Introduction

The potato cyst nematodes (PCN) Globodera rostochiensis and G. pallida are thought to have originated in the Andean region of South America (Evans & Stone, 1977) and to have been introduced into Europe after 1850 (Evans et al., 1975). They are major pests of potato, causing significant economic losses. Plant resistance is an effective means to control PCN, but its wide application is hampered by the occurrence of various virulent nematode populations. Determining whether there has been more than one introduction has practical importance for the management of these pests, since the genetic basis of virulence may differ.

PCNs are classified into pathotypes on the basis of their ability to multiply on different potato clones. However, the status of their pathotypes is still a matter of debate (Phillips & Trudgill, 1983; Mugniéry et al., 1989). The International Pathotype Scheme introduced by Kort et al. (1977) recognizes, among the European populations, five pathotypes within G. rostochiensis and three within G. pallida, while the Latin American Scheme (Canto & Scurrah, 1977) recognizes among the South American populations of these two species four and seven pathotypes, respectively. As outlined by Bakker et al. (1993), these schemes are not able to reflect all of the genetic diversity introduced into Europe.

Biochemical and molecular characterizations of PCN populations have been conducted to relate the genetic diversity to the virulence and to find markers that readily distinguish the pathotypes (Folkertsma et al., 1996b; Blok et al., 1997; Thiéry et al., 1997). Except to a certain degree for G. rostochiensis, no clear correlation has been found between the groupings obtained using various molecular techniques and the virulence. Moreover, some South American populations of G. pallida tend to form a distinct group, separate from the European G. pallida (Blok et al., 1997; Bendezu et al., 1998).

The objectives of this research were to look first at the taxonomic relationships among Globodera species. Following the work of Thiéry & Mugniéry (1996), a complementary study of the polymorphism of the rDNA Internal Transcribed Spacer (ITS) was conducted on the Peruvian populations. Highly repetitive satellite DNAs were also isolated and used as species-specific markers. Then, two-dimensional gel electrophoresis (2-DGE) of proteins and Random Amplified Polymorphic DNA (RAPD) techniques were used to produce two independent but complementary datasets for the assessment of the intraspecific genetic variation among both indigenous (Peruvian) and imported (European) populations identified as G. pallida.

Materials and methods

Nematode populations, genomic DNA and total soluble protein preparations

All the populations (Table 1) came from our laboratory collection and were reared on their common host. DNA was extracted from frozen females ground in liquid nitrogen by a phenol/chloroform procedure followed by ethanol precipitation as described in Thiéry & Mugniéry (1996). The proteins were extracted as described in Bossis & Mugniéry (1993) and stored at −70°C. DNA and protein extractions were carried out, respectively, on 2500 and 100 females per sample and the average yields were 10 μg of DNA and 150 μg of protein per sample.

Table 1 Names and related codes of the Globodera populations tested. The origins and the pathotype classification of each population are indicated when known. Pathotype determination was usually made by the institute which had sent us the population (i.e. CIP Lima, SCRI Dundee and BBA Münster)
Full size table

Cloning and sequencing of the satellite DNAs

Two micrograms of genomic DNA from G. pallida Guiclan and P6A populations were digested to completion with the SmaI and AluI restriction enzymes (Boehringer Mannheim), respectively. The smallest fragments (≈200 bp) visualized following agarose gel electrophoresis were first recovered onto DEAE paper (Sambrook et al., 1989) and then cloned into pUC 18 plasmid (Pharmacia). The ligation mixture was used to transform the competent Escherichia coli strain JM109 (Promega). The sequencing of the inserts selected after cloning was performed on an ABI Prism 310 automated sequencer (PE Applied Biosystems). The consensus sequences appear in the EMBL and GenBank databases under the accession numbers AF250033 and AF250034 for the SmaI and AluI satellite DNAs, respectively.

Dot blot procedure

Dot blots were prepared by denaturing 100 ng of genomic DNA of each population with 2 M NaOH for 10 min, and neutralizing it with ammonium acetate 4 M. Whole samples were then dotted by vacuum onto nitrocellulose membranes (Amersham) in a dot blotting apparatus (Bio-Rad).

