Intra and Inter-Spore Variability in Rhizophagus irregularis AOX Gene

Abstract

Arbuscular mycorrhizal fungi (AMF) are root-inhabiting fungi that form mutualistic symbioses with their host plants. AMF symbiosis improves nutrient uptake and buffers the plant against a diversity of stresses. Rhizophagus irregularis is one of the most widespread AMF species in the world, and its application in agricultural systems for yield improvement has increased over the last years. Still, from the inoculum production perspective, a lack of consistency of inoculum quality is referred to, which partially may be due to a high genetic variability of the fungus. The alternative oxidase (AOX) is an enzyme of the alternative respiratory chain already described in different taxa, including various fungi, which decreases the damage caused by oxidative stress. Nevertheless, virtually nothing is known on the involvement of AMF AOX on symbiosis establishment, as well on the existence of AOX variability that could affect AMF effectiveness and consequently plant performance. Here, we report the isolation and characterisation of the AOX gene of R. irregularis (RiAOX), and show that it is highly expressed during early phases of the symbiosis with plant roots. Phylogenetic analysis clustered RiAOX sequence with ancient fungi, and multiple sequence alignment revealed the lack of several regulatory motifs which are present in plant AOX. The analysis of RiAOX polymorphisms in single spores of three different isolates showed a reduced variability in one spore relatively to a group of spores. A high number of polymorphisms occurred in introns; nevertheless, some putative amino acid changes resulting from non-synonymous variants were found, offering a basis for selective pressure to occur within the populations. Given the AOX relatedness with stress responses, differences in gene variants amongst R. irregularis isolates are likely to be related with its origin and environmental constraints and might have a potential impact on inoculum production.

Introduction

The interaction of arbuscular endomycorrhizal fungi (AMF) (phylum Glomeromycota) and terrestrial plants is a highly ancient (~ 450 million years) and successful symbiosis maintained by the vast majority of plant species, including angiosperms, gymnosperms and some genus of bryophytes and pteridophytes [1,2]. AMF are obligate biotrophs and exist in the soil as extraradical mycelium, sporocarps and spores; following spore germination, the hyphal germ tube grows through the soil in search of a host root. Once contact between the symbionts is established, the fungus forms an appressorium on the root surface via which it enters the root forming highly branched structures (arbuscules) inside root cortical cells [2,3]. These arbuscules are connected to an extensive network of extra-radical mycelium providing multifunctional actions to the plant, from nutrient uptake to alleviation against a diversity of stresses including soil metal toxicity [4]. In return, the fungus acquires carbon from the plant, via the intraradical mycelium [2].

Due to the beneficial effects that AMF can have on plant performance, the interest on AMF application in agricultural systems has been increasing [5]. Nevertheless, it is not a widely accepted strategy and some mistrust is present within farmers as they often observe inconsistency of results in crop perform enhancement upon AMF application [6]. Also, companies dedicated to AMF inoculum production have been reporting a high heterogeneity of inocula quality from year to year. Whether several reasons can be behind this inconsistency (e.g. genetics, behavioural, ecologic), it is crucial to develop strategies that could tackle it. As a first step, at the level of inoculum production, tools that could allow maintaining and assessing the quality and uniformity of the inoculum would be of high interest. AMF present high genetic diversity which can contribute for unstable product quality of different AMF isolates [7]. AMF spores contain typically hundreds of nuclei [8] and can carry several genetic variants [9]. In this sense, the development of a marker, focused on a selected gene, directly linking a sequence motif (identified at genomic level) to phenotypic variation (inoculum quality) could represent a valid approach to be pursued.

The starting point of marker development through candidate gene approach is the knowledge of a gene with an assigned function. Following, it is required the existence of variability among allele sequences from which polymorphic, functional motifs affecting phenotype can be identified. Here, there are presented these two phases of functional marker development based on a hypothesis-driven approach previously presented by Vicente and co-workers [10] which focused on the alternative oxidase (AOX) as the target gene. According to that hypothesis, AOX could be a marker to characterise biodiversity/gene diversity and at the same time the functional integrity of AMF communities. AOX is a key enzyme in the alternative respiratory chain described in plants, fungi and also in some animals [11], that reportedly either decreases the damage caused by oxidative stress or minimises the production of reactive oxygen species (ROS) generation [12,13]. Research with plants confirmed that AOX fulfils physiological functions in growth, developmental events and in response to environmental stresses such as cold, drought, heat or nutrient deficiency (see review of Vanlerberghe (2013) [14]). For these reasons, AOX has been pointed out as a target gene for functional marker development related with plant plasticity [15,16]. Studies on fungi AOX are much more scarce; nevertheless, AOX was already isolated in several fungi species, and for instance has been related with defence against exogenous oxidative stress [17–19] or with the regulation of the biotrophic development and transition to necrotrophic stage in pathogenic fungi [20].

Tamasloukht et al. [21] showed that the respiratory activity of AMF is stimulated by root exudates, since an increase in oxygen consumption was noted after approximately 90 min of exposure. It was also observed that this stimulation of fungal respiration preceded an intense hyphal branching [21]. Therefore, these authors have speculated on the role of alternative respiration during the presymbiotic phase as enabling survival of AMF spores, being AOX the key enzyme in that pathway. More recently, Besserer and colleagues [22] further published important results showing the involvement of AOX activity in AMF as response upon the plant signalling exudate strigolactone.

In the present work, we report the isolation and characterisation of the AOX gene in the AMF Rhizophagus irregularis, the most AMF used as inoculant in agriculture, and we also show the involvement of RiAOX on symbiosis establishment by measuring the transcript accumulation over time at both host plant and AMF. The existence of RiAOX gene variability among different isolates of R. irregularis was further investigated by high throughput sequencing, considering the future development of functional markers to assess AMF inoculum quality.

Material and Methods

2.1. Isolation and characterisation of the RiAOX gene sequence

2.1.1. RiAOX complete gene isolation

R. irregularis single spores, collected from the in vivo inoculum INOQ69 provided by INOQ GmbH company (Solkau, Schnega, Germany) (http://inoq.de), were directly crashed in Whatman FTA paper (GE Healthcare, Little Chalfont, England) and immediately used for total DNA amplification using the REPLI-g Mini Kit (Qiagen, Hilden, Germany). A first approach to amplify the gene was based on the design of primers located at the 5’ and 3’ -UTRs in order to isolate the full-length sequence (based on the sequence available at NCBI, acc. no. GW102706). However, due to the inefficiency of that strategy, a new set of primers located within the ORF was designed for amplification of a fragment with a size of ca. 700 bp (RiFw_80: 5’-CATTTACTATCCACGCATCACGA-3’ and RiRev_651: 5’-TCCAACCACACGATGAGC-3’). PCR was conducted with Ready-To-Go PCR Beads (GE Healthcare, Little Chalfont, England) using 3 μl of the amplified DNA (previously diluted 1:20) as template and 0.2 μM of each forward and reverse specific primers. PCR was carried out for 35 cycles in a 2720 Thermocycler (Applied Biosystems, Foster City, CA, USA), consisting each cycle in 94°C for 30 s, 50°C for 30 s and 72°C for 1 min. An initial step of 94°C for 5 min and a final step of 72°C for 10 min were performed.

