Introduction

Verticillium wilt is a plant disease caused by two deuteromycete species: Verticillium albo-atrum and Verticillium dahliae (to be collectively referred to as Verticillium spp.). These soilborne plant pathogens exist worldwide, and are of major economic importance due to their broad host range and the extensive crop damage they cause. Yet, despite the substantial economic impact of this disease, little is known about host genes involved in resistance to verticillium wilt. Genotypes and varieties with varying levels of verticillium wilt resistance have been described in several plant species (Arbogast et al. 1999; Veronese et al. 2003; Vallad and Subbarao 2008), but specific genes required for its resistance have been characterized only in tomato. The Ve locus, which, was classically identified several decades ago (Schaible et al. 1951), confers resistance to Verticillium spp. Race 1, but not Race 2 (Alexander 1962). Its positional cloning was reported in 2001 (Kawchuk et al. 2001). A second gene, EDS1, is known to be required for Ve-mediated resistance in tomato on the basis that the absence of EDS1 function abolishes verticillium wilt resistance even in the presence of a functional Ve (Hu et al. 2005).

When sequenced, the tomato Ve locus was found to consist of two genes (Ve1 and Ve2) (Kawchuk et al. 2001). Each of these Ve genes contains a lengthy leucine-rich repeat (LRR) domain, a common plant disease resistance gene feature thought to play a role in recognition of pathogen-derived molecules. The Ve gene products are predicted to be cell surface receptor proteins based on the conserved amino acid sequence motifs (Kawchuk et al. 2001). Upon introduction via Agrobacterium-mediated transformation, each of the two Ve transgenes was reported to confer resistance to a susceptible potato variety (Kawchuk et al. 2001).

Studies of Ve-mediated resistance have been extended to other solanaceous species, focusing on identification of Ve homologs in diploid potato (Solanum lycopersicoides Dunal) (Chai et al. 2003) and turkeyberry (S. torvum Swartz) (Fei et al. 2004), and on development of Ve homolog markers for marker-assisted selection in cultivated, tetraploid potato (S. tuberosum L.) (Simko et al. 2004a, b). We herein describe the first targeted isolation of a putative Ve homolog from a non-solonaceouos species, Mentha longifolia, a wild diploid relative of the commercial mints (Mentha species: Lamiaceae family).

Two of the three commercially grown mint species in the United States, peppermint (Mentha × piperita L.) and Scotch spearmint (M. × gracilis Sole), are typically susceptible to verticillium wilt, while the third, Native spearmint (M. spicata L.) is considered to be relatively resistant (Berry and Thomas 1961; Lacy and Horner 1965; Sink and Grey 1999). Verticillium wilt has presented a particularly intractable problem for the United States peppermint industry, wherein the primary commercial cultivar, ‘Black Mitcham’, is a sterile hexaploid (2n = 6x = 72; Harley and Brighton 1977) and as such is not amenable to conventional breeding.

The genetics of verticillium wilt resistance has not been studied in any of the commercial mints. We are developing the wild, diploid species M. longifolia as a model for mint genetic and genomic research with initial emphasis on economically important characteristics such as disease resistance and essential oil quality (Vining et al. 2005). Toward this end, we identified M. longifolia accessions with varying phenotypic resistance to verticillium wilt (Vining et al. 2005) and cloned resistance gene analog (RGA) sequences from M. longifolia accessions (Vining et al. 2007). Furthermore, we used degenerate primers based on the tomato Ve1 gene to isolate a 445 bp Ve-like sequence from M. longifolia, then extended this sequence using inverse PCR, obtaining a 1,413 bp sequence having 56–57% predicted amino acid identity to the equivalent regions of tomato Ve1 and Ve2 (Vining et al. 2007). Here, we report the putatively complete coding sequence of a Mentha Ve homolog, mVe1, and describe its allelic diversity in a sampling of M. longifolia and commercial mint germplasm. As part of this study, the segregation patterns of mVe1 alleles were determined in M. longifolia genetic populations, including F1 and F2 populations derived from crosses of wilt-resistant and wilt-susceptible M. longifolia accessions. The possibility of association between mVe1 genotype and wilt resistance/susceptibility in these populations was also assessed.

