A mutant inLycopersicon esculentumMill. with highly reduced VA mycorrhizal colonization: isolation and preliminary characterisation


  • S. J. Barker,

    1. Department of Plant Science and
    Search for more papers by this author
    • Present address: Faculty of Agriculture, The Unversity of Western Australia, Nedlands, Western Australia, Australia 6009.

  • B. Stummer,

    1. Department of Soil Science, The University of Adelaide, Waite Campus, Glen Osmond, South Australia, 5064, Australia
    Search for more papers by this author
    • Present address: Department of Crop Protection, The University of Adelaide, Waite Campus, Glen Osmond, South Australia, 5064, Australia.

  • L. Gao,

    1. Department of Plant Science and
    Search for more papers by this author
    • §

      Permanent address: Department of Agronomy, The Shanxi Agricultural University, Taigu, Shanxi 030801, The Peoples Republic of China.

  • I. Dispain,

    1. Department of Soil Science, The University of Adelaide, Waite Campus, Glen Osmond, South Australia, 5064, Australia
    Search for more papers by this author
  • P. J. O’Connor,

    1. Department of Soil Science, The University of Adelaide, Waite Campus, Glen Osmond, South Australia, 5064, Australia
    Search for more papers by this author
  • S. E. Smith

    1. Department of Soil Science, The University of Adelaide, Waite Campus, Glen Osmond, South Australia, 5064, Australia
    Search for more papers by this author

*For correspondence (fax +61 8830 36511; e-mail ssmith@waite.


This paper reports the successful isolation and preliminary characterisation of a mutant ofLycopersicon esculentumMill. with highly reduced vesicular-arbuscular (VA) mycorrhizal colonization. The mutation is recessive and has been designatedrmc. Colonization byG. mosseaeis characterised by poor development of external mycelium and a few abnormal appressoria. Vesicles were never formed by this fungus in association with the mutant.Gi. margaritaformed large amounts of external mycelium, complex branched structures and occasional auxiliary cells. Small amounts of internal colonization also occurred. Laser scanning confocal microscopy (LSCM) gave a clear picture of the differences in development ofG. intraradicesandGi. margaritain mutant and wild-type roots and confirmed that the fungus is restricted to the root surface of the mutants. The amenability of tomato for molecular genetic characterisation should enable us to map and clone the mutated gene, and thus identify one of the biochemical bases for inability to establish a normal mycorrhizal symbiosis. The mutant represents a key advance in molecular research on VA mycorrhizal symbiosis.


The vesicular-arbuscular (VA) mycorrhizal symbiosis is a mutualistic association between fungi of the family Glomales (Zygomycota) and the majority of species of land plant. The fungi have an indefinite compatible and non-specific interaction with the plant. Bidirectional transfer of nutrients (mineral nutrients from fungus to plant and phytosynthetically fixed carbon from plant to fungus) between the symbionts is the main basis for mutualism ( Smith & Read 1997).

The symbiosis is obligate for the fungal partner, but not for the plants. A number of plant taxa are characteristically non-mycorrhizal. The non-mycorrhizal character, which has probably evolved several times ( Trappe 1987), is expressed at the high taxonomic levels of family (e.g. Cruciferae, Chenopodiaceae, Proteaceae) or genus (e.g. Lupinus), with little natural variation within species (see Gianinazzi-Pearson 1984;Tester et al. 1987 ). Mechanisms of exclusion have not been precisely identified. Cultivar or genotype differences exist only in the extent to which the fungus colonises the root or in the responsiveness of the plant to colonization ( Peterson & Bradbury 1995;Smith et al. 1992 ). Studies of the plant–fungus interactions during the establishment of the symbiosis have shown that the expression of genes involved in defence responses is limited and transient ( Harrison 1997). The mechanisms that enable the establishment of the association are more difficult to study, and only recently have genes relevant to this aspect of symbiotic development begun to be characterised ( Tagu & Barker 1997).

