The arbuscular mycorrhizal (AM) fungi have been elevated to the phylum Glomeromycota based on a ribosomal gene phylogeny. In order to test this phylogeny, we amplified and sequenced small subunit ribosomal RNA (SSUrRNA), actin and elongation factor 1 (EF1)-alpha gene fragments from single spores of Acaulospora laevis, Glomus caledonium, Gigaspora margarita, and Scutellospora dipurpurescens. Sequence variation within and among spores of an isolate was low except for SSUrRNA in S. dipurpurescens, and the actin amino acid sequence was more conserved than that of EF1-alpha. The AM fungal sequences were more similar to one another than to any other fungal group. Joint phylogenetic analysis of the actin and EF1-alpha sequences suggested that the sister group to the AM fungi was a Zygomycete order, the Mortierellales.
The arbuscular mycorrhizal (AM) fungi are obligately biotrophic root endosymbionts, and are ubiquitous in terrestrial ecosystems. Their biological properties have been studied intensively, but their evolutionary and taxonomic affinities remained unclear until recently. This was due in part to their morphological homogeneity, which provided few characters for analysis, and the absence of any identifiable stage of sexual reproduction. This has led to several analyses and revisions of their status [1–3].
The development of molecular techniques has transformed the study of AM fungi. The first studies of small subunit ribosomal RNA (SSUrRNA) genes showed that the Glomales (sensu Morton and Benny ) were a distinct and diverged lineage within the fungi . Field studies have shown that many sequences from AM fungi can be identified that are not allied to any known laboratory isolate, and that their global diversity may be considerably higher that was first thought [5,6]. The presence of multiple genomes in AM fungal spores was first suggested by data from Sanders et al.  and Lloyd-MacGilp et al.  and further evidence was recently presented by Kuhn et al. . These studies have shown that the genetic structure of AM fungi is complex, and that we need to know more about their genetics . A recent paper elevated the AM fungi and their relatives to the status of a phylum, the Glomeromycota . Using extensive SSUrRNA analysis the authors demonstrated clearly that the Glomeromycota are a well supported monophyletic clade. This delimitation is based on only a single gene, and it has been demonstrated that each individual gene, however robust the analysis, may present only part of the story [10,11]. The logical step, then, is to test the SSUrRNA phylogeny by analysing other genes .
In this paper, we present an analysis of protein coding sequences from single-copy genes that have been amplified from single spores of AM fungi. Single-copy genes have the advantage that any sequence variation within a spore can be attributed unambiguously to variation among nuclei. The elongation factor 1 (EF1)-alpha (tef) gene is usually present in a single copy and encodes the translation elongation factor that controls the rate and fidelity of protein synthesis . The actin (act) gene encodes actin, a cytoskeletal filament and has been found to be single copy in the majority of fungi tested . These genes have been used to study evolutionary relationships in the eukaryotes , including groups of fungi and especially the Zygomycota [14,15]. They therefore represent ideal candidate genes for studying the relationships of the Glomeromycota to other fungi. In addition, using these genes to estimate the variation within and among spores will reveal additional information about genetic structure in these fungi.
2Materials and methods
2.1The cultures used and their origins
Isolates of four AM fungi, Acaulospora laevis Gerdemann and Trappe, Glomus caledonium (Nicol. and Gerd.) Trappe and Gerd, Gigaspora margarita Becker and Hall and Scutellospora dipurpurescens Morton and Koske were maintained on Trifolium repens plants in a glasshouse. Spores were obtained from pots by mixing a sample obtained using a 0.5-cm diameter corer with water, allowing large particles to settle, and sieving through a 32- or 60-μm mesh sieve. Spores were transferred into distilled water and left for 24–48 h at 4°C. Any discoloured, infected or otherwise abnormal-looking spores were discarded at this stage. Spores were pre-sterilised in two washes of 2% Chloramine T, and rinsed in two washes of distilled water. DNA was extracted from spores by crushing with a sterile pipette tip in 30 μl molecular biology grade water (Eppendorf AG). Where necessary, samples were diluted and reconcentrated using a 100-kDa microfilter (Microcon, Millipore, UK). The origins of the cultures used and the accession numbers of sequences obtained from them are given in Table 1.