Materials and methods

Plant materials

Mentha accessions were obtained from the National Clonal Germplasm Repository (NCGR) in Corvallis, Oregon. Guided by prior screenings of diploid M. longifolia germplasm accessions (Vining et al. 2005), verticillium wilt-resistant CMEN 585 (PI 557767; South Africa) and CMEN 17 (PI 557755; Europe) and susceptible CMEN 584 (PI 557769; South Africa) and CMEN 516 (PI 557760; Europe), were used as parents in resistant (R) × susceptible (S) crosses. M. × piperita accession ‘Black Mitcham’ (PI 557971) and M. spicata (‘Native spearmint’) accession (PI 557793) represented the cultivated mint species.

mVe1 isolation and sequencing

Three site-specific primers based on an mVe1 sequence segment from M. longifolia CMEN 585 (Vining et al. 2007) and four generic, degenerate primers (Table 1) were used with template DNA from CMEN 585 in Thermal Asymmetric InterLaced (TAIL)-PCR (Liu et al. 1995; Liu and Whittier 1995), using the Accuprime™ Taq DNA Polymerase System (Invitrogen, Carlsbad, CA), and thermal cycling parameters described in Liu et al. (1995). Primary TAIL-PCRs contained 3 μM degenerate primer and 0.4 μM specific primer. Products from primary TAIL-PCR were diluted 1:50 with nuclease-free water for use as templates in secondary TAIL-PCR. Secondary and tertiary TAIL-PCRs contained 1.8 μM degenerate primer and 0.4 μM specific primer. Products from secondary PCR were used directly in tertiary PCR, without dilution. Secondary and tertiary TAIL-PCR products obtained using TAIL primers AD1 and AD6 in combination with mVe-specific primers were cloned and sequenced using methods described in Vining et al. (2007).

Table 1 Primers and probes used in this study
Full size table

Once a putatively complete mVe1 coding region was assembled, primers MA151 and MA148 (Table 1) were designed to capture this entire coding region from CMEN 585, CMEN 584, CMEN 17 and CMEN 516 and from ‘Black Mitcham’ peppermint and ‘Native spearmint’. The complete sequence of each PCR clone was then ascertained using a series of mVe-specific primers, plus standard vector (M13F, M13R) sequencing primers.

Sequence alignments, ORF translations and cluster analyses were performed using the software packages Lasergene MegAlign (DNAStar) and Clustal X (Thompson et al. 1997). GenBank BLAST searches were done on the National Center for Biotechnology Information web site (http://www.ncbi.nlm.nih.gov). Alignments of mVe1 predicted amino acid sequence with the tomato Ve-predicted amino acid sequences were used to delineate protein domains. Further structural analyses of the putative mVe1 protein were executed on the Swiss EMBnet node server (http://www.ch.embnet.org) and ExPASy Molecular Biology Server (http://www.expasy.org).

Genotyping of F1 and F2 populations

Primers MA151 and MA148 were used genotype parents and progeny populations that had been screened for verticillium wilt resistance. A primer pair (ML414_79F, sequence 5′GCATCACCAACAGCGAACA3′ and ML414_528R, sequence 5′CGGAGCAGACGATGAGGAC3′) targeting a mint chalcone synthase gene [(Lange et al. 2000): GenBank accession number AW255394] was designed and employed as a positive control to confirm template quality.

Alignments of completed allele sequences were used to identify polymorphic restriction sites. For population genotyping, 5 μl of PCR products from parents, F1 and F2 plants were digested in 20 μl reactions with 10 units of NcoI and XbaI for the South African populations, or 10 units of AgeI for the European populations (New England Biolabs, Ipswich, MA). All digests were incubated overnight at 37°C and digested products were electrophoretically separated on 2% agarose gels. mVe1 segregation data sets were evaluated by χ2 analysis for goodness-of-fit to appropriate segregation ratios at a 0.05 probability level.

