Phosphate in the arbuscular mycorrhizal symbiosis: transport properties and regulatory roles


M. J. Harrison. Fax: +(607) 2546779; e-mail:


In response to the colonization by arbuscular mycorrhizal (AM) fungi, plants reprioritize their phosphate (Pi)-uptake strategies to take advantage of nutrient transfer via the fungus. The mechanisms underlying Pi transport are beginning to be understood, and recently, details of the regulation of plant and fungal Pi transporters in the AM symbiosis have been revealed. This review summarizes recent advances in this area and explores current data and hypotheses of how the plant Pi status affects the symbiosis. Finally, suggestions of an interrelationship of Pi and nitrogen (N) in the AM symbiosis are discussed.


More than 80% of the vascular flowering plants can be colonized by arbuscular mycorrhizal (AM) fungi (Harley & Smith 1983; Schüßler, Schwarzott & Walker 2001). The AM fungus (order Glomeromycota), an obligate symbiont, relies on carbon provided by the plant and in exchange, it improves the mineral nutrition of the plant, in particular acquisition of phosphorus and to some extent, nitrogen (N) (Smith & Read 1997). These are essential mineral nutrients whose availability to the plant is frequently growth limiting; consequently, many plants exhibit some growth stimulation when colonized by AM fungi. Nutrient exchange between the two symbionts is at the core of the AM association and reciprocal transfer is a requirement for a functioning symbiosis. It is therefore not surprising that the exchanged molecules, phosphate (Pi) and carbon, also act as regulatory components of the symbiosis (Jakobsen 1995; Fitter 2006).


In many ecosystems, the phosphorus levels available to plants are limiting for growth, and this has a significant impact on agriculture, particularly in regions where low-input agriculture is practiced (Vance, Uhde-Stone & Allan 2003). Phosphorus exists in the environment as inorganic orthophosphate, primarily involved in inert complexes with cations such as iron phosphate (FePO4) and aluminum phosphate (AlPO4), and in organic molecules such as lecithin and phytate, the latter of which can account for up to 50% of total soil organic Pi (Brinch-Pedersen, Sorensen & Holm 2002). Pi is the only form directly accessible to plants, and the mechanisms they have evolved for Pi assimilation underscore the importance and difficulties of maintaining sufficient cellular levels of this nutrient.

While phosphorus is abundant in the environment, the negative charge of the ionic form makes it easily sequestered by cations such as Fe, Al and Ca, especially in acidic soils (Vance et al. 2003; Ticconi & Abel 2004). This leaves meager amounts of free Pi in the soil solution, where concentrations range from 1–10 µM, whereas cells require Pi in the millimolar range (Bieleski 1973). Moreover, in contrast to some other mineral nutrients, Pi is highly immobile in the soil. For this reason, its acquisition by the roots generates a depletion zone surrounding the epidermis and the root hairs (Shapiro, Armiger & Fried 1960).

Plants have evolved a number of physiological modifications to overcome scarce levels of Pi. One strategy is to increase the root–soil interface to maximize access to available Pi. Root hairs increase general surface area and their growth is influenced by Pi availability (Bates & Lynch 1996; Ma et al. 2001). In addition, because Pi concentration may be higher towards the soil surface, low Pi conditions stimulate a response in some plants known as topsoil foraging, a rearrangement of root development to favour lateral over downward growth (Williamson et al. 2001; Ticconi & Abel 2004).

A second strategy is to solubilize Pi trapped in complexes. Roots secrete organic acids such as malate and citrate, which compete with Pi for cation-binding partners (Hinsinger 2001; Vance et al. 2003; Johnson & Loeppert 2006) and phosphatases to mineralize Pi from organic compounds (Lefebvre et al. 1990; Duff, Sarath & Plaxton 1994; del Pozo et al. 1999; Zakhleniuk, Raines & Lloyd 2001). Recent experiments have reported enhanced tolerance to low Pi conditions in transgenic plants overproducing organic acids or phosphatases (Lopez-Bucio et al. 2000; Richardson, Hadobas & Hayes 2001; Xiao, Harrison & Wang 2005; Xiao et al. 2006).

A third strategy is to form a symbiotic association with AM fungi to benefit from their efficient Pi-acquisition capacity. Their fungal mycelia grow up to 100 times longer than root hairs (Jakobsen, Abbott & Robson 1992b; Bates & Lynch 1996) and branch, providing an efficientnutrient-absorbing net beyond the Pi-depletion zone. Fungal hyphae also have an enhanced ability to mineralize organic P (Joner, Ravnskov & Jakobsen 2000; Koide & Kabir 2000; Feng et al. 2003; Shibata & Yano 2003) and, in the case of a mycorrhizal interaction between Tagetes patula and Glomus etunicatum, it was additionally shown that AM fungal colonization induces expression and secretion of a plant-derived acid phosphatase in the rhizosphere, which further liberates Pi (Ezawa, Hayatsu & Saito 2005).


