Abstract

Keywords

Wolbachia, microsporidia, genetic diversity

Introduction

The diamondback moth, Plutella xylostella, is a major pest of crucifer production on all continents, leading to a strong impetus to control its populations in order to limit economic losses of food production (Talekar & Shelton 1993). Many different approaches have been used, including the introduction of natural enemies, IPM and many different insecticides. However, P. xylostella remains a major pest, largely due to the rapid emergence of resistance to a variety of different pesticides (e.g. Shelton et al. 1993). As with any pest species, our ability to control P. xylostella depends on our understanding of its fundamental biology. There has been considerable recent progress in some research areas, such as our understanding of the genetic basis of resistance to various pesticides, particularly Bacillus thuringiensis insecticidal endotoxins (Gahan et al. 2001, Ferré & van Rie 2002). However, we still have only a very limited grasp of the population structure and gene flow in this key pest species.

A better understanding of P. xylostella population biology would be valuable for a number of reasons. Given the propensity of P. xylostella to evolve pesticide resistance, and our increasing understanding of its genetic basis, we need to model the evolution and spread of different forms of resistance to consider and implement management alternatives (Peck et al. 1999). However, such models can only be realistic if we can characterise population structure and gene flow in P. xylostella populations. In particular, and in the light of recent theory, it will be important to quantify gene flow between selected and unselected sub-populations. In fact, dispersal in general is a very important issue in the management of insect pests, though notoriously difficult to study. At the two extremes there are reports of very long-distance dispersal by P. xylostella (Chu 1986, Talekar & Shelton 1993) and, in contrast, of very low dispersal distances from controlled release points (Shirai & Nakamura 1994). Clearly, we do not yet have a good understanding of average dispersal distances and how these vary with environmental factors, such as spatial distribution of oviposition sites. Such issues can be addressed using population genetic studies, but first require the development of reliable, polymorphic, molecular markers.

Many population genetic studies of Lepidoptera (including P. xylostella) have utilised allozymes. However, these markers generally have low variability, which limits the detection of population structure. They also require storage of insects at –70ºC prior to analysis, which is frequently impractical in field studies. In contrast, DNA-based techniques offer easier storage (ambient temperature in 95% ethanol), are not compounded by environmental or developmental post-translational or post-transcriptional changes and often reveal greater variability. The downside is the requirement for considerable effort in development and testing markers.

The advantages of microsatellites (e.g. codominance, species- and locus-specificity) are well documented (Goldstein & Schlötterer 1999, Sunnucks 2000). However, time and cost considerations have led to the popularity of “universal” anonymous locus techniques such as RAPDS and AFLPs (Vos et al. 1995, Suazo & Hall 1999). These methods may well be quicker and cheaper to develop, but the consistency of RAPDs is often unacceptably poor, while AFLPs remain relatively untested in insects. Since AFLPs are anonymous and (mostly) dominant markers, they may also be subject to the well-known problems that have plagued RAPDs. However, less attention has been given to the fact that AFLPs and RAPDs may also detect endosymbionts or parasites of the target organisms. For example, P. xylostella is host to both Wolbachia bacteria and microsporidia.

We have developed P. xylostella microsatellites and AFLPs and made preliminary studies of genetic diversity at both a local (Cameron Highlands, Malaysia) and global (populations from nine countries) scale. We also compared the markers within and between isofemale lines over several generations to test for consistency. Finally, we used inbred lines with different parasite infections to test whether some AFLP or microsatellite loci might belong to internal parasites.

Materials and methods

P. xylostella cultures and experimental lines

Stock cultures of P. xylostella were established from 300 insects collected from each of Blue, Mensum and Bertram Valleys in the Cameron Highlands (Malaysia) in December 2000 and maintained on Chinese cabbage. Subcultures were established on an artificial medium (details available by request). P. xylostella isofemale lines were established from microsporidia-free parents by single pair matings for six generations. Microsporidia-infected (M) sub-lines were then created by addition of 100 microsporidial spores cm^-2 to the surface of the larval food. All isofemale lines were initially Wolbachia-infected (W), so sub-lines were cured of Wolbachia by rearing larvae on a tetracyclin-HCl (3 mg ml^-1) supplemented diet for three generations and then maintained for a further 3 generations before screening markers. Infection status was confirmed using specific PCR (described below). In this way, three P. xylostella isofemale lines (termed A, B and O) of different geographic origin were established, each with sub-lines of different infection status (uninfected, W, M, W+M).