Verticillium wilt resistance screening and genotype–phenotype association testing

Verticillium wilt resistance screenings were performed on parents and progeny populations as described in Vining et al. (2005). The hypothesis of association between verticillium wilt resistance rating and allele composition in the European and South African F2 populations was tested using non-parametric, Kruskal–Wallis single-factor analysis of variance by ranks, performed with Systat© version 10 (SPSS Inc., Chicago, IL). A post hoc Tukey’s test was performed in the case of the South African F2 population, where pairwise comparisons were appropriate.

Results

mVe1

Using TAIL-PCR with template genomic DNA from verticillium wilt-resistant M. longifolia accession CMEN 585, we extended a previously obtained Ve-like fragment (Vining et al. 2007) to complete a draft sequence of a putative homolog, mVe1, of the tomato Ve genes. Initial sequence assembly of the degenerate, inverse and TAIL-PCR fragments produced a sequence of 3,328 nucleotides (nt), including 210 nt upstream of the putative start codon and 67 nt downstream of the putative stop codon. This draft sequence provided the basis for design of primers MA151 and MA148 (Table 1), which were targeted to putative 5′ and 3′ UTR sequences, respectively, allowing amplification of a single product spanning the entire mVe1 coding region and including some flanking UTR sequence. Using this primer pair to amplify and then clone mVe1 sequences from CMEN 585, two putative mVe1 alleles were identified based on the restriction digests of plasmid clones and subsequent sequencing. The allele best matching the original mVe1 draft sequence was allele II (mVe1-II), and is considered the canonical mVe1 sequence.

The canonical mVe1-II sequence has a predicted coding region of 3,054 nt, in one uninterrupted open-reading frame. The predicted stop codon (TAA) is followed immediately (in frame) by two more stop codons (TGATGA) (Fig. 1). The predicted mVe1 protein, 1,017 amino acids in length, is rich in leucine residues (15.2%). Five distinct sequence domains were identified in this predicted protein corresponding to the domains A through E of the tomato Ve proteins as defined by Kawchuk et al. (2001) (Fig. 2). The amino terminus of 24 amino acids (domain A) precedes an LRR region (amino acids 25–935, domain B) with 36 imperfect copies of the consensus sequence [XXIXNLXXLXXLXLSXNXLSGXIP]. Immediately adjacent to the LRR domain, a negatively charged stretch (amino acids 936–951, domain C) precedes a putative transmembrane domain (amino acids 952–972, domain D). The carboxy terminus (amino acids 973–1,017, domain E) is positively charged and contains a stretch of seven consecutive arginine residues.

Fig. 1
figure 1

Schematic drawing of Mentha longifolia mVe1 sequence (not to scale). The predicted coding region is shown in gray; 5′ and 3′ UTRs are white. a Initial assembly of mVe1 from degenerate, inverse and TAIL-PCR fragments. The 3′ 1,413 bp were obtained with degenerate PCR and inverse PCR (Vining et al. 2007). The 5′ 1,915 bp were obtained during the present study with TAIL-PCR. Relative locations of nested specific primers used in TAIL-PCR (sequences in Table 1) are indicated. b Locations of mVe1 primers used to clone coding regions from M. longifolia accessions. Clones included 23 bp of 5′ UTR and 54 bp of 3′ UTR

Full size image
Fig. 2
figure 2

Primary amino acid sequence of M. longifolia mVe1 deduced from genomic DNA sequence. The polypeptide has been divided into predicted domains A–E as described in the text, based on alignments with the tomato Ve putative protein sequences. Domain A amino terminus. Domain B leucine-rich repeat (LRR) domain, separated into individual, imperfect copies of the consensus motif: [XXIXNLXXLXXLXLSXNXLSGXIP]. Leucine residues are in boldface type. Domain C extracytoplasmic domain, domain D transmembrane domain, domain E cytoplasmic domain