Table 1. Origin of isolates used and sequences obtained
aIt was not possible to obtain multiple act sequences from the G. caledonium.1 spore. The act sequence accession here is a consensus of four identical clones from an extract of 10 spores. This sequence is identical to that of G. caledonium.3.
Origin and supplier
Elongation factor (tef)
Elphin, Scotland (Chris Walker)
Den of Fowlis, Scotland (John Dodd)
Unknown (Paola Bonfante)
Morgantown WV (Joe Morton, INVAM)
Partial SSUrRNA fragments (c. 550 bp, including the highly variable V4 region) were amplified using Pfu Hotstart DNA polymerase (Stratagene) with a universal eukaryotic primer NS31  and the primer AM1 that amplifies the traditional families of AM fungi, the Glomeraceae, Gigasporaceae and Acaulosporaceae . The reaction was performed in the presence of 0.2 mM dNTPs, 10 pmol of each primer and the manufacturer's reaction buffer. Polymerase chain reaction (PCR) was carried out for 30 cycles (10 cycles at 95°C for 1 min, 58°C for 1 min and 72°C for 2 min; 19 cycles at 95°C for 30 s, 58°C for 1 min and 72°C for 3 min; and 1 cycle at 95°C for 30 s, 58°C for 1 min and 72°C for 10 min) on a MJ Research PTC100. Bands of the correct length were directly sequenced using NS31 and AM1 as the sequencing primers according to the manufacturer's instructions using the BigDye terminator cycle sequencing kit with AmpliTaqFS DNA polymerase (ABI Perkin-Elmer).
Partial act gene fragments (c. 1000 bp) were amplified with a newly designed universal eukaryotic primer W2 5′GAC GAC ATG GAR AAR ATY TGG3′ corresponding to amino acid sequence DDMEKIW) and animal/fungal primer G2 5′AGC CGT GAT YTC CTT YTG CAT3′; corresponding to the reverse complement of amino acid sequence MQKEITA). Partial tef gene fragments (c. 480 bp) were amplified with universal eukaryotic primers 3F and 7R . The reactions were performed using Pfu Hotstart DNA polymerase (Stratagene) in the presence of 0.2 mM dNTPs, 50 pmol of each primer and the manufacturer's reaction buffer. For both fragments, PCR was carried out for 35 cycles (initial denaturation: 95°C for 2 min, 35 cycles at 95°C for 1 min, 48°C for 45 s and 72°C for 2 min) on a MJ Research PTC200 DNA Engine.
The resulting PCR products were often found to show multiple bands. Initially, it was possible only to predict a minimum length for the band, as the presence and size of introns was not known. In these cases, a sterile 200μl pipette tip was used to excise a small quantity of the selected bands from a gel stained with ethidium bromide and visualised under UV. The gel fragment was left in 40 μl of molecular biology grade water for 24 h at 4°C, to allow the DNA to diffuse. This solution was vortexed and used as template in a second round of PCR using the above conditions. This technique was also used where PCR products were found to be present in insufficient quantities for cloning, or where there were large numbers of mis-primed bands.
The resulting blunt-ended products were cloned into pCR-Script Amp SK(+) and transformed into Escherichia coli strain XL1-blue MRF′ (Stratagene). Putative positive transformants were selected and screened using standard T3/T7 amplification with a 48°C annealing temperature. PCR products were reamplified from selected clones using T3/T7, were cleaned using Qiaquick PCR purification spin columns (Qiagen) and sequenced using T3 and T7 as the sequencing primers according to the manufacturer's instructions using the BigDye terminator cycle sequencing kit with AmpliTaqFS DNA polymerase (ABI Perkin-Elmer). Ready-to-run sequencing reactions were analysed by Lark Technologies Ltd, Saffron Walden, UK.
