Research Article Received: 3 February 2014
Revised: 18 April 2014
Accepted article published: 28 April 2014
Published online in Wiley Online Library: 27 May 2014
(wileyonlinelibrary.com) DOI 10.1002/ps.3818
Using next-generation sequencing to detect mutations endowing resistance to pesticides: application to acetolactate-synthase (ALS)-based resistance in barnyard grass, a polyploid grass weed Christophe Délye,a* Romain Causse,a Véronique Gautier,b Charles Poncetb and Séverine Michela Abstract BACKGROUND: Next-generation sequencing (NGS) technologies offer tremendous possibilities for accurate detection of mutations endowing pesticide resistance, yet their use for this purpose has not emerged in crop protection. This study aims at promoting NGS use for pesticide resistance diagnosis. It describes a simple procedure accessible to virtually any scientist and implementing freely accessible programs for the analysis of NGS data. RESULTS: Three PCR amplicons encompassing seven codons of the acetolactate-synthase gene crucial for herbicide resistance were sequenced using non-quantified pools of crude DNA extracts from 40 plants in each of 28 field populations of barnyard grass, a polyploid weed. A total of 63 959 quality NGS sequence runs were obtained using the 454 technology. Three herbicide-resistance-endowing mutations (Pro-197-Ser, Pro-197-Leu and/or Trp-574-Leu) were identified in seven populations. The NGS results were confirmed by individual plant Sanger sequencing. CONCLUSION: This work demonstrated the feasibility of NGS-based detection of pesticide resistance, and the advantages of NGS compared with other molecular biology techniques for analysing large numbers of individuals. NGS-based resistance diagnosis has the potential to play a substantial role in monitoring resistance, maintaining pesticide efficacy and optimising pesticide applications. © 2014 Society of Chemical Industry Keywords: resistance; pesticide; herbicide; diagnosis; next-generation sequencing; Echinochloa
1
INTRODUCTION
Pest Manag Sci 2015; 71: 675–685
mutations can occur in gene regulatory regions and modify the expression of the pesticide target protein (i.e. overexpression, which compensates for the pesticide inhibitory action), of pesticide-metabolising enzyme(s) or of transporter proteins in a way causing an increase in pesticide degradation or compartmenting away from its site of action.2 – 4 Mutations are thus the basis of pesticide resistance, and past research has identified quite a few resistance-endowing mutations.3,8 Such mutations have long been a target of choice for DNA-based resistance detection assays, because such assays are reliable and allow rapid resistance diagnosis and within-season adaptation of the spraying programme to avoid further selection of resistance.3,9 However, the techniques
∗
Correspondence to: Christophe Délye, INRA, UMR1347 Agroécologie, 17 rue Sully, F-21000 Dijon, France. E-mail:
[email protected]
a INRA, UMR1347 Agroécologie, Dijon, France b INRA, UMR1095 Génétique, Diversité et Écophysiologie des Céréales, Clermont-Ferrand, France
www.soci.org
© 2014 Society of Chemical Industry
675
Weeds, pests and pathogens are the major causes of agricultural crop yield losses worldwide, and crop protection is essential to safeguard food production.1 Globally, crop protection currently relies largely on pesticide applications. However, the evolution of resistances to pesticides in a number of organisms is an increasing challenge to pesticide-based crop protection.2 – 4 Resistance detection for the purpose of guidance of crop protection and/or resistance monitoring is crucial for resistance management5 – 7 and demands effective and reliable methods. Resistance to pesticides evolves in weeds, pests and pathogens essentially as a result of adaptive selection of mutations conferring a decreased sensitivity to pesticides.2 – 4 In the case of pesticide resistance, these mutations can cause a structural modification in the spatial structure of a pesticide target protein that will lead to a decrease in the efficacy of pesticide(s) (e.g. mutations causing an amino acid substitution at the pesticide-binding site of a target protein). Alternatively, mutations at the active site of a metabolic enzyme or a transporter protein can improve the activity of these proteins in pesticide neutralisation. Other resistance-endowing
www.soci.org
676
implemented and the information generated by DNA-based resistance diagnosis assays have evolved little since the first assays were developed. Basically, the principle of most DNA-based assays is to reveal the presence of resistance-endowing mutations via variation in the amplicon pattern generated by a variant of the PCR technique. The vast majority of published DNA-based resistance diagnosis assays involve PCR-based mutation genotyping. Assays roughly fall into two categories: allele-specific PCR (one specific amplicon is generated only when a given mutation is present) or PCR followed by restriction enzyme digestion of one amplicon [PCR-RFLP (Restriction Length Fragment Polymorphism), CAPS (Cleaved Amplified Polymorphic Sequence) or dCAPS (derived CAPS): one specific amplicon is digested by a restriction enzyme according to the presence of a given mutation in the target DNA sequence]. Allele-specific PCR and PCR-RFLP/CAPS assays for pesticide resistance diagnosis have been continuously published from the late 1990s10 – 14 to this day.15 – 18 This is also true of dCAPS assays.19,20 Following technology evolution, allele-specific PCR has been at the basis of ‘hi-tech’, high-throughput assays implementing quantitative PCR.21,22 Other genotyping techniques allowing high-throughput detection of mutations at known positions have also been used to develop pesticide resistance diagnosis, such as MALDI-TOF,23 TaqMan,24,25 SimpleProbe,26 DNA microarrays27 or SNaPshot.28 A common feature of all the techniques described above is that they only provide information about the mutation(s) present at a given nucleotide position or codon, which generally requires identification of the mutation(s) in question beforehand. Thus, seeking resistance-endowing mutations scattered along a gene requires the development of a set of assays.20 Yet, only sequencing captures the full nucleotide variation in a DNA region of interest. Today, the rise of next-generation sequencing (NGS) technologies (described by Metzker29 ) has opened up the possibility of using sequencing for pesticide resistance diagnosis. The three currently dominant NGS technologies are Illumina’s (highly parallel sequencing by synthesis; Illumina, San Diego, CA), SOLiD (sequencing by ligation; Applied Biosystems, Foster City, CA) and 454 (pyrosequencing; Roche, Branford, CT).30 NGS generates massive amounts of short (currently, ca 50–600 nucleotides) sequence reads. NGS is generally used to sequence whole genomes or transcriptomes by sequencing a massive number of DNA or cDNA fragments in one or a few individuals.30 Conversely, NGS can also be used to sequence a few DNA fragments of interest in a massive number of individuals. NGS is accordingly increasingly used in the medical domain for the diagnosis of mutations of interest in human diseases31,32 or for the detection of resistance to antibiotics (reviewed by Lupo et al.33 ) or to antiviral drugs.34 In spite of its potential and its low cost compared with Sanger sequencing, NGS has not yet emerged as a tool for pesticide resistance diagnosis. To the best of the authors’ knowledge, only two studies have been published so far that have used pyrosequencing to detect nucleotide substitutions underlying herbicide resistance in plants analysed individually.35,36 This may be due to the difficulty in accessing this technology, which can now be overcome by a number of companies or institutions proposing affordable NGS services. Another barrier to widespread use of NGS technologies are the data generated, which may be considered too large and too complex to be analysed by non-specialists. This work is intended as a prototype to guide future NGS-based pesticide resistance diagnostic studies. Herein, the authors generated PCR amplicons encompassing regions of interest in a gene involved in pesticide resistance. A simple procedure based on the use of
wileyonlinelibrary.com/journal/ps
C Délye et al.
freely accessible programs was implemented to analyse NGS data for pesticide resistance diagnosis. As a case study, herbicide resistance in barnyard grass (Echinochloa crus-galli) was considered. This is a hexaploid grass weed species that has evolved resistance to herbicides inhibiting the acetolactate-synthase (ALS) enzyme in several countries.37,38 The NGS results were validated using Sanger sequencing.
