Evidence that Polymyxa species may infect Arabidopsis thaliana


  • Editor: Claire Remacle

  • Present address: Madeleine J. Smith, Texas AgriLife Research and Extension Center, 6500 Amarillo Blvd. West, Amarillo, TX 79106, USA

Correspondence: Elaine Ward, Department of Plant Pathology and Microbiology, Rothamsted Research, Harpenden, Hertfordshire AL5 2JQ, UK. Tel.: +44 1582 763 133, ext. 2495; fax: +44 1582 760 981; e-mail: elaine.ward@bbsrc.ac.uk


Polymyxa spp. are obligate biotrophs belonging to the plasmodiophorid group, responsible for transmitting a large number of plant viruses to many crop species. Their obligate nature makes them difficult to study. Controlled environment experiments were used to investigate the potential of infection of Arabidopsis thaliana by Polymyxa spp. to provide a more tractable system. Two ecotypes of Arabidopsis, Columbia and Landsberg erecta, were grown in soils known to be infested with Polymyxa. At the end of a 2-month growth period, both ecotypes were found to harbour Polymyxa-like structures or spores. These findings were confirmed by Polymyxa-specific PCR tests and rDNA sequencing, which positively identified the presence of Polymyxa in the roots of both ecotypes of Arabidopsis. Both Polymyxa graminis and Polymyxa betae were identified. This is the first report of infection of Arabidopsis by Polymyxa spp. and shows the possibility of using this system for studies of infection biology and host–parasite interactions.


Polymyxa spp. are a group of obligate root-infecting organisms belonging to the plasmodiophorid group that are important plant–virus vectors (Kanyuka et al., 2003). Polymyxa graminis transmits viruses such as soil-borne cereal mosaic virus (SBCMV), soil-borne wheat mosaic virus and wheat spindle streak mosaic virus to cereals. Polymyxa betae transmits beet necrotic yellow vein virus, the cause of rhizomania, to sugar beet. Polymyxa graminis has a wide host range including wheat, barley, rye, rice, sorghum, groundnut and various grasses, whereas P. betae is mostly restricted to beet and other plants in the family Chenopodiaceae.

A number of subgroups (ribotypes) of Polymyxa spp. have been identified according to rDNA sequence data (Ward et al., 1994, 2005; Ward & Adams, 1998; Legrève et al., 2002). Some of the P. graminis ribotypes appear to differ in host range and temperature requirements, leading to the suggestion that they should be classified as formae speciales (Legrève et al., 1998, 2002). Two groups of P. graminis isolates are found in temperate regions: ribotype I (f. sp. temperata) and ribotype II (f. sp. tepida). All internal transcribed spacer (ITS) rDNA sequences for P. betae reported to date fall into two types that differ by only one base pair (Ward & Adams, 1998; Legrève et al., 2002).

Because of their obligate nature and relatively long life cycle, Polymyxa spp. have been difficult to study. The development of a model system for studying Polymyxa–plant interactions would be extremely useful. Arabidopsis thaliana is an invaluable model system for several reasons: (1) short generation time, (2) the ability to grow large numbers in a relatively small space, (3) its ability to self-fertilize, (4) the large number of progeny that can be produced from a single plant, (5) its small haploid genome containing a relatively small number of repetitive genetic elements, (6) the availability of a fully sequenced genome, (7) the availability of mutagenized lines, (8) ease of transformation and (9) the large number of ecotypes exhibiting natural variation available (Meyerowitz, 1989). These features are in contrast to many crop species such as cereals, where genetic resources are less well advanced.