DNA extraction, genomic libraries and microsatellite screening

DNA (>50 Kbp fragments) from 40 freshly killed P. xylostella adults (with guts removed) was extracted and digested to completion with either Sau-3A, Alu-1, Hae3 or Rsa1. Complete genomic libraries (3.3–17.9 x10⁶ colonies; mean insert size 300–320 bp) were made using pBluescriptKS and electrocompetent DH10B (>10¹⁰ cells/µg) E. coli cells and screened for microsatellite-containing clones as described by Butcher et al. (2000). Probes to all possible mono- (2), di- (2) tri- (14) and tetra- (49) nucleotide microsatellite motifs (excluding self-homologous repeats and the restriction sites used in library constructions) were used to screen duplicate nylon membrane lifts. Positive clones were sequenced utilising Big Dye (Perkin Elmer) chemistry and resolved on an ABI 3700 sequencer.

PCR primers to the consensus flanking sequences of microsatellites were designed manually and loci were tested over three generations to confirm lack of sex linkage, reliable amplification and Mendelian segregation. Each primer set was then evaluated on P. xylostella samples from the Cameron Highlands and surrounding lowlands (Malaysia), Sarawak (Malaysia), Sumatra (Indonesia), China, England, Hawaii, France, Greece, Thailand and Japan, to assess allelic diversity. This also permitted detection of null allele prone primer pairs, which we redesigned further away from the core repeat unit and re-evaluated. To confirm species-specificity, primers were also used on DNA from the two principal P. xylostella endoparasitoids (Cotesia plutellae and Diadegma semiclausum), as well as an entomopathogenic fungus (Zoophthora radicans).

Microsatellite, AFLP and symbiont PCR

Prior to DNA extraction, all P. xylostella samples were cleaned and dissected to remove visible parasites (mites, parasitoid larvae, nematodes or fungal hyphae) and the intact gut was also removed. Wolbachia and microsporidia PCR analysis utilised DNA extracted from the gonad and surrounding fat body tissues, whilst microsatellite and AFLP PCR did not use adult abdomens as a DNA source to avoid false polymorphisms due to sperm. Long PCR with Taq/pfu was used to remove ambiguity due to incomplete Taq-based 3’ adenylation (stutter bands) (Butcher et al. in prep.) and involved an initial 60 s denaturation at 96ºC followed by 38 cycles of: 94ºC for 30 s, 55ºC for 30 s and 72ºC for 60 s. Microsatellite and AFLP amplicons were 45% formamide-heat denatured for 5 min at 96ºC and resolved by denaturing gel electrophoresis on 60 cm TBE-7.8 M urea-acrylamide gels (8% for microsatellites and 4–6% gradient for AFLP) at 55–60ºC and visualised by silver staining (Butcher et al. 2000). For microsatellite loci, GC and AT cycle sequencing tracts of PKS were resolved on each gel as size standards, while the cloned plasmid was also amplified and resolved on the gel as both a known size standard and stutter band control. For AFLPs, 100 bp and 1 Kbp ladders (Gibco) were used.

Wolbachia and microsporidia were detected by long PCR using primers specific to the Wolbachia wsp, fts-Z and gro-EL regions (Butcher et al. in prep.) with an estimated detection threshold of 10–50 bacteria (based on calibration with plasmid-cloned inserts) and microsporidia 16S rDNA (Weiss & Vossbrinck 1999). Amplicons were resolved using TAE-0.8% (w/v) agarose /2.5 gml^-1 ethidium bromide gels.

AFLP variation in two Cameron Highlands valleys

Fifteen individual P. xylostella larvae from each of the Blue and Bertram valleys (Cameron Highlands, Malaysia), ascertained to be symbiont and parasite free, served as DNA templates for AFLP analysis. Preliminary analysis revealed that single-stage amplification (Suazo & Hall 1999) produced too few bands, so a two-stage process was carried out as described by Vos et al. (1995), but with additional restriction enzyme (RE) combinations based upon Xba1, Xh01 or Pst-1 in place of EcoR1, and Alu-1 or Bfa1 in place of Mse1. Secondary amplification used the same primers as the first stage, but as six separate PCR reactions, with AC or CT (EcoR1, Pst-1, Xba1) and AGT, ATA or AGC (Mse1 and Bfa-1) 3’ additions. Alu-1 and Xho-1 based AFLP was problematical (over 8% inconsistent bands) and was not optimised further.