Genetic characterisation of the symbiosis began with the identification of nodulation-defective mutants of Pisum sativum and Vicia faba that were also nonmycorrhizal ( Duc et al. 1989 ). Similar results have been reported for Medicago sativa ( Bradbury et al. 1991 ; 1993a; b), M. truncatula ( Sagan et al. 1995 ) and Phaseolus vulgaris (Shirtliffe & Vessey 1996). In the majority of identified mutations colonization is blocked at the epidermis, prior to any significant penetration of the roots. There is no clearcut relationship between nodulation and mycorrhizal phenotypes. Some nodulation mutants are myc+ and conversely myc mutants can be nod or nod+/fix. As nod+/fix+ mutants have not been screened systematically for myc phenotypes it is not known if that phenotype can exist. Unfortunately, with the exception of M. truncatula ( Barker et al. 1990 ), these legume species are poorly developed for molecular genetic research and there is no information on the genetic defect responsible for the failure of these symbioses to develop.

Here we report identification and preliminary characterisation of a tomato mutant that does not form normal mycorrhizal associations. We chose this species because use of a non-legume avoids the complications of dealing with a tripartite symbiosis (Rhizobium, VA mycorrhizal fungus and plant). Furthermore, tomato is easy to grow and responsive to well-characterised VA mycorrhizal colonization. It is diploid, self-fertile and has a relatively small genome, with excellent genetic and molecular resources ( Tanksley et al. 1992 ).


Characteristics of normal mycorrhizal colonization in tomato

Light microscopy following clearing and staining of the roots with trypan blue indicated that mycorrhizal colonization of wild-type L. esculentum cv 76R by Glomus mosseae ( Fig. 1a) and Gigaspora margarita (data not shown) was normal and of the Arum-type (see Smith & Smith 1996). All steps in the colonization process were observed, including appressoria on the root surface, hyphal coils in hypodermal passage cells, intercellular hyphae and intracellular arbuscules (see below). G. mosseae and Gi. margarita produced vesicles or auxiliary cells, respectively. The percentage of the root length colonized by the fungi in different experiments ranged from 42 to 93% for Glomus mosseae and 29–60% for Gigaspora margarita. This variation is normal, taking into account differences in the infectivity of different batches of fungal inoculum and variations in environmental conditions. When grown in the same experiment, colonization of the near-isogenic L. esculentum 76S, and reciprocal crosses between 76R and 76S, did not differ significantly from L. esculentum 76R either in percentage internal colonization or in the relative development of external hyphae or arbuscules ( Table 1).

Figure 1.

Longitudinal squashes of roots of Lycopersicon esculentum cleared and stained with trypan blue.

(a) 76R, showing normal external and internal colonization by Glomus mosseae. Arbuscules (ar), external hyphae (eh) and plant cell walls (cw) arrowed. Scale bar = 100 μm.

(b) rmc, showing development of external hyphae of Glomus mosseae and complex appressoria (arrowed), but lack of internal colonization. Scale bar = 100 μm.

(c) Complex appressorium formed by G. mosseae on rmc, with many apparent attempts at root penetration (arrowed). Scale bar = 25 μm.

(d) Development of Gi. margarita on the surface of roots of rmc and lack of internal colonization. Auxiliary cells (au) arrowed. Scale bar = 100 μm.

Table 1.  Characteristics of mycorrhizal colonization in Lycopersicon esculentum rmc, wild-type genotypes 76R (parent) and 76S, and reciprocal crosses by (a) Glomus mosseae and (b) Gigaspora margarita, sampled at 42 days. Means and standard errors of means. Numbers of replicates in each treatment indicated as footnotes
 % colonizationrelative development of structures
  1. 1Three replicates; 2eight replicates.

  2. n.s. = not significant at P < 0.05.

76R × 76S134.110.2143.914.1836.311.000.840.020.81
76S × 76R139.68.2156.915.7453.217.370.900.080.76
76S ×rmc260.51.8263.23.9056.94.180.900.020.98
rmc× 76S252.85.4551.76.5047.46.040.920.011.06
76S ×rmc228.92.8335.03.2932.83.280.930.010.82
rmc× 76S230.34.0034.84.0930.34.070.860.040.87
Plant genotype  <0.001 <0.001 <0.001 <0.001
Fungal species  <0.001 <0.001 n.s. <0.001
Interaction<0.001 n.s. (0.085) n.s. n.s. n.s.