ITS amplification and restriction procedures

ITS–RFLP analyses were performed as described by Thiéry & Mugniéry (1996). Primers were the same as those described in Vrain et al. (1992) and PCR conditions were 30 cycles of a denaturation step at 94°C for 1 min, annealing at 60°C for 50 s and extension at 72°C for 1 min. A final extension was performed at the end for 5 min at 72°C. The amplified products were digested to completion with a set of 16 restriction enzymes (Table 2). Digestion products were separated by electrophoresis in a 1.5% agarose gel and visualized by ethidium bromide staining.

Table 2 Restriction enzymes used in the ITS–RFLP analysis of the Peruvian populations and digestion pattern comparison of the Peruvian with different Globodera species. GA, GR, GT, GM and GP stand for G. artemisiae, G. rostochiensis, G. tabacum, ‘ G. mexicana’ and European G. pallida, respectively. Presence of an asterisk (*) means that some Peruvian populations have produced other digestion pattern types than the ones mentioned
Full size table

RAPD procedure

PCR reactions were performed as described by Thiéry et al. (1997) in a final volume of 12.5 μL, using eight random primers from the kits G and L of Operon Technologies (USA). Genomic DNA (15 ng per reaction) was first denatured by heating at 94°C for 5 min and then processed through 40 cycles, each consisting of 1.0 min at 94°C, 1.0 min at 35°C and 1.0 min at 72°C. A final extension was performed at the end for 5 min at 72°C. PCR products were visualized by ethidium bromide staining on agarose gels after electrophoresis. Three replicates of each amplification were performed independently and a negative control was always included as a test for contamination. A population marker was chosen if it was present without ambiguity in the three repeated amplification reactions. Differences in band intensity were not a consideration.

Two-dimensional gel electrophoresis

2-DGE analyses were carried out with the mini Protean II equipment (Bio-Rad) mainly following O’Farrell’s procedure (O’Farrell, 1975). Modifications included the use of PDA as crosslinker and of ammonium thiosulphate as a reducer of the background (Hochstrasser et al., 1988). Staining was obtained with ammoniacal silver solution (Oakley et al., 1980) with slight modifications (Bossis & Mugniéry, 1993). Molecular weights and isoelectric points were determined with the 2D SDS-PAGE standards (Bio-Rad), consisting of six reduced denatured proteins: hen egg white conalbumin, bovine serum albumin, bovine muscle actin, bovine carbonic anhydrase, soybean trypsin inhibitor and equine myoglobin.

Analysis of protein patterns

Images of gels were compared on a Vax station 4000-60 computer (Digital Equipment Corporation) with the Kepler software (Large Scale Biology Corporation, Rockville, USA). Gels were digitized (2048 × 2048 elements) with a scanner (Eikonix EC78/99, Kodak) at a resolution of 100 μm/pixel. From each image file a fitted spotfile was generated and used as the basic data in all the analysis. This step was run with the procbille process (Zivy, 1990). Each protein spot encoded by a master number (MSN) was characterized by its coordinates (x, y), its width (sx, sy) and its amplitude (amp). The amplitude is dependent on the optical density of the elements of each spot. The volume (vol) of each modelled spot was calculated according to vol.=2 · π · sx · sy · amp.

Triplicates of protein samples were prepared from each population. Whatever its volume, a spot was considered to be present in a population when it was detected in at least two of the three gels of this population. A synthetic pattern called the master was built including the spots analysed in all the populations. Then, the matching was initiated in an interactive approach using 40 spots selected as the easiest to match. On this basis, automatic matching was performed and all the combinations were studied using logical operations, so that shared proteins were identified for each pair of populations.

Cluster analysis

Polymorphic markers were recorded as a binary matrix corresponding to their absence or presence. The genetic distances were estimated on the basis of the Nei similarity index (Nei, 1972) and trees were obtained with the Neighbour Joining algorithm using the PHYLIP software (Felsenstein, 1993) with the bootstrap and consensus tree options.