For each sample, three independent AFLP reactions were scored and non-consensus bands deemed artefacts were excluded. The pooled total number of different bands observed across all 30 samples is shown in Table 2, along with the number of bands that were observed to vary between individuals. AFLP analysis was also performed in the presence of realistic amounts of parasite template DNA (D. semiclausum, C. plutellae parasitoids or Z. radicans fungus) that might be found in a late instar P. xylostella larva (from dissection data).

Detecting symbiont loci

Twenty individuals from each of four sub lines (uninfected, W, M, W+M) were genotyped for each isofemale line. For example, the F₁ progeny of an A line male mated to each of the four B sublines (U, W, M, W+M) were assayed with AFLPs and microsatellites. Two-stage AFLP analysis after EcoR1/Mse1 restriction was performed with three replicates per sample. Differences in banding within a sample, but between replicates, were excluded as artefacts (<1.8%) and the remaining number of reliable bands is displayed as a pooled total from the different secondary amplifications (Table 3). Comparison of the bands observed with the isofemale line A profile to all the other samples revealed the number of informative (different) bands (Table 3) due to both host genetic differences (e.g. A versus B) and symbiont/parasite infection status (e.g. A versus B+W minus (A v. B).

Results

Microsatellite diversity in the Cameron Highlands

Screening of Sau-3A and Alu-1 genomic libraries, excluding mononucleotide motifs, yielded approximately 8300 putative microsatellite-containing clones. Sequencing of 1537 of these clones revealed 21% with either no (repeat motif size < 6) or a cryptic microsatellite and a 2.3 fold replication of clones, leaving 528 unique microsatellite clones. Of these, 91 loci contained microsatellite repeat unit sizes of over 10 and detect more than six alleles within the Cameron Highlands, while 15 are highly polymorphic with over 35 alleles. Taken together, the data reveal crude but conservative estimates of ~820 loci with >20 alleles and ~95 with >35 alleles and an average inter-locus distance of 12.3 Kbp (Butcher et al. 2000).

The first 12 microsatellite loci evaluated with more than 10 repeat units were used to estimate genetic variability. Each locus is polymorphic in the Cameron Highlands and has further alleles at a global scale (Table 1). As expected, the microsatellite primers did not amplify from associated parasites.

Table 1. Evaluation of twelve microsatellite loci for population genetic studies of Plutella xylostella

Locus	Core repeata	Sizeb	Primer sequence	PCRc	CH Allelesd	Globale
PxDi1	(GT)15	95	F: AGCAATGCACCTCTGCCTA R: GGAAAGTTAATATAACCGAAC	55ºC	8 (89–115)	17 (85–125)
PxDi2	(GT)12	182	F: TAGGTATACAAATTAGTTGTATT R: CAGCATAAATAAATTATTAAATG	54ºC	6 (174–200)	15 (170–228)
PxDi3	(AC)18	134	F: ATGCTAGTGCGACTTGCC R: TTCCTGATATAGCTGAAAAGC	54ºC	9 (128–144)	17 (114–146)
PxDi18	(GT)26	99	F: GCGTACATTAGTACAAGGC R: GATCGATATTAAATTTGTCCTA	54ºC	16 (87–119)	57 (65–165)
PxTri1	(GCG)3GTG (GCG)8(CCGGG)3	203	F: AAATCAAACCTGAAATGAGA R: AACAGTCGAGCCTCCGA	55ºC	5 (194–215)	11 (189–226)
PxTri2	(TCA)16	118	F: TCCTTAGGAGACGCCTATG R: CGCAAGCCTGTCAACCC	55ºC	6 (88–118)	16 (88–133)
PxTri3	(GTT)14(GCT)4GT(TGC)4	116	F: CCACATTCAAATCCGGATTC R: GATCGTGTGAGGCAGCAA	54ºC	7 (101–119)	23 (90–137)
PxTri4	(TGA)16	107	F: CCTGTGTCTAGCAGTTGAC R: ACTTTAGTAGGATTTTGGATAT	54ºC	8 (95–116)	21 (80–122)
PxTri5	(ATC)19	98	F: ACTGCCGCACGAGAAGAC R: GATCAGCGGGATGGGCT	52ºC	10 (77–117)	21 (62–125)
PxTri6	(CGC)12	100	F: ATTCAGAAAGTTGGTCCCC R: AAGAAGCGTTTAAGTAATTGC	54ºC	9 (88–121)	19 (81–124)
PxTet1	(CAGA)16	135	F: TCCGTTCAGTAGTTTTGG R: GTACTCAGGTGAGTGCTT	54ºC	6 (89–135)	17 (89–143)
PxTet2	(GTCT)11	155	F: CCAAATTTCATTCAAATCCGTT R: CACTTGACCATCCTTAATGTCGAA	55ºC	13 (137–167)	41 (137–203)