Following acquisition from the soil, the plant needs to distribute Pi to tissues throughout the plant, and Pi is cycled between the roots and the shoots in the xylem and phloem (Jeschke et al. 1997). Currently, little is known about transporters involved in long distance transport of Pi. In Arabidopsis thaliana, PHO1, a transmembrane (TM) protein, is required for loading Pi into the xylem and may also play a role in Pi signaling (Poirier et al. 1991; Hamburger et al. 2002; Wang et al. 2004). Within each tissue, Pi must be delivered to each cell, and then to the various subcellular compartments and organelles. While it is clear that these transfer processes require Pi transporters, in many cases, details of the actual transporters involved are not yet known. The first stage towards understanding these processes is the identification of potential players. As Pi concentrations impact many cellular reactions, Pi levels within the cytoplasm and within organelles must be tightly controlled, and plants have evolved various strategies to regulate and maintain Pi levels within strict limits, including the use of the vacuole to buffer fluctuations in cytoplasmic Pi content (Bieleski 1973). Pi often needs to be transported against unfavourable electrochemical gradients (Bieleski 1973), and active transporters are needed to transport Pi across the membranes into the root cells and also into subcellular compartments. Through their various substrate specificities and expression patterns, plant Pi transporters (symporters and translocators) further connect the metabolism of subcellular compartments and tissues (Weber, Schneidereit & Voll 2004; Toyota et al. 2006). They also provide flexible ways of altering metabolite fluxes throughout the plant when changes in Pi supply occur.

Pi:H+ symporters

Many Pi:H+ symporters have been identified in plants (Laloi 1999; Mudge et al. 2002; Poirier 2002; Versaw & Harrison 2002; Zhao et al. 2003; Picault et al. 2004; Glassop, Smith & Smith 2005; Nagy et al. 2006). Among those, the best described belong to the Pht1 family (TC No. 2.A.1.9; Saier 2000). Its members are homologues of the yeast PHO84 Pi transporter and the high-affinity Pi transporters identified in fungi, and together they belong to the major facilitator superfamily of proteins (Pao, Paulsen & Saier 1998). Of the Pht1 proteins examined, all were located in the plasma membrane (Chiou, Liu & Harrison 2001). They are predicted to have 12 TM domains, constituted of two partially duplicated subdomains of six TM segments (Lagerstedt et al. 2004). Heterologous expression in yeast or plant cells, coupled with Pi transport assays, indicated that the transporters usually show moderate to high-affinity for Pi (Table 1). Evaluation of the complete Pht1 family in A. thaliana revealed a large diversity of expression patterns throughout the plant, suggesting a broad involvement of this family in Pi transport within the plant (Mudge et al. 2002). Among the nine Pht1 genes encoded by the Arabidopsis genome, four were expressed in the root epidermis, suggesting that these may be involved in phosphate uptake from the soil. T-DNA insertions in two of those genes (ARAth;Pht1;1 and ARAth;Pht1;4) demonstrated their significant role in Pi acquisition (Shin et al. 2004).

Table 1.  Expression pattern and affinities of some fungal and plant phosphate (Pi) transporters
OrganismOfficial nomenclatureaOther namesbAccession numberExpression patterncApparent KmdReferencee
Neurospora crassa  PHO5AAA74899  1
Saccharomyces cerevisiae  PHO84P25297 µM2
Glomus versiforme  GvPTAAC49132 18 µM (yeast)3
Glomus intraradices  GiPTAAL37552  4
Glomus mosseae  GmosPTAAZ22389  5
Lycopersicon esculentum LYCes;Pht1;1LePT1O22548 31 µM (yeast)6, 7, 8, 9
L. esculentum LYCes;Pht1;2LePT2O22549  7, 9
L. esculentum LYCes;Pht1;3LePT3Unpub.AM + 9
L. esculentum LYCes;Pht1;4LePT4AAX85192AM S 9
L. esculentum LYCes;Pht1;5LePT5AAX85194AM + 9
Solanum tuberosum SOLtu;Pht1;1StPT1CAA67395AM −280 µM (yeast)9, 10, 11
S. tuberosum SOLtu;Pht1;2StPT2CAA67396AM −130 µM (yeast)9, 11
S. tuberosum SOLtu;Pht1;3StPT3CAC87043AM +64 µM (yeast)9, 11, 12
S. tuberosum SOLtu;Pht1;4StPT4AAW51149AM S 9
S. tuberosum SOLtu;Pht1;5StPT5AAX85195AM S 9
Medicago truncatula MEDtr;Pht1;1MtPT1AAB81346AM −192 µM (yeast)13, 14
M. truncatula MEDtr;Pht1;2MtPT2AAB81347AM − 13, 14
M. truncatula MEDtr;Pht1;4MtPT4AAM76744AM S493–668 µM (yeast)12, 14
Lotus japonicus LOTja;Pht1;1LjPT1BAE93351AM − 15
L. japonicus LOTja;Pht1;2LjPT2BAE93352AM − 15
L. japonicus LOTja;Pht1;3LjPT3BAE93353AM + 15
Oryza sativa ORYsa;Pht1;1OSPT1AAN39042AM − 16
O. sativa ORYsa;Pht1;2OSPT2AAN39043AM − 16
O. sativa ORYsa;Pht1;3OSPT3AAN39044AM − 16
O. sativa ORYsa;Pht1;4OSPT4AAN39045  16
O. sativa ORYsa;Pht1;5OSPT5AAN39046  16
O. sativa ORYsa;Pht1;6OSPT6AAN39047AM − 16
O. sativa ORYsa;Pht1;7OSPT7AAN39048  16
O. sativa ORYsa;Pht1;8OSPT8AAN39049  16
O. sativa ORYsa;Pht1;9OSPT9AAN39050AM − 16
O. sativa ORYsa;Pht1;10OSPT10AAN39051AM − 16
O. sativa ORYsa;Pht1;11OSPT11AAN39052AM S 16, 17
O. sativa ORYsa;Pht1;12OSPT12AAN39053  16
O. sativa ORYsa;Pht1;13OSPT13AAN39054AM + 16, 17
Hordeum vulgare HORvu;Pht1;1HvPT1AAN37900AM −µM (rice cells )18, 19, 20
H. vulgare HORvu;Pht1;2HvPT2AAO72433AM − 18, 19, 20
H. vulgare HORvu;Pht1;3HvPT3AAO72439  18, 19, 20
H. vulgare HORvu;Pht1;4 AAO72437  19, 20
H. vulgare HORvu;Pht1;5 AAO72435  19, 20
H. vulgare HORvu;Pht1;6 AAN37901 385 µM (rice cells )19, 20
H. vulgare HORvu;Pht1;7 AAO72436  19, 20
H. vulgare HORvu;Pht1;8 AAO72440AM + 19, 20
Triticum aestivum TRIae;Pht1;myc CAH25730AM S 20
Zea mays ZEAma;Pht1;1ZmPT2AAY42385AM − 21, 22
Z. mays ZEAma;Pht1;2 AAY42386  22
Z. mays ZEAma;Pht1;3 AAY42387  22
Z. mays ZEAma;Pht1;4ZmPT1AAY42388AM − 21, 22
Z. mays ZEAma;Pht1;5 AAY42389  22
Z. mays ZEAma;Pht1;6 CAH25731AM + 20, 22
Z. mays f ZmPT3AAT51692AM − 21