2
MATERIALS AND METHODS
2.1 Plant material Plant material was collected from a total of 28 rice fields in the French Camargue, an area that is located in the delta of the Rhône river, in the South of France. In the fields investigated, applications of ALS inhibitors gave poor or no control of barnyard grass populations. At the growers’ request, a study was conducted to check whether mutant ALS alleles conferring herbicide resistance were present in these populations. Two types of plant material were used for analysis (Table 1). One leaf per plant was collected on 40 fully grown plants in ten of the fields. The 40 leaves were placed in a paper bag and posted to the authors’ laboratory in Dijon, where DNA extraction was performed upon their arrival. Seeds from barnyard grass plants in the remaining 18 fields were collected (one bulk sample per field) and allowed to germinate. Forty young seedlings (1–2-leaf stage) per field were then used for DNA extraction. DNA was extracted from each leaf or seedling using a rapid method,39 and DNA samples were stored at −20 ∘ C prior to PCR amplification. 2.2 PCR amplification of ALS fragments Primer pairs were designed using barnyard grass ALS coding sequence (GenBank/EMBL accession JQ319776) to generate amplicons encompassing codons crucial for sensitivity to ALS inhibitors (codons 122, 197, 205, 376, 377, 574, 653 and 654, standardised to Arabidopsis thaliana sequence).8,40 As no amplicon carrying codon 122 could reliably be obtained in the present experiments, this codon was not investigated in this study. The 454 NGS technology (Roche) was used in this work. Primer pairs to be used for PCR followed by 454 sequencing must include specific sequences added at the 5′ end of the ‘classical’ PCR primers to allow sequencing. One primer in each pair must include sequence ‘Primer A-lib L’ (5′ -CCATCTCATCCCTGCGTGTCTCCGACTCAG) followed by a gene-specific sequence, while the other primer must include sequence ‘Primer B-lib L’ (5′ -CCTATCCCCTGTGTGCCTTGGCAGTCT CAG) followed by a gene-specific sequence. As amplicons originating from all populations studied were sequenced as a pool on a single 454 run, a ten-nucleotide multiplex identifier (MID) sequence specific to each population was also added between the ‘Primer A-lib L’ sequence and the gene-specific sequence of one primer per pair to allow automated software identification of the sequence reads corresponding to each population after sequencing. MID sequences are given in Table 1. The 3′ , ALS-specific sequence of the primers used is given in Table 2. As per recommendation from Roche, primer pairs were designed so that the expected size of each amplicon was between 200 and 600 nucleotides. The ‘Primer A-lib L’ sequence followed by a MID sequence was added to the 5′ end of primers ALE-A, ALE-BR and ALE-C, while the ‘Primer B-lib L’ sequence was added to the 5′ end of primers ALE-AR, ALE-B and ALE-CR (Table 2). PCR mixes were as described by Kati et al.41 The PCR programmes consisted
© 2014 Society of Chemical Industry
Pest Manag Sci 2015; 71: 675–685
Pesticide resistance diagnosis using next-generation sequencing
www.soci.org
Table 1. Plant material and corresponding multiplex identifier (MID) sequences used for 454 sequencing Sample codea
Plant materialb
ECH01 ECH03 ECH04 ECH05 ECH06 ECH07 ECH08 ECH09 ECH10 ECH12 ECH14 ECH15 ECH16 ECH17 ECH18 ECH19 ECH20 ECH21
Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings Seedlings
FDP SEY01 SARR PRIV
Field plants Field plants Field plants Field plants
BRU01 BRU02 SEY02 GOU01 GOU02 GOU03
Field plants Field plants Field plants Field plants Field plants Field plants
PCR poolsc 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 1 × 40 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 1 × 40 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants 10 × 4 plants
MID
MID sequence
MID01 MID02 MID03 MID04 MID05 MID06 MID07 MID08 MID10 MID11 MID13 MID14 MID15 MID16 MID17 MID18 MID19 MID20 MID32 MID21 MID22 MID23 MID24 MID31 MID25 MID26 MID27 MID28 MID29 MID30
ACGAGTGCGT ACGCTCGACA AGACGCACTC AGCACTGTAG ATCAGACACG ATATCGCGAG CGTGTCTCTA CTCGCGTGTC TCTCTATGCG TGATACGTCT CATAGTAGTG CGAGAGATAC ATACGACGTA TCACGTACTA CGTCTAGTAC TCTACGTAGC TGTACTACTC ACGACTACAG AGTACGCTAT CGTAGACTAG TACGAGTATG TACTCTCGTG TAGAGACGAG AGCGTCGTCT TCGTCGCTCG ACATACGCGT ACGCGAGTAT ACTACTATGT ACTGTACAGT AGACTATACT
a Populations used for comparison between Sanger and 454 sequencing are in bold type. b Plant material used for DNA extraction. c
Number of plants contained in DNA pools used for PCR amplification of the three ALS fragments, followed by 454 sequencing.
of 3 min at 95 ∘ C, followed by 37 cycles of 10 s at 95 ∘ C, 15 s at 60 ∘ C and 1 min at 72 ∘ C (amplicon encompassing codons 197 and 205) or by 40 cycles of 30 s at 95 ∘ C, 30 s at 57 ∘ C and 45 s at 72 ∘ C (amplicon encompassing codons 376 and 377) or by 37 cycles of 10 s at 95 ∘ C, 30 s at 60 ∘ C and 45 s at 72 ∘ C (amplicon encompassing codons 574 to 654). All primers were used at 0.2 μM
final concentration each. Primer specificity was checked using PCR followed by agarose gel electrophoresis and amplicon Sanger sequencing. In order to reduce the workload necessary to seek mutations endowing resistance in a given population, PCRs were performed on pools of DNA samples rather than on individual DNA samples.
Table 2. 3′ ALS-specific sequences of the primers used for PCR
Primer codea
Expected amplicon size (bp)
ALS codons of interest targetedc
Primer sequence (5′ –3′ )
Targetb
ALE-A ALE-AR
CATGGTCGCCATCACCGGCCAGG ATCTGCTGCTGGATGTCCTTGGG
276–298 497–475
222
197, 205
ALE-B ALE-BR
GTTCTTCCTCGCCTCCTCTGG CTGGATCAATATCAATGTGCACAAT
426–446 907–883
482
376, 377
ALE-C ALE-CR
GCTGCTGGTGCTGCTGTGGC ACACGGTCCTGCCATCACCATCC
1270–1289 1720–1689
451
574, 653, 654
a
Pest Manag Sci 2015; 71: 675–685
© 2014 Society of Chemical Industry
wileyonlinelibrary.com/journal/ps
677
Codes in bold type, ‘Primer A-lib L’ + MID sequence added at the 5′ end of the primer sequence for 454 sequencing. ‘Primer B-lib L’ sequence was added to the 5′ end of the second primer in each primer pair. b Nucleotides in GenBank/EMBL accession JQ319776 (E. crus-galli ALS sequence). c Numbered after Arabidopsis thaliana ALS sequence (GenBank/EMBL accession X51514).
