a differentiation of black-grass populations much higher than expected for an allogamous plant. We demonstrated that within-population and between-population distribution of polymorphisms in ACCase sequences are a consequence of local, recent, positive, herbicide-driven selection.
Materials and Methods
Plant Materials
Seeds of blackgrass were collected in 2000 in 17 agricultural fields in France (fig. 1). All populations except population SA were collected in fields that have been sprayed with ACCase-inhibiting herbicides during at least four years. To better assess the effects of the selective pressure exerted by ACCase-inhibiting herbicides, population LF85 was used as a reference population. This population was collected in 1985, before the intensive use of ACCase-inhibiting herbicides in France. For DNA extraction, seeds were deposited on blotting paper laid down on two layers of 6-mm diameter glass beads in a plastic box (9 cm x 9 cm x 9 cm). Water (25 ml) was added to each box, which was closed and placed at 22°C, 18 h light, during 1 week. The first leaf of one to 15 seedlings per population was then cut, placed into a 0.5-ml microcentrifuge tube and stored at –20°C until used for DNA extraction.
FIG. 1. Geographical origins of the 18 French black-grass populations analyzed in this study
ACCase Sequencing Experiments
DNA was extracted as described by Délye, Wang, and Darmency (2002). PCR amplification of a DNA fragment, including nucleotide positions 4368 to 7329 in A. myosuroides chloroplastic ACCase coding sequence (EMBL accession number AJ310767 [Délye, Calmès and Matéjicek 2002]) with a proofreading polymerase was as described (Délye et al. 2003). The amplified DNA fragment encompasses the entire CT domain of black-grass ACCase. The amplicon obtained from at least five independent PCR reactions put together and purified using a Nucleospin Extract kit (Macherey-Nagel) was directly sequenced on both strands using gene-specific primers. When seedlings contained two different ACCase alleles, the amplicon was cloned in plasmid pDrive (Qiagen), and three different DNA inserts were sequenced for each ACCase allele. All differences that were observed after aligning the sequences were visually rechecked from chromatograms. Sequence assembling and alignments were performed using the BioEdit software (Hall 1999) and the Multalin software (Corpet 1988), respectively.
Data Analysis
Eighteen black-grass populations were investigated. A total of 86 black-grass seedlings were used for sequencing experiments. Forty-seven seedlings contained two distinct ACCase alleles. The 39 remaining seedlings contained two identical ACCase alleles. For each seedling, the two ACCase sequences obtained were used for analysis. A total of 172 sequences were thus used for analysis. To achieve adequate sample sizes for statistical analyses, only sequences obtained from the eight populations collected in 2000 in which at least four individual plants have been studied were considered for detailed between-population and within-population analyses (tables 1 and 2).
Table 1 Nucleotide Polymorphism in Black-Grass ACCase.
Table 2 Results of Tajima's D Statistic and Fay and Wu's Tests of Neutrality.
Estimates of nucleotide diversity ( [Nei 1987]) were computed using DnaSP3.51 (Rozas and Rozas 1999). The degree of codon bias was measured by the effective number of codons (ENC [Wright 1990]). The ENC value can range from 20 (strong codon bias) to 61 (no codon bias). To test whether the frequency distribution of observed polymorphisms conformed to a neutral mutation-drift equilibrium, we calculated the value of two test statistics, Tajima's D statistic (Tajima 1989) and Fay and Wu's H test (Fay and Wu 2000). For the H test, the "ancestral state" of ACCase sequence was determined from the comparison of coding sequences of chloroplastic ACCase from black grass with those from wheat, maize, and foxtail millet (GenBank accession numbers AF029895, U19183, and AF294805, respectively). Those three accessions are the only other multidomain-type, chloroplastic ACCase sequences available so far. The P values for D and H were estimated from 104 coalescent simulations with the assumption of no recombination. This assumption is the most conservative one (Gilad et al. 2002).
Differentiation between populations was measured by FST as implemented in the MEGA version 2.1 software (Kumar S., K. Tamura, I.B. Jakobsen, and M. Nei. 2001. MEGA2: molecular evolutionary genetics analysis software Bioinformatics 17:1244–1245). Pairwise genetic distances between populations were computed using MEGA 2.1 as the mean number of sites differing between sequences (Nei and Kumar 2000). Genetic distances were used in the nearest neighbor statistic (Snn) permutation test (Hudson 2000) to assess whether there was a significant pairwise and global differentiation between the populations. Correlation between pairwise genetic distances and pairwise geographical distances between populations was assessed using a Mantel test with 1,000 random permutations.
Recombination detection within the alignment of ACCase sequences was performed using the "phylogenetic profiles" method (Weiller 1998). This method computes the correlation coefficient of pairwise distances between a given sequence and all other sequences upstream and downstream of each sequence position using a sliding-window procedure. Phylogenetic profiles were computed by the software PhylPro (Weiller 1998). The width of the sliding window for comparisons was set to 100 nucleotides. Haplotypes that gave the smallest phylogenetic correlations, and thus the strongest recombination signals, were discarded from further phylogenetic analysis as described (Weiller 1998). The program AMOVA in the software Arlequin version 2.00 (Schneider S., D. Roessli, and L. Excoffier. 2000. Arlequin ver. 2.00: a software for population genetics data anlaysis. Genetics and Biometry laboratory, University of Geneva, Geneva, Switzerland) was used with 5,000 repetitions to construct a minimum-spanning network connecting all nonrecombinant haplotypes.
Results
Polymorphism of Black-Grass ACCase CT Domain
Alignment of the 172 sequences was 3,396 bp and included the full CT domain coding sequence, four short introns, and 136 nucleotides in the 3'-flanking region (fig. 2). The protein sequence encoded was 895 amino acids long and ranged from position 1426 to 2320 in black-grass ACCase sequence. In the following, all amino acid positions correspond to those in the full-length, black-grass ACCase sequence (EMBL accession number AJ310767 [Délye, Calmès, and Matéjicek 2002]). All nucleotide positions correspond to those in the alignment generated in this study. Among the 86 seedlings studied, a total of 72 haplotypes were identified (fig. 2), of which 46 were found only once (singleton haplotypes). The distribution of haplotypes is shown in table 1. The haplotypes differed by 92 polymorphisms, consisting of 82 nucleotide substitutions, four of which had three alleles, and of six short indels (fig. 2). Among the 49 substitutions found in the coding sequence, 13 were nonsynonymous, two of which had three alleles. For each of these two positions, a single amino-acid change was predicted for both nucleotide substitutions (fig. 2). The ENC value was 52.04, suggesting that no codon bias was present. The amino acid sequence encoded by the region of the black-grass ACCase gene we studied was 91.4%, 87.7%, and 86.3% identical to the corresponding regions in the sequences from wheat, foxtail millet, and maize, respectively. Identity values were of 94.0%, 92.2%, and 91.0%, respectively, when considering ACCase CT domain only.
FIG. 2. Summary of DNA polymorphisms in black-grass ACCase sequences. Dots indicate nucleotides identical to those in haplotype Am01 (most frequently found haplotype). Plus (+) or minus (–), respectively, indicates the presence or absence of indels. Nuc. indicates nucleotide position within alignment. Aa. indicates amino acid position within black-grass ACCase sequence (EMBL accession number AJ310767). Indel 1 = TATGTGACATA; Indel 2 = TTACTACTCCCTCCGATCCATAAATTAATAGGTGCTGAAGGAG; (CT)n indicates Indel 3, n is the number of times the dinucleotide CT is repeated into a haplotype. Indel 4 = AAATGCTGCTATCTTCATATACACTTGTA; Indel 5 = CTA; and Indel 6 = T
The overall nucleotide diversity computed for all 17 black-grass populations collected in 2000 was 0.00388 (table 1). Synonymous and nonsynonymous diversity values were, respectively, 0.01008 and 0.00065, yielding a a/S ratio of 0.0645. Sliding-window analysis indicated a lower level of nucleotide diversity along ACCase CT domain when compared with its flanking regions (fig. 3). However, a cluster of amino acid changes (six out of the 13 recorded) occurred between amino acid positions 2027 and 2096, within ACCase CT domain (fig. 3).