In order to isolate the 5’ and 3’ -UTRs, total RNA was extracted from 1 cm³ of an in vitro culture of R. irregularis strain INOQ69, which included spores and mycelium. Total RNA was extracted using the RNeasy ® Micro Kit (Qiagen, Hilden, Germany) following manufacturer’s protocol. RNA purity and concentration were assessed in a NanoDrop-2000C spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Ten ng of total RNA were used as template for cDNA synthesis using the SMARTer^TM RACE cDNA Amplification Kit (Clontech, Mountain View, USA). For 5’and 3’ -UTRs isolation, a reverse and a forward specific primers were designed (GSP2: 5’- GAAACTGTTGCCGCAGTTCCTGGTATG-3’ and GSP1a: 5’- GCGTTCCCACCAGGTAGGTTGTGAAAT-3’, respectively) to be used in combination with a universal primer provided by the manufacturer. PCR was performed according to the protocol of the RACE cDNA Amplification Kit using the Advantage^® 2 PCR Kit (Clontech, Mountain View, USA).

For complete gene isolation, a forward and a reverse primer set were designed in the previously obtained 5’ and 3’ -UTRs regions (RiORF105Fw: 5’-AATGATTCGCTCGGTTTTTC-3’ and RiORF1137Rev: 5’-TTATTTACTCCATTTCCATTTTTC-3’). Total DNA of a single in vitro spore of R. irregularis strain INOQ69 was amplified as described above and 1 μl was directly used as template for PCR using 0.2 μM of each specific primer. The enzyme Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific, Waltham, MA, USA) was used according to the manufacturer’s protocol. The amplification consisted in 35 cycles of 98°C for 10 s, 50°C for 20 s and 72°C for 1 min. An initial step of 98°C for 2 min and a final step at 72°C for 10 min were performed. The obtained amplicons were cloned onto a pGEM^® –T Easy vector (Promega, Madison, Wisconsin, USA) and used for transformation of E. coli JM109 (Promega, Madison, WI, USA) competent cells. Selected recombinant clones were send for commercial sequencing (Macrogen, Seoul, Korea: www.macrogen.com) using T7 and SP6 primers (Promega, Madison, WI, USA).

2.1.2. RiAOX sequence analysis

CLC Genomics workbench 7.5.1 platform (CLCbio, Aarhus N, Denmark) was used to edit the isolated AOX sequences. Homology of obtained sequences was confirmed in NCBI GenBank database (National Center for Biotechnology Information, Bethesda, MD, http://www.ncbi.nlm.nih.gov/) using BLASTn and BLASTp algorithm [23].

The exon-intron structure and the location of the AOX (Ferritin-like super family) conserved domain was compared between R. irregularis and other fungal species using the Fancy Gene v1.4 software (freely available at http://bio.ieo.eu/fancygene/). The genomic AOX sequences of Rozella allomycis, Agaricus bisporus, Laccaria bicolor, Macrophomina phaseolina, Cryptococcus neoformans, Coccidioides posadasii, Neurospora crassa, Lichtheimia corymbifera, Candida albicans, Talaromyces stipitatus and Spraguea lophii were retrieved from EMBL-EBI (http://www.ebi.ac.uk/) or NCBI (http://www.ncbi.nlm.nih.gov/).

The prediction of the fungal AOX N-termini that can support mitochondrial targeting sequence was performed on the translated peptide sequence of R. irregularis AOX and on other fungal AOX sequences using MITOPROT software (http://ihg.gsf.de/ihg/mitoprot.html).

A phylogenetic analysis of AOX proteins from R. irregularis and from other fungi and plant species was performed. AOX protein sequences from plant Monocots, Eudicots and Gymnosperm species were retrieved from PLAZA (http://bioinformatics.psb.ugent.be/plaza/), UniProt (http://www.uniprot.org/), Ensembl Plants (http://plants.ensembl.org/index.html) and/or NCBI (see accessions on S1 List). Fungi AOX sequences were retrieved from EMBL-EBI, UniProt and/or NCBI (accessions on S1 List). Sequence alignments were performed on MAFFT (http://mafft.cbrc.jp/alignment/server/) under the default parameters and a selection of best-fit models of amino acid (aa) replacement was performed using ProtTest 2.4 (http://darwin.uvigo.es/software/prottest2_server.html). Using the selected best-fit model (LG+I+G) a Maximum Likelihood phylogenetic analysis was performed on MEGA V6.0 [24], with 150 bootstraps. Two bacterial AOX sequences were used as outgroup (Rhodanobacter sp. and Afipia felis).

To identify distinctive features of fungi and plant AOX proteins, multiple alignments were performed using the CLC software on the helices α2, α3, α5 and α6, which form the four-helix bundle accommodating the diiron center, on the helices α1 and α4, which form the hydrophobic region on the AOX molecular surface, and on region 3, previously defined as possibly important for plant AOX activity [25]. The regions involving the two conserved cysteine residues (CysI and CysII) which in plants are responsible for the α-keto acid regulation were also compared between plant and fungal sequences.

2.2. Study of AOX transcript variation during symbiosis establishment on AMF and host plant

2.2.1. Experimental design and sample collection

To study the involvement of AOX at transcriptional level on both AMF and host plant during mycorrhizal colonisation, an experimental trial was set up using tomato (Solanum lycopersicum L. cv. MoneyMaker), which is often selected as model plant for experiments with AMF, and R. irregularis (BEG 121 isolate) (former Glomus intraradices).

Tomato seeds were surfaced-sterilised in 4% sodium hypochlorite containing 0.02% (v/v) Tween 20, rinsed thoroughly with sterile water and germinated for 2 days in a plate on moistened filter paper at 25°C in darkness. Subsequently, tomato seedlings were grown hydroponically in 3 L plastic containers with Long Ashton nutrient solution and at constant aeration. The nutrient solution was replaced once a week. After 2 weeks, plantlets were individually transferred to 0.45 L pots with sterile sand:soil:vermiculite (3:1:1) mixture (soil had a pH of 7.2, 1.6% organic matter, 2.1 μg Kg^-1 of N, 1.7 μg Kg^-1 of P and 0.8 μg Kg^-1 of K as nutrient concentration).

Half of the plants (24 plants) were inoculated by adding 10% (v:v) of R. irregularis inoculum, which consisted in R. irregularis kept in a soil-sand mixture containing extradical mycelium and spores, and mycorrhizal root fragments of Trifolium repens L. and Sorghum vulgare. This AMF had been isolated from a Mediterranean agricultural soil. The same amount of soil:sand mix, but free of AMF, was added to control non- mycorrhizal plants. Non-mycorrhizal plants received an aliquot of AM inoculum filtrated (< 20 μm) (meaning without propagules) to homogenize the microbial populations. Plants were randomly distributed in a greenhouse and grown at 24/16°C with 16 h photoperiod and 70% humidity, watered three times a week with Long Ashton nutrient solution [26] containing 25% of the standard P concentration.

Plants were harvested at different time points: 1, 2, 4 and 6 weeks post inoculation. Three plants (biological replicates) were collected per time point for both inoculated and non-inoculated plants (control). Samples taken from the root system were immediately frozen in liquid nitrogen and stored at -80°C for further analysis. A sample of each individual root system was also taken for mycorrhizal quantification through microscopy observation.