www.soci.org Pools were prepared by mixing identical volumes of individual DNA samples. The 40 DNA samples analysed in each of the 28 barnyard grass populations were divided into ten pools of four DNA samples each, and two independent PCRs were performed per pool and per amplicon. In two populations (Table 1), a single pool was also obtained from the 40 DNA samples to assess the possibility of further reduction in the resistance diagnosis workload. In this case, six independent PCRs obtained from the same DNA sample using the same primer pair were pooled. One specific MID was used for each population, except for the two populations in which two types of amplicon pool were tested; in each of these populations, one specific MID was used per type of amplicon pool (Table 1). All PCRs obtained from the same DNA pool using the same primer pair were pooled, and 1 μL of each PCR pool was visualised by agarose gel electrophoresis. Amplicon pools were then prepared by mixing similar amounts of amplicons obtained from each DNA pool on the basis of visual assessment of the respective amplicon intensities on agarose gels. One amplicon pool contained amplicons obtained from all DNA pools in the same population using one primer pair; three amplicon pools were thus obtained for each of 26 populations, and six amplicon pools were obtained for each of the two remaining populations for which two types of DNA pool were analysed (Table 1). 2.3 454 sequencing Amplicons in each amplicon pool were purified separately using Agencourt Ampure XP kit (Beckman Coulter, Pasadena, CA) and subsequently quantified with an Infinite 200 (Tecan, Männedorf, Switzerland) fluorimeter using the Quant-iT PicoGreen dsDNA Assay kit (Invitrogen, Carlsbad, CA). Amplicon quality was checked using a TapeStation system (Agilent, Santa Clara, CA) with DNA1000 kit (Agilent). All amplicons were then mixed in an equimolar pool. The equimolar pool was amplified to generate an amplicon library using the GS Junior Titanium emPCR kit (Lib-A) (Roche) and following the manufacturer’s instructions. 454 sequencing was subsequently performed on a 454 GS Junior v.2.7 system (Roche) using the GS Junior Titanium Sequencing kit (Roche). This configuration was expected to deliver approximately 70 000 sequence reads. The quality of the sequence reads was checked using the default parameters of the Run Browser tool in GS Data Analysis software (Roche). Sequence reads shorter than 40 nucleotides or not containing a ‘Primer A-lib L’ + MID sequence were filtered out. The reads associated with each barnyard grass population analysed were separated after their specific MID sequence using the GS reference Mapper tool included in the GS Data Analysis software (Roche), yielding one standard flowgram format file (.sff, a binary file format used to encode the results of 454 sequencing from Roche platforms) for each sample analysed. One fasta file including the nucleotide sequences of all reads (.fasta) and one fasta quality file including the sequence quality information for every nucleotide in each read sequence (.qual) were generated from each .sff file using GS reference Mapper. 2.4 Procedure for sequence analysis The entire procedure is summarised in Fig. 1.
678
2.4.1 Filtering sequence reads In a first step, Prinseq 0.20.342 (accessible at http://edwards. sdsu.edu/cgi-bin/prinseq/prinseq.cgi) was used to eliminate
wileyonlinelibrary.com/journal/ps
C Délye et al.
poor-quality sequence reads. The .fasta and .qual files corresponding to a given population were uploaded, and the ‘basic statistics’ option was selected, which allowed visualisation of sequence read length and GC content distribution, and sequence read quality data (occurrence of ambiguous base calls, base quality distribution). In the next step, the input sequence data were processed by filtering out sequence reads containing more than one ambiguous base call (Ambiguity-code-related filters, maximum number of allowed Ns = 1) and trimming poor-quality sequence at the 5′ ends of the reads (trim ends by quality scores, quality score threshold for 5′ end = 22). Trimming was implemented because the sequence quality data analysis performed in the previous step revealed poor 5′ end sequence quality for some of the sequence reads obtained. In this work, the amplicons analysed were in two very distinct size classes (i.e. 222 bp and 451 and 482 bp) (Table 2). The sequence reads were thus first filtered by length. Setting the length range to 150–250 bp generated a fasta file containing sequences mostly expected to correspond to the 222 bp amplicon, while setting the length range to 400–485 bp generated a fasta file containing the sequences expected to correspond to the other two amplicons (451 and 482 bp). 2.4.2 Aligning sequence reads The sequences in the fasta files generated in the previous step were aligned using Clustal Omega43 (accessible at http://www. ebi.ac.uk/Tools/msa/clustalo). The number of combined iterations was set to 1 to decrease the time required for computation, and the selected output format was fasta. A Clustal Omega output file contains a global alignment of all input sequences where sequences are grouped according to similarity. In the present case, in the output file containing the sequence reads mostly expected to correspond to the 222 bp amplicon, the authors expected a major cluster of sequence reads corresponding to this amplicon and minor clusters including flawed reads corresponding to incomplete sequences of the other two longer amplicons. In the output file containing the sequence reads expected to correspond to the two longest amplicons, the authors expected two major clusters of sequence reads, each one corresponding to one of the longest amplicons, and possibly minor clusters of flawed sequence reads. The rough alignments in the Clustal Omega output files were visualised with BioEdit44 (available at http://www.mbio.ncsu.edu/bioedit/bioedit.html). Sequences were selected, copied and pasted using BioEdit. Each read cluster containing all good-quality sequences corresponding to one amplicon were combined in a single fasta file. The ‘–’ symbols inserted in the sequence reads to compensate for gaps in the overall alignment were removed using a text editor with an automated character string replacement tool (Word 2010, Microsoft). A total of three fasta files containing the unaligned quality sequence runs obtained for each amplicon sequenced was thus obtained for each barnyard grass population studied. 2.4.3 Identifying and quantifying mutations endowing resistance The number of reads containing the wild-type sequence of the codon of interest was determined in each population for each codon investigated (197, 205, 376, 377, 574, 653 and 654, standardised to A. thaliana ALS sequence) by seeking a short sequence including the wild-type codon sequence in the fasta files containing the unaligned quality sequence runs, using a text editor with an automated character string research function displaying the
© 2014 Society of Chemical Industry
Pest Manag Sci 2015; 71: 675–685
Pesticide resistance diagnosis using next-generation sequencing
www.soci.org
Crude DNA extracts DNA extract pools PCR Amplicon ‘equimolar’pools (one per population) NGS (454 sequencing) Sequence reads One .fasta+ one .qual files (or one .fastaq file) per population Prinseq -Eliminates poor-quality sequence reads -Sorts reads by size (if working on amplicons with different sizes) Quality sequence reads in .fasta files (one file per amplicon size class and per population) Clustal Omega -Roughly aligns all sequence reads in a single, overall alignment in a .fasta file -Clusters sequence reads by sequence similarity Roughly aligned quality sequence reads in .fasta files (one file per amplicon size class and per population) BioEdit -Pastes aligned sequence reads in one .fasta file per amplicon sequence Roughly aligned quality sequence reads in .fasta files (one file per amplicon and per population) Text editor (e.g. Microsoft Word) -Eliminates the gap marks introduced in the sequences of the reads by Clustal Omega when generating the alignments Quality sequence reads in .fasta files (one file per amplicon and per population) Text editor (e.g. Microsoft Word) -Searches for short sequence stretches including a variant of a codon of interest -Counts the occurrence of each codon variant Frequencies of mutant and wild-type alleles
Figure 1. Flowchart of the NGS data analysis procedure.
number of occurrences detected. The following sequences were sought (the codon of interest is underlined): AGGTGCCC (codon 197), GCCTTCCAG (codon 205), TCGATGAT (codon 376), CGTGTGAC (codon 377), TGGGAGG (codon 574), TCCCGAGC (codon 653) and GGTGGCGC (codon 654). When the number of reads containing the wild-type sequence of a given codon was lower than the total number of reads in the fasta file, sequences containing one of the possible nucleotide changes at the codon in question were successively sought until the total number of reads containing wild-type and mutant sequences equalled the total number of reads.
Pest Manag Sci 2015; 71: 675–685
2.6 Confirmation of barnyard grass species Two barnyard grass species infest rice crops in Europe:45 the hexaploid barnyard grass (E. crus-galli) and rice barnyard grass (Echinochloa phyllopogon = Echinochloa oryzicola), a tetraploid species closely related to E. crus-galli. These species are rather difficult to distinguish on the basis of morphological traits.45,46 The nucleotide sequences encoding ALS have been determined in both barnyard grass37,38 and rice barnyard grass.47 They are 99% identical between the two species, which does not enable species identification. PCR-RFLP assays targeting chloroplastic DNA sequences, allowing the differentiation of barnyard grass and rice barnyard grass,46 were thus used to assign the plants in each population to a species.