FIG. 3. Sliding window analysis of nucleotide diversity (). The analysis is based upon a 100-bp window with a step size of 25 bp. The exon-intron structure is shown below the plot. Exons are shown as boxes with the CT domain of ACCase shaded. Arrowheads indicate polymorphic amino acid residues
Within-Population and Between-Population Nucleotide Diversity
Between two and 14 distinct haplotypes were found within each of the eight populations considered for detailed analysis (table 1; see Materials and Methods). A high proportion of those haplotypes was private or singleton haplotypes (table 1). The mean within-population nucleotide diversity value was 0.00228, with the extreme values being 0.00011 and 0.00325 (table 1). The ratio of nonsynonymous to synonymous diversity (a/S) within each population did not vary greatly from the value computed over all sequences.
The genetic differentiation between the eight populations was very high, with an FST value of 0.416. The Snn test for differentiation was highly significant when applied to the whole set of sequences obtained from the eight populations (Snn = 0.6415, P < 10–3). Application of the Snn test to all possible pairs of populations revealed significant differentiation for 27 pairs out of 28. Mantel test detected no significant association between pairwise genetic and geographical distances.
The reference population LF85 had a level of nucleotide diversity similar to those of the other populations (table 1). However, population LF85 contained a low number of private alleles, all being singletons (table 1). Population LF85 shared three, one, and one alleles with one, two, and five other populations, respectively. Pairwise Snn tests computed between population LF85 and each of the eight populations used for detailed analysis were significant for only three pairs out of eight.
Linkage Disequilibrium and Selection
To obtain a high power of detection, recombination and linkage disequilibrium were analyzed using the whole set of 172 sequences. All four gametes were found in 12.7% of all pairs of sites analyzed. The minimum number of recombination events needed to explain these data was Rm = 20. Despite this indication for frequent occurrence of recombination, significant linkage disequilibrium was detected by Fisher's exact test with Bonferroni correction in 284 pairs of sites out of 2,145 tested.
Tajima's D statistic was either close to 0 or slightly negative, as expected for an excess of low-frequency polymorphisms. However, D values were never significantly different from 0 (table 2). Fay and Wu's H test uses the frequency distribution of derived polymorphisms to test for an excess of high-frequency variants compared with equilibrium neutral expectations. It was applied to the coding region in the sequences we studied. Among the 49 positions where substitutions were observed in the coding sequence in the region we investigated in the gene encoding black-grass ACCase, 32 contained the same nucleotide in the sequences from wheat, maize, and foxtail millet. This nucleotide was, therefore, deemed "ancestral." The "ancestral state" of the 17 remaining positions was assumed to be that found in the sequence from wheat, which is most closely related to that from black grass. In contrast with Tajima's D statistic, H test values were always negative. They were significantly different from 0 in five out of the eight populations, as well as for the overall set of 17 populations. The H test was not significant for the reference population LF85 (table 2).
Phylogenetic Relationships Between Haplotypes
Phylogenetic profile analysis revealed strong recombination signals in 34 out of the 72 haplotypes. Removing any of those haplotypes from the analysis improved the phylogenetic correlation values for at least 85% of the remaining sequences. The minimum-spanning haplotype network connecting the 38 remaining haplotypes revealed three main clusters that centered on haplotypes Am22, Am37, and Am64 (fig. 4). When considering the distribution of haplotypes containing nonsynonymous substitutions using black-grass, herbicide-sensitive ACCase sequence (EMBL accession number AJ310767) as a reference, we found that haplotypes containing a Lys-to-Arg substitution at position 2264 all clustered around haplotype Am64. In contrast, haplotypes containing Ile-to-Leu and Ile-to-Asn substitutions at positions 1781 and 2041, respectively, were found in several clusters (fig. 4).
FIG. 4. Minimum-spanning network of black-grass, nonrecombinant ACCase haplotypes. Circles represent haplotypes. All populations containing a given haplotype are listed inside the corresponding circle. Nonsingleton, nonsynonymous substitutions, when present, are indicated above haplotypes (the reference sequence is EMBL accession number AJ310767). Haplotypes Am21, Am45, and Am70 also contained singleton, nonsynonymous substitutions (see figure 2)
Discussion
Purifying Selection at ACCase
Comparison of known chloroplastic ACCase amino acid sequences showed that the coding sequence contained in the section of the ACCase gene we investigated, and particularly the CT domain, is very highly conserved between species (92.4% identity on average). This is also true within black grass. Of the 13 amino-acid substitutions we recorded, eight fell within the CT domain, with six of them clustered into a 70–amino acid region (fig. 3). Three out of the eight nonsynonymous substitutions found within the CT domain were rare mutations that were only found in two singleton and one private haplotype, respectively. Two other substitutions (Ile to Leu at position 1781 and Ile to Asn at position 2041) were previously shown to confer resistance to herbicides not only in black grass (Délye, Calmès, and Matéjicek 2002; Délye et al. 2003) but also in rye grass (Lolium sp. [Zagnitko et al. 2001; Délye et al. 2003]), green foxtail (Setaria viridis L. Beauv. [Délye, Wang, and Darmency 2002]) and wild oat (Avena sativa L. [Christoffers, Berg, and Messersmith 2002]). The three remaining changes (Trp to Cys at position 2027, Asp to Gly at position 2078, and Gly to Ala at position 2096), all clustered into the variable 70–amino acid region, are also involved in herbicide resistance (Délye, unpublished data). These five changes are, thus, clearly adaptive in the context of herbicide application. This indicates that nonsynonymous mutations may not occur anywhere within ACCase CT domain, and that most of them are potentially maintained by positive selection exerted by ACCase-inhibiting herbicides. This data, together with the very low a/S ratio we found in black grass (0.0645), strongly suggests that, before herbicide application, variation within ACCase sequence has been shaped by purifying selection, which is consistent with ACCase being a vital metabolic enzyme (Harwood 1988).
High Level of Population Differentiation in an Allogamous, Annual Plant
In the literature, few studies of gene nucleotide diversity have considered both within-population and between-population variation in allogamous plants (table 3). Most studies only considered a very small number of populations; hence, there may be high variation in parameter estimation. However, most computed FST values ranged between 0 and 0.3 (table 3). From these data, the FST value we computed for eight black-grass populations (0.416) is exceptionally high for an allogamous, annual plant. Furthermore, this very high FST value is conflicting with a previous survey of population diversity in black grass (Chauvel and Gasquez 1994). In this study, genetic differentiation between 19 black-grass populations from various countries assessed at seven isoenzyme loci was only 0.023. This value is in agreement with the biology of black grass (allogamous mating system, wind pollination, and large populations), in the absence of selective pressure. In contrast, we found high FST values for ACCase, which were confirmed by significant pairwise and overall Snn tests. These values are in contradiction with neutral expectations, suggesting diversifying and/or local selection exists between populations. Two possible explanations can be proposed: (1) diversifying and/or local selection may act only indirectly upon ACCase (i.e., the ACCase gene is in strong disequilibrium with another gene undergoing diversifying selection), or (2) ACCase itself may be the target for diversifying and/or local selection. The second hypothesis is not conflicting with the occurrence of purifying selection at ACCase proposed in the previous section. It only implies that diversifying and/or local selection targets the few nonsynonymous mutations, most of which are involved in resistance to herbicides.
Table 3 Nucleotide Polymorphism in Allogamous Plants.