Identification and preliminary characterisation of the reduced mycorrhizal phenotype

We screened a fast neutron mutagenised population of tomato ( Salmeron et al. 1994 ). A total of 215 M2 families were screened for variations from this typical pattern of colonization, using G. mosseae as the fungal symbiont. Several families had one or more members that displayed some difference in phenotype. One such M2 plant produced M3 progeny that were uniform in expressing reduced mycorrhizal colonisation, suggesting that it was homozygous for a mutation. To reflect the phenotype and preliminary genetic characterisation (see below) we named the mutant line rmc.

The colonization phenotype in M3 and M4 rmc plants was investigated using G. mosseae and Gi. margarita. G. mosseae developed sparse external hyphae ( Fig. 1b and Table 1). Occasionally, complex branching appressoria were formed ( Fig. 1c), but actual penetration of roots was not found. Gi. margarita produced extensive, but highly variable, mycelium on the surface of roots ( Table 1), which occasionally bore auxiliary cells ( Fig. 1d). Complex branching structures were also observed, especially on plants grown in nurse pots to produce a very high fungal inoculum potential (see below). Small amounts of internal colonization by this fungus were observed, but arbuscule development was very infrequent.

Inheritance of the mutation was investigated in F1 progeny of reciprocal crosses between rmc and 76S. Colonization by both G. mosseae and Gi. margarita in these plants was compared with that of the parents and 76R, and by G. mosseae alone for F1 progeny from reciprocal crosses between 76R and 76S ( Table 1). The development of total internal colonization and arbuscules was significantly different between rmc and the other plant genotypes. Interactions between plant genotype and fungal species were only significant for the development of external hyphae. Figure 2 shows the effect of plant genotype (rmc or 76R) on total internal and arbuscular colonization at 42 days, combining data for the two fungal species. The differences were highly significant (P < 0.001) and emphasise the abnormal VAM developmental pattern in rmc plants. These data are consistent with the mutation rmc being recessive.

Figure 2.

Internal colonization (▪) and development of arbuscules (□) on roots of Lycopersicon esculentum 76R and the mutant rmc.

Means and standard errors of means of six replicate plants. The main effect of plant genotype (blocking on fungal species) is significantly different (P < 0.001). Regression analysis by Genstat 5 Release 3.2 (Lawes Agricultural Trust).

In several of the screens infection of both wild-type 76R and rmc plants by non-mycorrhizal fungi (a binucleate Rhizoctonia and a Fusarium sp.) was observed. These observations indicate that the rmc mutation has not resulted in general exclusion of fungi.

Laser scanning confocal microscopy (LSCM) observation of fungal growth on rmc roots

We examined the resilience of the rmc phenotype using a very high inoculum potential of the fungi G. intraradices and Gi. margarita in nurse pots containing mycorrhizal leek plants ( Rosewarne et al. 1997 ). Tomato plants transplanted into nurse pots are rapidly and almost synchronously colonized, with maximum (plateau) values of percentage colonization reached within about 8 days. This contrasts with the much slower and progressive colonization, in which the stages overlap, which occurs in soil containing propagules of the fungi, as in our screening conditions. Plants were harvested repeatedly between 4 and 28 days after transplanting, so that material at all stages of colonization was available for wild-type plants, and that the mutant plants had a long exposure to inoculum and would therefore become colonized if they were susceptible. Total and arbuscular colonization in 76R at the timepoints corresponding to Fig. 3 were, respectively, 87% and 31% for G. intraradices (13 d, Fig. 3a) and 43% and 25% for Gi. margarita (11 d, Fig. 3b). The values for G. intraradices are very similar to those observed by Rosewarne et al. (1997) . There is no previous information on colonization by Gi. margarita in nurse pots. No internal colonization of roots of rmc grown in nurse pots occurred.

Figure 3.

Confocal microscopy of colonisation of Lycopersicon esculentum grown in nurse pots.