Arabidopsis has already been used very successfully to study the interactions of another plasmodiophorid: Plasmodiophora brassicae (Koch et al., 1991). The ability to separate host sequences from those of Plasmodiophora by bioinformatics analysis has simplified the interpretation of data, for example from suppressive subtractive hybridization experiments to study gene structure and expression (Bulman et al., 2006, 2007). Sources of resistance and factors important for the infection of Plasmodiophora have been studied by exploring the responses to both natural and induced (mutagenic) variation in host genes affecting infection (Siemens et al., 2002; Alix et al., 2007). Arabidopsis has been used to visualize infection biology of P. brassicae (Mithen & Magrath, 1992). The availability of synteny maps between Arabidopsis and Brassica spp. has allowed the identification of resistance loci in Brassica spp. first identified in Arabidopsis (Suwabe et al., 2006). Global analysis of host gene expression at different time points postinfection has been possible using Arabidopsis genome arrays, and this has allowed the identification of host genes that may be important for infection by Plasmodiophora (Siemens et al., 2006). Genes of interest can then be studied further by transforming into Arabidopsis or by utilizing the bank of insertion lines available in Arabidopsis (Puzio et al., 2000; Siemens et al., 2006).

Many of the host plants that Polymyxa spp. infect are not well characterized genetically, have fewer genetic tools available and they have long generation times. Also, the roots of cereals can be difficult to visualize by microscopy as they are thicker in diameter than those of Arabidopsis. This can sometimes make the visual detection of Polymyxa in roots difficult. Therefore, if infection of Arabidopsis by Polymyxa spp. can be demonstrated, this could be a valuable tool in increasing our understanding of plant–Polymyxa interactions. This study aimed to look at the potential for infection of Arabidopsis by Polymyxa spp. under controlled environment conditions using Polymyxa-infested soils.

Materials and methods

Arabidopsis thaliana ecotypes Landsberg erecta (Ler-0) and Columbia (Col-0) were used for this study (supplied by A. Cuzick, Rothamsted Research, UK). These ecotypes were chosen because they are genetically distinct and mapping populations are available. Seeds were sown into sterile Levingtons No. 2 compost containing sand and stratified for 4 days in the dark at 4 °C. Pots were then removed and placed in a greenhouse under short-day length conditions (8 h day at 20 °C, 16 °C night, light levels 200–300 μmol m−2 s−1). Once the seedlings had produced their first true leaves, they were transferred to 10 cm pots containing infectious soils diluted 1 : 2, soil to sterile sand and grown as before. Two UK soils were used: one from Wiltshire, which was infested with SBCMV (Lyons et al., 2008), and one from Woburn, where Polymyxa was present, but no associated virus had ever been identified (Ward et al., 2005; R. Lyons, pers. commun.). For each soil, five seedlings of each ecotype were planted. Plants were then allowed to grow for 2 months. Flowering bolts were removed upon development to prolong vegetative growth.

Roots were removed from pots and vigorously washed in sterile, distilled water. Portions of root were then mounted in sterile water under a coverslip and examined using an Axiophot (Zeiss) light microscope with bright field illumination. Roots were separated into three 3-cm sections: one from the base of the plant, one from the middle of the root and one from the root tip. Around 20 pieces of each section of root were examined for each of the five plants from each ecotype– soil combination (i.e. approximately 60 root pieces per plant).

DNA was extracted from approximately 0.5 g freeze-dried and ground root material (one root system for each ecotype–soil combination) as described by Ward et al. (2005). Polymyxa-specific rDNA primers Pxfwd1 (5′-CTG CGG AAG GAT CAT TAG CGT T-3′) and Pxrev7 (5′-GAG GCA TGC TTC CGA GGG CTC T-3′) were used in PCR (Ward & Adams, 1998). Plasmodiophora-specific PCR was performed as in Cao et al. (2007) using primers TC1F/TC1R. For sequencing studies, the Polymyxa-specific forward primer Pxfwd1 and the generic fungal ITS4 reverse primer (5′-TCC TCC GCT TAT TGA TAT GC-3′) (White et al., 1990) were used to amplify rDNA. Each reaction mix (50 μL) contained 0.2 μM primers, 1 U Taq DNA polymerase (MBI), 0.2 mM dNTPs (Sigma), 1 × PCR buffer NH4 (MBI) and 0.02 mg μL−1 bovine serum albumin. Cycling conditions were 2 min at 95 °C, and then 30 cycles of 94 °C for 30 s, 50 °C for 1 min and 72 °C for 2 min, followed by 72 °C for 10 min. Products were analysed in 1% agarose gels.