Results

Comparison of the ITS restriction fragment length polymorphisms

The amplified ITS of all the Globodera species and populations tested was 1200 bp. More than 90 markers were revealed after digestion of the ITS (Table 2). Only 44 were used in the cluster analysis of Fig. 1 as some were not polymorphic (BglI, HincII and PstI), or others, when summed, resulted in a much higher size than the expected 1200 bp due to additional digestion of some markers. As expected, G. artemisiae –– as representative of the Compositae Globodera species –– is distant from the Solanaceae Globodera species. ‘G. mexicana’ and the European G. pallida are extremely close, whereas G. rostochiensis and G. tabacum are more distinct. The Peruvian G. pallida split into three groups without clear correlation with their pathotypes. These three groups are closely related to each other and to a lesser extent to the ‘G. mexicana’ and European G. pallida cluster. The Peruvian populations are separated from the European G. pallida, however, the bootstrap values are quite low.

Fig. 1
figure 1

Unrooted boot-strapped tree of Globodera species and of the Peruvian populations inferred by Neighbour Joining from the ITS–RFLP data. Bootstrap values are indicated besides each branch and were obtained after 500 iterations.

Full size image

Isolation and primary structure of two satellite DNA families

Digestion of genomic DNA from the G. pallida Guiclan and P6A populations has revealed intensely staining bands of approximately 350 bp and 150 bp in, respectively, the SmaI digest of Guiclan and the AluI digest of P6A. After cloning, 10 suitable clones displaying an insert of the expected size (five of 350 bp and five of 150 bp) were selected for further analysis. Unambiguous consensus sequences of 367 bp and 160 bp were deduced from the alignment of the SmaI and AluI fragments, respectively (Fig. 2).

Fig. 2
figure 2

Consensus sequence and monomer sequences of the SmaI (A) and AluI (B) satellite DNAs. All the monomer repeats were aligned manually, point mutations relative to the consensus sequence are shown in each repeat, insertions are indicated by brackets and deletions by (–). Alignment with GenBank accession number X85444, corresponding to Burrows et al. (1995) satellite DNA, is also shown in A.

Full size image

The SmaI and AluI satellite DNA families have an adenine + thymine content of 63% and 58%, respectively. The five monomers cloned in each of these families differ mainly by single base substitutions, but two- and three-point deletions plus one insertion are found in the SmaI and AluI satellite DNA families, respectively. Sequence variation among the cloned monomers of each family relative to the derived consensus sequence is quite low, ranging from 0 to 15 variant nucleotides per cloned repeat for the SmaI satellite DNA family and 4 to 16 for the AluI satellite DNA family. The average divergences from the consensus sequence for these two satellite DNA families are 1.7% and 5.5%, respectively.

No significant similarity was detected between these two satellite DNA families (only 31% DNA identity). Comparison with other sequences recorded in the GenBank and EMBL databases revealed no substantial similarities in the case of the AluI satellite DNA family, but 94% of DNA identity with a repetitive element of G. pallida (GenBank accession number X85444) in the case of the SmaI satellite DNA family.

Species-specificity of the isolated satellite DNA sequences

The SmaI and AluI repeated elements were hybridized to genomic DNA of various Globodera populations. Fig. 3A indicates that the SmaI satellite DNA isolated from Guiclan hybridizes strongly with the European populations of G. pallida but more weakly with the Peruvian ones. Hybridization was never detected with genomic DNA of G. rostochiensis and ‘G. mexicana’ even with long exposure times, but a slight signal was consistently observed with a subset of the G. tabacum populations classified as G. t. virginiae.

Fig. 3
figure 3

Species-specificity of the isolated satellite DNA sequences. Genomic DNAs of populations of different Globodera species were dot blotted onto a membrane and hybridized with the SmaI (A) or AluI (B) satellite DNAs.