Full size image

When subjected to Blastx homology searches of the GenBank nr database using the translated, canonical mVe1 sequence as a search query, the top hit was an “SlVe2 precursor” sequence from S. lycopersicoides (Chai et al. 2003) (51.7% identities, score = 863, e value = 0.0), while tomato Ve2 was the second best Blastx hit (score = 861, e value = 0.0). The remaining top 10 Blastx hits included tomato Ve1, other “SlVe1 and SlVe2 precursor” sequences from S. lycopersicoides and sequences simply titled “verticillium wilt resistance protein” from S. torvum. Identities between the translated mVe1 sequence and the Solanum sequences were 48–50%. Other mVe1 Blastx hits were LRR proteins and predicted proteins from a variety of plant taxa, including a number of unnamed and hypothetical protein products from Vitis vinifera (≤41% identities).

Pairwise alignments of the predicted mVe1 gene product with tomato Ve1 and Ve2 predicted gene products showed 51.6 and 51.1% amino acid identities, respectively. When each of the tomato Ve amino acid sequences was used as query in a Blastp search of the nr database, the top hits were all sequences from other solanaceous species (scores ≤ 2,149, e values = 0.0). Outside of the Solanaceae, the closest Blastp matches for the tomato Ve1 and Ve2 translations were to V. vinifera contigs (scores ≤ 509, e values = 0.0), followed by the Mentha mVe1 translations (scores ≤ 247, e values = 0.0). These results show that possible homologs of the tomato Ve genes exist in plant families other than the Solanaceae.

Again using PCR primers MA151 and MA148, mVe1 alleles were PCR-amplified, cloned and sequenced from the four studied M. longifolia accessions used in this study, as well as peppermint (M. × piperita) cultivar Black Mitcham, and a Native spearmint (M. spicata) accession, resulting in the identification of 13 distinguishable mVe1 alleles. From M. longifolia, two alleles each were identified from CMEN 585 (alleles I and II) and CMEN 584 (alleles III and IV), one allele was identified from CMEN 17 (allele V) and two were identified from CMEN 516 (alleles VII and VIII) (Fig. 3). Four mVe1 alleles were identified from ‘Black Mitcham’ peppermint (alleles BM5, BM7, BM8 and BM10), and two were identified from the Native spearmint accession (alleles NS1 and NS2). These 13 distinct mVe1 alleles are listed in the GenBank sequence database under accession nos. EU587375–EU587387. Overall, these alleles have 94.7–99.9% predicted amino acid identity with each other, 50.2–51.4% predicted amino acid identity with tomato Ve1, and 50.0–50.8% predicted amino acid identity with tomato Ve2. All coding regions end in three consecutive stop codons. Eleven of the allele sequences encode predicted proteins of 1,017 amino acids, while two sequences from ‘Black Mitcham’ (mVe1BM7, mVe1BM10) encode predicted proteins of 1,016 amino acids (Fig. 3). Polymorphisms among the mVe1 alleles are limited to base substitutions, with the exception of a single trinucleotide deletion in alleles mVe1BM7 and mVe1BM10 corresponding to nucleotides 1,251–1,253 in the other sequences.

Fig. 3
figure 3figure 3figure 3figure 3

ClustalX alignment of mVe1 deduced amino acid sequences from ‘Black Mitcham’ peppermint (Mentha xpiperita) (alleles BM5, BM7, BM8, BM10), Native spearmint (Mentha spicata) (alleles NS1, NS2) and Mentha longifolia (alleles I, II, III, IV, V, VII, VIII). Periods represent conserved amino acid residues; other letters show differences from the concensus

Full size image

Sequence alignments of mVe1 alleles revealed strong similarities among accessions from common geographic origins, as well as striking differences between alleles of South African versus European accessions (Fig. 3). In pairwise comparisons, the two most similar alleles were from South African accessions CMEN 585 and CMEN 584 (alleles I and III, respectively), which differed at only three nucleotides and one amino acid. Pairs of alleles from South African plants averaged nine nucleotide differences, and alleles from European accessions differed from each other by an average of four nucleotides. In contrast, there was an average of 73 nucleotide differences in pairwise comparisons of South African with European alleles.