2.2.2. Determination of mycorrhizal colonisation

Root samples of R. irregularis inoculated tomato plants were stained with trypan blue [27] and examined using a Nikon Eclipse 50i microscope and bright-field conditions. Root colonisation by the AMF was determined as previously described [28], using the MYCOCALC software (http://www.dijon.inra.fr/mychintec/Mycocal-prg/download.html). The parameters measured were the mycorrhizal colonisation frequency (F %) and intensity (M %), arbuscule (A %) and vesicle (V %) abundance, and the relative colonisation intensity (m %). At least five slides, containing 30 root pieces of 1 cm length, were analysed per biological replicate. Five replicates were analysed at each time point.

2.2.3. RNA isolation and gene expression analysis by quantitative real time RT-PCR (qPCR)

Total RNA was extracted using Tri-Reagent (Sigma-Aldrich, MO, USA) according to manufacturer’s instructions. RNA was treated with RQ1 DNase (Promega, Wisconsin, USA), purified through a silica column using the NucleoSpin RNA Clean-up kit (Macherey-Nagel, PA, USA). Before storage at -80°C, RNA was quantified using a NanoDrop-2000C spectrophotometer (Thermo Scientific, Wilmington, DE, USA) and its integrity was checked by gel electrophoresis. First strand cDNA was synthesised with 1 μg of purified total RNA using the iScript cDNA Synthesis kit (Bio-Rad, CA, USA) according to the manufacturer’s instructions. Real time qPCR was performed using 1 μl of single stranded cDNA (diluted 1:10) and specific primers for the four tomato AOX genes (SlAOX1a, SlAOX1b, SlAOX1c and SlAOX2), and the AOX gene of R. irregularis. The tomato phosphate transporter SlPT4, usually used as molecular marker for arbuscular functionality with direct link with mycorrhiza development in the roots [29] was also evaluated (see primer details on S1 Table). Reactions were carried out in a iCycler iQ5 system (Bio-Rad) for 40 cycles, consisting each cycle in 94°C for 15 s, 55°C for 20 s and 72°C for 20 s. An initial step at 94°C for 2 min was considered. Three independent biological replicates were analysed per treatment. Three independent biological replicates were analysed per treatment and each sample was analysed in duplicate. The specificity of the different amplicons was checked by a melting curve analysis at the end of the amplification protocol.

Two genes for tomato (SlEF-1 and SlActin2) and two genes for R. irregularis (RiTEF1a and RiBTub1) were used as reference genes for further normalisation of the transcript data achieved on the target genes. Evaluation of expression stability for the reference genes was done using the statistical application geNorm [30]. Expression of target genes was evaluated by relative quantification using the geometric normalisation factors obtained from geNorm. Standard curves of a 4-fold dilution series from pooled cDNAs were used for PCR efficiency calculations [31].

Differences in the expression level of target genes between non-mycorrhizal and mycorrhizal roots were examined by a t-test, or when the data did not meet the normality and/or equal variance requirements with a Mann-Whitney Rank Sum test. Transcript levels between weeks post-inoculation were examined by a one-way analysis of variance (ANOVA) with Holm–Sidak post-hoc tests. When the data did not meet the normality and/or equal variance requirements, a Kruskal–Wallis one-way ANOVA on ranks with a Dunn’s test for post-hoc comparisons was performed instead. All expression data were analysed with the SigmaStat statistical package (Systat software, London, UK). Significance levels were set at P < 0.05.

2.3. Analysis of RiAOX sequence diversity at single spore level by high-throughput sequencing

2.3.1. Isolate selection and DNA extraction at single spore level

Three single spores from three different R. irregularis isolates were used to investigate the RiAOX genomic sequence variability. The isolates used were BEG72 (provided by the Institute of Agrifood Research and Technology, IRTA, Spain), BEG144 (provided by Prof. Daniel Wipf, Agroécologie INRA-Université de Bourgogne, Dijon, France), and one isolate provided by INOQ GmbH, hereafter designated by INOQ isolate. Those AMF were, respectively, originally isolated in 1996 in Northeastern Spain [32], in 1979 in St. Petersburg, Russia (according to the International Bank for the Glomeromycota (http://www.i-beg.eu/)) and in 1996 in Breiningerberg/Aachen, Germany. All three inocula have been propagated on in vivo cultures ever since.

In vivo inoculum of BEG72 used in the experiment consisted in a mixture of mycorrhizal root fragments and spores on Terragreen® (Oil-Dri, UK) carrier substrate. The BEG144 and INOQ isolates used in this experiment consisted of in vitro AMF cultures (in carrot root organ cultures, ROC) established in 2013 from the original in vivo inoculum. Inoculum samples were observed under a binocular lens and spores were individually picked (three spores per isolate) with a pipette tip and placed in 0.2 ml tube containing 1 μl of sterile tap water. Spores were then crushed using a sterile pipette tip and the entire DNA content was amplified using the GenomiPhi V2 Whole-Genome Amplification Kit (GE Healthcare, Canada) according to manufacturers’ instructions. Synthesised DNA was stored at – 20°C until further use.

2.3.2. Molecular identification of R. irregularis

To confirm the identity of all 9 individual spores as belonging to R. irregularis, the ribosomal rRNA sequences including the ITS-LSU-SSU regions [33], were amplified by PCR using the enzyme Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific), cloned onto a pGEM^® –T Easy vector and send for commercially Sanger sequencing. Sequences from the rRNA gene of several R. irregularis and R. intraradices isolates together with rRNA sequences of Glomus clarum described in Stockinger et al. [34] and publically available in the NCBI database, were used to perform the clustering analysis with the sequences from the INOQ, BEG72 and BEG144 isolates. The region used for the analysis comprised ~1400 bp of the small subunit, ITS, and large subunit of the rRNA which is suitable to differentiate very closely related species such as R. irregularis and R. intraradices [35].

The multiple sequence alignment was performed using the CLUSTALW program in the software MEGA V6.0 [24]. Genetic distances were inferred by the Neighbour-Joining method [36] and the bootstrap consensus tree was inferred from 100 replicates.

2.3.3. High throughput sequencing of RiAOX in single spores

In order to obtain enough amount of RiAOX amplicon in the 9 single spores regarding the high throughput sequencing using the Ion Torrent™ platform, it was necessary to follow a nested PCR approach.

Two sets of specific primers were used: Ri_F1: ATGATTCGCTCGGTTTTTC and Ri_R1: TTATTTACTCCATTTCCATTTTTC for first PCR, followed by a second PCR using the Ri_F2: CATAACACATTTCTTCGTAACACAAC and Ri_R2: CAGGAGCCTCATTAGTCAAACCG primers. The obtained amplicon had a size of 1209 bp. In both cases the reactions were conducted with the enzyme Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific) for 35 cycles, consisting each cycle in 98°C for 10 s, 59°C/63°C (first/second PCR, respectively) for 20 s and 72°C for 1 min. An initial denaturation step at 98°C for 5 min and a final extension of 10 min at 72°C were also performed.

A minimum of 500 ng of PCR products were directly sent for sequencing using the Ion Torrent™ platform at STAB VIDA (http://www.stabvida.com/). The obtained FASTQ reads have been deposited in the Sequence Read Archive (SRA) from NCBI database.