To confirm that sequences amplified using degenerate primers are derived from the AM fungal spores, primer pairs specific to the act (GcActF: 5′GAT TTG GCT GGT CGT GAT TT 3′; GcActR: 5′ AGC GGT TTG CAT TTC TTG TT 3′; 160 bp) and tef (GcTefF: 5′TTG CTT TCG TCC CAA TAT CC 3′; GcTefR: 5′ GAC GGT ACC GAT ACC ACC AA 3′ 210 bp) sequences of G. caledonium were identified. PCR amplification was carried out using the same conditions as the SSU PCR above, but using a 50°C annealing temperature and a 45 s extension time. PCR fragments amplified from a G. caledonium spore template were directly sequenced using the PCR primers on a Beckman CEQ 8000 Genetic Analysis System. The resulting sequences were 100% identical to the cloned sequences for both genes.
Forward and reverse sequences were aligned using Lasergene SeqMan (DNAStar Inc.). ClustalX  was used for multiple alignment of SSUrRNA sequences and neighbor-joining phylogeny . Act and tef sequences were aligned by eye using BioEdit . Intron sequences were identified and removed from the sequences before translation and other analyses. An alignment of concatenated act (224 positions) and tef (148 positions) amino acid sequences was constructed using a consensus from each spore analysed (nine in total) and 22 reference sequences. Partition homogeneity testing was carried out using PAUP . A neighbor-joining tree was calculated using Kimura's correction for multiple substitution. Bayesian inference of protein trees was carried out using Markov chain Monte Carlo simulation implemented in MrBayes . The Dayhoff and Jones models of amino acid evolution were used. Trees were sampled every 100 generations out of 100 000. Likelihoods stabilised around tree 70 of 1000. The first 75 trees were discarded (burnin=75) and all trees generated thereafter were used to construct a half-compatible consensus tree.
Pairwise similarities of sequences were estimated using Blast2Sequences (http://www.ncbi.nlm.nih.gov/blast/bl2seq/bl2.html). Gap opening and extension penalties were reduced until the sequence pairs were aligned in a single length to allow comparison of more divergent regions. Only the exons of the act (672 bp) and tef (444 bp) genes were used in this analysis. The nucleotide sequence lengths of the SSUrRNA genes are given in Table 1.
3Results and discussion
3.1Identification of single spores using SSUrRNA sequences
In the absence of reliable morphological characters that can be analysed non-destructively, the ribosomal genotype of each spore was determined in order to identify it. Fig. 1 shows how each spore is resolved using representative sequences from the analysis of Schüßler et al. . The direct sequence of NS31-AM1 SSUrRNA fragments confirmed that each of the spores was correctly identified, and that there was no significant contamination. Ambiguities in the direct sequences of A. laevis and S. dipurpurescens spores corresponded to point mutations that have previously been identified in the clones from these isolates included in Fig. 1. G. caledonium and Gi. margarita sequences show greatest similarity to groups containing isolates of the same name. The direct sequences were readable in all cases showing that, within a spore, there were no variants of different length (resulting in sequences that are out of phase), and the lack of ambiguities suggests that, for these isolates at least, highly divergent SSUrRNA sequences are not present.
3.2Actin and EF1-alpha sequences
The act sequences derived from the four fungal isolates were most similar to one another, and matched zygomycete sequences most closely on BLAST searching. G. caledonium, Gi. margarita and one A. laevis spore each yielded multiple identical clones. Variation among sequences obtained from the other A. laevis spore and each S. dipurpurescens spore derived mainly from indels in the introns. The amino acid sequence was highly conserved among the four isolates. Pairwise comparison showed that the number of codons with a synonymous substitution between isolates ranged from 16% to 40% and was lowest between the two isolates from the same family, Gi. margarita and S. dipurpurescens (Table 2).
Table 2. Pairwise comparisons of synonymous (below diagonal) and non-synonymous (above diagonal) substitution frequencies per codon in (left) act sequences and (right) tef sequences from AM fungi
The tef gene sequences are shorter, but show much higher amino acid variation (16/146 codons) than the act genes (4/225). By contrast, the rate of synonymous substitution at the third position of codons is similar (Table 2). Tef clones from different spores of A. laevis were identical, as were those from S. dipurpurescens spores. Most tef sequences from the two G. caledonium spores were identical, though one spore produced one clone that showed a single synonymous substitution, and another that showed a point mutation causing an amino acid change. Gi. margarita also yielded mostly identical clones, but again one spore produced one clone with a synonymous substitution and another with a single non-synonymous change. The tef sequences were all free of introns.