In the phylogenetic analysis, the mVe1 alleles grouped into two major clades (Fig. 4). The two alleles from Native spearmint grouped with the alleles from the European accessions of M. longifolia, as did three of the four alleles from ‘Black Mitcham’ peppermint. The fourth ‘Black Mitcham’ mVe1 allele grouped with those from the South African accessions.

Fig. 4
figure 4

Clustering analysis of mVe1 sequences based on the predicted amino acid sequence alignments from ClustalX. The Solanum lycopersicum Ve2 sequence is used as an out group. The sequences are from ‘Black Mitcham’ peppermint (Mentha xpiperita) (BM5, BM7, BM8, BM10), Native spearmint (Mentha spicata) (NS1, NS2) and Mentha longifolia (I, II, III, IV, V, VII, VIII). Cluster SA includes sequences from South African M. longifolia accessions and from Black Mitcham peppermint; cluster E includes sequences from European M. longifolia accessions, Black Mitcham peppermint and Native spearmint

Full size image

mVe1 genotyping

The inheritance of mVe1 alleles, and the possibility of association of an mVe1 genotype with a resistant or susceptible phenotype, was investigated using progeny populations from resistant × susceptible crosses made using pairs of plants from South Africa and Europe. The South African (SA) cross (CMEN585 × CMEN 584) produced an F1 population (SAF1) of 55 plants, and self-pollination of one Verticillium-resistant F1 plant (#SAF1-1) produced an F2 generation population (SAF2) of 120 plants. From the European (E) cross (CMEN17 × CMEN 516), self-pollination of one wilt-resistant F1 plant (#EF1-8) produced an F2 generation population (EF2) of 18 plants. Self-pollination of CMEN 17 produced an S1 population (E17S) of 31 plants. Owing to the logistical considerations, including the death of some plants during the term of the study, most of the population genotyping and wilt resistance screening assays were done on population subsets.

Based on the comparison of allele sequences from South African parent plants CMEN 585 (genotype I/II) and CMEN 584 (genotype III/IV), informative restriction site polymorphisms were identified and then used to determine mVe1 genotypes in the SAF1 and SAF2 populations. An NcoI cut site was present in alleles II, III and IV at 2,507–2,512 bp; an XbaI cut site was present in allele IV at 1,163–1,168 bp. Allele I did not contain a restriction site for either enzyme. Digestion of mVe1 PCR products from CMEN 584 and CMEN 585 with either NcoI or XbaI produced the fragment sizes expected from the known restriction site positions in each allele (Fig. 5). Given the heterozygous genotypes of the South African parent plants, there were four possible F1 genotypes, all of which were represented in a subset of 40 SAF1 plants that were subjected to genotyping: 13 plants were genotype I/IV; 12 were II/IV; 8 were II/III; and 7 were I/III.

Fig. 5
figure 5

Restriction digest patterns used to distinguish mVe1 alleles. Lane 1 1 Kb Plus ladder (Invitrogen), lane 2 CMEN 585, undigested; lane 3 CMEN 584, undigested, lane 4 CMEN 585, NcoI digested, lane 5 CMEN 584, NcoI digested, lane 6 CMEN 585, XbaI digested, lane 7 CMEN 584, XbaI digested

Full size image

The SAF1 plant (SAF1-1) used to produce the SAF2 generation had allele composition I/IV. In the SAF2 population, 107 plants were genotyped. As inferred from gel banding patterns, 33 were homozygous for the CMEN 585-derived allele (I/I), 61 were heterozygous (I/IV), and 13 were homozygous for the CMEN 584-derived allele (IV/IV). These results deviated significantly from a 1:2:1 monogenic segregation ratio (χ2 = 9.58, P = 0.0083). Allele I had a frequency (0.59) higher than the 0.5 frequency expected in a population segregating 1:2:1, while allele IV had a correspondingly lower frequency (0.41).