Obtained reads were uploaded into the CLC Genomics workbench. The quality of sequencing reads was enhanced by trimming and discarding low quality sequences and only high quality FASTQ sequences were used for further analysis. Reads were mapped to a consensus reference sequence previously obtained from cloned Sanger sequences of RiAOX from the 3 isolates, using the following parameters: masking mode = no masking, mismatch cost = 2, insertion cost = 3, deletion cost = 3, length fraction = 0.5 and similarity fraction = 0.8, global alignment = no, non-specific match handling = map randomly, output mode = create read tracks, create report = yes, collect unmapped reads = no.

Once the reads were mapped to the reference, the mapped reads files were used as input to discover low frequency variants using quality score with the following parameters: ignore broken pairs = yes, minimum coverage = 10, minimum variant frequency (%) = 1, neighborhood radius = 5, minimum neighborhood quality = 15, minimum central quality = 20, remove pyro-error variants = yes, in homopolymer regions with minimum length = 3, with frequency below = 0.8, create track = yes, create annotated table = yes, create report = yes.

A comparison of all variants was then performed, being classified as Single Nucleotide Variation (SNV), Multiple Nucleotide Variation (MNV), Deletion (Del), Insertion (Ins) and Replacement (Repl). Comparison of all possible variants from the 9 spores was performed setting a frequency threshold (%) = 0. With this threshold it was possible to identify variants that are present in only one spore. On the contrary, to identify variants that are common to all 9 spores, a frequency threshold of 100% was set.

The frequency (defined as read count divided by read coverage) distribution of all variants, in relation to the total number of variants found for each spore was performed. The SNV frequency (read count divided by read coverage) along the gene sequence was performed for the three isolates.

The prediction of variants (SNVs and MNVs) that could putatively induce aa substitutions in the protein sequence was performed for each spore.

Results

3.1. RiAOX sequence and structure analysis

The full-length gene sequence of R. irregularis AOX has been deposited in GenBank (accession numbers KT423114 and KT423115). At the genomic level, RiAOX is composed by 4 exons interrupted by 3 introns, with a total size of 1250 bp (Fig 1). Variability among the exons and the introns within the RiAOX sequence is observed. The gene coding sequence, with 1032 bp and coding for a putative peptide of 343 aa, is composed by exon 1 (444 bp), exon 2 (116 bp), exon 3 (188 bp) and exon 4 (281 bp). The introns are considerably shorter, having 69, 81 and 68 bp for introns 1, 2 and 3, respectively.

Fig 1. Comparison of exon-intron structure of AOX gene in R. irregularis and fungi from other phyla.

Introns are represented by dashed lines and exons by blue rectangles. The green rectangles represent the location of the AOX encoded conserved domain, within the exons.

The AOX genomic structure of R. irregularis was compared with those of other fungi (Fig 1). The structure of AOX gene varies considerably amongst fungal species and phyla: for instance, T. stipitatus (Ascomycota), C. albicans (Ascomycota) and S. lophii (Microsporidia) have no introns, whereas R. allomycis (Cryptomycota) has 7 introns A. bisporus (Basidiomycota) and L. bicolor (Basidiomycota) have 10 introns (Fig 1). Also, the location of the sequence encoding the conserved domain for AOX (Ferritin-like super family) differs between fungi: while many have it starting on exon 1, like R. irregularis, L. corymbifera (Zygomycota), M. phaseolina (Ascomycota), N. crassa (Ascomycota) and C. posadasii (Ascomycota), others such as C. neoformans (Basidiomycota) and A. bisporus have it starting on exon 2. R.allomycis differs from all the others since the AOX domain only starts on exon 5 (Fig 1).

The predicted N-terminal region of RiAOX protein that can support mitochondrial targeting sequence was determined as 52 bp length. Nevertheless, a great variation on the predicted mitochondrial targeting peptide was observed amongst fungi (Fig 2). The largest N-terminal was found in Ascomycota fungi (67 bp) and the smallest ones in R. allomycis, B. dendrobatidis and some Zygomycota species (12 bp). In some Zygomycota (R. delemar and L. corymbifera), Microsporidia (S. lophii and N. parisii) and Basidiomycota (C. subvermispora) the N-terminal was not predictable.

Fig 2. Predicted size of the N-terminal sequence that can support mitochondrial targeting and the cleavage site, in AOX proteins of different fungi.

The N-terminal is indicated by blue rectangles.

3.2. Phylogenetic analysis of plant and fungal AOX

The phylogenetic analysis clearly differentiated between fungal and plant AOX protein sequences, which was supported by a high ML bootstrap value (0.95) (Fig 3). On the plant side, AOX sequences clustered into AOX2 and AOX1, and within the AOX1-subfamily the division in three main groups (AOX1a-c/1e, AOX1d from monocots, and AOX1d from eudicots) is evident, agreeing with the recent classification scheme proposed by Costa and co-workers [37]. However, while the monocot and eudicot AOX1d clusters were well supported by high bootstrap values, their positions relatively to the AOX1a-c/1e clade, as well as several positions inside this group were unresolved. Within the AOX2-subfamily, the AOX2d clade is also possible to identify, albeit with a lower branch support.

Fig 3. Maximum likelihood phylogenetic analysis of AOX protein sequences in fungi and plants.

Two bacterial AOX sequences were placed as outgroup (Rhodanobacter sp. and Afipia felis). Location of R. irregularis is shaded. Fungal and plant AOX protein sequences were clearly separated by a ML bootstrap value of 0.95. The fungal sequences were grouped on two major clades, supported by a ML bootstrap value of 0.61: the Basidiomycota/Ascomycota clade, and a clade containing the Chytridiomycota, Cryptomycota, Glomeromycota, Microsporidia and Zygomycota phyla. Basidiomycota were separated from Ascomycota by a ML bootstrap of 1.

Fungal AOX sequences were grouped in two major clades: the Basidiomycota/Ascomycota clade, and another containing the Chytridiomycota, Cryptomycota, Glomeromycota, Microsporidia and Zygomycota phyla (Fig 3). Moreover, the Basidiomycota AOX sequences were completely separated from the Ascomycota.

R. irregularis AOX sequence clustered with the clade including B. dendrobatidis (Chytridiomycota) and with R. allomycis (Cryptomycota) (Fig 3). The sister group of R. irregularis clade comprised members of the Zygomycota (order Mucorales) and included M. circinelloides, R. delemar, L. corymbifera and A. idahoensis. The second closest clade to R. irregularis was composed by the Microsporidia members Nematocida parisii, Trachipleistophora hominis and S. lophii.

Multiple alignments of the helices α1 and α4 forming the hydrophobic region on the AOX molecular surface revealed that fungal and plant sequences are quite divergent N-terminally, especially regarding helix α1 and the region containing the plant conserved regulatory cysteine residue I (CysI) (S1 Fig). CysI is located N-terminal to α1. The helix α4 is fairly more conserved than α1. Some plant AOX lack the conserved CysI, having a serine residue instead (S1 Fig). The conserved CysI was not found in any of the fungal sequences.

The analysis of the region 3 [25] showed that whereas plant generally present the glutamic acid/aspartic acid, asparagine and valine (E/DNV) motif (the most evident exceptions go to AOX1d from monocots and some AOX1d from eudicots), the fungal sequences are much more divergent in this regard. The most similar to plant sequences were R. allomycis and R. irregularis. RiAOX showed the ENS motif, identical to that of some AOX1d from eudicots. R. allomycis presented the ENV motif, identical to most of the analysed plant species (S2 Fig).