Selection in Action: Resistance to ACCase-Inhibiting Herbicides in Black Grass
The frequency distribution of segregating sites in a set of sequences is useful information for inferring about selective processes. Tajima's D statistic detects an excess of low-frequency variants, which yields significantly negative D values. However, observing significantly negative D values is consistent with several processes such as past hitchhiking event(s), background selection, or a recent expansion in population size. In contrast, Fay and Wu's H test detects an excess of high-frequency–derived variants, which is consistent with recent hitchhiking event(s) but not with background selection. While the values of Tajima's D statistic were never significantly different from 0 in our data set, Fay and Wu's H test was significant for the overall data set (table 2). However, the presence of population structure or unequal sampling from the different populations can lead to a significant H test (Gilad et al., 2002). Because we found H values significantly negative within five out of eight populations, as well as for our overall sequence sample (table 2), we considered that population structure was not responsible for the result. The D and H values we observed here would, thus, be consistent with recent, or even ongoing, hitchhiking, when no positively selected mutation has yet been fixed.
Reference population LF85 was collected before the use of ACCase-inhibiting herbicides. The H test was not significant for population LF85. Both the presence of several shared alleles and the nonsignificant pairwise Snn tests involving population LF85 indicated a low level of differentiation between population LF85 and the populations collected in 2000. Population LF85 may, thus, be considered as representing preherbicide ACCase allelic diversity. Several distinct substitutions associated with herbicide resistance are found in haplotypes occurring in most of the populations sampled in 2000 but not in population LF85. The most likely scenario for selection consistent with this data is that positive selection targeting haplotypes resistant to ACCase-inhibiting herbicides occurred after population LF85 was collected (i. e., after 1985). This time lag represents at most 15 black-grass generations. Among the eight populations studied in detail, all except population Lux contained at least one of the five mutations associated with resistance to ACCase-inhibiting herbicides (see Purifying Selection at ACCase above). These mutations were found in intermediate to high frequencies in the seven remaining population (0.10 and 0.20 in populations 010 and 006, and from 0.60 to 0.80 in the five remaining populations). Population Lux contained one haplotype with an Ala-to-Thr substitution at position 2241 and one with a Cys-to-Ser substitution at position 1610. The effects of both substitutions upon sensitivity to ACCase-inhibiting herbicides remain to be characterized. The origin of resistant ACCase haplotypes may be caused by de novo mutation or by the introduction of resistant alleles from migration. Assuming an initial frequency of 10–6, complete dominance and no cost for resistance in a deterministic model (Wright 1977), a net selection coefficient (s) value of at least 0.85 would be sufficient to attain an observed resistant ACCase haplotype frequency as high as 0.80 within 10 generations (not shown). Such an s value would correspond to a herbicide killing rate (mortality of herbicide-sensitive individuals) of 85%. This is largely consistent with literature. The average killing rate for herbicides ranges from 90% to 99% (Jasieniuk, Br?lé-Babel, and Morrison 1996), with ACCase-inhibiting herbicides displaying a killing rate of 95% to 97% (Foster, Ward, and Hewson 1993).
On the minimum-spanning network connecting nonrecombinant haplotypes, herbicide-resistant haplotypes containing Leu and Asn at positions 1781 and 2041 were found in two and three distinct clusters, respectively (fig. 4). This implies several independent appearances of each of those nonsynonymous changes. One given resistant ACCase haplotype most often occurs within a single population. The multiple, independent appearance of resistant haplotypes would, thus, be a consequence of population structure, suggesting that herbicide selection is a local, population-level process. However, resistant haplotypes with clearly distinct origins were sometimes recorded in a single population (e.g., haplotypes Am07 and Am36 carrying an Asn residue at position 2041 in population 099) (fig. 4). This would imply either that some populations have large enough effective sizes to allow several independent appearances of identical amino acid substitutions or that some resistant haplotypes may have been introduced from migration. Clearly, appearance and spread of resistant ACCase alleles is a complex process strongly dependent on the population structure of black grass.
Our study is the first gene-level analysis carrying evidence for an ongoing selection process resulting from anthropic action upon natural plant populations. Studies considering the effects of pesticide-mediated selection pressure at the gene level are scarce and most often considered insects or human parasites, which are prone to rapid, active, or passive (i.e., human-mediated) long-range dissemination. In most cases, one or a few distinct resistance gene(s) appeared once and spread throughout all populations of the insect or parasite undergoing drug or pesticide-based selection pressure (e.g., Raymond et al. [1998], Daborn et al. [2002], and Wootton et al. [2002]). Here, we demonstrated that, because of the very strong associated selection pressure, the use of ACCase-inhibiting herbicides, although very recent, had clearly visible effect upon the genetic variation at the ACCase gene. However, in contrast with the studies mentioned above, our black-grass population sampling showed that herbicide selection pressure is a local selection process leading to the selection of several independently arisen, resistant ACCase alleles. This is consistent with herbicide-spraying programs being designed at the field level and with the absence of known, efficient, long-range black-grass dissemination. Local adaptation at the molecular level has already been documented in short-lived plants, especially in A. thaliana, with respect to flowering time (e.g., Le Corre, Roux, and Reboud [2002]) and resistance to herbivores or pathogens (e.g., Bergelson et al. [2001] and Berger, Mitchell-Olds, and Stotz [2002]), but never for human-mediated selection pressure. Direct studies of nucleotide sequence variability of a major adaptive gene, thus, provided insight into the ecological genetics of short-lived plant populations and showed that adaptive response to a drastic selection pressure can arise within very few generations.
Literature Cited
Bamshad, M., and S. P. Wooding. 2003. Signatures of natural selection in the human genome. Nat. Rev. Genet. 4:99-111.
Bergelson, J., M. Kreitman, E. A. Stahl, and D. Tian. 2001. Evolutionary dynamics of plant R-genes. Science 292:2281-2285.
Berger, S., T. Mitchell-Olds, and H. U. Stotz. 2002. Local and differential control of vegetative storage protein expression in response to herbivore damage in Arabidopsis thaliana. Physiol. Plant. 114:85-91.
Chauvel, B., and J. Gasquez. 1994. Relationships between genetic polymorphism and herbicide resistance whithin Alopecurus myosuroides. Huds. Heredity 72:336-344.
Christoffers, M. J., M. L. Berg, and C. G. Messersmith. 2002. An isoleucine to leucine mutation in acetyl-CoA carboxylase confers herbicide resistance in wild oat. Genome 45:1049-1056.
Corpet, F. 1988. Multiple sequence alignment with hierarchical clustering. Nucl. Acids Res. 16:10881-10890.
Daborn, P. J., J. L. Yen, and M. R. Bogwitz, et al. (13 co-authors). 2002. A single P450 allele associated with insecticide resistance in Drosophila. Science 297:2253-2256.
Délye, C., é. Calmès, and A. Matéjicek. 2002. SNP markers for black-grass (Alopecurus myosuroides Huds.) genotypes resistant to acetyl CoA-carboxylase inhibiting herbicides. Theor. Appl. Genet. 104:1114-1120.
Délye, C., T. Wang, and H. Darmency. 2002. An isoleucine-leucine substitution in chloroplastic acetyl-Co A carboxylase from green foxtail (Setaria viridis L. Beauv.) is responsible for resistance to the cyclohexanedione herbicide sethoxydim. Planta 214:421-427.
Délye, C., X.-Q. Zhang, C. Chalopin, S. Michel, and S. B. Powles. 2003. An isoleucine residue within the carboxyl-transferase domain of multidomain acetyl-CoA carboxylase is a major determinant of sensitivity to aryloxyphenoxypropionate, but not to cyclohexanedione inhibitors. Plant Physiol. 132:1717-1724.