Genotyping of the European (E) populations illuminated the allele composition of CMEN 17, for which only one allele (allele V) had been identified by sequencing. No mVe1 PCR product was generated in 6 out of 19 ES1 plants and 8 out of 21 EF2 plants. Template quality was confirmed in all plants by successful amplification with the positive control chalcone synthase primer pair, suggesting the presence of an mVe1 “PCR-null” (non-amplifying) allele in CMEN 17. Given the allele designation VI(n) for its PCR-null allele, the deduced genotype of CMEN 17 was V/VI(n). Subsequently, a variety of primer pairs spanning different subregions of mVe1 were used to test for the presence of an indel polymorphism. Primer pairs produced PCR products of the expected size, with the exception that pairs having the 3′ primer either within ~200 bp of the 3′ end of the ORF or in the 3′ UTR did not produce a product in individuals known to be homozygous for the VI(n) allele (results not shown). This result indicated that either a deletion (eliminating a priming site) or an insertion (increasing the primer-to-primer distance beyond amplification range) of indeterminate length was present in the downstream region of allele VI(n).

Based on their allele sequences, the AgeI restriction enzyme was expected to cut at a site spanning base pairs 1,768–1,773 in alleles V, VII and VIII and also at a site spanning base pairs 1,197–1,202 in allele VII. The AgeI restriction digest banding patterns expected and obtained in the E parents, EF1 and EF2 populations are diagrammed in Fig. 6. The banding pattern obtained for CMEN17 conformed to the known presence of a single restriction site in allele V, its only amplified allele. The CMEN 516 pattern conformed to the known presence of one restriction site in allele VIII (resulting in two bands of 1,769 bp and 1,362 bp, equal in size to the two bands characteristic of allele V) and two restriction sites in allele VII (resulting in bands of 1,198, 571 and 1,362 bp). Because alleles VII and VIII each produce a band of 1,362 bp, CMEN 516 has only four gels bands (Fig. 6).

Fig. 6
figure 6

Restriction digest patterns of mVe1 genotypes observed and expected in EF1 and EF2 populations

Full size image

Given the defined genotypes of CMEN 17 and CMEN 516, four possible genotypes, and three possible banding patterns, were expected in the EF1 population (Fig. 6). Of the nine EF1 plants, six had banding pattern A and three had banding pattern B. Expected pattern C did not occur, but its absence could be due to sampling bias in this small population subset. In the EF2 population, five plants had genotype VII/− and eight plants were homozygous for the null allele VI(n)/VI(n).

Verticillium wilt resistance screening and association study

When a standardized root dip inoculation technique was used to determine verticillium wilt phenotypes of progeny plants, a more or less continuous response spectrum was observed in SAF1 and SAF2 progeny populations (results not shown). A significant genotype effect was seen only in the SAF2 population (P = 0.046) and a post hoc, pairwise Tukey’s test detected a significant (P = 0.037) difference between the phenotypes associated with genotypes I/I (mean disease rating = 2.37) and IV/IV (mean disease rating = 0.50). The EF1 progeny population had a distinctly bimodal distribution, with six highly resistant and ten highly susceptible plants, while the EF2 population had a continuous distribution of phenotypes. No significant phenotype–genotype associations were seen in either population. However, it is noteworthy that of the nine EF2 plants with genotype VI(n)/VI(n), four had disease ratings of 0.0 (highly resistant).