The multiple alignments of the helices α2, α3, α5 and α6, which form the four-helix bundle accommodating the diiron center, confirmed the four universally conserved glutamate residues (E) and the two universally conserved histidine (H) residues across plant and fungal sequences (S3 Fig). Amongst Angiosperms, only the AOX1d sequences from Monocots lacked the conserved CysII, having a serine residue instead. The conserved plant CysII was not found in any of the fungal sequences (S3 Fig).

3.3. AOX expression analysis in tomato root plants and R. irregularis

The gene expression analysis revealed that tomato AOX genes were stable during the first two weeks of the experiment, although mycorrhizal colonisation was already visible one week post-inoculation (S2 Table). At four weeks post-inoculation, two genes were up-regulated by mycorrhization with R. irregularis (Fig 4): SlAOX1a and SlAOX1c gene transcript levels increased (Fig 4A and 4C), coinciding with the appearing of the first arbuscules and vesicules (S2 Table). By 6 weeks post-inoculation, when more than a half of the roots were colonized by the AMF (S2 Table), SlAOX1a and SlAOX1c transcript accumulation was significantly higher in mycorrhizal plant roots (Fig 4A and 4C) (P < 0.05), and a similar trend was found for SlAOX1b despite the non-significant results (Fig 4B). Non-significant differences were observed for SlAOX2 along the entire experiment (Fig 4C). The expression of SlAOX genes in non-mycorrhizal plant roots remained stable during the 6 weeks of the trial. The tomato phosphate transporter SlPT4 was 14.5 fold up-regulated between 1 and 6 weeks post-inoculation with R. irregularis (P < 0.001) (Fig 4E).

Fig 4. Expression levels of AOX gene family in mycorrhizal and non-mycorrhizal tomato roots.

Expression levels were analysed at 1, 2, 4 and 6 weeks post-inoculation with R. irregularis. The expression levels of SlAOX1a (A), SlAOX1b (B), SlAOX1c (C), SlAOX2 (D) and SlPT4 (E) were analysed in roots of mycorrhizal and non-mycorrhizal tomato plants. The transcript levels of RiAOX were analysed on mycorrhizal roots of tomato (F). Asterisks (*) indicate significant differences (P < 0.05) in expression between mycorrhizal and non-mycorrhizal roots at one time point. Different superscript letters ^{(a, b)} indicate significant differences (P < 0.05) between weeks on mycorrhizal roots. No expression of SlPT4 and RiAOX was found in non-mycorrhizal roots.

The AOX transcript levels of R. irregularis were relatively stable during the first two weeks post-inoculation, but then decreased by 2.8 fold until 6 weeks (P < 0.05) (Fig 4F). No expression of SlPT4 and RiAOX was found in non-mycorrhizal plant roots.

3.4. High-throughput sequencing of RiAOX

3.4.1. Molecular identification of R. irregularis spores

The phylogenetic analysis of the cloned sequences placed all spores in the R. irregularis clade, together with sequences retrieved from databases and already recognised as belonging to R. irregularis (S4 Fig). The R. intraradices group was completely separated from R. irregularis.

3.4.2. High throughput sequencing of AOX in R. irregularis single spores

A total of 385 379 high quality reads were obtained from the 9 individual spores, with a mean percentage of mapped reads to the reference sequence of 97%. The mean length of mapped read in the 9 spores was of 143 bp (see details on Table 1). The mean variant coverage was of 3 856 and the mean SNV coverage was of 3 482 (Table 1).

Table 1. Reads and variants statistics of the 9 R. irregularis single spores.

Sample	# of reads	% of mapped reads	Average mapped read length	Average variant coverage	Average SNV coverage
INOQ 2	71 940	98	138	5 598	4 714
INOQ 3	4 456	98	105	276	226
INOQ 5	112 690	98	142	9 308	8 707
BEG144_2	5 521	95	136	436	381
BEG144_3	49 936	97	156	4 847	4 297
BEG144_8	36 707	96	144	3 282	3 028
BEG72_1	41 944	98	155	4 509	3 810
BEG72_2	28 258	98	156	2 861	2 689
BEG72_4	33 927	98	152	3 585	3 485
TOTAL	385 379
Mean	42 820	97	143	3 856	3 482

When considering all 9 spores, a total of 288 nucleotide variants (including SNVs, MNVs, insertions, deletions and replacements) were found within the 1209 bp sequence length (Fig 5A). Of these 288 variants, 93 corresponded to SNVs (Fig 5A). Many polymorphisms accumulated in the region corresponding to intron 2: 92 of which 38 were SNVs; however, a large number were also found dispersed throughout the sequence, including all exons (Fig 5A, S5 Fig). Only 13 polymorphisms, consisting mostly of insertions or deletions of 1 bp and one SNV were identical between spores.

Fig 5. Distribution of variants along the 1209 bp AOX sequence of R. irregularis.

All the 288 variants (SNVs, MNVs, deletions, insertions and replacements) and only the SNVs present in the 9 spores are shown in A). The variants present in INOQ, BEG144 and BEG72 isolates plus the variants unique to each isolate are presented in B), C) and D), respectively.

In the three spores of the INOQ isolate, a total of 123 variants were identified, of which 47 were unique to this isolate (16% of the 288 variants) (Fig 5B). Regarding the BEG144 isolate, a total of 167 variants were found and 80 were exclusive to this isolate (28% of the 288 variants) (Fig 5C). For BEG72, 141 polymorphisms were found and 54 (19% of the 288 variants) were unique to this isolate (Fig 5D). When comparing the three spores from the same isolate, the percentage of unique polymorphisms in each spore varied with isolate: 33% (13% corresponding to SNVs), 37% (17% for SNVs) and 26% (7% for SNVs) for INOQ, BEG144 and BEG72, respectively.

The frequency (read count divided by read coverage) distribution of all variants (in relation to the total number of variants in one spore) was plotted on histograms, ranging from rare (<2% of frequency of occurrence) until variants that showed a frequency of occurrence of almost 100% (Fig 6). All isolates presented highly frequent polymorphisms (>80–100% of frequency of occurrence), particularly SNVs (Fig 6). Nevertheless, most of variants showed until 20% of frequency of occurrence. The frequency of SNVs was relatively similar amongst isolates, ranging from low frequency SNVs until highly frequent SNVs, with several of them showing frequencies close to 100% (Fig 7A–7C).

Fig 6. Variants frequency (deletions, insertions, SNVs, MNVs and replacements) distribution on single spores of INOQ, BEG144 and BEG72 isolates.

Variants can range from less than 2% until almost 100% of frequency of occurrence (relatively to the total number of variants in one spore).

Fig 7. Frequency of SNVs along the RiAOX sequence on the INOQ, BEG144 and BEG72 isolates.