Dvornyk, V., A. Sirvi?, M. Mikkonen, and O. Savolainen. 2002. Low nucleotide diversity at the pa1 locus in the widely distributed Pinus sylvestris. Mol. Biol. Evol. 19:179-188.
Fay, J. C., and C.-I. Wu. 2000. Hitchhiking under positive darwinian selection. Genetics 155:1405-1413.
Filatov D. A., and D. Charlesworth. 1999. DNA polymorphism, haplotype structure and balancing selection in the Laevenworthia PgiC locus. Genetics 153:1423-1434.
Foster, D. K., P. Ward, and R. T. Hewson. 1993. Selective grass-weed control in wheat and barley based on the safener fenchlorazole-ethyl. Pp. 1267–1272 in British Crop Protection Council, ed. Proceedings of the Brighton crop protection conference—weeds. British Crop Protection Council, Surrey, England.
García-Gil, M. R., M. Mikkonen, and O. Savolainen. 2003. Nucleotide diversity at two phytochrome loci along a latitudinal cline in Pinus sylvestris. Mol. Ecol. 12:1195-1206.
Gilad, Y., S. Rosenberg, M. Przeworski, D. Lancet, and K. Skorecki. 2002. Evidence for positive selection and population structure at the human MAO-A gene. Proc Natl Acad Sci USA 99:862-867.
Gornicki, P., J. Podkowinski, L. A. Scappino, J. DiMaio, E. Ward, and R. Haselkorn. 1994. Wheat acetylcoenzyme A carboxylase: cDNA and protein structure. Proc Natl Acad Sci USA 91:6860-6864.
Gronwald, J. W. 1991. Lipid biosynthesis inhibitors. Weed Sci. 39:435-449.
Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acids Symp. Ser. 41:95-98.
Harwood, J. L. 1988. Fatty acid metabolism. Annu. Rev. Plant Physiol. 39:101-138.
Hudson, R. R. 2000. A new statistic for detecting genetic differentiation. Genetics 155:2011-2014.
J?rvinen, P., J. Lemmetyinen, O. Savolainen, and T. Sopanen. 2003. DNA sequence variation in BpMADS2 gene in two populations of Betula pendula. Mol. Ecol. 12:369-384.
Jasieniuk, M., A. L. Br?lé-Babel, and I. N. Morrison. 1996. The evolution and genetics of herbicide resistance in weeds. Weed Sci. 44:176-193.
Kuittinen, H., D. Salguero, and M. Aguadé. 2002. Parallel patterns of sequence variation within and between populations at three loci of Arabidopsis thaliana. Mol. Biol. Evol. 19:2030-2034.
Konishi, T., K. Shinohara, K. Yamada, and Y. Sasaki. 1996. Acetyl-CoA carboxylase in higher plants: most plants other than gramineae have both the prokaryotic and the eukaryotic forms of this enzyme. Plant Cell Physiol. 37:117-122.
Le Corre, V., F. Roux, and X. Reboud. 2002. DNA polymorphism at the FRIGIDA gene in Arabidopsis thaliana: extensive nonsynonymous variation is consistent with local selection for flowering time. Mol. Biol. Evol. 19:1261-1271.
Nei, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York.
Nei, M., and S. Kumar. 2000. Molecular evolution and phylogenetics. Oxford University Press, New York.
Podkowinski, J., G. E. Sroga, R. Haselkorn, and P. Gornicki. 1996. Structure of a gene encoding a cytosolic acetyl-CoA carboxylase of hexaploid wheat. Proc. Natl. Acad. Sci. USA 93:1870-1874.
Raymond, M., C. Chevillon, T. Guillemaud, T. Lenormand, and N. Pasteur. 1998. An overview of the evolution of overproduced esterases in the mosquito Culex pipiens. Phil. Trans. R. Soc. Lond. B Biol. Sci. 353:1707-1711.
Rozas, J., and R. Rozas. 1999. DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175.
Savolainen, O., C. H. Langley, B. P. Lazzaro, and H. Freville. 2000. Contrasting patterns of nucleotide polymorphism at the alcohol dehydrogenase locus in the outcrossing Arabidopsis lyrata and the selfing Arabidopsis thaliana. Mol. Biol. Evol. 17:645-655.
Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585-595.
Weiller, G. F. 1998. Phylogenetic profiles: a graphical method for detecting genetic recombination in homologous sequences. Mol. Biol. Evol. 15:326-335.
Wootton, J. C., X. Feng, M. T. Ferdig, R. A. Cooper, J. Mu, D. I. Baruch, A. J. Magill, and X.-Z. Su. 2002. Genetic diversity and chloroquine selective sweeps in Plasmodium falciparum. Nature 418:320-323.
Wright, S. 1977. Evolution and the genetics of populations. Vol. 3. Experimental results and evolutionary deductions. University of Chicago Press, Chicago.
Wright, F. 1990. The "effective number of codons" used in a gene. Gene 87:23-29.
Wright, S. I., B. Lauga, and D. Charlesworth. 2003. Subdivision and haplotype structure in natural populations of Arabidopsis lyrata. Mol. Ecol. 12:1247-1263.
Zagnitko, O., J. Jelenska, G. Tevzadze, R. Haselkorn, and P. Gornicki. 2001. An isoleucine/leucine residue in the carboxyltransferase domain of acetyl-CoA carboxylase is critical for interaction with aryloxyphenoxypropionate and cyclohexanedione inhibitors. Proc. Natl. Acad. Sci. USA 98:6617-6622.(Christophe Délye1, Cécile)
E-mail: delye@dijon.inra.fr.
Abstract
Acetyl coenzyme A carboxylase (ACCase) is the target of highly effective herbicides. We investigated the nucleotide variability of the ACCase gene in a sample of 18 black-grass (Alopecurus myosuroides [Huds.]) populations to search for the signature of herbicide selection. Sequencing 3,396 bp encompassing ACCase herbicide-binding domain in 86 individuals revealed 92 polymorphisms, which formed 72 haplotypes. The ratio of nonsynonymous versus synonymous substitutions was very low, in agreement with ACCase being a vital metabolic enzyme. Within black grass, most nonsynonymous substitutions were related to resistance to ACCase-inhibiting herbicides. Differentiation between populations was strong, in contrast to expectations for an allogamous, annual plant. Significant H tests revealed recent hitchhiking events within populations. These results were consistent with recent and local positive selection. We propose that, although they have only been used since at most 15 black-grass generations, ACCase-inhibiting herbicides have exerted a positive selection targeting resistant haplotypes that has been strong enough to have a marked effect upon ACCase nucleotide diversity. A minimum-spanning network of nonrecombinant haplotypes revealed multiple, independent apparitions of resistance-associated mutations. This study provides the first evidence for the signature of ongoing, recent, pesticide selection upon variation at the gene encoding the targeted enzyme in natural plant populations.
Key Words: acetyl coenzyme A carboxylase ? Alopecurus myosuroides ? herbicide ? local selection ? population structure
Introduction
Genetic variation is the basis for adaptation to environmental variation. Environment-driven, positive selection leads to the fixation of adaptive mutations in populations. However, the overall picture that emerges from studies of molecular variation is that most mutations are either neutral, or slightly deleterious. Their frequencies, thus, depend on stochastic effects (genetic drift). So far, relatively few studies of local positive selection within species have been carried out. Although there has recently been increasing interest in characterizing patterns of genetic variation within and between populations in human (Bamshad and Wooding 2003), only few studies concerning plants have considered sequence variation at both the between-population and within-population levels. Even when there are well-characterized phenotypic differences between populations, the underlying genes are often not clearly identified. Furthermore, selection pressures generally studied are ancient, and their action in shaping molecular variation cannot easily be discriminated from that of demography, (e.g., postglacial recolonization [see Dvornyk et al. {2002}, Kuittinen, Salguero, and Aguadé {2002}, and J?rvinen et al. {2003}]). In contrast, highly managed environments such as agricultural fields are of interest to study adaptation of short-lived plant populations. In agricultural fields, herbicides targeting well-identified genes are used to kill weed populations. Herbicides generally represent the most powerful selective pressure exerted upon weed populations. Such populations, thus, provide the basis for studying recent but drastic selection pressures targeting one or a few metabolic genes.