Two other families of Pi:H+ symporters have been identified in plant organelles (plastids and mitochondria), but none have yet been identified in the tonoplast. The plastid symporters comprise the Pht2 family and exhibit a structure similar to the Pht1, except for a long hydrophilic loop between the eighth and ninth TM domains, and a long hydrophilic N terminus that includes the plastid signal sequence (Poirier 2002). Although they share sequence similarity with the Pi:Na+ symporters from Neurospora crassa (PHO4) and Saccharomyces cerevisiae (PHO89), all the plant members characterized so far exhibited only Pi:H+ transport properties in yeast (Mann et al. 1989; Daram et al. 1999; Versaw & Harrison 2002; Zhao et al. 2003).

The Pht3 Pi:H+ symporters (Rausch & Bucher 2002), also known as mitochondrial phosphate carriers (PiC), are represented by three members in A. thaliana (Picault et al. 2004). They display the typical structure of the larger mitochondrial carrier family, containing six TM domains composed of three repeated segments of two TM alpha-helices separated by a hydrophilic loop (Laloi 1999; Nakamori et al. 2002; Karandashov & Bucher 2005). Based on their mammalian homologues, they are predicted to function via Pi:OH− antiport and to catalyse Pi:Pi exchange between the matrix and the cytosol in addition to Pi:H+ symport (Stappen & Kramer 1994; Wohlrab & Briggs 1994; Takabatake et al. 1999).

Pi antiporters/translocators

Genome-sequencing data from Arabidopsis suggest that plant mitochondria possess yet another class of carrier proteins capable of transporting Pi. Plant homologues of dicarboxylate carriers (DIC) have been identified but direct evidence of their activity is still lacking. In animals, DICs catalyse the exchange between Pi and dicarboxylates such as malate, succinate and malonate (Laloi 1999; Picault et al. 2004). Further experiments are required to confirm the properties of plant DICs.

Plastid translocators are significantly better described than their mitochondrial counterparts. Because of their central role in plant metabolism, plastids are equipped with a large number of translocators capable of exchanging Pi with various phosphorylated compounds (Knappe, Flugge & Fischer 2003; Toyota et al. 2006). In the absence of a crystal structure, the molecular structure of the translocators (between six and nine TM domains) is still a matter of debate (Flügge 1999; Knappe et al. 2003; Weber, Schwacke & Flugge 2005). Plastid Pi translocators fall into four families based on substrate specificities: Triose phosphate:Pi translocators (TPT; Flügge et al. 1989), PEP:Pi translocators (PPT; Fischer et al. 1997), Glc-6P:Pi translocators (GPT; Kammerer et al. 1998) and xylulose-5-phosphate:Pi translocators (XPT; Eicks et al. 2002). Many predicted plastid translocators have not yet been assigned a function (Knappe et al. 2003), making it very likely that the array of substrates known to be transported across the plastid inner membrane in exchange for Pi will be expanded in the future.


The AM symbiosis is accompanied by a dramatic reorganization of Pi fluxes in the plant but other cellular adjustments are also necessary to allow successful colonization of the root by the fungus. The sequential steps of the colonization process are well known (Harrison 1997). Following the spore germination and growth of the hyphal germ tube, the fungus invades the root. Once inside the cortex, the fungal hyphae penetrate the cortical cells, and in each cell, the fungus differentiates to form a highly branched structure called an arbuscule. Although the arbuscule can fill most of the cell space, it does not compromise the integrity of the plant plasma membrane, and cortical cells respond to the invasion by enveloping the arbuscule in a specialized host membrane known as the periarbuscular membrane. The large interface generated between the fungal arbuscular membrane and the plant periarbuscular membrane has been proposed to be the site of solute exchange between the two symbionts (Harley 1969; Cox & Tinker 1976).