Variants (SNVs and MNVs) that could induce putative aa substitutions were compared amongst spores (Fig 8, S3 Table, S6 Fig). These SNVs and MNVs ranged from low until highly frequent (S3 Table). Most of other type of variants induced frameshift mutations and not aa changes. In INOQ 2, 7 SNVs and one MNV that could lead to aa changes were identified, being located in exons 1, 3 and 4. One SNV was located in the AOX ferritin-like domain on exon 3. Five SNVs were identified in INOQ 3 spore (one within the AOX ferritin-like domain), and 10 variants (8 SNVs and 2 MNVs) in the INOQ 5 (two SNVs within the AOX ferritin-like domain) (Fig 8). BEG144 isolate showed 24, 12 and 13 SNVs or MNVs in spores 2, 3 and 8, respectively. Of these, several were located in the same positions of exons 1, 3 and 4 (Fig 8) and 9 were within the AOX domain. BEG72_1 showed 8 variants that could induce aa changes, and two were located in exon 3, inside the AOX ferritin-like domain. BEG72_2 and BEG72_4 showed 10 and 7 variants, respectively, of which two (either SNV or MNV) were located in the same position as in BEG72_1, in the exon 3 (Fig 8).

Fig 8. Location of the putative amino acid changes induced by SNVs or MNVs within the AOX ferritin-like domain, in the INOQ, BEG72 and BEG144 isolates.

Exons are indicated by different colours. The 52 bp of the predicted N-terminal are underlined. The AOX ferritin-like domain, obtained by nucleotide search on CDD database is italicised. The putative amino acid changes are bold and underlined. The location of the four universally conserved glutamate residues (E) and the two universally conserved histidine residues (H) are presented in bold and enlarged.

Discussion

4.1. Characterisation of RiAOX gene and corresponding translated peptide

The comparison of R. irregularis AOX genomic sequence with other fungi revealed a high variability in the gene structure, not only related with the number of exons/introns but also with their sizes (Fig 1). This differs greatly with respect to plant AOX genes, with a well-known structure typically composed by four exons with well conserved sizes. Only exceptionally, AOX plant genes present variation on exons number, and consequently on their size, which is due to the occurrence of events related with intron loss or gain [38,39]. The cause and the effects of the variation on number and length of fungal AOX introns that we observed in the present study is unknown; but in fact, fungi have been shown to possess extensive and rapid intron gains and losses, even in closely related species [40]. Data obtained from Cryptococcus and Fusarium genomes showed very recent intron gains and losses, indicating that internal intron gain and loss might be a general trend affecting fungal intron evolution over short evolutionary timeframes [41]. Moreover, a high number of variable intron positions were also found in several genes of the Cryptococcus and Fusarium clades [41].

RiAOX has four exons and three introns (Fig 1). This agrees with the mean number of 2.46 introns in R. irregularis genome [42]. However, for instance the microsporidian S. lophii does not have any intron on the AOX gene. Microsporidians seem to have almost [43] or completely [44] lost introns in their genomes, concomitantly with a low efficiency of splicing of the remaining ones. In the genome of S. lophii only 8 putative intron-containing genes were found [45], and AOX is not among them. In contrast, AOX gene of A. bisporus showed 10 introns. A. bisporus has been described as the preferred model for adaptation, persistence, and growth in a humic-rich environment, and has particular genetic and enzymatic mechanisms governing adaptation to the niche formed during plant degradation [46]. Since introns enable alternative splicing and often encode small RNAs such as miRNAs [47], thus allowing a finer level of regulatory control, differences in intron number might be related with individual lifestyle and particular adaptations of fungal species. New introns may be fixed by selection if part of the coding sequence has negative effects on the function of the protein (e.g., deleterious mutation). In this case, splicing of the affected sequence may remedy the deleterious effects [48]. Also, the presence of introns may provide benefits under stress conditions and play an important role in adapting growth to various conditions [49]. Whether this is an explanation for the high number of introns in A. bisporus AOX we cannot say; however it might be that variations in fungal AOX gene intron number relate not only with genetic similarity but also with habitat and its constraints.

The predicted amino acid sequence of the RiAOX gene revealed one of the largest N-termini analysed in the present study, comparable to that of some Ascomycota. This region, located N-terminally to the first hydrophobic domain, has long been recognised as variable compared to the rest of the protein amongst different organisms [50,51]. There is also a length difference between prokaryotic and eukaryotic AOX sequences since N-termini of eukaryotic AOX are much longer than their prokaryotic counterparts [51,52]. In the present study, only eukaryotic AOX proteins were analysed, and the predicted N-termini that can support mitochondrial targeting sequence and the cleavage site showed a great size range amongst fungal species (Fig 2). Interestingly, from the three Microsporidia species analysed, only one displayed a predictable AOX N-terminal (T. hominis), which was also the smallest N-terminal found along with that of R. allomycis (Fig 2). Microsporidians lack canonical mitochondria, having mitosomes instead [53], and also lack Krebs cycle and a respiratory chain, but retain an alternative respiratory pathway which however seems to be differentially regulated along their life cycle [54]. These facts might be related with the non-existence or the short size of the predicted AOX N-terminal that we found in microsporidians. However, and similarly to bacterial AOX [52], the essential information in AOX of microsporidians, R. allomycis and some Zygomycota might be contained in a minimal N-terminal or then is unnecessary for the cell-membrane targeting.

4.1.1. Regulatory motifs in fungi AOX

Multiple alignments of AOX sequences revealed that N-termini of fungal and plant proteins are highly divergent, particularly on helix α1 and on the region containing the plant conserved CysI (S1 Fig). In plants, CysI and CysII are involved in the mechanism of α-keto acid (particularly pyruvate) regulation [55,56]. The more N-terminal CysI acts as a site for intermolecular bond formation and α-keto acid activation [57], but CysII has a less well defined role in this activation. None of these cysteines were found in fungi, indicating a different mode of AOX regulation. Also animals lack the N-terminal cysteine, important for plant AOX regulation [11]. Crichton et al. [25] showed also that the E/DNV motif may be crucial during pyruvate activation in non-thermogenic plants. It was suggested that the presence or not of a conserved CysI residue and/or E/DNV motif, and a preferential stimulation by pyruvate or succinate may be related to the thermogenic (or not) phenotype of plants [58]. Interestingly however, our results showed that R. allomycis has the ENV motif, identical to most of the plant species, and RiAOX presents the ENS motif, identical to some eudicot AOX1d. Contrarily to plants, AOXs of fungi are generally monomeric and subjected to post-translational regulation by purine nucleoside 5’ monophosphates GMP and/or AMP, and its induction is much dependent on cytochrome pathway restriction and triggering by ROS [51]. Therefore, the reasons for this similarity with plants remain elusive, but it does not seem likely that it indicates any similarity in regulation with plants.

4.1.2. Phylogenetic analysis of R. irregularis AOX

The results obtained in the present study showed a clear division between plant and fungal AOX protein sequences (Fig 3). In plants, AOX is encoded by a small gene family composed by a maximum of six gene members distributed in two subfamilies, the AOX1 and AOX2-subfamily (Cardoso et al. 2015). Our results are in accordance with the recent plant AOX classification scheme proposed by Costa et al. [37]. AOX sequences were completely separated between the two AOX subfamilies, the AOX2 (only described in eudicot plant species) and AOX1-subfamily, and amongst the AOX1 it created three main groups (AOX1a-c/1e, AOX1d monocots and AOX1d eudicots).

Concerning fungi AOX, our results clearly show the existence of two major clades: the Dikarya subkingdom, which includes Basidiomycota and Ascomycota phyla and a clade comprising all the other studied phyla (Fig 3). The fungal AOX phylogenetic tree therefore resembles the general phylogeny constructed for fungi [59], thus seeming that AOX phylogeny echoes fungi lineage-specific events.