Extended focus images of (a) 76R colonised by Glomus intraradices after 13d, (b) Gigaspora margarita after 11d, and (c and d) rmc colonised by Glomus intraradices after 11 and 28 d. Surface colonisation of rmc by a highly branched appressorium of Gigaspora margarita 11d after transplantation to nurse pots; (e-j) montage of six optical sections 5 μm apart in the Z axis; (k) extended focus image and (h) rotated extended focus image of 10 optical sections. ap, appressorium; c, hyphal coil; a, arbuscule; v, vesicle. Arrows: (c) penetration point; (d) appressorium; (k) complex appressarium. Scale bars: a = 50 μm, b-l = 20 μm.

We used LCSM to characterise the spatial relationships between plant and fungus in roots of 76R and rmc grown in nurse pots. This technique sections material optically as the laser scans thick sections in x, y and z axes. Colonization of 76R was normal for both fungi used ( Fig. 3a,b), with characteristic appressoria, intercellular hyphae, arbuscules and (for G. intraradices) vesicles.

LSCM of fungal structures on the surface of roots of rmc confirmed the lack of penetration of the roots. The extended focus images of colonisation by G. intraradices ( Fig. 3c,d) show the sparse development of undifferentiated hyphae after 28 days in a nurse pot, with an unsuccessful attempt at entry ( Fig. 3c), and a single hypha and simple appressorium on the root surface, which failed to penetrate ( Fig. 3d). Figure 3(e–j) shows a series of six optical sections through a complex branching structure produced by Gi. margarita on the surface of rmc. The sections pass from the outside, through the fungus to the root surface ( Fig. 3i,j) and demonstrate the complete lack of penetration of epidermal cells. The extended focus and rotated images ( Fig. 3k,l) shows the location of this structure in a depression on the root surface.


We report the first identification of a mutation affecting VA mycorrhizal colonization in a non-legume plant. It is apparently recessive, stably inherited and markedly reduces mycorrhizal colonization by three species representing the two suborders of VA mycorrhizal fungi. The identification of mutations affecting mycorrhizal establishment indicates that some of the genes involved are neither genetically redundant nor essential for plant function, reflecting the fact that the symbiosis is not genetically obligate for the plants.

In the mutant rmc, colonization remained very low or absent, compared with wild-type plants, up to 8 weeks after inoculation and the phenotype was stable when plants were challenged with very high levels of inoculum in nurse pots up to 28 days. Thus, the mutation does not simply represent very slow colonization. Differences in development of external mycelium and appressoria by the different fungi on the roots of mutant plants (as shown here) have also been observed on the roots of ‘non-mycorrhizal’ genotypes of M. sativa, ( Bradbury et al. 1991 , 1993a).

The mutation rmc blocks colonization at the root surface, like the myc–1 mutants in legumes ( Duc et al. 1989 ;Shirtliffe & Vessey 1996;Sagan et al. 1995 ). The occasional penetration of the roots of rmc was not detected in M4 plants. This may represent phenotypic variation due to segregation of other mutations. With G. mosseae and Gi. margarita, appressoria were often abnormal and sometimes highly branched, as though repeated penetrations were being attempted. This phenotype is similar to the overproduced appressoria on myc genotypes of M. sativa ( Bradbury et al. 1991 ), and both responses indicate morphogenetic changes in fungal branching pattern, similar to those that occur when contact between the fungus and plant is prevented by a Millipore membrane ( Giovannetti et al. 1993, 1994 ; see Smith & Read 1997). In Pisum mutants the penetration step is apparently blocked by production of callose and phenolics in the cells below the appressoria ( Gollotte et al. 1993 ), but the mechanism in others, including rmc, has yet to be investigated.

In summary, colonization in rmc is normally blocked at the surface of the root so that it is apr+pen in the developmental framework proposed by Smith (1995). The mutation in tomato was identified without pre-screening for defective nodulation in legumes, which means that the rmc mutation is likely to be in a biochemical pathway common to all mycorrhizal plants.

Our objective is to identify the biochemical basis for the mutation by more detailed phenotypic characterisation of the mutant and by identifying and cloning the affected gene. This should be achievable using the high density molecular marker map ( Tanksley et al. 1992 ). Isolation of the gene will allow us to make progress on its function, and probe a range of hosts and non-hosts to determine its distribution and expression. It will also be important to expand the collection of mutations identified in order to build up a comprehensive picture of the molecular genetic basis for a successful mycorrhizal symbiosis.