Discussion

Starting with a 3′ gene segment of 1,413 bp which we had previously obtained via degenerate PCR and inverse PCR (Vining et al. 2007) and which extended 64 bp downstream of the putative mVe1 stop codon, we used TAIL-PCR to acquire an additional 1,915 bp in the 5′ direction, thereby extending our mVe1 sequence far enough to encompass the putative mVe1 start codon. The putative mVe1 start codon environment (ACAATGG) is consistent with the Kozak consensus sequence (RCCATGG: Kozak 1986). Thus, we have acquired a total of 3,328 of mVe1 sequence from M. longifolia CMEN 585, encompassing the putatively complete 3,054 bp mVe1 coding sequence.

Using primers targeted to the 5′ and 3′ UTRs of this initial, canonical sequence, we then amplified and cloned additional mVe1 sequences from CMEN 585 and three other M. longifolia accessions, as well as from ‘Black Mitcham’ peppermint and Native spearmint. All of the mVe1 sequences obtained from these accessions contain uninterrupted ORFs, translations of which align well, and have relatively high homology with the tomato Ve-predicted proteins, indicating the mint sequences could also encode functional proteins. Given its uninterrupted ORF, the mVe1 gene evidently has no introns, as is the case for the tomato Ve genes (Kawchuk et al. 2001).

The mVe1 gene is predicted to encode a protein with 36 LRRs, in contrast to the 38 LRR units of tomato Ve1 and Ve2 genes (Kawchuk et al. 2001). The amino acid alignment of the predicted mVe1 and tomato Ve1 and Ve2 proteins showed two gaps in the mVe1 sequences relative to the tomato sequences, which corresponded to LRR units 14 and 24 in both tomato Ve1 and Ve2. These gaps could represent the loss or gain of individual repeats in the respective mint and tomato genes, as compared to their common ancestor. Interestingly, when Ve homolog sequences from other solanaceous species were included in alignments, LRR units 14 and 24 were present in all Solanaceae sequences (data not shown). Therefore, if the absence of LRR units 14 and 24 was the ancestral state in the last common ancestor of the Lamiaceae and Solanaceae, then the gain of these LRR units evidently preceded divergence among Solanaceae species.

The tomato Ve locus contains two adjacent genes, both of which reportedly conferred resistance to Verticillium spp. Race 1 when introduced into a susceptible potato variety (Kawchuk et al. 2001). To date, there are no reports of functionality tests of potato or turkeyberry Ve homologs. It is not known which, if any, amino acids in the LRR domain of Ve are key determinants of wilt resistance.

The cloning of the canonical mVe1 sequence from M. longifolia accession CMEN 585 facilitated the cloning of mVe1 sequences from other Mentha accessions, just as the previous identification of the tomato Ve genes provided a foundation for the homology-based cloning of full-length and partial potential Ve homologs from other members of the Solanaceae: SlVe1 from diploid potato (Chai et al. 2003), 11 StVe1 partial alleles from the cultivated, tetraploid potato (Simko et al. 2004a) and StVe from turkeyberry (Fei et al. 2004). The 11 StVe1 sequences share 76.2–99.6% predicted amino acid identity with each other, and 82.9–90.8% and 74.2–90.8% to tomato Ve1 and Ve2, respectively. In mint, among the mVe1 alleles identified in this study, overall amino acid similarity is much higher from 94.7 to 100%. The mVe1 alleles from European M. longifolia accessions are more similar to each other than are the alleles from South African M. longifolia accessions. This is not surprising since the two South African accessions are considered different subspecies: CMEN 585 is M. longifolia subsp. capensis, while CMEN 584 is M. longifolia subsp. polyadena (Tucker and Naczi 2005).

In addition to sequencing alleles from M. longifolia, alleles were sequenced from an M. spicata Native spearmint accession and from Mentha × piperita peppermint cultivar Black Mitcham for comparative purposes. These sampled alleles are not assumed to comprise all of the mVe1 alleles in the two cultivated accessions. The triploid (USDA Agricultural Research Service 2008) Native spearmint accession used in this study could possess up to three different mVe1 alleles. Black Mitcham peppermint, a hexaploid, may have as many as six different alleles of mVe1. In a cluster analysis, three of the four mVe1 alleles from peppermint grouped with alleles from the European M. longifolia accessions, while the fourth peppermint allele grouped with those from South African M. longifolia accessions. Given that M. longifolia is a presumed ancestor of M. × piperita (Tucker and Naczi 2005), our results draw attention to both European and South African M. longifolia germplasm as possible allele contributors to hexaploid peppermint, and indicate that mVe1 sequences could be phylogenetically informative if a broader geographic sampling of allelic diversity could be compiled.