The transfer of Pi from the fungus to the plant has been demonstrated in compartmented pot systems via the use of 32P- or 33P-labelled Pi sources accessible only to the fungus (Jakobsen, Abbott & Robson 1992a; Pearson & Jakobsen 1993; Smith, Smith & Jakobsen 2003, 2004). This system revealed that the fungal contribution to the plant Pi acquisition ranges from a small percentage to almost all of the acquired Pi, and varies depending on each plant/fungus combination (Pearson & Jakobsen 1993; Smith et al. 2003, 2004). It was shown that even when Pi was delivered to the root system, all plants tested would obtain at least some Pi via the fungus (Smith et al. 2003, 2004). In specific cases, such as when colonized with Glomus intraradices, some plants acquired their entire Pi through the fungal pathway. These results included symbioses in which colonized plants showed no increase in biomass or total shoot P concentration in comparison to non-mycorrhizal plants (Smith et al. 2004). It had long been assumed that AM associations showing no plant growth improvement, so-called ‘non-responsive’ plants, made no contribution to plant Pi uptake. These results showing the predominance of the fungal pathway in those interactions overturn this long-held belief and broaden the number of species likely receiving fungal Pi.

Transport proteins potentially involved in symbiotic Pi transport in the fungi have been identified. In particular, AM fungi possess high-affinity Pi:H+ symporters, which are homologues of the yeast PHO84 Pi transporter and belong to the major facilitator superfamily similar to plant Pi transporters (Harrison & van Buuren 1995; Pao et al. 1998; Maldonado-Mendoza, Dewbre & Harrison 2001; Benedetto, Magurno & Lanfranco, 2005). The three AM Pi transporters identified so far (GvPT, GiPT and GmosPT, from Glomus versiforme, G. intraradices and Glomus mosseae, respectively) are all expressed in extraradical hyphae, while GmosPT also shows significant expression in intraradical hyphae.

Once Pi enters the fungal cytoplasm, it accumulates quickly in the vacuole (Solaiman et al. 1999; Ezawa, Smith & Smith 2001). It was recently demonstrated that when a fungus is transferred from a Pi-deprived to a Pi-rich medium, Pi accumulates and is converted very rapidly (less than 3 h) into large quantities of polyphosphates (poly-P) (Ezawa et al. 2003). Poly-P is a linear polymer of three to thousands of Pi connected by high-energy phospho-anhydrate bonds (Kornberg, Rao & Ault-Riche 1999). This molecule is ubiquitous and fulfills multiple functions ranging from Pi storage to improvement of translation fidelity (Kornberg et al. 1999; McInerney, Mizutani & Shiba 2006). Although the range of functions assigned to poly-P is not yet known in AM fungi, their involvement in Pi storage and translocation is well established. Poly-P accumulated in the vacuolar compartment can be translocated from the extraradical hyphae to the intraradical hyphae possibly via cytoplasmic streaming and/or along a motile tubular vacuole system (Cooper & Tinker 1981; Smith & Read 1997; Olsson et al. 2002; Uetake et al. 2002).

The observation of a direct correlation between root poly-P content and root colonization level (with Gigaspora margarita) gives an indication of the massive translocation of poly-P occurring from the extraradical hyphae to the intraradical hyphae (Ohtomo & Saito 2005). Although a broad distribution of poly-P chain length has been observed in the extraradical hyphae (Ezawa et al. 1999; Rasmussen et al. 2000), it was revealed that the poly-P chains are shorter in intraradical hyphae compared to extraradical hyphae of G. margarita (Solaiman et al. 1999; Ohtomo & Saito 2005). This observation suggests that poly-P is hydrolysed in the intraradical hyphae in order to release Pi to the plant. The exact mechanism of poly-P breakdown in the intraradical hyphae has not been uncovered, yet the detection of intense phosphatase activities around arbuscules indicates a possible role of these enzymes in this phenomenon. Alkaline and acid phosphatase activities have been found in intra- and extraradical hyphae (Smith & Gianinazzi-Pearson 1990; van Aarle, Olsson & Söderström 2001; Ezawa et al. 2001). In a study combining Marigold and Glomus sp., Ezawa et al. (2001) revealed that extraradical hyphae contained an alkaline phosphatase activity (pH 7.5) demonstrating a higher affinity for long-chain poly-P, whereas the intraradical hyphae showed an active acid phosphatase activity (pH 5.0) with a higher affinity for short-chain poly-P. Although those results suggested that an acid phosphatase would be the most likely candidate for releasing Pi from poly-P, Kohjima & Saito (2004) also detected a strong alkaline phosphatase activity correlated with fungal structures. The recent cloning of fungal alkaline phosphatases (GiALP and GmALP from G. intraradices and G. margarita, respectively) confirmed that this type of enzyme was expressed more highly in intraradical hyphae than in extraradical hyphae (Aono et al. 2004). However, biochemical data suggested that the alkaline phosphatase activity was involved in fungal sugar metabolism rather than poly-P breakdown (Ezawa et al. 1999). In the absence of data regarding substrate specificities of the cloned fungal alkaline phosphatases, no definitive answer has been formulated regarding which phosphatases are responsible for the poly-P breakdown in the intraradical hyphae.

Once Pi has been released in the intraradical hyphae, it is assumed to be transferred to the periarbuscular apoplastic compartment, by an as yet unidentified mechanism. From here, it is available to the plant. This new source of Pi presented within the root cortex is accompanied by large rearrangements in plant Pi transport and a coincident change in Pi transporter regulation.


Transcription level of Pht1 transporters is affected by the AM symbiosis

The expression of the Pht1-type Pi transporters was evaluated in mycorrhizal roots of a broad range of plant species and in many cases, modifications in expression were noted. The data are summarized in Table 1. Each Pht1 member is indicated according to the official nomenclature (Karandashov & Bucher 2005), as well as their previously reported names. Here, we refer to the Pi transporters according to this official nomenclature.