© 2014 Society of Chemical Industry
wileyonlinelibrary.com/journal/ps
679
2.5 Checking 454 sequencing results using Sanger sequencing The present 454 sequencing experiment was designed to analyse each barnyard grass population as a bulk of 40 plants, and to provide frequencies of mutant ALS alleles in each population. Therefore, sequence data were not recovered for individual plants. Seven populations were used to check the reliability of the frequencies of mutant ALS alleles assessed using 454 sequencing (Table 1). In each of these populations, the amplicon(s) analysed
were generated using primers in Table 2 from each of the 40 individual DNA samples that were extracted from field plant leaves (three populations) or from seedlings (four populations) (Table 1). Each amplicon was obtained from at least three independent PCRs from each DNA sample, and subsequently sequenced on both strands using Sanger sequencing.
www.soci.org
3
RESULTS
3.1 Primer specificity In spite of several attempts, no amplification of a fragment of the ALS gene carrying codon 122 that is involved in herbicide resistance8,40 could be obtained. Difficulties in amplifying the ALS 5′ end had previously been reported.37 Codon 122 was therefore not investigated herein. Specific amplification of the expected amplicon was confirmed for the three primer pairs shown in Table 2 and for the derived primers intended for PCR followed by 454 sequencing that allowed amplicons labelled with MID01 to be generated (not shown). Primer specificity is determined by the primer 3′ gene-specific sequence (corresponding in the present case to the ALS-specific primer sequences displayed in Table 2). The extra sequence added to the 5′ end of the primers intended for PCR followed by 454 sequencing only serves for 454 sequencing and sequence read indexing, and does not influence primer specificity. As the only differences between all primer pairs intended for PCR followed by 454 sequencing resided in the MID sequence at the 5′ end of one primer in each pair, the results obtained for the three MID01 primer pairs were considered to be extrapolatable to the other primer pairs. 3.2 Species identification Species-specific PCR-RFLP assays revealed that three of the 28 populations investigated only contained barnyard grass plants, while the remaining 25 populations only contained rice barnyard grass plants (Table 3).
680
3.3 454 sequencing results Barnyard grass is a hexaploid species, while rice barnyard grass is tetraploid.48 There is one copy of the ALS gene with two alleles per barnyard grass genome, i.e. a total of six or four ALS copies in a hexaploid or tetraploid species respectively. Thus, 240 or 160 copies of the ALS gene are expected to be present in each of the 40-plant populations of barnyard grass or rice barnyard grass that were investigated, respectively. Overall, a total of 87 333 sequence reads were obtained for the 30 DNA pools (Table 1) sequenced. After filtering, 63 959 reads were available to seek resistance-endowing mutations (Table 3). The number of quality sequence reads that could be analysed per DNA pool ranged from 1175 to 4429 (Table 3), with an average value of 2132. The number of quality sequence reads that could be analysed per DNA pool for amplicons carrying codons 197 and 205, codons 376 and 377 and codons 574, 653 and 654 ranged from 206 to 2347, from 308 to 1359 and from 197 to 2979 respectively (Table 3), with average values of 2027, 526 and 579 respectively. This means that, under the hypothesis of homogeneous distribution of the sequence reads among all plants in one DNA pool, amplicons carrying codons 197 and 205, codons 376 and 377 or codons 574, 653 and 654 were sequenced with an average sequencing depth of 4.6×, 2.4× or 1.7× in barnyard grass respectively. They were sequenced with an average sequencing depth of 6.2×, 3.2× or 3.5× in rice barnyard grass respectively The sequencing depth per amplicon was never below 1.0× (Table 3). PCR amplification of DNA and the NGS technologies themselves generate sequence errors by nucleotide misincorporation or misdentification.29,30 Thus, only nucleotide substitutions detected with a frequency of >0.4% or >0.6% were considered to be relevant in barnyard grass or in rice barnyard grass respectively. These thresholds were based on the expected frequency of one nucleotide substitution present in the heterozygous state in one
wileyonlinelibrary.com/journal/ps
C Délye et al.
out of 40 hexaploid barnyard grass individuals (i.e. one mutant copy out of 240) or in one out of 40 tetraploid rice barnyard grass individuals (i.e. one mutant copy out of 160) respectively. 3.4 Identification of mutations endowing resistance The search for mutations at the seven ALS codons investigated identified non-synonymous mutations in nine of the 28 populations investigated (Table 3). The mutations observed were a C-to-T transition at the first position in codon 197 (CCC to TCC) that caused a Pro-to-Ser replacement (detected in nine populations and in both species), a C-to-T transition at the second position in codon 197 (CCC to CTC) that caused a Pro-to-Leu replacement (detected in two populations and in both species) and a C-to-T transition at the second position in codon 574 (TGG to TTG) that caused a Trp-to-Leu replacement (detected in one rice barnyard grass population) (Table 3). The Ser197 ALS allele was by far the most frequently observed mutant allele (2366 sequence reads in nine populations). The other two mutant ALS alleles were observed with a much lower number of sequence reads (150 and eight sequence reads for the Leu197 and Leu574 alleles respectively). Populations containing these alleles also contained the Ser197 allele (Table 3). The PCRs used to generate the amplicons intended for 454 sequencing were performed using pools of DNA samples obtained by mixing the same volume of several DNA samples extracted using a rapid, crude protocol. Ten pools of four DNA samples were used for each population, and an additional single pool including all 40 DNA samples was also used for two populations (populations ECH21 and PRIV) (Table 1). No differences in the frequency of mutant ALS alleles detected in the two populations in question were observed when starting from the ten four-sample pools or from the single 40-sample pool (Table 3). 3.5 Checking 454 sequencing results using Sanger sequencing In seven of the populations where mutant ALS alleles had been detected (Table 3), Sanger sequencing of the amplicon(s) carrying the mutation(s) identified was performed for each plant individually, starting from individual DNA samples. The amplicon carrying codons 197 and 205 was sequenced in all 280 plants from these seven populations, and the amplicon carrying codons 574 to 654 was also sequenced in all 40 plants from population ECH19. The presence of mutations at codons 197 and 205 and the presence of mutations at codons 574, 653 and 654 in population ECH19 were visually checked in the sequence chromatograms. In every population analysed, the occurrence of all mutant ALS alleles detected by 454 sequencing was confirmed by Sanger sequencing (Table 3). In the barnyard grass population SEY01, all plants carrying Leu197 ALS also carried Ser197 ALS, while in the rice barnyard grass population ECH19, Ser197 and Leu574 ALS alleles were carried by different plants. No additional mutant ALS alleles could be detected in the seven populations analysed. 454 sequencing yields a single set of sequences for a whole population, and individual sequence reads cannot be attributed to individual plants. Conversely, Sanger sequencing yields a single ‘consensus’ sequence resulting from the superimposition of the sequences of the individual DNA fragments sequenced. On the sequencing chromatograms, superimposed peaks are observed at any variable nucleotide position. In the case of polyploid species, when superimposed peaks are observed, it is not possible reliably to determine how many genomic copies of the sequence
© 2014 Society of Chemical Industry
Pest Manag Sci 2015; 71: 675–685