a - The core microsatellite repeat unit of the most common clone, b - PCR amplicon size (bp) of most common clone, c - Optimal annealing temperature in direct 3 stage long PCR at 2 mM Mg²⁺, d - Number (and amplicon size range) of resolvable alleles from 450 larvae from 3 valleys in Cameron Highlands, Malaysia, e - As for d but with samples from nine different countries. Note that allelic sizes observed do not always match integer changes in microsatellite repeat numbers, suggesting indels, which have been confirmed in two alleles by sequencing.

AFLP diversity in the Cameron Highlands

In the two-stage amplification AFLP analysis, all paired restriction enzyme combinations tested revealed resolvable and informative bands (Table 2). However, the paired combinations of Pst-1 and EcoR1 with Mse-1 yielded the most informative bands when used in conjunction with three nucleotide selective 3’ extensions. Within any sample, ignoring ghost bands, 98% of the bands were reproducible with DNA equivalents of 20–30% of a III instar larva, falling to 94–95% with 5% DNA equivalence. In addition, AFLP analysis revealed many informative bands associated with low levels of endoparasitoid or entomopathogenic fungal contamination (Table 2).

Table 2. AFLP bands resolved with different primers and DNA templates in Plutella xylostella

Template DNA	Preamp. primers	Bands	Informative
P. xylostella	EcoR1/Mse-1	421	142 (34%)
	EcoR1/Bfa-1	387	103 (27%)
	Pst-1/Mse-1	465	156 (34%
	Pst-1/Bfa-1	424	121 (29%)
	Xba-1/Mse-1	229	54 (24%)
	Xba-1/Bfa-1	252	58 (23%)
P. xylostella +	EcoR1/Mse-1	458	216 (47%)
D. semiclausum	Pst-1/Mse-1	471	245 (52%)
P. xylostella +	EcoR1/Mse-1	438	265 (61%)
C. plutellae	Pst-1/Mse-1	462	247 (54%)
P. xylostella +	EcoR1/Mse-1	428	208 (49%)
Z. radicans	Pst-1/Mse-1	399	209 (52%)

Comparing microsatellites and AFLPs using isogenic lines with different parasite infections

Within the isogenic A line (expected heterozygosity <0.1%), neither AFLPs nor microsatellites detected variation, as expected. However, addition of endoparasites led to about 5% variable AFLP bands, but no microsatellite variation (Table 3). When comparing uninfected host lines there were major differences in both AFLPs and microsatellites. For example, lines B and O were fixed for different alleles to line A at eight and 12 microsatellite loci, respectively.

However, only the AFLP method revealed increased variability when infected individuals were assayed. In the most extreme case, the double infected B+W+M sub-line, 16/72 = 22% variable bands were potentially attributable to infection status (Table 3).