Full size image

The AluI satellite DNA isolated from P6A hybridized equally with the European and Peruvian populations of G. pallida (Fig. 3B), but a signal was also clearly observed with the closely related species ‘G. mexicana’. With longer exposure times, hybridization signals were even observed with G. rostochiensis and G. t. virginiae.

Two-dimensional gel electrophoresis of proteins

On the master, 542 spots were scored including the very faint ones (vol. < 500). A mean of 286 spots per population was detected. Fig. 4 gives the master on which all spots are displayed and where some of them (in dark) are identified by their MSN as in Table 3.

Fig. 4
figure 4

Master image obtained after 2-DGE analysis of the European and Peruvian Globodera pallida populations. Black spots represent spots listed in Table 3 with their master numbers.

Full size image
Table 3 Presence or absence of some 2-DGE spots characteristic of some populations or group of populations. The characteristic 2-DGE spots are identified by their master number, MSN and their presence in each population is indicated by X. Population-specific spots and their absence in the different populations is indicated by an asterisk. For easier reading, all the European populations were grouped on the left side of the table and all the Peruvians on the right side
Full size table

Some of these discriminating spots were characteristic of only one population and could be used as markers for these populations: MSNs 537 for Audierne, 565 for Luffness, 582 for Pukekoe and 502 for P6A. The absence of one spot can also be used in the identification of five populations: MSNs 42 and 409 for Luffness, 62 for Duddingston, 218 for Perpignan, 254 for Chavornay and 322 for Pukekoe. Of the 18 spots strongly represented in the European populations, three (MSNs 531, 535 and 541) were never observed in the Peruvians. But, among the 19 spots strongly represented in the Peruvian populations, none were absent from the Europeans (Table 3). However, MSN 453, which was never detected in the European populations, was present in two populations of each Peruvian pathotype.

The mean genetic distance in the Peruvian populations was 0.39 [0.23–0.50] whereas the mean genetic distance in the European populations was 0.45 [0.29–0.61]. The mean genetic distance between the two groups was 0.53 [0.42–0.59] (Table 4(a). The consensus tree (Fig. 5A) derived from the genetic distances shows that the European and Peruvian populations can clearly be separated and that this clustering is supported by strong bootstrap values. In the Peruvian group, two pairs of populations were observed with bootstrap values above 96 (P4A 6 and P4A 1, P6A 14 and P6A 15). In the European group, two pairs of populations were also found: (Saint Malo, Pas de Calais) and (Perpignan, BB1), with respective bootstrap values of 84 and 79.

Table 4 Genetic distances obtained from 2-DGE (a) and RAPD (b) data. Genetic distances were calculated on the basis of the Nei similarity index (Nei, 1972)
Full size table
Fig. 5
figure 5

Unrooted boot-strapped trees of the European and Peruvian Globodera pallida populations estimated by Neighbour Joining from the 2-DGE (A) and RAPD (B) genetic distances presented in Table 4. Bootstrap values obtained after 500 iterations are indicated beside each branch.

Full size image

RAPD fingerprinting

An average of 25 markers per primer combination was observed in a size range between 225 and 3000 bp, and 180 polymorphic markers were retained for the analysis. Clearly the Peruvian populations are closer to themselves than to those from Europe (Table 4(b)). The mean genetic distance in the Peruvian populations was 0.41 [0.12–0.57] whereas the mean genetic distance in the European populations was 0.25 [0.10–0.40]. The mean genetic distance between the two groups was 0.57 [0.33–0.58]. The P5A 8 population seems intermediate between the Peruvian and European populations. If this population is removed from the analysis, the genetic distance between the two groups increases from 0.57 to 0.59. Similarly the genetic distance within the Peruvian group drops from 0.41 to 0.38. With this set of populations there are two groups with a higher diversity within the Peruvian group than within the European one.