Experimental procedures

Plant material

Screening for phenotypes defective in stages of mycorrhizal colonization was carried out using a population of tomatoes (Lycopersicon esculentum Mill. cv 76R; Peto Seed Company, CA, USA), mutagenised by fast neutron bombardment ( Salmeron et al. 1994 ). The near isogenic line L. esculentum Mill. cv 76S was used as the wild-type parent in genetic crosses because it differs from 76R at several DNA marker loci with respect to the presence (76R) or absence (76S) of the Pto locus and several closely linked DNA markers ( Carland & Staskawicz 1993).

Putative mutants were allowed to self-fertilise and up to 50 M3 progeny were re-screened for inheritance of the mutant phenotype. Any M2 plant that bred true for a mutant phenotype was considered to be homozygous for a mutation at a locus involved in mycorrhizal symbiosis. The plants were early generation material from a mutagenesis treatment and to reduce the potential phenotypic variability resulting from segregation of additional mutations, one rmc M3 plant was selected for single seed descent and all further analyses were of M4 progeny from that individual. Reciprocal crosses were confirmed using the Pto locus RFLP marker TG 475 ( Martin et al. 1993 ) (results not shown).

Potting mix and inoculum

Two potting mixes were used in the screening program and crossing experiments. In early screens, washed and sterilised sand (9 parts) was mixed with autoclaved low P soil from Mallala, South Australia (1 part), mixed with inoculum (see below) and brought to a 12% soil moisture. For secondary screens and genetic analysis, a defined medium, based on a University of California recipe (UC mix), was prepared as follows: 400 l of washed, coarse sand was sterilised at 100°C for 30 min in a sterilising mixer; 400 l of peat was added and mixed for 10 sec. After cooling to 70°C (about 10 min), nutrients were added and mixed for 20 sec. For complete UC mix, nutrients were: calcium hydroxide, 700 g; calcium carbonate, 480 g; calcium sulphate, 400 g; magnesium carbonate, 120 g; potassium sulphate, 60 g; potassium nitrate, 60 g; blood meal (approximately 16%N), 700 g; dicalcium phosphate, 560 g; final pH approximately 6.8. The low phosphate (low P) Waite-UC mix, used for the secondary genetic analysis was prepared in the same way, but superphosphate and blood meal were omitted.

Inocula of Glomus mosseae (Nicol. and Gerd.) Gerdemann and Trappe (obtained from Dr G.D. Bowen, CSIRO Division of Soils, Adelaide, Australia), G. intraradices Schenck and Smith (obtained from NPI, Utah, USA) and Gigaspora margarita Becker and Hall (obtained from Dr V. Gianinazzi-Pearson, INRA, Dijon, France) were prepared as pot cultures on Trifolium subterraneum L. cv Mt Barker. Soil, colonised roots and spores from the pot cultures were mixed through the potting mixes (see above).

Screening procedure

Twenty M2 seeds from each family and eight 76R control seeds were prepared by sterilisation in 2.7% sodium hypochlorite (30 min), rinsed in sterile reverse osmosis (RO) water and germinated on moist filter paper. Usually 14 M2 seedlings from each family and three or four 76R seedlings were screened, allowing for poor germination and survival, with 11 plants statistically likely to permit detection of mutations within a single M2 family ( Sedecole 1977). Eight to 12 families were scored in each round of screening.

Seedlings were planted individually in 4-inch pots containing Mallala soil/sand mix and inoculum of G. mosseae. Four additional seeds of 76R were grown without inoculum. This provided a check for infectivity of the inoculum, extent and characteristics of colonization in the parent plants and the presence of mycorrhizal fungi and pathogens in the sterilised mix. Plants were grown for 4–5 weeks in a glasshouse and watered three times per week with RO water. Plants were removed from the pots, approximately 1/3 root mass was taken for evaluation of colonization (see below) and the plants returned to the pots. After 2–3 further weeks, plants which were identified as having potentially abnormal colonization were re-sampled to check colonization and transferred to complete UC mix.