Although the transcript levels of many Pht1 transporters decrease with an increase in Pi status and also in the AM symbiosis (Liu et al. 1998b; Rae et al. 2003; Glassop et al. 2005; Nagy et al. 2005) expression of a small subgroup of Pht1 transporters is actually induced in mycorrhizal roots. Mycorrhiza-specific Pht1 members, expressed strictly in response to AM symbiosis, have been identified in the following species: Medicago truncatula (MEDtu;Pht1;4: Harrison, Dewbre & Liu 2002); rice (ORYsa;Pht1;11: Paszkowski et al. 2002); potato (SOLtu;Pht1;4 and SOLtu;Pht1;5: Nagy et al. 2005); wheat (TRIae;Pht1;myc: Glassop et al. 2005) and tomato (LYCes;Pht1;4: Nagy et al. 2005). Those can be distinguished from mycorrhiza up-regulated Pi transporters, which are strongly induced by AM symbiosis but have a basal expression in uninoculated roots (Table 1). SOLtu;Pht1;3 from Potato (Rausch et al. 2001; Nagy et al. 2005); LOTja;Pht1;3 from Lotus japonicus (Maeda et al. 2006); ORYsa;Pht1;13 from rice (Guimil et al. 2005); HORvu;Pht1;8 from barley and ZEAma;Pht1;6 from maize (Glassop et al. 2005; Nagy et al. 2006) belong to the second category. In some cases (SOLtu;Pht1;3 and ZEAma;Pht1;6), low expression of the AM-induced genes could also be detected in shoots (Nagy et al. 2005, 2006).

For seven plant species analysed, in situ hybridization and promoter::GUS fusion studies indicated that the corresponding mycorrhiza-induced Pht1 genes are predominantly or exclusively expressed in cells containing arbuscules (Rausch et al. 2001; Harrison et al. 2002; Glassop et al. 2005; Nagy et al. 2005; Maeda et al. 2006). Although tomato LYCes;Pht1;1 was initially reported to be induced in cells with arbuscules, it also had an ubiquitous expression level throughout the plant (Rosewarne et al. 1999). The later identification of LYCes;Pht1;4, strongly induced in arbuscules, as well as the detection of two other mycorrhiza-induced Pht1 members in tomato (LYCes;Pht1;3 and 5: Nagy et al. 2005) suggested that the original LYCes;Pht1;1 signal observed around the arbuscule might have resulted from cross-hybridization. In M. truncatula, immunolocalization of MEDtr;Pht1;4 protein revealed that the Pi transporter was detected solely in the plant periarbuscular membrane (Fig. 2; Harrison et al. 2002). The signal was stronger around the fine arbuscular branches and was absent from regions around the hyphal trunk. Furthermore, it could only be detected in developing and mature arbuscules, reinforcing the idea that arbuscules could be the major site of Pi exchange between the two symbionts (Harrison et al. 2002). This expression pattern corroborates previous results suggesting intense active transport on the periarbuscular membrane: plasma membrane H+-ATPases are abundant around fine branches and disappear in decaying arbuscules (Gianinazzi-Pearson et al. 2000). The crucial role played by arbuscules in plant Pi uptake is also reflected by the results of a study on the reduced mycorrhizal colonisation (rmc) tomato mutant (Poulsen et al. 2005). This mutant is characterized by allowing certain AM fungi to develop arbuscules, while halting other AM fungi at various developmental stages of the symbiosis. Results of the study indicated that only those fungi that were able to develop arbuscules triggered the expression of AM-induced LYCes;Pht1;3 and 4, and thus enabled the symbiotic Pi transfer. Therefore, activation of mycorrhiza-induced Pi transporters can be used as a marker for a functional symbiotic Pi uptake pathway.

The heterologous expression of mycorrhiza-inducible Pht1 transporters in yeast has enabled estimates of substrate affinity (Table 1). The mycorrhiza-induced SOLtu;Pht1;3 from potato revealed an apparent Km of 64 µM, whereas the mycorrhiza-specific MEDtrPht1;4 from M. truncatula exhibited a lower affinity for Pi, between 493 and 668 µM (Rausch et al. 2001; Harrison et al. 2002). The biological relevance of these data would be further strengthened if heterologous expression data combined the use of yeast and plant cell culture systems, such as the one developed with tobacco and rice cells (Mitsukawa et al. 1997; Rae et al. 2003).

Formation of the symbiosis and expression of mycorrhiza-induced Pht1 members in arbuscule-containing cells is often accompanied by the down-regulation of other Pht1 transporters, in particular those located at the epidermis (Table 1). To date, down-regulation of Pht1 transporters has been observed in potato (SOLtu;Pht1;1 and 2: Rausch et al. 2001); M. truncatula (MEDtr;Pht1;1 and 2: Liu et al. 1998b); L. japonicus (LOTja;Pht1;1 and 2: Maeda et al. 2006); rice (ORYsa;Pht1;1,2,3,6,9 and 10: Paszkowski et al. 2002); barley (HORvu;Pht1;1 and 2: Glassop et al. 2005) and maize (ZEAma;Pht1;1 and 1;4: Wright et al. 2005). These results reflect the fine balance maintained between fungal and root epidermis uptake pathways. They can also partially explain the results from Smith et al. (2003) showing that even in the absence of any detectable stimulation of the plant Pi content, a significant amount of Pi could be delivered through the fungus. Because the down-regulated Pht1 transporters are also generally responsive to the Pi status of the plant, it is not clear currently whether their down regulation results indirectly from the fungal-induced improvement of the plant Pi status, or from a direct regulation of their expression triggered by the plant in response to the symbiosis. Indeed, in M. truncatula, it was shown that many genes activated by Pi starvation including acid phosphatases and Mt4, are rapidly down-regulated in the AM symbiosis (Burleigh & Harrison 1997, 1998; Liu et al. 1998b, 2003). The Arabidopsis homologue of Mt4 is involved in Pi allocation and its expression may be under the control of a member of the MicroRNA399 family (Shin et al. 2006), which suggests that regulation by MicroRNAs might also operate in the AM symbiosis. A role for the MicroRNA399 in Pi-signaling was very recently demonstrated (Aung et al. 2006; Bari et al. 2006) and in silico analysis, suggests that Pht1 Pi transporters are potential candidates for a regulation by members of the MicroRNA399 family (Jones-Rhoades 2004 #2228; Chiou et al. 2006).