We considered the selection pressure exerted by herbicides in the grass weed Alopecurus myosuroides Huds. (black grass). Black grass is a very widespread, annual, allogamous (Chauvel and Gasquez 1994) weed of cereal crops in northwestern Europe. To control this weed, herbicides targeting the chloroplastic acetyl coenzyme A carboxylase (ACCase, EC 6.1.4.2) have been broadly used since the end of the 1980s. Chloroplastic ACCase is a vital enzyme that catalyzes the first committed step in fatty acid biosynthesis (Harwood 1988). In Poaceae (grasses), the chloroplastic ACCase is a multidomain enzyme (Konishi et al. 1996) encoded by a nuclear gene (Gornicki et al. 1994; Podkowinski et al. 1996) of about 12,500 bp, with a complex intron-exon structure (Podkowinski et al. 1996). It encodes a protein of about 250 kDa, containing three distinct functional domains, biotin-carboxylase, biotin-carboxyl carrier protein, and carboxyl-transferase (CT) (Gornicki et al. 1994). Herbicides targeting ACCase are known to specifically bind to the CT domain of grass chloroplastic ACCase, thus blocking fatty acid biosynthesis and causing plant death (Gronwald 1991). Recently, mutant ACCase alleles with reduced sensitivity to ACCase-inhibiting herbicides were identified in black-grass populations (Délye, Calmès, and Matéjicek 2002; Délye et al. 2003), suggesting that, in agricultural fields, strong selection pressure can drive fast adaptation of weed populations. Hence, the ACCase gene in black grass is an excellent candidate for use in the exploration of the selective effects of herbicides in a weed.
In this work, we examined nucleotide variation in ACCase CT domain within and between 18 black-grass populations to find out whether a signature of the selective pressure exerted by herbicides specifically targeting ACCase could be detected. We found that ACCase CT domain was highly conserved both between and within grass species. The distribution of DNA polymorphisms found within ACCase sequences revealed a differentiation of black-grass populations much higher than expected for an allogamous plant. We demonstrated that within-population and between-population distribution of polymorphisms in ACCase sequences are a consequence of local, recent, positive, herbicide-driven selection.
Materials and Methods
Plant Materials
Seeds of blackgrass were collected in 2000 in 17 agricultural fields in France (fig. 1). All populations except population SA were collected in fields that have been sprayed with ACCase-inhibiting herbicides during at least four years. To better assess the effects of the selective pressure exerted by ACCase-inhibiting herbicides, population LF85 was used as a reference population. This population was collected in 1985, before the intensive use of ACCase-inhibiting herbicides in France. For DNA extraction, seeds were deposited on blotting paper laid down on two layers of 6-mm diameter glass beads in a plastic box (9 cm x 9 cm x 9 cm). Water (25 ml) was added to each box, which was closed and placed at 22°C, 18 h light, during 1 week. The first leaf of one to 15 seedlings per population was then cut, placed into a 0.5-ml microcentrifuge tube and stored at –20°C until used for DNA extraction.
FIG. 1. Geographical origins of the 18 French black-grass populations analyzed in this study
ACCase Sequencing Experiments
DNA was extracted as described by Délye, Wang, and Darmency (2002). PCR amplification of a DNA fragment, including nucleotide positions 4368 to 7329 in A. myosuroides chloroplastic ACCase coding sequence (EMBL accession number AJ310767 [Délye, Calmès and Matéjicek 2002]) with a proofreading polymerase was as described (Délye et al. 2003). The amplified DNA fragment encompasses the entire CT domain of black-grass ACCase. The amplicon obtained from at least five independent PCR reactions put together and purified using a Nucleospin Extract kit (Macherey-Nagel) was directly sequenced on both strands using gene-specific primers. When seedlings contained two different ACCase alleles, the amplicon was cloned in plasmid pDrive (Qiagen), and three different DNA inserts were sequenced for each ACCase allele. All differences that were observed after aligning the sequences were visually rechecked from chromatograms. Sequence assembling and alignments were performed using the BioEdit software (Hall 1999) and the Multalin software (Corpet 1988), respectively.
Data Analysis
Eighteen black-grass populations were investigated. A total of 86 black-grass seedlings were used for sequencing experiments. Forty-seven seedlings contained two distinct ACCase alleles. The 39 remaining seedlings contained two identical ACCase alleles. For each seedling, the two ACCase sequences obtained were used for analysis. A total of 172 sequences were thus used for analysis. To achieve adequate sample sizes for statistical analyses, only sequences obtained from the eight populations collected in 2000 in which at least four individual plants have been studied were considered for detailed between-population and within-population analyses (tables 1 and 2).
Table 1 Nucleotide Polymorphism in Black-Grass ACCase.
Table 2 Results of Tajima's D Statistic and Fay and Wu's Tests of Neutrality.
Estimates of nucleotide diversity ( [Nei 1987]) were computed using DnaSP3.51 (Rozas and Rozas 1999). The degree of codon bias was measured by the effective number of codons (ENC [Wright 1990]). The ENC value can range from 20 (strong codon bias) to 61 (no codon bias). To test whether the frequency distribution of observed polymorphisms conformed to a neutral mutation-drift equilibrium, we calculated the value of two test statistics, Tajima's D statistic (Tajima 1989) and Fay and Wu's H test (Fay and Wu 2000). For the H test, the "ancestral state" of ACCase sequence was determined from the comparison of coding sequences of chloroplastic ACCase from black grass with those from wheat, maize, and foxtail millet (GenBank accession numbers AF029895, U19183, and AF294805, respectively). Those three accessions are the only other multidomain-type, chloroplastic ACCase sequences available so far. The P values for D and H were estimated from 104 coalescent simulations with the assumption of no recombination. This assumption is the most conservative one (Gilad et al. 2002).
Differentiation between populations was measured by FST as implemented in the MEGA version 2.1 software (Kumar S., K. Tamura, I.B. Jakobsen, and M. Nei. 2001. MEGA2: molecular evolutionary genetics analysis software Bioinformatics 17:1244–1245). Pairwise genetic distances between populations were computed using MEGA 2.1 as the mean number of sites differing between sequences (Nei and Kumar 2000). Genetic distances were used in the nearest neighbor statistic (Snn) permutation test (Hudson 2000) to assess whether there was a significant pairwise and global differentiation between the populations. Correlation between pairwise genetic distances and pairwise geographical distances between populations was assessed using a Mantel test with 1,000 random permutations.
Recombination detection within the alignment of ACCase sequences was performed using the "phylogenetic profiles" method (Weiller 1998). This method computes the correlation coefficient of pairwise distances between a given sequence and all other sequences upstream and downstream of each sequence position using a sliding-window procedure. Phylogenetic profiles were computed by the software PhylPro (Weiller 1998). The width of the sliding window for comparisons was set to 100 nucleotides. Haplotypes that gave the smallest phylogenetic correlations, and thus the strongest recombination signals, were discarded from further phylogenetic analysis as described (Weiller 1998). The program AMOVA in the software Arlequin version 2.00 (Schneider S., D. Roessli, and L. Excoffier. 2000. Arlequin ver. 2.00: a software for population genetics data anlaysis. Genetics and Biometry laboratory, University of Geneva, Geneva, Switzerland) was used with 5,000 repetitions to construct a minimum-spanning network connecting all nonrecombinant haplotypes.