PCR products were cloned into the pGEM®-T Easy vector (Promega Corporation, Madison, WI). Plasmid DNA was prepared using the QIAprep spin miniprep kit (Qiagen, Crawley, UK) and sequenced using the ABI PRISM Big-Dye version 1.1 kit using M13 sequencing primers and run at the Geneservice sequencing facility (http://www.geneservice.co.uk). ITS rDNA sequences were aligned by clustalx and manually adjusted. Phylogenetic analysis was performed using the neighbour-joining method (maximum composite likelihood distances) in mega4 (Tamura et al., 2007) with 10 000 bootstrap replications.

Results and discussion

Examination by microscopy showed the presence of Polymyxa-like spores in numerous root hairs (but not the main root) of all five Arabidopsis ecotype Ler-0 plants grown in the Woburn soil (Fig. 1). Two of the Col-0 plants grown in the Woburn soil contained structures that resembled Polymyxa zoosporangia (Fig. 2). Three of these structures were seen in total and they were all located in the main root system rather than the root hairs. No spore clusters were observed. In the root sections examined from Arabidopsis plants grown in the Wiltshire soil, no clusters of Polymyxa-like resting spores or zoosporangia were identified.

Figure 1.

 Root hair from Arabidopsis Ler-0 plant grown in Woburn soil containing plasmodiophorid-like spore clusters.

Figure 2.

Polymyxa-like zoosporangial structure in a main root cell of an Arabidopsis Col-0 plant grown in Woburn soil.

PCR with the Polymyxa-specific primers Pxfwd1/Pxrev7 demonstrated the presence of Polymyxa spp. in the roots of all four combinations of Arabidopsis ecotypes and soils (Fig. 3). Using a Plasmodiophora-specific PCR assay, we also demonstrated that Plasmodiophora was not present in these samples (Fig. 3). A total of 28 clones were sequenced following the amplification of rDNA products from Arabidopsis roots using primers Pxfwd1/ITS4. These included representatives of all of the restriction fragment length polymorphism types obtained by prescreening the clones using DdeI. Eleven of these sequences showed significant identity to P. graminis F1 ITS ribosomal DNA (Table 1), one to P. betae F67 ITS rDNA and nine to Arabidopsis rDNA. For the remaining seven sequences, the closest matches were to uncultured Basidiomycetes (two clones) and an uncultured Helotiales, and there were partial matches (short regions of high identity in a limited part) to Urostyla grandis, Anguina agropyri and an ectomycorrhizal fungus. The nucleotide sequence of one clone showed no significant identity to any sequence in GenBank. The identification of Arabidopsis and other non-Polymyxa sequences in the roots is not unexpected, as only one of the primers used (Pxfwd1) is Polymyxa specific, whereas the ITS4 primer is a generic, ‘fungal’ rDNA primer.

Figure 3.

 Testing of DNA from Arabidopsis plants and soil with specific PCRs for Polymyxa (lanes 1–10) and Plasmodiophora (lanes 11–20). Lanes 1 and 11, Arabidopsis Ler-0 from Wiltshire soil; lanes 2 and 12, Arabidopsis Col-0 from Wiltshire soil; lanes 3 and 13, Arabidopsis Ler-0 from Woburn soil; lanes 4 and 14, Arabidopsis Col-0 from Woburn soil; lanes 5 and 15, healthy uninfected Arabidopsis Ler-0; lanes 6 and 16, healthy uninfected Arabidopsis Col-0; lanes 7 and 17, soil from Woburn; lanes 8 and 18, Polymyxa DNA control (F1); lanes 9 and 19, Plasmodiophora DNA control (isolate N); lanes 10 and 20, no DNA control; M, size marker (100 bp ladder, Fermentas).