Sequence comparison between characterized fungal Pi transporters, Pht1 members from A. thaliana and AM plant species (Fig. 1) illustrates some interesting features of the Pi transporter family. First, a large majority of mycorrhiza-specific and mycorrhiza-up-regulated Pi transporters cluster into their own subgroups. These subgroups contain only one mycorrhiza ‘non-responsive’ Pht1 member (Zeama;Pht1;5) and do not contain any Arabidopsis Pht1 members. Because Arabidopsis cannot be colonized by AM fungi, this observation suggests that plant species capable of forming mycorrhizas possess specialized types of Pi transporters adapted to the AM symbiosis. Two mycorrhiza-up-regulated Pht1 members (LOTja;Pht1;3 and SOLtu;Pht1;3) do not cluster with this subgroup and cluster with the other Pht1 members, many of which are down-regulated in mycorrhizal roots. Because more than one Pht1 member is induced in mycorrhizal roots of rice, tomato and potato, additional induced transporters will likely be identified in the other plants included in this dendrogram. It will be particularly interesting to see how these new Pht1 members relate to the current groups.

Figure 1.

Unrooted dendrogram of plant and fungal phosphate (Pi) transporters. The tree was generated using alignment of full-length protein sequences by ClustalW. Mycorrhiza-specific, up-regulated and down-regulated Pi transporters are highlighted in red, orange and blue, respectively. Transporters and corresponding plant species are: Arabidopsis thaliana, ARAth;Pht1;1 to 9 (Mudge et al. 2002); Tomato, LYCes;Pht1;1,2 and 4 (Liu et al. 1998a; Nagy et al. 2005); potato, SOLtu;Pht1;1 to 4 (Leggewie, Willmitzer & Riesmeier 1997; Rausch et al. 2001; Karandashov et al. 2004; Nagy et al. 2005); Medicago truncatula, MEDtu;Pht1;1, 2 and 4 (Liu et al. 1998b; Harrison et al. 2002); Lotus japonicus, LOTja;Pht1;1 to 3 (Maeda et al. 2006); rice, ORYsa;Pht1;1 to 13 (Paszkowski et al. 2002; Guimil et al. 2005); barley, HORvu;Pht1;1, 2, and 4 to 8 (Rae et al. 2003; Glassop et al. 2005); wheat, TRIae;Pht1;myc (Glassop et al. 2005); maize, ZEAma;Pht1;1to 6 (Wright et al. 2005; Glassop et al. 2005; Nagy et al. 2006). Fungal transporters are Neurospora crassa, PHO5 (Versaw 1995); Saccharomyces cerevisiae, PHO84 (Bun-ya et al. 1991); Glomus versiforme, GvPT (Harrison & van Buuren 1995); Glomus intraradices, GiPT (Maldonado-Mendoza et al. 2001).

The fact that the family of mycorrhiza-induced Pi transporters includes members from the legumes, solanaceae and grasses, and that down-regulation of Pi transporters in the root epidermis occurs in all these groups, is consistent with the finding that AM symbiosis is a very ancient interaction predating the divergence between mono and dicotyledonous plants (Wolfe et al. 1989; Remy et al. 1994).

Pht2 and other plant transporters

Although the large majority of data regarding Pi transport during AM symbiosis focus on the regulation of the Pht1 type of transporters, it is expected that other components of the Pi transport network play an active part into the reorganization of the Pi fluxes through the plant.

Members of the Pht2 family of Pi transporters have been shown to be involved in the Pi allocation at the whole plant level (Versaw & Harrison 2002), and could therefore be affected during the AM symbiosis. However, all Pht2 members characterized so far (MEDtr;Pht2;1, ARAth;Pht2;1 and SOLtu;Pht2;1) have no, or very moderate responses to Pi level or AM symbiosis (Daram et al. 1999; Zhao et al. 2003; Rausch et al. 2004). Furthermore, the alteration of SOLtuPht2;1 expression did not impact the expression level of a Pht1 Pi transporter expressed in the same cell (SOLtu;Pht1;1: Rausch et al. 2004), raising the question of whether or not these two transporter families are co-regulated at the transcriptional level. Still, a lack of altered transcriptional regulation does not preclude a role for this family in regulating symbiotic Pi allocation. Further assessing the various functions of Pht2 transporters may give rise to new discoveries concerning the reorganization of Pi fluxes within a plant that occurs in response to AM symbiosis. In Arabidopsis, the overexpression of MicroRNA399b down-regulates Pht2;1, and has a differential effect on Pht3;2 and Pht3;3 that is dependant on Pi status (Aung et al. 2006). It will be interesting to see if the MicroRNA399 family plays a role in the regulation of transporters and Pi-signaling pathways in the AM symbiosis.