Results
Polymorphism of Black-Grass ACCase CT Domain
Alignment of the 172 sequences was 3,396 bp and included the full CT domain coding sequence, four short introns, and 136 nucleotides in the 3'-flanking region (fig. 2). The protein sequence encoded was 895 amino acids long and ranged from position 1426 to 2320 in black-grass ACCase sequence. In the following, all amino acid positions correspond to those in the full-length, black-grass ACCase sequence (EMBL accession number AJ310767 [Délye, Calmès, and Matéjicek 2002]). All nucleotide positions correspond to those in the alignment generated in this study. Among the 86 seedlings studied, a total of 72 haplotypes were identified (fig. 2), of which 46 were found only once (singleton haplotypes). The distribution of haplotypes is shown in table 1. The haplotypes differed by 92 polymorphisms, consisting of 82 nucleotide substitutions, four of which had three alleles, and of six short indels (fig. 2). Among the 49 substitutions found in the coding sequence, 13 were nonsynonymous, two of which had three alleles. For each of these two positions, a single amino-acid change was predicted for both nucleotide substitutions (fig. 2). The ENC value was 52.04, suggesting that no codon bias was present. The amino acid sequence encoded by the region of the black-grass ACCase gene we studied was 91.4%, 87.7%, and 86.3% identical to the corresponding regions in the sequences from wheat, foxtail millet, and maize, respectively. Identity values were of 94.0%, 92.2%, and 91.0%, respectively, when considering ACCase CT domain only.
FIG. 2. Summary of DNA polymorphisms in black-grass ACCase sequences. Dots indicate nucleotides identical to those in haplotype Am01 (most frequently found haplotype). Plus (+) or minus (–), respectively, indicates the presence or absence of indels. Nuc. indicates nucleotide position within alignment. Aa. indicates amino acid position within black-grass ACCase sequence (EMBL accession number AJ310767). Indel 1 = TATGTGACATA; Indel 2 = TTACTACTCCCTCCGATCCATAAATTAATAGGTGCTGAAGGAG; (CT)n indicates Indel 3, n is the number of times the dinucleotide CT is repeated into a haplotype. Indel 4 = AAATGCTGCTATCTTCATATACACTTGTA; Indel 5 = CTA; and Indel 6 = T
The overall nucleotide diversity computed for all 17 black-grass populations collected in 2000 was 0.00388 (table 1). Synonymous and nonsynonymous diversity values were, respectively, 0.01008 and 0.00065, yielding a a/S ratio of 0.0645. Sliding-window analysis indicated a lower level of nucleotide diversity along ACCase CT domain when compared with its flanking regions (fig. 3). However, a cluster of amino acid changes (six out of the 13 recorded) occurred between amino acid positions 2027 and 2096, within ACCase CT domain (fig. 3).
FIG. 3. Sliding window analysis of nucleotide diversity (). The analysis is based upon a 100-bp window with a step size of 25 bp. The exon-intron structure is shown below the plot. Exons are shown as boxes with the CT domain of ACCase shaded. Arrowheads indicate polymorphic amino acid residues
Within-Population and Between-Population Nucleotide Diversity
Between two and 14 distinct haplotypes were found within each of the eight populations considered for detailed analysis (table 1; see Materials and Methods). A high proportion of those haplotypes was private or singleton haplotypes (table 1). The mean within-population nucleotide diversity value was 0.00228, with the extreme values being 0.00011 and 0.00325 (table 1). The ratio of nonsynonymous to synonymous diversity (a/S) within each population did not vary greatly from the value computed over all sequences.
The genetic differentiation between the eight populations was very high, with an FST value of 0.416. The Snn test for differentiation was highly significant when applied to the whole set of sequences obtained from the eight populations (Snn = 0.6415, P < 10–3). Application of the Snn test to all possible pairs of populations revealed significant differentiation for 27 pairs out of 28. Mantel test detected no significant association between pairwise genetic and geographical distances.
The reference population LF85 had a level of nucleotide diversity similar to those of the other populations (table 1). However, population LF85 contained a low number of private alleles, all being singletons (table 1). Population LF85 shared three, one, and one alleles with one, two, and five other populations, respectively. Pairwise Snn tests computed between population LF85 and each of the eight populations used for detailed analysis were significant for only three pairs out of eight.
Linkage Disequilibrium and Selection
To obtain a high power of detection, recombination and linkage disequilibrium were analyzed using the whole set of 172 sequences. All four gametes were found in 12.7% of all pairs of sites analyzed. The minimum number of recombination events needed to explain these data was Rm = 20. Despite this indication for frequent occurrence of recombination, significant linkage disequilibrium was detected by Fisher's exact test with Bonferroni correction in 284 pairs of sites out of 2,145 tested.
Tajima's D statistic was either close to 0 or slightly negative, as expected for an excess of low-frequency polymorphisms. However, D values were never significantly different from 0 (table 2). Fay and Wu's H test uses the frequency distribution of derived polymorphisms to test for an excess of high-frequency variants compared with equilibrium neutral expectations. It was applied to the coding region in the sequences we studied. Among the 49 positions where substitutions were observed in the coding sequence in the region we investigated in the gene encoding black-grass ACCase, 32 contained the same nucleotide in the sequences from wheat, maize, and foxtail millet. This nucleotide was, therefore, deemed "ancestral." The "ancestral state" of the 17 remaining positions was assumed to be that found in the sequence from wheat, which is most closely related to that from black grass. In contrast with Tajima's D statistic, H test values were always negative. They were significantly different from 0 in five out of the eight populations, as well as for the overall set of 17 populations. The H test was not significant for the reference population LF85 (table 2).
Phylogenetic Relationships Between Haplotypes
Phylogenetic profile analysis revealed strong recombination signals in 34 out of the 72 haplotypes. Removing any of those haplotypes from the analysis improved the phylogenetic correlation values for at least 85% of the remaining sequences. The minimum-spanning haplotype network connecting the 38 remaining haplotypes revealed three main clusters that centered on haplotypes Am22, Am37, and Am64 (fig. 4). When considering the distribution of haplotypes containing nonsynonymous substitutions using black-grass, herbicide-sensitive ACCase sequence (EMBL accession number AJ310767) as a reference, we found that haplotypes containing a Lys-to-Arg substitution at position 2264 all clustered around haplotype Am64. In contrast, haplotypes containing Ile-to-Leu and Ile-to-Asn substitutions at positions 1781 and 2041, respectively, were found in several clusters (fig. 4).
FIG. 4. Minimum-spanning network of black-grass, nonrecombinant ACCase haplotypes. Circles represent haplotypes. All populations containing a given haplotype are listed inside the corresponding circle. Nonsingleton, nonsynonymous substitutions, when present, are indicated above haplotypes (the reference sequence is EMBL accession number AJ310767). Haplotypes Am21, Am45, and Am70 also contained singleton, nonsynonymous substitutions (see figure 2)
Discussion
Purifying Selection at ACCase
Comparison of known chloroplastic ACCase amino acid sequences showed that the coding sequence contained in the section of the ACCase gene we investigated, and particularly the CT domain, is very highly conserved between species (92.4% identity on average). This is also true within black grass. Of the 13 amino-acid substitutions we recorded, eight fell within the CT domain, with six of them clustered into a 70–amino acid region (fig. 3). Three out of the eight nonsynonymous substitutions found within the CT domain were rare mutations that were only found in two singleton and one private haplotype, respectively. Two other substitutions (Ile to Leu at position 1781 and Ile to Asn at position 2041) were previously shown to confer resistance to herbicides not only in black grass (Délye, Calmès, and Matéjicek 2002; Délye et al. 2003) but also in rye grass (Lolium sp. [Zagnitko et al. 2001; Délye et al. 2003]), green foxtail (Setaria viridis L. Beauv. [Délye, Wang, and Darmency 2002]) and wild oat (Avena sativa L. [Christoffers, Berg, and Messersmith 2002]). The three remaining changes (Trp to Cys at position 2027, Asp to Gly at position 2078, and Gly to Ala at position 2096), all clustered into the variable 70–amino acid region, are also involved in herbicide resistance (Délye, unpublished data). These five changes are, thus, clearly adaptive in the context of herbicide application. This indicates that nonsynonymous mutations may not occur anywhere within ACCase CT domain, and that most of them are potentially maintained by positive selection exerted by ACCase-inhibiting herbicides. This data, together with the very low a/S ratio we found in black grass (0.0645), strongly suggests that, before herbicide application, variation within ACCase sequence has been shaped by purifying selection, which is consistent with ACCase being a vital metabolic enzyme (Harwood 1988).