The tree shown in Fig. 5B clearly separates the Peruvian and European populations. In the Peruvian populations, the clusters are in relative good accordance with the pathotype scheme of Canto & Scurrah (1977) who separated the P4A pathotype from the others. In contrast, the European populations cluster independently of the pathotype scheme. The Pa1 and Pa2 pathotypes cluster within a group of Pa2/3 pathotypes. The Luffness population, which is supposed to be different (Phillips et al., 1992; Phillips & Trudgill, 1998), is not clearly distinct from other European populations.

Discussion

Two repeated DNA sequences, representing two different satellite DNA families from European and Peruvian G. pallida, were isolated. While no substantial similarities were detected between the AluI sequences and any other sequence recorded in the GenBank and EMBL database, the SmaI sequences belong undoubtedly to the same satellite DNA family as that isolated by Burrows et al. (1995) using the HindIII enzyme. The average divergence from the consensus sequence is estimated at only 1.7% for the SmaI family and 5.5% for the AluI family. Compared with other satellite DNAs, the SmaI family seems to be extremely conserved. Such a high degree of sequence similarity was shown for satellite DNAs of meiotic parthenogenetic nematodes like Meloidogyne fallax (2.7%) (Castagnone et al., 1998) and M. hapla (3%) (Piotte et al., 1994). Two explanations can be provided: (i) the SmaI satellite DNA family is younger than the AluI family; and (ii) the impact of the homogenizing process is stronger in the SmaI family. It has been suggested that during the evolution of satellite DNAs, C to T and G to A transitions are more frequent than the reverse transitions, which results in A + T enrichment of these sequences (Ugarkovic et al., 1989). Under this hypothesis, the low A + T content observed within the AluI satellite DNA family is not in accordance with the first explanation while the observation of a higher number of multivariant positions substituted with the same nucleotide in the SmaI family is in accordance with the second explanation.

Satellite DNAs that evolve in a concerted manner are often described as species-specific sequences but some may be restricted to a specific group of related species (Mantovani et al., 1997; Mestrovic et al., 1998). This seems to be the case with the isolated AluI satellite DNA, which recognizes both G. pallida and ‘G. mexicana’, two closely related species that cluster together in our ITS analysis. Though not intense, the consistent hybridization of both probes to the same subset of G. tabacum populations, all classified as G. t. virginiae, was confirmed in a more extensive dot blot (data not shown). This result gives some support to the subspecies grouping adopted inside the G. tabacum group but raises questions about the possible relationships which exist between some G. tabacum populations and the G. pallida and ‘G. mexicana’ species.

The Peruvian populations are more complex. Their ITS heterogeneity is shown to be greater than that observed between populations inside each of the other groups (Thiéry & Mugniéry, 1996). Blok et al. (1998) showed little sequence variation between repeat units of the rDNA of a P5A population compared to a P4A population and one can notice that this last pathotype appears also more variable in our ITS analysis. Moreover, it is tempting to speculate that some of the Peruvian populations are mixtures of individuals with different types of rDNA or even that different types of repeat units occur in the rDNA of an individual, as it was shown to be the case in some imported populations of G. pallida (Blok et al., 1998). ITS–RFLP and satellite DNA sequences are used as taxonomic markers for nematodes (Grenier et al., 1997; Powers et al., 1997). However, the expected clarification of the taxonomic status of the Peruvian populations is still far from being obtained. Although the SmaI satellite DNA groups the Peruvian and the European G. pallida, the Peruvian populations appear to be distinct from all the other species and especially from the European G. pallida in the ITS–RFLP analysis. On one hand, we know that satellite DNA can be unable to distinguish closely related species (Castagnone et al., 1998) but, on the other hand, because of the low number of unambiguous markers revealed by the ITS–RFLP analysis, the robustness of the tree inferred from these data is not well supported. In fact, only 44 markers were found in this study, but this number would have dropped drastically to 26 or even to 22 if, respectively, G. artemisiae or both of the non PCN species (G. artemisiae and G. tabacum) had not been included. If the ITS region is a good candidate for species determination, then including the Peruvian populations in G. pallida would be questionable. Our results emphasize the need to augment these molecular data with biological and morphological studies before addressing the exact taxonomic position of these Peruvian populations.