4.1.3. Phylogeny of AOX in fungi

The position of R. irregularis AOX and of its sister group formed by members of the Zygomycota (sub-phylum Mucoromycotina, order Mucorales) (Fig 3) reflects the phylogenetic relation described by Ebersberger et al. [59], Tisserant et al. [42] and Lin et al. [60], where the Glomeromycota and particularly R. irregularis are referred as close to Mucorales. Interestingly, Tisserant and colleagues [42] have compared the genome of R. irregularis with Mucoromycotina and Chytridiomycota genomes and found similar patterns of absence of several genes related to diverse metabolisms and detoxification/stress responses, suggesting that this loss is a common feature of the early diverging fungi. However, whereas for instance Lin and colleagues define Mucoromycotina as the closest fungi to R. irregularis, when regarding only AOX sequences R. irregularis seems more related to Chytridiomycota and Cryptomycota. R. irregularis is an obligate biotrophic symbiont and R. allomycis is an endobiotic biotrophic parasite; whether the relative proximity between AOX of R. irregularis and R. allomycis indicates some resemblance in lifestyle is unknown but can be hypothesised. In any case, R. irregularis AOX sequence seems to relate with that of ancient fungi, given the fact that R. allomycis has been placed on the putative primary branch of the fungi, basal to Chytridiomycota [61].

Although the division between the clade containing R. irregularis and the sister clade of Zygomycota was clear, their position relatively to Microsporidia was unresolved. Nevertheless, a relatively close AOX phylogenetic relationship between ancient fungi and microsporidians seems to be supported by findings of James et al. [62], who related R. allomycis with Microsporidia through shared signatures of parasitism and genomic features.

4.2. Putative role of AOX genes in AM symbiosis

AOX has already been described as participating in the regulation of growth, development and resistance against oxidative stress in the Ascomycota plant pathogen Sclerotinia sclerotiorum [19] and in the transition from mycelium to the yeast form in the human pathogen Paracoccidioides brasiliensis [18]. Our results from the tomato mycorrhization trial also point to a dynamic involvement of RiAOX in a crucial part of R. irregularis lifecycle–plant root mycorrhization. This relates to the reported by Besserer and co-workers [22], who showed that AMF AOX respiratory chain is involved in early steps of symbiosis, mainly during spore germination and hyphal branching of Gigaspora rosea. These pre-contact phases are followed by the infection of roots via fungal appressoria, the subsequent colonisation of the cortex by fungal hyphae, and finally the formation of arbuscules. The further progress of the symbiosis is characterised by permanent arbuscule build-up and break-down [2,3]. In our study, by 6 weeks post-inoculation the number of arbuscules was high and transcript levels of RiAOX had decreased (Fig 4F), suggesting that RiAOX is implicated in the early phases of the symbiotic process.

In tomato, AOX family is composed by four gene members, three AOX1 (SlAOX1a, SlAOX1b, SlAOX1c) and one AOX2 subfamily members [63]. Similarly to other plants, also in tomato different reports show the involvement of SlAOX1 in the response to biotic and abiotic stress conditions [63,64]. AOX2, only found in eudicot plant species has been described as typically constitutive or related to development (see review in Cardoso et al. [65]). In the present work, a constitutive expression was found for all the four SlAOX genes analyzed, though at lower levels in the non-mycorrhizal plant roots, (Fig 4A–4D). SlAOX1a and SlAOX1c expression (and at a lower extent SlAOX1b) in mycorrhizal plants seemed to correlate with the development of R. irregularis arbuscules within the root, as transcript levels increased after 4 and especially after 6 weeks of inoculation, whereas SlAOX2 showed a similar pattern between mycorrhizal and non-mycorrhizal plants. The pattern of transcript accumulation observed in mycorrhizal plant roots for both SlAOX1a and SlAOX1c genes, which seems to occur simultaneously with the arbuscules establishment, is coincident with the pattern observed for the AM-specific tomato phosphate transporter SlPT4, which also correlated with mycorrhizal development. Phosphate (Pi) plays a central role in virtually all major metabolic processes in higher plants and algae, including photosynthesis and respiration [66,67] and enhanced uptake of inorganic Pi is one of the major benefits that plants derive from AMF. Therefore, the increase in the transcript accumulation of SlPT4 here identified would be related with the development of mycorrhiza and the synthesis of phosphate transport proteins by the plant to uptake the Pi supplied by the AMF. Yip and Vanlerberghe [68] investigated the existence of a potential involvement of AOX during resupply of P to P-limited cells, which is accompanied by a dramatic stimulation of respiratory O₂ consumption. They measured O₂ consumption rates upon the use of cytochrome or alternative pathway inhibitors in wild type and transgenic cell lines lacking AOX, and provided an in vivo evidence that AOX has a role during the period of high respiration supporting the Pi uptake, which seems to dampen the mitochondrial generation of active oxygen species. Therefore, the increase in both SlAOX1a and SlAOX1c observed in the present work can be hypothesized as linked to an increase in cell respiration in the arbusculated cells to support the continuous increment in Pi released into the cell as a consequence of root colonisation, while the AMF AOX would be mainly associated to the early stages of the plant-fungus interaction.

4.3. High throughput sequencing of RiAOX reveals intra and inter-isolate spore variability

AMF do not pass through a uninucleate stage in any part of their life cycle and mature AMF spores contain hundreds of nuclei obtained directly from the parent mycelium (e.g. [7]). In the AMF Glomus etunicatum a very high level of allelic variation per single spore was found for a low copy number gene [69]. Interestingly, these authors found that an individual spore always contains fewer alleles than the overall allelic diversity observed when eight spores were pooled together. This is in agreement with the results obtained in the present work. When considering all the 9 spores from the three independent isolates, we found a total of 288 AOX nucleotide variants (Fig 5), but the maximum number of variants found within one spore was of 132, for BEG144_2. Moreover, the total number of variants found for the three spores of BEG144 was of 167, which is superior to the number of variants of the spore BEG144_2. Similar results were found for the other isolates. Such differences in AMF genetic variation among sister spores may arise from a genetic bottleneck occurring during sporulation [69]. Indeed, segregation of genetically differentiated nuclei can lead to different genomic contents between subcultures from the same isolate. R. irregularis isolate (DAOM197198) has been maintained in several different in vitro culture collections since 1992 [70]. Cardenas-Flores et al. [71] found that subcultures from the same isolate which had been maintained in separate laboratories for 69 generations anastomosed (or exchanged nuclei through hyphal fusion) at a much lower rate than spores taken from within the same isolate and subculture. This could indicate that subcultures no longer share the same genomic content at a given gene, since decreased rates of anastomosis were observed between genetically diverse isolates [8,72].

In the present study, the three R. irregularis isolates showed different patterns and numbers of exclusive AOX polymorphisms’ for each spore. The BEG72 isolate showed a lower percentage (26%) of unique variants for each of the spores than INOQ (33%) or BEG144 (37%). The three isolates have distinct geographic origins and BEG144 was originally isolated in 1979, BEG72 in 1996 [32] and INOQ in 1999 [73]. The higher or lower homogeneity within isolates is likely to be highly related with its origin and environmental constraints, particularly important due to AOX relatedness with biotic and abiotic stress responses. Also, when comparing the number of variants from the three isolates, it is not possible to establish any relation with time since initial isolation or in vitro propagation, as the number of variants from the initial population is unknown. Without replenishment of genetic variation during repeated propagation, segregation may lead to a loss of nuclear variants with each new generation. The only means of countering this loss is through the action of anastomosis, in which hyphal strands fuse and nuclei are exchanged [74]. Interestingly, it was shown that genetic exchange in AMF can also be beneficial for plant growth, and that symbiosis-specific gene transcription is altered by such genetic exchange [75].