High Level of Population Differentiation in an Allogamous, Annual Plant
In the literature, few studies of gene nucleotide diversity have considered both within-population and between-population variation in allogamous plants (table 3). Most studies only considered a very small number of populations; hence, there may be high variation in parameter estimation. However, most computed FST values ranged between 0 and 0.3 (table 3). From these data, the FST value we computed for eight black-grass populations (0.416) is exceptionally high for an allogamous, annual plant. Furthermore, this very high FST value is conflicting with a previous survey of population diversity in black grass (Chauvel and Gasquez 1994). In this study, genetic differentiation between 19 black-grass populations from various countries assessed at seven isoenzyme loci was only 0.023. This value is in agreement with the biology of black grass (allogamous mating system, wind pollination, and large populations), in the absence of selective pressure. In contrast, we found high FST values for ACCase, which were confirmed by significant pairwise and overall Snn tests. These values are in contradiction with neutral expectations, suggesting diversifying and/or local selection exists between populations. Two possible explanations can be proposed: (1) diversifying and/or local selection may act only indirectly upon ACCase (i.e., the ACCase gene is in strong disequilibrium with another gene undergoing diversifying selection), or (2) ACCase itself may be the target for diversifying and/or local selection. The second hypothesis is not conflicting with the occurrence of purifying selection at ACCase proposed in the previous section. It only implies that diversifying and/or local selection targets the few nonsynonymous mutations, most of which are involved in resistance to herbicides.
Table 3 Nucleotide Polymorphism in Allogamous Plants.
Selection in Action: Resistance to ACCase-Inhibiting Herbicides in Black Grass
The frequency distribution of segregating sites in a set of sequences is useful information for inferring about selective processes. Tajima's D statistic detects an excess of low-frequency variants, which yields significantly negative D values. However, observing significantly negative D values is consistent with several processes such as past hitchhiking event(s), background selection, or a recent expansion in population size. In contrast, Fay and Wu's H test detects an excess of high-frequency–derived variants, which is consistent with recent hitchhiking event(s) but not with background selection. While the values of Tajima's D statistic were never significantly different from 0 in our data set, Fay and Wu's H test was significant for the overall data set (table 2). However, the presence of population structure or unequal sampling from the different populations can lead to a significant H test (Gilad et al., 2002). Because we found H values significantly negative within five out of eight populations, as well as for our overall sequence sample (table 2), we considered that population structure was not responsible for the result. The D and H values we observed here would, thus, be consistent with recent, or even ongoing, hitchhiking, when no positively selected mutation has yet been fixed.
Reference population LF85 was collected before the use of ACCase-inhibiting herbicides. The H test was not significant for population LF85. Both the presence of several shared alleles and the nonsignificant pairwise Snn tests involving population LF85 indicated a low level of differentiation between population LF85 and the populations collected in 2000. Population LF85 may, thus, be considered as representing preherbicide ACCase allelic diversity. Several distinct substitutions associated with herbicide resistance are found in haplotypes occurring in most of the populations sampled in 2000 but not in population LF85. The most likely scenario for selection consistent with this data is that positive selection targeting haplotypes resistant to ACCase-inhibiting herbicides occurred after population LF85 was collected (i. e., after 1985). This time lag represents at most 15 black-grass generations. Among the eight populations studied in detail, all except population Lux contained at least one of the five mutations associated with resistance to ACCase-inhibiting herbicides (see Purifying Selection at ACCase above). These mutations were found in intermediate to high frequencies in the seven remaining population (0.10 and 0.20 in populations 010 and 006, and from 0.60 to 0.80 in the five remaining populations). Population Lux contained one haplotype with an Ala-to-Thr substitution at position 2241 and one with a Cys-to-Ser substitution at position 1610. The effects of both substitutions upon sensitivity to ACCase-inhibiting herbicides remain to be characterized. The origin of resistant ACCase haplotypes may be caused by de novo mutation or by the introduction of resistant alleles from migration. Assuming an initial frequency of 10–6, complete dominance and no cost for resistance in a deterministic model (Wright 1977), a net selection coefficient (s) value of at least 0.85 would be sufficient to attain an observed resistant ACCase haplotype frequency as high as 0.80 within 10 generations (not shown). Such an s value would correspond to a herbicide killing rate (mortality of herbicide-sensitive individuals) of 85%. This is largely consistent with literature. The average killing rate for herbicides ranges from 90% to 99% (Jasieniuk, Br?lé-Babel, and Morrison 1996), with ACCase-inhibiting herbicides displaying a killing rate of 95% to 97% (Foster, Ward, and Hewson 1993).
On the minimum-spanning network connecting nonrecombinant haplotypes, herbicide-resistant haplotypes containing Leu and Asn at positions 1781 and 2041 were found in two and three distinct clusters, respectively (fig. 4). This implies several independent appearances of each of those nonsynonymous changes. One given resistant ACCase haplotype most often occurs within a single population. The multiple, independent appearance of resistant haplotypes would, thus, be a consequence of population structure, suggesting that herbicide selection is a local, population-level process. However, resistant haplotypes with clearly distinct origins were sometimes recorded in a single population (e.g., haplotypes Am07 and Am36 carrying an Asn residue at position 2041 in population 099) (fig. 4). This would imply either that some populations have large enough effective sizes to allow several independent appearances of identical amino acid substitutions or that some resistant haplotypes may have been introduced from migration. Clearly, appearance and spread of resistant ACCase alleles is a complex process strongly dependent on the population structure of black grass.
Our study is the first gene-level analysis carrying evidence for an ongoing selection process resulting from anthropic action upon natural plant populations. Studies considering the effects of pesticide-mediated selection pressure at the gene level are scarce and most often considered insects or human parasites, which are prone to rapid, active, or passive (i.e., human-mediated) long-range dissemination. In most cases, one or a few distinct resistance gene(s) appeared once and spread throughout all populations of the insect or parasite undergoing drug or pesticide-based selection pressure (e.g., Raymond et al. [1998], Daborn et al. [2002], and Wootton et al. [2002]). Here, we demonstrated that, because of the very strong associated selection pressure, the use of ACCase-inhibiting herbicides, although very recent, had clearly visible effect upon the genetic variation at the ACCase gene. However, in contrast with the studies mentioned above, our black-grass population sampling showed that herbicide selection pressure is a local selection process leading to the selection of several independently arisen, resistant ACCase alleles. This is consistent with herbicide-spraying programs being designed at the field level and with the absence of known, efficient, long-range black-grass dissemination. Local adaptation at the molecular level has already been documented in short-lived plants, especially in A. thaliana, with respect to flowering time (e.g., Le Corre, Roux, and Reboud [2002]) and resistance to herbivores or pathogens (e.g., Bergelson et al. [2001] and Berger, Mitchell-Olds, and Stotz [2002]), but never for human-mediated selection pressure. Direct studies of nucleotide sequence variability of a major adaptive gene, thus, provided insight into the ecological genetics of short-lived plant populations and showed that adaptive response to a drastic selection pressure can arise within very few generations.