Pesticide resistance diagnosis using next-generation sequencing
www.soci.org
Table 3. 454 sequencing results Number of reads Number of reads containing a mutant codon (%)c
Corresponding to amplicons carrying codons Populationa
MID
197–205b
376–377b
574–654b
Total
Ser197
Leu197
Leu574
Echinochloa crus-gallid FDP
MID21
1022 (4.3×)
469 (2.0×)
228 (1.0×)
1719
0 (0.0)
0 (0.0)
0 (0.0)
SEY01
MID22
1140 (4.8×)
539 (2.2×)
499 (2.1×)
2178
305 (26.8) (34/40)e
70 (6.1) (9/40)e
0 (0.0)
SARR
MID23
1119 (4.7×)
699 (2.9×)
495 (2.1×)
2313
0 (0.0)
0 (0.0)
0 (0.0)
Echinochloa phyllopogond ECH01
MID01
846 (5.3×)
416 (2.6×)
203 (1.3×)
1465
0 (0.0)
0 (0.0)
0 (0.0)
ECH03
MID02
1053 (6.6×)
400 (2.5×)
290 (1.8×)
1743
0 (0.0)
0 (0.0)
0 (0.0)
ECH04
MID03
970 (6.1×)
499 (3.1×)
316 (2.0×)
1785
0 (0.0)
0 (0.0)
0 (0.0)
ECH05
MID04
708 (4.4×)
389 (2.4×)
338 (2.1×)
1435
0 (0.0)
0 (0.0)
0 (0.0)
ECH06
MID05
206 (1.3×)
590 (3.7×)
667 (4.2×)
1463
0 (0.0)
0 (0.0)
0 (0.0)
ECH07
MID06
1166 (7.3×)
542 (3.4×)
453 (2.8×)
2161
0 (0.0)
0 (0.0)
0 (0.0)
ECH08
MID07
1198 (7.5×)
1395 (8.7×)
504 (3.2×)
3097
0 (0.0)
0 (0.0)
0 (0.0)
ECH09
MID08
1121 (7.0×)
329 (2.1×)
378 (2.4×)
1828
0 (0.0)
0 (0.0)
0 (0.0)
ECH10
MID10
1259 (7.9×)
464 (2.9×)
683 (4.3×)
2406
0 (0.0)
0 (0.0)
0 (0.0)
ECH12
MID11
1133 (7.1×)
315 (2.0×)
458 (2.9×)
1906
0 (0.0)
0 (0.0)
0 (0.0)
ECH14
MID13
1831 (11.4×)
330 (2.1×)
458 (2.9×)
2619
929 (50.7) (40/40)e
0 (0.0)
0 (0.0)
ECH15
MID14
1301 (8.1×)
660 (4.1×)
650 (4.1×)
2611
0 (0.0)
0 (0.0)
0 (0.0)
ECH16
MID15
2347 (14.7×)
1046 (6.5×)
1036 (6.5×)
4429
316 (13.5)
75 (3.2)
0 (0.0)
ECH17
MID16
1107 (6.9×)
594 (3.7×)
417 (2.6×)
2118
0 (0.0)
0 (0.0)
0 (0.0)
ECH18
MID17
1259 (7.9×)
464 (2.9×)
432 (2.7×)
2155
82 (6.5) (7/40)e
0 (0.0)
0 (0.0)
ECH19
MID18
967 (6.0×)
448 (2.8×)
383 (2.4×)
1798
38 (3.9) (5/40)e
0 (0.0)
8 (2.1) (2/40)e
ECH20
MID19
897 (5.6×)
423 (2.6×)
2979 (18.6×)
4299
0 (0.0)
0 (0.0)
0 (0.0)
ECH21
MID20
1104 (6.9×)
495 (3.1×)
697 (4.4×)
2296
0 (0.0) (0/40)e
0 (0.0)
0 (0.0)
MID32f
787 (4.9×)
668 (4.2×)
603 (3.8×)
2058
0 (0.0)
0 (0.0)
0 (0.0)
MID24
712 (4.5×)
493 (3.1×)
713 (4.5×)
1918
26 (3.7) (4/40)e
0 (0.0)
0 (0.0)
MID31f
685 (4.3×)
508 (3.2×)
436 (2.7×)
1629
23 (3.4)
0 (0.0)
0 (0.0)
PRIV BRU01
MID25
659 (4.1×)
319 (2.0×)
197 (1.2×)
1175
282 (42.8)
0 (0.0)
0 (0.0)
BRU02
MID26
808 (5.1×)
516 (3.2×)
672 (4.2×)
1996
358 (44.3)
0 (0.0)
0 (0.0)
SEY02
MID27
835 (5.2×)
400 (2.5×)
541 (3.4×)
1776
7 (0.8) (1/40)e
0 (0.0)
0 (0.0)
GOU01
MID28
631 (3.9×)
568 (3.6×)
521 (3.3×)
1720
0 (0.0)
0 (0.0)
0 (0.0)
GOU02
MID29
1306 (8.2×)
504 (3.2×)
419 (2.6×)
2229
0 (0.0)
0 (0.0)
0 (0.0)
GOU03
MID30
621 (3.9×)
308 (1.9×)
705 (4.4×)
1634
0 (0.0)
0 (0.0)
0 (0.0)
30 798
15 790
17 371
63 969
2366
145
8
Total a
Populations used for comparison between Sanger and 454 sequencing are in bold type. Numbered after Arabidopsis thaliana ALS sequence (GenBank/EMBL accession X51514). Numbers in parentheses represent the sequencing coverage depth for the amplicon in the population considered (i.e. the average number of times the region corresponding to the amplicon is expected to have been sequenced in each of the 160 or 240 ALS gene copies present in each population). c Percentage of the 454 sequence reads containing the mutant codon Ser197, Leu197 or Leu574 relative to the total number of sequence reads corresponding to the amplicon carrying codon 197 or 574 in a given population. d Species identified using species-specific molecular markers.45 e Number in italics represents the number of plants carrying the mutation in the population, as identified by Sanger sequencing. f 454 sequencing from PCRs performed on a single pool of 40 DNA samples. b
Pest Manag Sci 2015; 71: 675–685
carrying a mutant allele were homozygous for this allele in one of their genomes, i.e. they carried two copies of the mutant ALS allele in one of their genomes. Expected proportions of mutant ALS alleles as a function of the number of mutant plants identified by Sanger sequencing were computed for a hexaploid and for a tetraploid species (Fig. 2). For the seven populations considered, the observed proportion of each mutant allele in a population plotted as a function of the number of plants carrying two
© 2014 Society of Chemical Industry
wileyonlinelibrary.com/journal/ps
681
investigated contain each of the nucleotides observed at the variable position. In accordance with barnyard grass species being polyploid,48 superimposed peaks were observed at the variable nucleotide positions for all mutant plants identified. To compare Sanger and 454 results, the number of mutant plants identified using Sanger sequencing was converted into an expected proportion of mutant ALS alleles in each population of 40 plants. As barnyard grass species are selfing,48,49 it was postulated that plants
www.soci.org mutant alleles in the population could be considered to match that expected for a hexaploid species (population SEY01) or for a tetraploid species (the other six populations) (Fig. 2).
4
DISCUSSION
4.1 NGS-based identification of mutant ALS alleles in barnyard grass populations NGS technology was used to detect mutant ALS alleles in barnyard grass or in rice barnyard grass. The occurrence of plants carrying ALS alleles known to be involved in resistance to ALS-inhibiting herbicides was detected in nine of the populations investigated and confirmed by Sanger sequencing (Table 3). Nine plants in the barnyard grass population SEY01 were shown by Sanger sequencing to carry two types of mutant ALS allele (Leu197 and Ser197). The frequency of each mutant allele in the population measured by 454 sequencing was consistent with that expected for hexaploid plants carrying two copies of the mutant ALS allele in one of their three genomes (Fig. 2). The simplest explanation for this finding is that these nine plants all contain two Leu197 ALS in one of their genomes, and two Ser197 ALS in another genome. Similarly, the appearance of mutant alleles endowing herbicide resistance in different genomes and the accumulation of mutant alleles in different genomes in the same individual plant had previously been reported in the hexaploid, selfing weeds Avena sterilis50 and Avena fatua.51 The accumulation of mutant ALS alleles in several genomes of the same individual barnyard grass plant is likely to confer an increased resistance level to ALS-inhibiting herbicides.
C Délye et al.
The frequency of mutant ALS alleles assessed by 454 sequencing could be considered to match the number of plants carrying mutant ALS alleles identified by Sanger sequencing for a tetraploid species in all rice barnyard grass populations (Fig. 2). This shows that the PCR primers designed for NGS-based herbicide resistance diagnosis in barnyard grass can be reliably used on rice barnyard grass, which is useful considering that resistance to ALS-inhibiting herbicides has evolved in both species. Using these primers and NGS, three mutant ALS alleles were identified in this study (Table 3). Allele Leu574 had already been observed in both barnyard grass37 and rice barnyard grass.45 ALS alleles carrying an amino acid replacement at codon 122 had also been previously reported in barnyard grass,38 but their occurrence could not be investigated in the present work. Alleles Ser197 and Leu197 identified herein had not been previously reported in either species. No mutant ALS allele could be detected in 19 of the 28 populations investigated. It may be that ALS alleles carrying an amino acid replacement at codon 122 were present in these populations. Alternatively, non-ALS-based resistance could be present in these populations rather than mutant ALS alleles, or failure of barnyard grass control in the corresponding fields could have been due to improper herbicide applications. 4.2 Feasibility of NGS-based pesticide resistance diagnosis and key points In this work, the feasibility of NGS-based pesticide resistance diagnosis from pools of individuals was demonstrated for the first time, using the application of 454 sequencing to the diagnosis of ALS-based resistance to herbicides in barnyard grass as an example.