A recent study from Lin and co-workers [60] brought new insights to the question of how genetic diversity is organized within R. irregularis spores; interestingly, they have found low levels of polymorphisms genome wide between nuclei of R. irregularis, but in contrast, within a single nucleus, the 45S rDNA repeat unit was found to be quite divergent. They found evidence of a basically homokaryotic genome of the R. irregularis studied strain, though it might happen that slightly divergent nuclei can still coexist [60]. The gene copy number of AOX in R. irregularis genome is unknown, so we cannot say whether the variants found in the present study come from single or multiple nuclei, or from a combination of both. Nevertheless, given the fact that the number of polymorphisms found when considering the three spores from each isolate was higher than in individual spores, it seems that there is diversity loss in individual spores. Also in our study, due to the amplicon target approach, the average variant coverage was much higher than the average sequencing depth reported by these authors so likely we were prone to find more polymorphisms.

In spite of the high polymorphic diversity that was found in the spores, it seems that a high number of AOX variants are under functional constraints, since many occurred in intron 2 or induced frameshift mutations. However, since some putative amino acid changes induced by SNVs or MNVs were found within the AOX conserved domain (Fig 8), it might happen that some of these variants may be subjected to positive selective pressure within the population. AMF show sufficient phenotypic variation and fitness differentials for selection to occur [76]. It has been proposed that direct selection on fungal traits related to their survival and performance in the soil independent of the host is likely the major driver of differentiation for AMF, with evidences for direct fungal responses to soil conditions such as pH, hypoxia, and temperature [76]. However, the precise origins of selective pressures on RiAOX sequences and possible consequences of sequence variability for protein function remain to be determined.

Considerations

The drive of this research was the confirmation of assumptions required for functional marker development: gene involvement and ample sequence variability. First, our results suggest that R. irregularis AOX is involved in the establishment of the symbiosis with plant roots. At plant level, AOX appears more involved on the later stages, when arbuscules are already established and the Pi uptake inside the cell is high. Second, we found relatively high levels of allelic diversity within isolates that could further be tested as functional or related with phenotypic differences. Indeed it seems that AOX genetic diversity exists not only inter but also intra-spore: thus different spores do not necessarily contain the same genetic variation. Also, given AOX relatedness with biotic and abiotic stress responses, the differences in gene variants amongst different R. irregularis isolates are likely to be related with its origin and environmental constraints. Considering these findings, and from an applied perspective, the next steps would be (1) functionally testing the variants and try to establish a link between genotype and phenotype (for a potential technological approach, see [77,78] and (2) analyse the scope for selection at spore level for inoculum production towards a high quality and homogeneous inoculum.

Supporting Information

S1 Fig. Multiple alignments of the hydrophobic region of AOX in plants and fungi (helices α1 and α4).

The conserved regulatory cysteine residue (CysI) is located N-terminal to α1. Some sequences lack the conserved CysI, having a serine residue instead. The conserved CysI is not found in any of the fungal sequences (boxed). The fungal and plant sequences are fairly divergent N-terminal, whereas the helix α4 is more conserved. R. irregularis AOX is dashed.

(TIF)

S2 Fig. Multiple alignments on region 3 of AOX in fungi and plants.

Plants generally present the glutamic acid/aspartic acid, asparagine and valine (E/DNV) motif (with exceptions of AOX1d from monocots and some AOX1d from eudicots). The fungal sequences are much more divergent. RiAOX showed the ENS motif, identical to that of some AOX1d from eudicots.

(TIF)

S3 Fig. Conserved Fe-Fe ligands in the AOX helices α2, α3, α5 and α6.

The four universally conserved glutamate residues and the two universally conserved histidine residues are indicated by asterisks (*). The conserved cysteine residue seen in Angiosperms (CysII) is located on helix α2. Amongst Angiosperms, only the AOX1d sequences from Monocots lack the conserved CysII, having a serine residue instead. The conserved CysII is not found in any of the fungal sequences (boxed).

(TIF)

S4 Fig. Phylogenetic analysis (NJ) of clones from ribosomal rRNA sequences of the ITS-LSU-SSU region from the 9 R. irregularis single spores.

All spores placed in the R. irregularis clade, together with sequences retrieved from databases and already recognized as belonging to R. irregularis. The R. intraradices group was completely separated from R. irregularis.

(PDF)

S5 Fig. Location of all variants (SNVs, MNVs, insertions, deletions, replacements) of RiAOX in the INOQ, BEG144 and BEG72 spores.

(TIF)

S6 Fig. Location of the variants that putatively can induce changes on R. irregularis AOX protein sequence, in the INOQ, BEG144 and BEG72 isolates.

The location of the SNVs is indicated with red lines.

(TIF)

S1 List. Accession numbers of fungal and plant AOX sequences used for phylogenetic analysis.

(PDF)

S1 Table. Specific primers used for qPCR in tomato (Sl) and R. irregularis (Ri).

Sequences accession numbers and amplicon sizes are presented.

(PDF)

S2 Table. Mycorrhizal colonisation of tomato roots.

Tomato roots were inoculated with R. irregularis over a period of six weeks and evaluated microscopically with trypan-blue. Scores according to the method of Trouvelot et al. (1986) are given for frequency (F %), intensity (M %) and relative intensity (m %) of mycorrhizal colonisation. Scores for the total number of vesicles (V %) and arbuscules (A %) are also given. Data represent the means of 4 replicates (± SE). Data not sharing a common superscript letter differ significantly (P < 0.05).

(PDF)

S3 Table. Variants (SNVs and MNVs) that could induce putative amino acid substitutions in the 9 single spores.

The variants within the AOX ferritin-like domain are in red.

(PDF)

Data Availability

The genomic and mRNA AOX gene sequences are available from the NCBI database (accession numbers: KT423114, KT423115).

Funding Statement

This work was supported by the European Commission through the project AGRO-AMF-AOX within the program Industry-Academia Partnerships and Pathways (IAPP, FP7). Still in the frame of the AGRO-AMF-AOX project, JS further acknowledges his recruitment, and CC, HC, AN, CS and BAS appreciate the further support given in the form of a secondment. Inoq GmbH provided support in the form of salary for CS, but did not have any additional role in the study design, data collection and analyses, decision to publish, or preparation of the manuscript. The specific role of CS is articulated in the ‘author contributions’ section. Part of this work was also supported by grant AGL2012-39923 from the Spanish National R&D Plant of the Ministry of Science and Innovation (MICINN). TN was supported by a Marie Curie fellowship (FP7-PEOPLE-2012-CIG Project Reference 321725) and by the Portuguese Foundation for Science and Technology – FCT (SFRH/BCC/52187/2013). The authors further thank the program POPH—Operational Program for Human Potential—and Fundo Europeu de Desenvolvimento Regional (FEDER) funds through the Operational Program for Competitiveness Factors—COMPETE, and national funds through FCT under the strategic project PEst-C/AGR/UI0115/2011 and PEst-OE/AGR/UI0115/2014.