Literature Cited
Bamshad, M., and S. P. Wooding. 2003. Signatures of natural selection in the human genome. Nat. Rev. Genet. 4:99-111.
Bergelson, J., M. Kreitman, E. A. Stahl, and D. Tian. 2001. Evolutionary dynamics of plant R-genes. Science 292:2281-2285.
Berger, S., T. Mitchell-Olds, and H. U. Stotz. 2002. Local and differential control of vegetative storage protein expression in response to herbivore damage in Arabidopsis thaliana. Physiol. Plant. 114:85-91.
Chauvel, B., and J. Gasquez. 1994. Relationships between genetic polymorphism and herbicide resistance whithin Alopecurus myosuroides. Huds. Heredity 72:336-344.
Christoffers, M. J., M. L. Berg, and C. G. Messersmith. 2002. An isoleucine to leucine mutation in acetyl-CoA carboxylase confers herbicide resistance in wild oat. Genome 45:1049-1056.
Corpet, F. 1988. Multiple sequence alignment with hierarchical clustering. Nucl. Acids Res. 16:10881-10890.
Daborn, P. J., J. L. Yen, and M. R. Bogwitz, et al. (13 co-authors). 2002. A single P450 allele associated with insecticide resistance in Drosophila. Science 297:2253-2256.
Délye, C., é. Calmès, and A. Matéjicek. 2002. SNP markers for black-grass (Alopecurus myosuroides Huds.) genotypes resistant to acetyl CoA-carboxylase inhibiting herbicides. Theor. Appl. Genet. 104:1114-1120.
Délye, C., T. Wang, and H. Darmency. 2002. An isoleucine-leucine substitution in chloroplastic acetyl-Co A carboxylase from green foxtail (Setaria viridis L. Beauv.) is responsible for resistance to the cyclohexanedione herbicide sethoxydim. Planta 214:421-427.
Délye, C., X.-Q. Zhang, C. Chalopin, S. Michel, and S. B. Powles. 2003. An isoleucine residue within the carboxyl-transferase domain of multidomain acetyl-CoA carboxylase is a major determinant of sensitivity to aryloxyphenoxypropionate, but not to cyclohexanedione inhibitors. Plant Physiol. 132:1717-1724.
Dvornyk, V., A. Sirvi?, M. Mikkonen, and O. Savolainen. 2002. Low nucleotide diversity at the pa1 locus in the widely distributed Pinus sylvestris. Mol. Biol. Evol. 19:179-188.
Fay, J. C., and C.-I. Wu. 2000. Hitchhiking under positive darwinian selection. Genetics 155:1405-1413.
Filatov D. A., and D. Charlesworth. 1999. DNA polymorphism, haplotype structure and balancing selection in the Laevenworthia PgiC locus. Genetics 153:1423-1434.
Foster, D. K., P. Ward, and R. T. Hewson. 1993. Selective grass-weed control in wheat and barley based on the safener fenchlorazole-ethyl. Pp. 1267–1272 in British Crop Protection Council, ed. Proceedings of the Brighton crop protection conference—weeds. British Crop Protection Council, Surrey, England.
García-Gil, M. R., M. Mikkonen, and O. Savolainen. 2003. Nucleotide diversity at two phytochrome loci along a latitudinal cline in Pinus sylvestris. Mol. Ecol. 12:1195-1206.
Gilad, Y., S. Rosenberg, M. Przeworski, D. Lancet, and K. Skorecki. 2002. Evidence for positive selection and population structure at the human MAO-A gene. Proc Natl Acad Sci USA 99:862-867.
Gornicki, P., J. Podkowinski, L. A. Scappino, J. DiMaio, E. Ward, and R. Haselkorn. 1994. Wheat acetylcoenzyme A carboxylase: cDNA and protein structure. Proc Natl Acad Sci USA 91:6860-6864.
Gronwald, J. W. 1991. Lipid biosynthesis inhibitors. Weed Sci. 39:435-449.
Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl. Acids Symp. Ser. 41:95-98.
Harwood, J. L. 1988. Fatty acid metabolism. Annu. Rev. Plant Physiol. 39:101-138.
Hudson, R. R. 2000. A new statistic for detecting genetic differentiation. Genetics 155:2011-2014.
J?rvinen, P., J. Lemmetyinen, O. Savolainen, and T. Sopanen. 2003. DNA sequence variation in BpMADS2 gene in two populations of Betula pendula. Mol. Ecol. 12:369-384.
Jasieniuk, M., A. L. Br?lé-Babel, and I. N. Morrison. 1996. The evolution and genetics of herbicide resistance in weeds. Weed Sci. 44:176-193.
Kuittinen, H., D. Salguero, and M. Aguadé. 2002. Parallel patterns of sequence variation within and between populations at three loci of Arabidopsis thaliana. Mol. Biol. Evol. 19:2030-2034.
Konishi, T., K. Shinohara, K. Yamada, and Y. Sasaki. 1996. Acetyl-CoA carboxylase in higher plants: most plants other than gramineae have both the prokaryotic and the eukaryotic forms of this enzyme. Plant Cell Physiol. 37:117-122.
Le Corre, V., F. Roux, and X. Reboud. 2002. DNA polymorphism at the FRIGIDA gene in Arabidopsis thaliana: extensive nonsynonymous variation is consistent with local selection for flowering time. Mol. Biol. Evol. 19:1261-1271.
Nei, M. 1987. Molecular evolutionary genetics. Columbia University Press, New York.
Nei, M., and S. Kumar. 2000. Molecular evolution and phylogenetics. Oxford University Press, New York.
Podkowinski, J., G. E. Sroga, R. Haselkorn, and P. Gornicki. 1996. Structure of a gene encoding a cytosolic acetyl-CoA carboxylase of hexaploid wheat. Proc. Natl. Acad. Sci. USA 93:1870-1874.
Raymond, M., C. Chevillon, T. Guillemaud, T. Lenormand, and N. Pasteur. 1998. An overview of the evolution of overproduced esterases in the mosquito Culex pipiens. Phil. Trans. R. Soc. Lond. B Biol. Sci. 353:1707-1711.
Rozas, J., and R. Rozas. 1999. DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15:174-175.
Savolainen, O., C. H. Langley, B. P. Lazzaro, and H. Freville. 2000. Contrasting patterns of nucleotide polymorphism at the alcohol dehydrogenase locus in the outcrossing Arabidopsis lyrata and the selfing Arabidopsis thaliana. Mol. Biol. Evol. 17:645-655.
Tajima, F. 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123:585-595.
Weiller, G. F. 1998. Phylogenetic profiles: a graphical method for detecting genetic recombination in homologous sequences. Mol. Biol. Evol. 15:326-335.
Wootton, J. C., X. Feng, M. T. Ferdig, R. A. Cooper, J. Mu, D. I. Baruch, A. J. Magill, and X.-Z. Su. 2002. Genetic diversity and chloroquine selective sweeps in Plasmodium falciparum. Nature 418:320-323.
Wright, S. 1977. Evolution and the genetics of populations. Vol. 3. Experimental results and evolutionary deductions. University of Chicago Press, Chicago.
Wright, F. 1990. The "effective number of codons" used in a gene. Gene 87:23-29.
Wright, S. I., B. Lauga, and D. Charlesworth. 2003. Subdivision and haplotype structure in natural populations of Arabidopsis lyrata. Mol. Ecol. 12:1247-1263.
Zagnitko, O., J. Jelenska, G. Tevzadze, R. Haselkorn, and P. Gornicki. 2001. An isoleucine/leucine residue in the carboxyltransferase domain of acetyl-CoA carboxylase is critical for interaction with aryloxyphenoxypropionate and cyclohexanedione inhibitors. Proc. Natl. Acad. Sci. USA 98:6617-6622.(Christophe Délye1, Cécile)