60
Percentage mutant ALS alleles
50
40
30
20
10
0
0
10 20 30 Number of plants carrying two mutant ALS alleles
40
682
Figure 2. Expected proportion of mutant alleles in a population of 40 plants as a function of the number of plants carrying two mutant alleles in the population. A selfing, polyploid species was considered (i.e. mutant plants carrying two mutant ALS alleles in one of their genomes). Open squares: tetraploid species; open circles: hexaploid species; solid or grey triangles: data observed in the present study where the percentage of mutant ALS alleles was determined by 454 sequencing and the number of plants carrying mutant ALS alleles was determined by Sanger sequencing; solid triangles: populations identified as barnyard grass (E. crus-galli); grey triangles: populations identified as rice barnyard grass (E. phyllopogon). Each type of mutant allele detected in a population is plotted individually [e.g. one of the solid triangles corresponds to the Leu197 and the other to the Ser197 ALS allele detected in population SEY01 (Table 3)].
wileyonlinelibrary.com/journal/ps
4.2.1 DNA sample pools The procedure used for analysing 28 (rice) barnyard grass populations was meant to be as basic as possible (Fig. 1). DNA was roughly extracted using a rapid procedure yielding non-quantifiable DNA solutions. PCRs were performed from pools constituted by mixing equal volumes of rough DNA solutions extracted from individual plants. ‘Equimolar’ pools of amplicons intended for NGS analysis were constituted on the basis of visual assessment of the respective amplicon concentrations on agarose electrophoresis gels. In spite of the absence of precise DNA or amplicon quantification, the results obtained were consistent with those yielded by Sanger sequencing of individual plants in terms of both mutation identification and mutant ALS allele quantification (Fig. 2, Table 3). Mutation detection and quantification were achieved with similar accuracy when starting from four pools of ten individual DNA extracts or from a single pool of 40 individual DNA extracts in one population (Table 3). Thus, increasing the number of samples in the DNA sample pools is clearly a feasible way to reduce the number of PCR reactions necessary for NGS-based resistance diagnosis. Pooling tissues of individual plants prior to extraction may also decrease the workload required for resistance diagnosis. Yet, as DNA extraction yields may be uneven among individuals, it may be a good option to extract DNA from similar amounts of similar tissue for all individuals in the same population, in order to obtain similar DNA concentrations for all individuals in the DNA pool used for PCR. 4.2.2 Sequencing and sequencing analysis procedure In this study, 454 sequencing was implemented for pesticide resistance diagnosis. It should be emphasised that any other NGS technology could have been used for this purpose, and that the
© 2014 Society of Chemical Industry
Pest Manag Sci 2015; 71: 675–685
Pesticide resistance diagnosis using next-generation sequencing choice of this particular NGS technology was only dictated by the equipment available at INRA. Not every group involved in pesticide resistance diagnosis may have access to in-house NGS facilities, but quite a few sequencing companies offer NGS services including basic analyses and are able to deliver files including the sequence reads corresponding to each population analysed or to each amplicon in each population analysed, at continuously decreasing costs.52 The analysis procedure implemented herein (Fig. 1) was kept to maximum simplicity and was uniquely based on programs with free access and no programming required so as to be usable by virtually any scientist with an interest in pesticide resistance diagnosis. Obviously, there are other programs that can be used for NGS data analysis (e.g. Muscle53 for aligning sequence reads). After mutations have been identified, it must be kept in mind that PCR and NGS can generate sequencing errors.29,30,52 For this purpose, a threshold was set for mutation detection that was based on the expected frequency of one nucleotide substitution present in the heterozygous state in one out of 40 hexaploid or tetraploid individuals, depending on the species considered. In the present experiments, a C-to-G transversion was observed at the first position in codon 197 (CCC to GCC) in ca 0.3% of the sequence reads in the rice barnyard grass populations ECH14 and ECH16. This transversion was not considered, because its frequency was below the threshold applied to identify relevant nucleotide substitutions in rice barnyard grass (>0.6% of the reads), and it was not observed by Sanger sequencing in these two populations. Conversely, the occurrence of a C-to-T transition observed at the first position in codon 197 (CCC to TCC) in 0.8% of the sequence reads in rice barnyard grass population SEY02 was confirmed by Sanger sequencing (Table 3). To avoid false positive detection, it is thus clearly necessary to set a frequency threshold below which the occurrence of a mutation will be considered to be a sequencing error. This threshold should be adapted to the NGS technology and to the species considered, taking into account its mating system, its ploidy level and the numbers of copies of the gene(s) targeted per individual. 4.2.3 Sequencing coverage depth Successful detection of resistance-endowing mutations was achieved in this work with a sequencing coverage depth ranging
www.soci.org from 1.7× to 18.6× according to the amplicon and the population (Table 3). In the present work there was a clear bias in sequencing coverage depth in favour of the shortest amplicon sequenced (amplicon carrying codons 197 and 205) (Table 3). Depending on the barnyard grass species, this 222-nucleotide amplicon had an average sequencing coverage depth of 6.2× or 4.6×, while the 482-nucleotide and 451-nucleotide amplicons had average sequencing coverage depths of 3.2× or 2.4× and 3.5× or 1.7× respectively. This bias most likely can be solved by reducing the concentration in the shortest amplicon in the sequencing mixes, or by sequencing amplicons with similar sizes. A sequencing coverage depth of only 2.4× allowed detection of a mutation present in two plants out of 40 in the rice barnyard grass population ECH19 (Table 3). Thus, a high sequencing coverage depth does not seem to be critical to detecting mutations in low frequencies. This is consistent with a previous study indicating that 1× sequencing coverage depth generates the most information about a population, and that the best strategy for optimising nucleotide polymorphism assessment in populations is to increase the sample size rather than the sequencing coverage depth.54 4.2.4 Workload Considering that the authors used a very crude and rapid DNA extraction procedure, the most time-consuming steps in this study were creating the amplicon pools and preparing the amplicons for 454 sequencing, which was performed in five and three working days respectively. This workload could clearly be decreased by analysing larger pools of DNA extracts, and by subcontracting the execution of the NGS. Sequence analysis using the simple procedure described (Fig. 1) was performed within less than two working days for the 28 populations studied (Table 1). NGS-based detection of mutant ALS alleles in the 28 populations studied as four pools of ten individuals required 672 individual PCRs. The present results indicate that using one pool of 40 plants per population is a viable option and would decrease the number of PCRs to 168 or 252, depending on whether two or three independent PCRs are performed for each amplicon and each pool. Analysing these 28 populations using PCR-based genotyping of codons involved in resistance in individual plants would require 7840 PCRs and, in the case of (d)CAPS assays, 7840 digestion
Table 4. Comparison of NGS with the techniques most frequently used for pesticide resistance diagnosis considering the detection of mutations at ALS codons 197, 205, 376, 377, 574, 653 and 654 in 40 individuals
Technique
Mininum number of PCRs
Basis for resistance diagnosis
PCR-based genotypinga One per individual and Amplicon pattern or per targeted codon restriction pattern (here, 7 × 40 = 280) Sanger sequencing Two per individual and Sequence of targeted per amplicon codons (here, 2 × 40 × 3 = 240) NGS Two per DNA pool and Sequence of targeted per amplicon codons (here, 2 × 3 = 6)c
Identification of amino acid substitution(s)
Identification of heterozygous plants
Number of resistant plants in population
Detection of new mutations
Yes/nob
Yes
Yes
No
Yes
Yes
Yes
Possible
Yes
No
Estimationd
Possible
a Allele-specific PCR, CAPS, dCAPS. b Depends on the technique and on the number of possible mutant alleles at a given codon.