Two independent data sets were used to determine the intraspecific variation among European and South American G. pallida populations: both 2-DGE and RAPD analysis confirm the clear distinction between the Peruvian and the European populations. With RAPD the P5A 8 population is the only one to be somewhat intermediate, although it belongs clearly to the Peruvian group when analysed by ITS–RFLP. The same type of South American/European clustering was observed by Blok et al. (1997) who found, using RAPDs with a set of two Peruvian and 17 British G. pallida populations, an average similarity of 51.8% between these two groups. Our mean RAPD and 2-DGE genetic distances can also be compared to those reported in Folkertsma et al. (1996a) on a set of 36 European G. pallida field populations. While RAPD genetic distances are similar, our 2-DGE distances are higher. We have scored a large number of faint spots which undoubtedly have tended to increase the protein variability. Our 2-DGE genetic distances are also probably overestimated due to some artefacts of the technique, particularly the failure to detect very faint spots on some gels. It would have been possible to eliminate these faint spots, but we have preferred to keep all the detected and modelled spots in the patterns because some of them were scored in all the populations. In addition to and in contrast to the 2-DGE data of Folkertsma et al. (1996a), we chose to consider all the proteins as independent markers, although some are probably produced by alleles at polymorphic loci.

Inside each group, no clear correlation between pathotype is evident. The Pa1 pathotype (the Duddingston population) clusters within the Pa2 and Pa2/3 pathotypes. While 2-DGE associates the Peruvian populations in a way that does not reflect the virulence groups, the RAPD data show that a P4A and a P5A/P6A groupings tend to be observed. This difference is expected, as, biologically, the P4A pathotype differs strongly from the P5A and the P6A pathotypes by its inability to develop on Solanum vernei (VTn)2 62.33.3 (Canto & Scurrah, 1977; Phillips & Trudgill, 1998). However, as the Pa1 pathotype also differs from the Pa2/3 pathotype by its inability to develop on the H2 gene from S. multidissectum P55/7 (Kort et al., 1977) the same type of difference would also be expected for the European Duddingston population, but was not found. Some pairs of populations appear to be closely related. These populations might have the same origin, with the same founders, but more probably are the result of secondary founding events. The slight differences observed between each member of a pair could be explained by the reduction of the diversity associated to the founder effect and also by occurrence of subsequent random genetic drift.

Apart from the Peruvian/European distinction, considerable discrepancies are observed between the trees obtained by 2-DGE and RAPD. Folkertsma et al. (1996a) did a comparison of the two techniques with a same set of G. pallida populations; they reported such inconsistencies and concluded that when clusters of populations are not well differentiated from each other, different types of data set may result in clusters with differing compositions. In our case, the results obtained are surprising in that if one assumes that the European populations were imported from the Andean region, it would be expected that the variability in the South American populations would be greater than the variability in the European populations. RAPD data reveal, in the Peruvian group, high genetic distances and a completely unresolved branching, consistent with this great variability. However, 2-DGE data reveal high genetic distances in the European populations and several characteristic proteins are observed in these populations. If we assume that the European populations were introduced from other South American populations that were not investigated in this study, then the characteristic proteins observed can be explained. Why RAPD and 2-DGE display opposite results concerning the most variable group remains to be further investigated. But, the codominant nature of 2-DGE markers, in contrast to the usual dominant nature of RAPD markers, can be part of the explanation. Assuming that very few nematodes were at the origin of the primary founders of the European populations and that the genetic drift has resulted in the fixing of different alleles in these imported populations, it is possible that our inability to identify products of alleles at polymorphic loci has led to an overestimation of the genetic distances by 2-DGE in the European group.

The European populations seem therefore to have come from a different part of the Peruvian area than that investigated in this study. The PCN samples in this study may not be representative of the whole of Peru. Further samples, both of Peruvian PCN populations and of other South American populations such as Bolivian and Chilean populations, may reveal the origin of European PCN.