There is virtually no information regarding the possible participation of the Pht3 type of transporters, or the Pi antiporters/translocators in the reorganization of the Pi transport in response to AM symbiosis. Considering the known effects of AM fungi on plastid and mitochondrial biosynthetic pathways, these transporters are likely involved in channeling Pi transport during AM symbiosis. For example, stimulation of photosynthesis (probably to compensate for the allocation of carbon to the fungus), or carotenoid biosynthesis is observed in many colonized plants, in the shoots and roots, respectively (Jakobsen 1995; Fester et al. 2005). Numerous studies also describe the direct effect of Pi level fluctuations on biosynthetic pathway regulation (Plaxton 2004). The study of Pi translocators is a relatively new area of research, and the involvement of these proteins in the AM association has not yet been investigated. Consequently, the question of their involvement in the AM symbiosis is open, but they are strong candidates for a role in Pi reallocation during this symbiosis.


Regulation of colonization of the roots by Pi status

The negative effects of high soil Pi levels on colonization of roots by AM fungi have been described in multiple experiments (Menge et al. 1978; Jasper, Robson & Abbott 1979). A detailed analysis in potato with a split root experiment (Rausch et al. 2001) demonstrated that when high Pi (1 mM) is applied to a part of the root system, colonization in the other part of the root system is altered, and fungal structures show a predominance of internal hyphae and a reduced number of arbuscules. In these experiments, the expression level of SOLtu;Pht1:3 (mycorrhiza up-regulated), as well as the other root Pi transporters (i.e. SOLtu;Pht1;1 and 2) was diminished, which suggests that another Pi transport system must be operating under high Pi conditions.

To explain the decreased root colonization rates at high soil Pi levels, it has been hypothesized that plants supplied with a non-limiting Pi level will not deliver carbon to the fungus and will instead continue to acquire Pi via transport systems in the root epidermis. Olsson et al. (2002) verified this hypothesis by taking advantage of an in vitro experimental design that allows the growth of external hyphae in a compartment separate from that containing the roots (with transgenic carrot root culture and G. intraradices; Bago et al. 1998;Bago, Pfeffer & Shachar-Hill 2000). They added 13C-labelled glucose to the colonized root compartment and followed the incorporation of labelled carbon into fungal fatty acids. In this way, they demonstrated an inverse correlation between the root Pi content and the accumulation of fatty acids in the fungus. Recent experiments showed that although the inhibitory effect of high Pi level on C transfer is true on the long term, high Pi has a stimulatory effect on the short term (Olsson, Hansson & Burleigh 2006). This could indicate that roots maintain C transfer to the fungus until they reach a satisfactory intracellular Pi status and do not directly respond to the external Pi level.

Altering Pi transfer between symbionts

Whereas initial investigations analysed the interdependency of the AM symbiosis and plant Pi status, recent experiments focused on altering the expression of the mycorrhiza-induced Pi transporters (Nagy et al. 2005; Maeda et al. 2006). The first mycorrhiza-inducible Pi transporter mutant described was a tomato mutant with a transposon insertion in the LYCes;Pht1;4 gene (Nagy et al. 2005). The mutant displayed no detectable phenotype associated with this insertion including no alteration of fungal Pi transfer to the plant. Expression of the other known Pht1 tomato transporters was also unchanged in the mutant, including the other mycorrhiza-inducible and mycorrhiza-specific Pi transporters. Partial redundancy of the three mycorrhiza-inducible Pi transporters, LYCes;Pht1;3, 4 and 5, may explain the lack of phenotype. Consequently, it may be necessary to create double or triple Pi transporter mutants to observe a phenotype.

Using RNAi technology in transgenic roots of L. japonicus (Kumagai & Kouchi 2003), Maeda et al. (2006) were able to induce a significant, although not complete (15 % of the vector control), knockdown of the mycorrhiza-up-regulated LOTja;Pht1;3 gene. This construct did not affect the expression of LOTja;Pht1;1 or 2, root Pi transporters. LOTja;Pht1;3 mutants still showed an increase in shoot growth in the symbiosis, but to a lower extent than the corresponding control plants. Although no effect on root Pi content was observed when the pots were fertilized with 33P, a moderate but significant reduction in shoot P content was measured for one of the two RNAi constructs. In addition, 33P was observed to accumulate in the root area containing arbuscules, consistent with impaired Pi transfer at the arbuscule interface. The RNAi lines displayed an increased number of idioblasts and epidermal entry points and a modest reduction in the number of arbuscules. In addition, when the plants were co-inoculated with AM fungi and rhizobium, the RNAi lines displayed an increased occurrence of necrotic nodules. Based on these observations, the authors suggested that in the absence of Pi acquisition from the fungus, the plant may activate defense responses. Another explanation could derive from the widespread observation that nutrient transport stimulation by mycorrhiza positively affects nodulation (He, Critchley & Bledsoe 2003). Alternatively, in the absence of Pi delivery, plants may not transfer any carbon to the root colonized by the fungus, thereby preventing the development of the hyphae and also accelerating the senescence of nodules. Further experiments will be necessary to fully understand the nature of this phenotype.