For both act and tef, all the sequences were more similar to each other than to those reported for other fungi. This gives us confidence that the sequences derived from these spores are glomeralean in origin and not due to the presence of contaminants. The pairwise similarity of the coding sequences ranged from 0.85 to 0.94 (Table 3). This is slightly lower than the values for SSUrRNA sequences. The amino acid sequences for both genes were very highly conserved, and this was maintained for multiple spores within an isolate. This suggests that there is the potential for these genes to be robust markers for molecular identification, though more data need to be collected in order to identify the level of discrimination.
Table 3. Pairwise DNA sequence similarity of SSUrRNA, act and tef sequences excluding introns
A. laevis–G. caledonium
A. laevis–Gi. margarita
A. laevis–S. dipurpurescens
G. caledonium–Gi. margarita
G. caledonium–S. dipurpurescens
Gi. margarita–S. dipurpurescens
The Glomeromycota sequences were aligned with the reference sequences shown in Fig. 2. The additional fungal species were selected to reflect a broad range of fungal taxa from which sequences of both genes are available. Animals and mycetozoans were included to give a clear Opisthokonta (animal/fungal) clade. Plants were selected as outgroups. Partition homogeneity testing showed that random partitions of the concatenated peptide sequences produced trees of a similar length to that of the gene-partitioned tree (P=0.436 for 500 replicates). It was therefore legitimate to combine the act and tef data in the analysis. Neighbor-joining using concatenated sequences produced a tree in which the AM fungal taxa form a highly supported clade (Fig. 2a). The other fungal groups were also well supported but higher order support was low. Phylogenetic trees were generated from the same alignment using Bayesian inference, a maximum likelihood approach. In both the consensus trees, the Zygomycete group Mortierellales was the sister group to the Glomerales/Diversisporales clade. The Mortierellales do not cluster with the Mucorales, supporting the conclusion of Voigt and Wöstmeyer  that the Zygomycota are paraphyletic (Fig. 2b).
The Glomerales/Diversisporales are a well supported monophyletic group in all the analyses. S. dipurpurescens and Gi. margarita, the two Gigasporaceae isolates, group together. This supports the molecular analyses of the SSUrRNA gene  and the current family level taxonomy . Within the Glomales clade, the four isolates are less differentiated than they are in the ribosomal gene phylogeny due to the low level of amino acid variation.
In all the analyses, the sister group of the AM fungi is the Mortierellales, which contradicts the suggestion of Schüßler et al.  that the Ascomycete–Basidiomycete clade is the sister group to the Glomeromycota. This result is well supported by Bayesian inference, but weakly in the neighbor-joining analysis. It has been suggested that the posterior probabilities in Bayesian analysis are too liberal and that bootstrap values may be conservative . However, the trees in Fig. 2 were generated using partial gene fragments, and the topology an the support may change with additional informative sites. The act sequence is nearly complete, but the tef fragment is only 146 aa of 460, and this is also the more phylogenetically informative gene. No sequences have yet been obtained from the more divergent groups of the Glomeromycota, the Archaeosporales and Paraglomerales. These are needed to resolve the placement of these taxa relative to the Zygomycete sister group.
We have demonstrated that it is possible to clone two single-copy genes from a single spore of an AM fungus. This enables a rigorous approach to phylogenetic analysis as each sequence can be unambigously assigned to an SSUrRNA genotype. The variability of amino acid sequences both within and among spores of an isolate are lower than that reported for ribosomal genes. This suggests that using these highly conserved genes, it is possible that molecular identification of AM fungi may be significantly improved.
Thanks to Paola Bonfante, John Dodd, Joe Morton and Chris Walker for providing the original inoculum for the cultures used in this study. Allen Mould and Celina Whalley assisted with sequencing. This study was funded by the BBSRC grant 87/P13883.