Nurse pots

Plants which were to be used for analysis of colonization using LSCM were grown individually in seedling trays containing a sterilized mix (autoclaved for 1 h at 121°C on 2 successive days, oven dried at 105°C for 3–4 d) of washed sand (9 parts) and low P soil (1 part) and transplanted, when 14 d old, to ‘nurse pots’ of mycorrhizal leek (Allium porrum L.) plants colonised either by G. intraradices or Gi. margarita ( Rosewarne et al. 1997 ). The same soils and mixes were used in seedling trays and nurse pots. For G. intraradices the soil came from Mallala, South Australia (pH 7.1) and for Gi. margarita from Kuitpo, South Australia (pH 5.0). Plants were grown as above. Long Ashton nutrient solution (– P) was applied weekly as described by Rosewarne et al. (1997) .

Evaluation of mycorrhizal colonization

Root samples were cleared in 10% KOH and stained with trypan blue using a modification of the method of Phillips & Hayman (1970), omitting phenol. Colonization in the inoculated control 76R plants was determined as percentage root length colonized, under a dissecting microscope using the grid intersect method ( Tennant 1975) and subjectively ranked on a scale of 0–3 for development of arbuscules and vesicles. Mutagenised plants were subjectively screened and if the colonization appeared abnormal, percentage colonization was determined as above. More detailed assessment of colonization was carried out by the method of McGonigle et al. (1990) . Briefly, cleared, stained whole root segments were mounted on slides. Intersects between the roots and an ocular cross hair were scored for the presence of different mycorrhizal structures at ×100 magnification. A disadvantage of using root squashes was that it was sometimes difficult to be certain that the fungus had actually penetrated the root surface, particularly in the mutant plants. Results are expressed as percentage intersects having external, internal and arbuscular colonization. The relative development of the different structures was calculated as the ratio of arbuscular or external colonization to total internal colonization. This allowed identification of abnormal patterns of colonisation, irrespective of overall percentage colonization. Results of the experiment to determine the genetic characterisation of the mutation were subjected to regression analysis using Genstat 5 Release 3.2 (1995, Lawes Agricultural Trust) to determine main effects of fungal species and plant genotype and interactions between them.

Laser scanning confocal microscopy (LSCM)

Roots were harvested from nurse pots, carefully washed and treated in one of the following ways: (a) segments of root approximately 1 cm long were embedded in cold 15% gelatin blocks containing 2% glycerol, frozen on a freezing stage (Zeiss) and sectioned (120 μm) in the longitudinal plane using a Leitz freezing microtome (based on Smith & Dickson 1991). Sections were stained with 1% acid fuchsin overnight and examined under a dissecting microscope. Those that showed mycorrhizal colonization were mounted on slides in lactoglycerol and the coverslips sealed with nail polish; (b) roots of rmc plants were stained with 1% acid fuchsin without sectioning (30 min) and mounted directly onto slides. Images were visualised using a BioRad MRC 1000 Laser Scanning Confocal Microscope system combined with a Nikon Diaphot 300 inverted microscope with fluorescence optics. Images were captured as computer files using 488/10 nm excitation and 522/32 emission wavelengths and ×40 water immersion lens NA 1.15 and analysed with Comos Image analysis software (Biorad) and Confocal Assistant Version 4.02 (Todd Clarke Brelje). The data were too complex for construction of 3D images and information is presented as montages of simple confocal pictures and extended focus and rotated images composed of a varying number of optical sections in the z axis.


We would like to express our thanks to many people for their help in this project: Peter Kolesik and Sandy Dickson for confocal microscopy and advice on staining; Michelle Lorrimer for statistical advice; Eileen Scott and Rina Sri Kasiamdari for identification of fungal parasites. All of the following helped with advice and/or technical help in the very time consuming stages of the screening and preliminary genetic characterisation: David Hein, Debbie Miller, Marg Pallotta, Jenny Ling, Heather Fraser, Garry Rosewarne and Andrew Barker. Funding from the Waite Research Committee and the Australian Research Council Small Grants Scheme is gratefully acknowledged. Lingling Gao is also grateful for support of a grant by the Provincial Government of Shanxi, PRC.