Preventing fungal growth unless Pi is effectively delivered from the fungus would ensure protection for the plant against unbeneficial symbionts. This control is important because even beneficial associations have a high carbon cost. Colonized plants allocate an estimated 4–20% of their photosynthate to support AM fungi, which translates into roughly 5 billion tons of globally fixed carbon per year (Bago et al. 2000). In a recent review focusing on the link between Pi and C fluxes in the AM symbiosis (Fitter 2006), the author suggested that the Pi delivery into the cortical cells would constitute a non-forgeable signal that would ensure plants of the reciprocal benefits of the symbiosis. According to this model, Pi delivery to the cortical cell would trigger sugar allocation to the colonized root. Thus, a fungus unable to deliver significant levels of Pi would only have access to the typically low levels of sugars available in the root apoplast. Although some components of the Pi-signaling pathways have been identified (Abel, Ticconi & Delatorre 2002; Fujii et al. 2005; Bari et al. 2006; Chiou et al. 2006), the Pi-sensing mechanism is still unknown. Its identification will be important to further understand the interconnection between Pi and sugar fluxes at the cellular level.


Although the role of Pi in AM symbiosis has been a primary focus of the field, many experiments suggest that N could also play a key role in this interaction. Significant levels of N can be transferred from an AM fungus to a root, although it may not cause a change in the total N content of the plant (Ames et al. 1983; He et al. 2003; Govindarajulu et al. 2005). Sources of N available to AM fungi in the environment are diverse. It can be taken up as ammonium (NH4+: Johansen, Jakobsen & Jensen 1992; Frey & Schuepp 1993; Johansen, Finlay & Olsson 1996); nitrate (NO3-: Bago et al. 1996; Johansen et al. 1996) and amino acids (Hawkins, Johansen & George 2000). AM fungi can also accelerate the decomposition of organic matter, thereby increasing nitrate availability and uptake (Hodge, Campbell & Fitter 2001). Regarding NH4+transport in particular, high-affinity ammonium transporters have been identified in extra- and intraradical hyphae of G. intraradices (Govindarajulu et al. 2005; Lopez-Pedrosa et al. 2006), but in the absence of precise identification of their location, their role in the fungal ammonium transport is not yet clear (Chalot, Blaudez & Brun 2006).

Recent data combining metabolic profiling and enzyme transcript level studies suggest that once N has been transferred to the fungal cytoplasm, it is assimilated into arginine, transferred via the vacuole into the intraradical hyphae, and released to the plant as NH4+ (Govindarajulu et al. 2005). The importance of arginine in N transfer is reflected by the fact that this amino acid represents at least 70–90% of the total free amino acids in the extraradical hyphae (Johansen et al. 1996; Govindarajulu et al. 2005; Jin et al. 2005). It is not yet known if arbuscules are involved in symbiotic N transfer, or how the plant takes up ammonium released by the fungus. Plant mycorrhiza-induced nitrate and ammonium transporters have been identified (Frenzel et al. 2005; Guimil et al.2005; Hohnjec et al. 2005), and it will be interesting to identify their precise expression patterns to explore possible parallels with mycorrhiza-induced Pi transporters.

The impact of the plant N status on AM fungal N uptake and growth has been studied in only a few systems, and it is not yet possible to build a consensus from these data (Chambers, Smith & Smith 1980; Hawkins et al. 2000; Mader et al. 2000; Govindarajulu et al. 2005). However, it was shown recently that, similar to the Pi effect, high external N level around the mycorrhizal roots can reduce the carbon allocation to the fungus (Olsson, Burleigh & van Aarle 2005). The fact that N sources have distinct effects may be problematic when comparing the results from different experiments. In particular, NH4+ and NO3- are not interchangeable. For example, NH4+ will be preferentially taken up by the plant (compared with NO3-), but it has negative effects when supplied as sole N source (von Wiren et al. 2000). As a consequence, experimental result may vary depending on the N source supplied to the plant or the fungus.

Some aspects of N and Pi transfer in the symbiosis are very similar. Following uptake in the extraradical hyphae, they are translocated to the intraradical hyphae via the vacuole in a storage form (poly-P or arginine), and the ionic forms (Pi or NH4+) are released to the plant apoplast, and subsequently taken up by mycorrhiza-up-regulated transporters. Experimental data suggest that N transfer through the fungus is closely interrelated to Pi. In a study of an ectomycorrhizal fungus, nuclear magnetic resonance (NMR) data indicated that arginine was bound to another compound, suspected to be poly-P (Martin 1985). The accumulation of poly-P and arginine is also correlated in many non-AM fungi: a ratio of 1/1 for these two compounds has been measured in the vacuole (Davis 1986). The potential link between N and P is also supported by the observation that the up-regulation of the fungal Pi-transporter GiPT is dependent on the presence of N (Olsson et al. 2005). It is not yet clear to what extent P and N transfer to the plant are interrelated during the AM symbiosis, but future studies, taking advantage of plant mutants impaired in nutrient transport or assimilation, may provide an answer to this question.


In recent years, there has been a significant increase in our understanding of the physiology and the molecular mechanisms of Pi transport in the AM symbiosis. Development of the symbiosis is accompanied by alterations in the expression patterns of root Pi transporters, and Pi flow in the mycorrhizal root can include Pi uptake through the root epidermis and through the symbiotic pathways. In most mycorrhizal associations, some Pi is delivered via the fungus, but in some cases, the plant appears to depend completely on Pi delivered by the symbiosis. Further research is needed to determine whether carbon allocation to the fungus is linked to the symbiotic delivery of Pi and to fully understand the basis of a successful AM symbiosis.


Financial support for this work was provided by the U.S. National Science Foundation (DBI-0421677).