Pest Manag Sci 2015; 71: 675–685
© 2014 Society of Chemical Industry
wileyonlinelibrary.com/journal/ps
683
c Considering a single pool of 40 individuals. d Estimation from the frequencies of the mutations detected, based on the ploidy level and the mating system of the species considered.
www.soci.org reactions, followed by agarose gel electrophoresis (see Table 4). Using Sanger sequencing of amplicons obtained from individual plants would require 6720 PCRs and 3360 Sanger sequencing runs (single-end sequencing). NGS-based resistance diagnosis thus clearly outperforms other molecular biology techniques used routinely for pesticide resistance diagnosis. An additional advantage of NGS-based diagnosis is obtaining sequence data, which can be useful in identifying new mutations potentially involved in resistance. Although in the present study browsing sequence alignments using BioEdit to seek mutations outside the targeted codons did not identify non-synonymous substitutions in frequencies above the detection threshold, this possibility can be useful to identify new mutations potentially involved in resistance.
5
CONCLUSIONS
This work illustrated the feasibility of implementing NGS technologies for the detection of mutations endowing pesticide resistance without the need for extensive training in NGS data analysis, and the benefit of using NGS diagnosis when large numbers of individuals are to be analysed. The authors’ work is expected to promote interest in the use of NGS technologies in pesticide resistance diagnosis. Currently, NGS technologies generate millions of sequencing runs with increasing individual read lengths52 (e.g. up to 25 million 300-nucleotide reads for the last evolution of Illumina’s MiSeq sequencer). NGS-based resistance diagnosis therefore opens up the possibility of rapidly analysing a tremendous number of individuals in populations of weeds, pests or pathogens, provided nucleotide polymorphisms at the basis of resistance have been identified beforehand. This will enable the detection of resistant individuals at very early stages in the evolution of resistance in the field, when resistance evolution can be hampered or even prevented by a rapid adaptation of the crop protection programme.55 In the near future, NGS-based resistance diagnosis or monitoring will most likely play a substantial role in sustaining crop protection strategies aimed at maintaining pesticide efficacy and optimising pesticide applications.
ACKNOWLEDGEMENTS This study was funded by France Agrimer (France) as grant SIVAL No. 2011-1995. The authors thank É Oudard (DRAAF/SRAL PACA, Montfavet, France) and C Thomas (Centre Français du Riz, Arles, France) for providing the barnyard grass populations.
REFERENCES
684
1 Oerke E-C, Crop losses to pests. J Agric Sci 144:31–43 (2006). 2 Délye C, Jasieniuk M and Le Corre V, Deciphering the evolution of herbicide resistance in weeds. Trends Genet 29:649–658 (2013). 3 Ma Z and Michailides TJ, Advances in understanding molecular mechanisms of fungicide resistance and molecular detection of resistant genotypes in phytopathogenic fungi. Crop Prot 24:853–863 (2005). 4 ffrench-Constant R, The molecular genetics of insecticide resistance. Genetics 194:807–815 (2013). 5 Furuya S, Mochizuki M, Saito S, Kobayashi H, Takayanagi T and Suzuki S, Monitoring of Qol fungicide resistance in Plasmopara viticola populations in Japan. Pest Manag Sci 66:1268–1272 (2010). 6 Beckie HJ, Leeson JY, Thomas AG, Brenzil CA, Hall LM, Holzgang G. et al, Weed resistance monitoring in the Canadian prairies. Weed Technol 22:530–543 (2008).
wileyonlinelibrary.com/journal/ps
C Délye et al.
7 Slater R, Ellis S, Genay J-P, Heimbach U, Huart G, Sarazin M. et al, Pyrethroid resistance monitoring in European populations of pollen beetle (Meligethes spp.): a coordinated approach through the Insecticide Resistance Action Committee (IRAC). Pest Manag Sci 67:633–638 (2011). 8 Beckie H and Tardif FJ, Herbicide cross resistance in weeds. Crop Prot 35:15–28 (2012). 9 Corbett C-AL and Tardif FJ, Detection of resistance to acetolactate synthase inhibitors in weeds with emphasis on DNA-based techniques: a review. Pest Manag Sci 62:584–597 (2006). 10 Délye C, Laigret F and Corio-Costet MF, A mutation in the 14𝛼-demethylase gene of Uncinula necator that correlates with resistance to a sterol biosynthesis inhibitor. Appl Environ Microbiol 63:2966–2970 (1997). 11 Frey JE, Muller-Scharer H, Frey B and Frei D, Complex relation between triazine-susceptible phenotype and genotype in the weed Senecio vulgaris may be caused by chloroplast DNA polymorphism. Theor Appl Genet 99:578–586 (1999). 12 Berticat C, Dubois M-P, Marquine M, Chevillon C and Raymond M, A molecular test to identify resistance alleles at the amplified esterase locus in the mosquito Culex pipiens. Pest Manag Sci 56:727–731 (2000). 13 Délye C, Matéjicek A and Gasquez J, PCR-based detection of resistance to Acetyl-CoA carboxylase-inhibiting herbicides in black-grass (Alopecurus myosuroides Huds) and ryegrass (Lolium rigidum Gaud). Pest Manag Sci 58:474–478 (2002). 14 Délye C, Zhang X-Q, Chalopin C, Michel S and Powles SB, An isoleucine residue within the carboxyl-transferase domain of multidomain acetyl-coenzyme A carboxylase is a major determinant of sensitivity to aryloxyphenoxypropionate but not to cyclohexanedione inhibitors. Plant Physiol 132:1716–1723 (2003). 15 Malandrakis AA, Markoglou AN and Ziogas BN, PCR-RFLP detection of the E198A mutation conferring resistance to benzimidazoles in field isolates of Monilinia laxa from Greece. Crop Prot 39:11–17 (2012). 16 Low VL, Chen CD, Lim PE, Lee HL, Lim YAL, Tan TK. et al, First molecular genotyping of insensitive acetylcholinesterase associated with malathion resistance in Culex quinquefasciatus Say populations in Malaysia. Pest Manag Sci 69:1362–1368 (2013). 17 Stenhouse SA, Plernsub S, Yanola J, Lumjuan N, Dantrakool A, Choochote W. et al, Detection of the V1016G mutation in the voltage-gated sodium channel gene of Aedes aegypti (Diptera: Culicidae) by allele-specific PCR assay, and its distribution and effect on deltamethrin resistance in Thailand. Parasite Vector 6:253 (2013). 18 Yin YN and Xiao CL, Molecular characterization and a multiplex allele-specific PCR method for detection of thiabendazole resistance in Penicillium expansum from apple. Eur J Plant Pathol 136:703–713 (2013). 19 Kaundun SS and Windass JD, Derived cleaved amplified polymorphic sequence, a simple method to detect a key point mutation conferring acetyl CoA carboxylase inhibitor herbicide resistance in grass weeds. Weed Res 46:34–39 (2006). 20 Délye C, Pernin F and Michel S, ‘Universal’ PCR assays detecting mutations in acetyl-coenzyme A carboxylase or acetolactate-synthase that endow herbicide resistance in grass weeds. Weed Res 51:353–362 (2011). 21 Kaundun SS, Cleere SM, Stanger CP, Burbidge JM and Windass JD, Real-time quantitative PCR assays for quantification of L1781 ACCase inhibitor resistance allele in leaf and seed pools of Lolium populations. Pest Manag Sci 62:1082–1091 (2006). 22 Chen CJ, Zhao W, Lu YJ, Wang JX, Chen Y, Li HX. et al, High-throughput detection of highly benzimidazole-resistant allele E198A with mismatch primers in allele-specific real-time polymerase chain reaction. Pest Manag Sci 65:413–419 (2009). 23 Menchari Y, Camilleri C, Michel S, Brunel D, Dessaint F, Le Corre V. et al, Weed response to herbicides: regional-scale distribution of herbicide resistance alleles in the grass weed Alopecurus myosuroides. New Phytol 171:861–874 (2006). 24 Délye C, Michel S, Bérard A, Chauvel B, Brunel D, Guillemin J-P. et al, Geographical variation in resistance to acetyl-coenzyme A carboxylase-inhibiting herbicides across the range of the arable weed Alopecurus myosuroides (black-grass). New Phytol 186:1005–1017 (2010). 25 Billard A, Laval V, Fillinger S, Leroux P, Lachaise H, Beffa R. et al, The allele-specific probe and primer amplification assay, a new real-time