Table 1.   Results of blast hits for sequenced clones with significant nucleotide identity to Polymyxa rDNA
Arabidopsis ecotypeSoilClone no.blast hit* (distmat value)GenBank accession no.
  • *

    Closest match showing the percentage nucleotide identity between the query and the database sequences.

  • Percentage nucleotide identity calculated from the clustalx alignment using the emboss program distmat (Rice et al., 2000).

  • Accession number: Y12824.

  • §

    Accession number: Y12827.

LandsbergWoburnLeWob34PgF1 98% (99.4%)FN393973
LandsbergWiltshireLeWil3PgF1 99% (99.4%)FN393974
LandsbergWiltshireLeWil7PgF1 99% (99.2%)FN393967
LandsbergWiltshireLeWil34PbF67§ 99% (99.4%)FN393976
LandsbergWoburnLeWob8PgF1 97% (98.2%)FN393968
ColumbiaWoburnCoWob3PgF1 93% (94.7%)FN393971
ColumbiaWoburnCoWob10PgF1 99% (99.0%)FN393972
ColumbiaWoburnCoWob11PgF1 99% (99.4%)FN393966
ColumbiaWoburnCoWob29PgF1 99% (99.4%)FN393975
ColumbiaWiltshireCoWil1PgF1 99% (99.2%)FN393969
ColumbiaWiltshireCoWil7PgF1 99% (99.6%)FN393970

Sequences from these experiments (approximately 430–500 bp) were aligned with existing Polymyxa rDNA sequences and phylogenetic analyses were performed in mega4 (Fig. 4). With the exception of LeWil clone 34, which grouped with P. betae, all of the other Polymyxa sequences obtained from Arabidopsis root samples formed a clade with the P. graminis F1 (ribotype I) isolate (Y12824, 96% support from bootstrapping). There was strong bootstrap support (98%) separating the Col-0 Woburn clone 3 sequence from the other sequences in this clade. The sequence identity between P. graminis type I sequences and those of P. betae was around 80%.

Figure 4.

 Phylogenetic relationships between Polymyxa ITS rDNA sequences from Arabidopsis and other isolates. Sequences previously deposited in public databases are PgF1 (Polymyxa graminis ribotype I, accession no. Y12824), PgF51 (ribotype II, Y12826), PgI1-229 (ribotype III, Y12825), PgS6 (ribotype III, AJ311580), PgMeta (ribotype V, AJ010424), PbF67 (Polymyxa betae, Y12827), Pbr (Plasmodiophora brassicae, Y12831), Ligniera (Y12828), Ssn (Spongospora subterranea f. sp. nasturtii, AF310907) SssI (S. subterranea f. sp. subterranea type I, AY604171) and Sss II (S. subterranea f. sp. subterranea type II, AY604172). Other sequences were obtained in the current study. The neighbour-joining method (maximum composite likelihood distances) was used in mega4 (Tamura et al., 2007) with 10000 bootstrap replications.

The range of Polymyxa sequences obtained from the Arabidopsis roots was diverse, but not unexpected as previous work has demonstrated that plants can contain more than one ribotype of Polymyxa in their roots (Ward et al., 2005; Vaïanopoulos et al., 2007; Smith, 2008). The diversity seen could also be due to the heterogeneity between rDNA repeat units in the same Polymyxa spore or cell.