In Verticillium dahliae screens, three of the four F1 and F2 M. longifolia populations exhibited more or less continuous patterns of segregation for resistance versus susceptibility to verticillium wilt, while the small EF1 progeny population had a distinctly bimodal distribution.

Segregation distortion favoring the mVe1-I allele, as evident from the deviation from a 1:2:1 Mendelian expectation was observed in the overall SAF2 population of 107 plants and in the subpopulation of 52 plants used in the association analysis. Segregation distortion is a common phenomenon in progenies derived from wide crosses (Sargent et al. 2004), but unless the distortion is extreme it does not preclude the quantification of linkage or the detection of genotype–phenotype association.

The detection of association between phenotype (symptom rating) and mVe1 genotype in the SAF2 population, but not the SAF1, EF1 and EF2 populations provides conflicting yet tantalizing evidence relevant to the possible role of mVe1 as a potential component of a multigenic resistance system in M. longifolia. Weighing against a possible role for mVe1 in resistance is the finding that four plants of the EF2 population having genotype VI(n)/VI(n) were rated as highly resistant. If mVe1 is a necessary component of resistance, and if the functionality of the VI(n) allele has been eliminated by a downstream deletion or insertion, then VI(n) homozygotes would not be expected to be highly resistant. These considerations point to the need for investigation of the expression and functionality of mVe1 gene products, about which nothing is currently known.

Overall, the association study results suggest that one or more resistance genes other than mVe1 are segregating and account for the genotypic components of phenotypic variation in the SAF1, EF1 and EF2 populations and probably for a portion of phenotypic variation in the SAF2 population. Certainly, the lack of association with resistance/susceptibility in the SAF1, EF1 and EF2 populations does not in any way rule out an mVe1 gene product as a component of a mint verticillium wilt resistance system. One possibility is that all of the mVe1 alleles segregating in the tested populations are functional alleles, and that a major effect of the mVe1 locus will only be seen in populations segregating for a non-functional mVe1 allele. mVe1-VI(n) may prove to be such an allele, but as previously noted, we as yet know nothing about its functionality or lack thereof.

Another possibility is that an effect of any functional mVe1 allele will only be apparent when that allele is in homozygous form in an appropriate genetic background, in which other phenotypically relevant genes are present in a particular homo or heterozygous state. In the M. longifolia SAF2 population, allele IV, derived from the susceptible accession CMEN 584, was associated with a resistant phenotype. The effect of other genes, and the allelic state of those genes relative to that of mVe1, is not known in this population. Hence, the development of inbred lines and expanded association studies focused on evaluating the segregation patterns of various mVe1 allele combinations in different genetic backgrounds is warranted. Toward this end, we have developed three F2 populations as a basis for continuing research: one segregating for allele VI(n), one segregating for alleles I and III, and one segregating for alleles II and IV.

Efforts to dissect the genetics of verticillium wilt resistance in mint are at an early stage. We have isolated and sequenced the mVe1 gene, a candidate gene that ranks among the top homology matches in the GenBank database to the tomato Ve genes. A diverse sampling of mVe1 alleles has also been sequenced. Our association analysis of several mVe1 alleles in the studied populations leaves the relationship between this gene and verticillium wilt resistance in mint unresolved, but encourages continued investigation. It is also imperative to identify other genetic components of the mint verticillium defence system, and, as part of this effort, we have initiated development an M. longifolia linkage map as a framework for future association studies. Thus, the results reported here provide a basis for continuing investigations along several indicated lines.