© 2014 Society of Chemical Industry
Pest Manag Sci 2015; 71: 675–685
Pesticide resistance diagnosis using next-generation sequencing
26
27
28
29 30 31
32
33
34
35 36
37 38
39
PCR method for fine quantification of single-nucleotide polymorphisms in pooled DNA. Appl Environ Microbiol 78:1063–1068 (2012). Sarkar M, Baruah I, Srivastava RB, Borkotokic A and Bhattacharyyad IK, High-throughput approach to detection of knockdown resistance (kdr) mutation in mosquitoes, Culex quinquefasciatus, based on real-time PCR using single-labelled hybridisation probe/melting curve analysis. Pest Manag Sci 67:156–161 (2011). Chung I-H, Kang S, Kim YR, Kim J-H, Jung JW, Lee S. et al, Development of a low-density DNA microarray for diagnosis of target-site mutations of pyrethroid and organophosphate resistance mutations in the whitefly Bemisia tabaci. Pest Manag Sci 67:1541–1548 (2011). Marshall R, Hanley SJ, Hull R and Moss SR, The presence of two different target-site resistance mechanisms in individual plants of Alopecurus myosuroides Huds., identified using a quick molecular test for the characterisation of six ALS and seven ACCase SNPs. Pest Manag Sci 69:727–737 (2013). Metzker ML, Sequencing technologies – the next generation. Nat Rev Genet 11:31–46 (2010). Ward JA, Ponnola L and Weber CA, Strategies for transcriptome analysis in nonmodel plants. Am J Bot 99:267–276 (2012). De Brakeleer S, De Greve J, Lissens W and Teugels E, Systematic detection of pathogenic Alu element insertions in NGS-based diagnostic screens: the BRCA1/BRCA2 example. Hum Mutat 34:785–791 (2013). Shanks ME, Downes SM, Copley RR, Lise S, Broxholme J, Hudspith KAZ. et al, Next-generation sequencing (NGS) as a diagnostic tool for retinal degeneration reveals a much higher detection rate in early-onset disease. Eur J Hum Genet 21:274–280 (2013). Lupo A, Papp-Wallace KM, Sendi P, Bonomo RA and Endimiani A, Non-phenotypic tests to detect and characterize antibiotic resistance mechanisms in Enterobacteriaceae. Diagn Micr Infect Dis 77:179–194 (2013). Sahoo MK, Lefterova MI, Yamamoto F, Waggoner JJ, Chou S, Holmes SP. et al, Detection of cytomegalovirus drug resistance mutations by Next-Generation Sequencing. J Clin Microbiol 51:3700–3710 (2013). Ellison MT, Fischinger F, Zemetra RS and Kuhl JC, Use of pyrosequencing technology to genotype imidazolinone-tolerant wheat. Crop Sci 53:2021–2027 (2013). Collavo A, Strek H, Beffa R and Sattin M, Management of an ACCase-inhibitor-resistant Lolium rigidum population based on the use of ALS inhibitors: weed population evolution observed over a 7 year field-scale investigation. Pest Manag Sci 69:200–208 (2013). Panozzo S, Scarabel L, Tranel PJ and Sattin M, Target-site resistance to ALS inhibitors in the polyploid species Echinochloa crus-galli. Pestic Biochem Physiol 105:93–101 (2013). Riar DS, Norsworthy JK, Srivastava V, Nandula V, Bond JA and Scott RC, Physiological and molecular basis of acetolactate synthaseinhibiting herbicide resistance in barnyardgrass (Echinochloa crus-galli). J Agric Food Chem 61:278–289 (2013). Délye C, Boucansaud K, Pernin F and Le Corre V, Variation in the gene encoding acetolactate-synthase in Lolium species and proac-
www.soci.org
40 41
42 43 44 45
46
47
48 49 50
51
52 53 54 55
tive detection of mutant, herbicide-resistant alleles. Weed Res 49:326–336 (2009). Tranel PJ, Wright TR and Heap IM, Mutations in Herbicide-resistant Weeds to ALS Inhibitors. [Online]. Available: http://www.weedscience.com [11 January 2014]. Kati V, Le Corre V, Michel S, Jaffrelo L, Poncet C and Délye C, Isolation and characterisation of 11 polymorphic microsatellite markers in Papaver rhoeas L. (corn poppy), a major annual plant species from cultivated areas. Int J Mol Sci 14:470–479 (2013). Schmieder R and Edwards R, Quality control and preprocessing of metagenomic datasets. Bioinformatics 27:863–864 (2011). Sievers F, Wilm A, Dineen D, Gibson TJ, Karplus K, Li W. et al, Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol 7:539 (2011). Hall TA, BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucl Acids Symp Ser 41:95–98 (1999). Kaloumenos NS, Chatzilazaridou SL, Mylona PV, Polidoros AN and Eleftherohorinos IG, Target-site mutation associated with cross-resistance to ALS-inhibiting herbicides in late watergrass (Echinochloa oryzicola Vasing.). Pest Manag Sci 69:865–873 (2013). Yasuda K, Yano A, Nakayama Y and Yamaguchi H, Molecular identification of Echinochloa oryzicola Vasing. and E. crus-galli (L.) Beauv. using a polymerase chain reaction–restriction fragment length polymorphism technique. Weed Biol Manag 2:11–17 (2002). Iwakami S, Uchino A, Watanabe H, Yamasue Y and Inamura T, Isolation and expression of genes for acetolactate synthase and acetyl-CoA carboxylase in Echinochloa phyllopogon, a polyploid weed species. Pest Manag Sci 68:1098–1106 (2012). Mulligan GA, Chromosome numbers of Canadian weeds. Can J Bot 39:1057–1066 (1961). Maun MA and Barrett SCH, The biology of Canadian weeds. 77. Echinochloa crus-galli L. (Beauv.). Can J Plant Sci 66:739–759 (1986). Liu W, Harrison DK, Chalupska D, Gornicki P, O’Donnell CC, Adkins SW. et al, Single-site mutations in the carboxyltransferase domain of plastid acetyl-CoA carboxylase confer resistance to grass-specific herbicides. Proc Natl Acad Sci USA 104:3627–3632 (2007). Yu Q, Ahmad-Hamdani MS, Han H, Christoffers MJ and Powles SB, Herbicide resistance-endowing ACCase gene mutations in hexaploid wild oat (Avena fatua): insights into resistance evolution in a hexaploid species. Heredity 110:220–231 (2013). Thudi M, Li Y, Jackson SA, May GD and Varshney RK, Current state-of-art of sequencing technologies for plant genomics research. Brief Funct Genom 11:3–11 (2012). Edgar RC, MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucl Acids Res 32:1792–1797 (2004). Buerkle CA and Gompert Z, Population genomics based on low coverage sequencing: how low should we go? Mol Ecol 22:3028–3035 (2013). Neve P, Vila-Aiub M and Roux F, Evolutionary-thinking in agricultural weed management. New Phytol 184:783–793 (2009).
685
Pest Manag Sci 2015; 71: 675–685
© 2014 Society of Chemical Industry
wileyonlinelibrary.com/journal/ps