Collectively, our results indicate that Arabidopsis is susceptible to infection by Polymyxa spp. Polymyxa-like spore clusters were identified in root hairs of Arabidopsis Ler-0 plants and structures resembling young Polymyxa-like zoosporangia in the roots of Col-0 plants. The putative zoosporangium is not like that of any of the other plasmodiophorid genera. Although these structures were not observed in all plants, it is possible that they were present in parts of the root system other than those examined by microscopy. The spores, although similar in appearance to Plasmodiophora, were aggregated together in clusters, whereas Plasmodiophora spores do not form aggregates. Additionally, no galls were observed in the roots of these plants, as would occur in Plasmodiophora infections, and a Plasmodiophora-specific PCR assay showed that Plasmodiophora was not present in the Arabidopsis or soil samples.

Using Polymyxa-specific PCR assays, Polymyxa was detected in all four combinations of Arabidopsis ecotypes and soils, and this was confirmed by rDNA sequencing; sequences either had high nucleotide identity to the rDNA sequence from ribotype I P. graminis or to P. betae. None showed close identity to P. graminis type II despite this ribotype being present in both soils (Ward et al., 2005; Lyons et al., 2008). Although temperate ribotypes of P. graminis have been shown mainly to infect monocotyledonous plants, P. betae and tropical isolates of P. graminis have been shown to infect dicotyledonous plants (Barr, 1979; Ratna et al., 1991; Barr & Asher, 1992; Legrève et al., 2000).

The observation of spores in the root hairs of the Arabidopsis ecotype Ler-0 plants is interesting as Polymyxa spp. are not routinely reported infecting root hairs, although this has been observed infrequently (M. Smith & M.J. Adams, unpublished data). Because this is a new and distinctive host, it is not unreasonable to expect that that the localization of Polymyxa within the plant or aspects of its morphology might differ. This could result for example from spatial constraints within the cells. There is support for this from anatomical studies of P. graminis infection in sorghum and wheat (Littlefield et al., 1997).

Unfortunately, we cannot confirm absolutely that the structures observed in the roots of the Arabidopsis plants correspond to the Polymyxa detected using molecular methods. In hindsight, we should have selected infected root tissue before DNA extraction to provide additional support for this, but conclusive proof would require a technique such as laser capture microdissection (Day et al., 2005). These techniques are technically challenging and have rarely been successfully used in these types of study.

There are problems associated with the use of soil to infect the plants rather than resting spores or zoospores from previously characterized Polymyxa isolates. There is a possibility of detection of Polymyxa from soil adhering to the root, which could confuse the issue of whether detection in the plant has occurred. However, the roots were washed thoroughly before use and this was facilitated by growth in a mixture of soil and sand (1 : 2), rather than soil alone. Also, from our previous experience of this system, we feel that it is unlikely that loosely attached Polymyxa spores would be responsible for the detection. Infection using Polymyxa-infected material would also have been superior in that it would have allowed a demonstration of Koch's postulates. However, it is generally more difficult to infect plants using zoospores or resting spores, than using soil and we felt that, to establish the system, it would be better to bait plants with the mixture of ribotypes that are present in the soil, rather than test individually zoospores/resting spores from a wide range of different isolates, some of which may not be well adapted to the new host.

Further experiments would be required to optimize the system, to demonstrate completion of the life cycle and onward transmission to further plants, to establish the range of Polymyxa isolates capable of infecting Arabidopsis and to determine whether there were any links between the type of infection seen (location and developmental stage) and the Arabidopsis ecotype used.

This is the first report to demonstrate that infection of Arabidopsis by Polymyxa spp. is possible. Both P. graminis and P. betae sequences were found in infected Arabidopsis roots and extends the range of known hosts for both species. This important finding opens up the exciting possibility of using a model system for studying Polymyxa infections with a wide range of available tools, and that is much more amenable to study than using sugar beet or cereal hosts.


The authors would like to thank A. Cuzick for providing seed and A. Tymon and K. Kanyuka for assisting with soil sampling. M.J.S. was supported by a BBSRC PhD studentship; John Walsh is thanked for his supervision and encouragement. Rothamsted Research receives grant-aided support from the Biotechnology and Biological Sciences Research Council.