Table 3. AFLP detection of symbiont genomes in Plutella xylostella

Isofemale line A	AFLPs			Microsatellites (12 loci)
Relative toa	Bands	Variable	Symbiont bandsb,c	Bands	Variable	Symbiont bands
A	370	0	control	12	0	control
A + W	368	17	17	12	0	0
A + M	371	19	19	12	0	0
A + W + M.	373	25	25	12	0	0
B	357	56	control	12	8	control
B + W	361	64	8	12	8	0
B + M	360	69	13	12	8	0
B + W + M	364	72	16	12	8	0
O	378	96	control	12	12	control
O + W	376	103	7	12	12	0
O + M	381	105	9	12	12	0
O + W + M	380	111	15	12	12	0

a - Different isofemale lines are denoted by A, B and O and different infection status by W (Wolbachia) and M (microsporidia), b - The uninfected comparison provides the control number of variable host bands (e.g. A relative to B = 56) and subtraction of this number from the total seen in an infected treatment (e.g. A relative to B+W = 64) gives the number attributed to symbionts (e.g. 8), c–In each of the three lines, the number of putative symbiont bands for the (W+M) case is less than the combined total for W and M tested separately (e.g. for B 8+13<16). This may reflect competitive amplification effects when more target genomes are available.

Discussion

Reliable and informative genetic markers are needed to investigate the population genetics of P. xylostella. Studies to date have mostly utilised allozymes (e.g. Caprio & Tabashnik 1992, Kim et al. 1999, Pichon et al. 2003) which have generally failed to detect clear evidence of population differentiation. Despite the advantages of codominance, relative species and locus specificity, and frequently high levels of polymorphism, the cost and time required to develop microsatellites has led to increased popularity of alternative markers such as AFLPs. The situation is compounded in Lepidoptera since this Order has been viewed as deficient in polymorphic microsatellites (Meglécz & Solignac 1998, Neve & Meglécz 2000). However, there is no coherent hypothesis to explain why this should be the case and we suggest that it may not be true. Other DNA-based mutation rates do not appear aberrant in Lepidoptera and Reddy et al. (1999) reported (GT) repeats in Bombyx mori at frequencies similar to those of other animals. Nevertheless, most lepidopteran studies have found few microsatellite loci, each with relatively few alleles (Meglécz & Solignac 1998, Saccheri et al. 1999, Harper et al. 2000, Bogdanowicz et al. 1997, Tan et al. 2001, Keyghobadi et al. 1999, Reddy et al. 1999). An exception is the nymphalid, Speyeria nidalia, which has four very polymorphic loci (Williams et al. 2002). Our library screening suggests that there are about 820 very polymorphic microsatellite loci in the diamondback moth genome. We have isolated 81 of these and described here twelve evaluated loci that should prove valuable for population studies.

The great utility of AFLPs for genome and pedigree mapping is unequivocal. However, their (usually) dominant nature, lack of species-specificity and dependence on template-quality and quantity are drawbacks for population genetics studies. Our AFLP analysis certainly revealed a wealth of informative bands, illustrating the potential of the method for population genetic studies. However, in addition to well-known problems with AFLPs, we have highlighted the possibility that some bands are primed in symbiont or parasite genomes. This calls for some care in using AFLPs for studies between samples with a low genetic diversity, especially when one considers spatial and temporal variation in parasite infections. We must emphasise that the symbiont origin of the extra bands now requires direct testing and has so far only been inferred from the experimental design. Nevertheless, this caveat is important.

It is interesting that so many putative symbiont AFLP bands were detected (Tables 2 and 3). The symbionts in question have genome sizes crudely estimated at around 2.5% (Wolbachia) to 7.5% (microsporidia) of the P. xylostella genome, perhaps suggesting that around 10% of all bands could be of symbiont origin. The number of informative bands of symbiont origin may be much lower than this because microbial genomes are deficient in satellite DNA, one source of polymorphism resolved by AFLPs. Nevertheless, when comparing infected and uninfected individuals all (even if there are few) symbiont/parasite bands will be “informative” by definition.

This could be considered a significant “nuisance factor” for studies of host population genetics. On the other hand, comparisons of banding patterns of infected and uninfected members of the same host isofemale line could offer a novel method for identifying and cloning new symbiont gene fragments. However, with regard to P. xylostella population genetic studies, we currently favour the use of the polymorphic microsatellite loci described here and further loci that are being evaluated by ourselves and others.

Acknowledgements

We thank the numerous generous Cameron Highland farm owners for access to their fields, Mrs. Ong (Malaysian Agricultural Research and Development Institute, Tanah Rata) for support in field work and Liz Canning (Imperial College) for helpful advice on all aspects of microsporidia biology. The study was funded by a BBSRC research grant to JC and DW.

References