Could the urea cycle be translocating nitrogen in the arbuscular mycorrhizal symbiosis?

Arbuscular mycorrhizal (AM) fungi take up nitrogen from the soil, translocate it and transfer it to the host plant (Jennings, 1995; Smith & Read, 1997), often in significant quantities. Depending on N availability and mobility this can represent an important benefit to the plant, and thus the mechanisms involved are a matter of considerable interest (Ames et al., 1983; Martin & Botton, 1993; Johansen et al., 1996; Smith & Read, 1997; Botton & Chalot, 1999). We know the forms of N taken up by different types of mycorrhizal fungi, and something of the fungal enzymes involved in assimilation, but almost nothing of the translocation and transfer to the roots of host plants. As a result there are few integrated models (Martin et al., 1986) of the movement of N through mycorrhizas – a situation that contrasts with a well developed picture of the flow of phosphorus (Ashford et al., 1994; Smith & Read, 1997; Rasmussen et al., 2000) and the emerging models for the flow of carbon (Hampp & Schaffer, 1999; Bago et al., 2000). Here we propose a simple model for N flow from soil to host via the fungus that involves the urea cycle and the translocation of arginine from extraradical to intraradical mycelium and then to the host plant. The mechanisms involved in N movement in mycorrhizas must transfer N efficiently from heterotroph to autotroph with minimal net transfer of C. For AM fungi in particular, the possible transport mechanisms are further constrained by the necessity for vectorial translocation along coenocytic (nonseptated) hyphae. The model postulated here could meet these demands and is consistent with previous observations of N metabolism in the AM symbiosis (Fig. 1). A proposed model by which the urea cycle together with arginine translocation could act to move nitrogen in the arbuscular mycorrhizal symbiosis. The flow of N is indicated in red, of phosphorus in blue, and of carbon in black; membrane transporters/channels and enzymes are shown in green. Individual steps in the figure labeled a–g refer to the same steps discussed in the text. N acquired from the soil (a) is metabolized to Gln in the extraradical hyphae, then to Arg by the anabolic arm of the urea cycle (b). The Arg is imported into vacuoles by means of a specific Arg tonoplast carrier (c). The cationic Arg binds to polyanionic polyphosphates within fungal vacuoles (d), and these move to the intraradical fungal mycelium by cytoplasmic streaming or a peristaltic tubular vacuolar system. Once within the root, Arg and P are released into the fungal cytoplasm (e). The N is further metabolized through the catabolic arm of the urea cycle producing urea and Orn, and finally ammonium which is then transferred to the host root (f). Resulting C skeletons in the intraradical fungus are re-incorporated into fungal C pools (g). Both nitrate (Tobar et al., 1994; Bago et al., 1996) and ammonium (Ames et al., 1983; Johansen et al., 1992, 1996; Frey & Schüepp, 1993) can be taken up and used by AM fungi. There is no direct evidence for mobilization and uptake of organic N (Smith & Read, 1997; Hodge et al., 2000), though dilution of exogenously supplied inorganic 15N label has been interpreted as suggesting it (Ames et al., 1984; Cliquet et al., 1997). Accordingly only NH4+ and NO3− are shown as being taken up in step (a) of Fig. 1. Nitrate reductase (NR) activity has been found in germinating AM spores (Ho & Trappe, 1975) and in mycorrhizal root extracts (Subramanian & Charest, 1998), and high levels of a transcript with close homology to fungal NR genes were found in arbuscules of the AM fungus Glomus intraradices (Kaldorf et al., 1998). Several lines of evidence (e.g. the inhibition of 15N labelling from ammonium by albizzin, an inhibitor of glutamate synthase) indicate that the glutamine synthetase/glutamate synthase (GS/GOGAT) cycle is responsble for ammonium assimilation in AM extraradical hyphae (Johansen et al., 1996). However, the involvement of glutamate dehydrogenase (GDH) has not been experimentally excluded. Indeed the primary amination step in filamentous fungi is often catalysed by NADPH-dependent GDH, with synthesis of glutamate (Glu) (Pateman & Kinghorn, 1975), although, depending on the plant–fungal system considered, GS/GOGAT may also play a role (Smith & Read, 1997; Botton & Chalot, 1999). The model, therefore, shows nitrate and ammonium being assimilated into glutamine (Gln), without specifying enzymes other than nitrate reductase. There is no direct evidence as to the chemical form in which N is transferred along AM fungal hyphae to the plant, although the finding of nitrate reductase (NR) activity in arbuscules led to the suggestion that at least some of the N is transferred as nitrate (no assay for NR activity was reported in this study for extraradical hyphae; Kaldorf et al., 1998). We think that amino acids, and specifically Arg, are more likely to be transferred because of the active assimilation of ammonium and nitrate into amino acids by AM roots (Johansen et al., 1996) and the observation of Arg at substantial levels in AM fungi (Bago et al., 1999). Thus, for example, HPLC analyses showed that Arg is the dominant free amino acid in external hyphae of the AM fungus G.claroideum (Johansen et al., 1996). The usual pathway of Arg synthesis appears to be active in G. intraradices (Bago et al., 1999), with synthesis of ornithine (Orn) from Glu, and of carbamoyl P with fixation of CO2; and conversion of these two compounds to Arg (Pateman & Kinghorn, 1975; Davis, 1986; Jennings, 1995). Also in favour of Arg as a candidate for translocation is the requirement for a substantial flow of cations from soil to AM roots to accompany the polyanionic polyphosphates (polyP) translocated by AM fungi. It is tempting to suggest that Arg acts in part as this counterion in the AM symbiosis and that anionic polyP is a counterion for the N, as has been suggested for other mycorrhizal symbioses (Smith & Read, 1997; Bücking & Heyser, 1999). This should not be taken to mean that the model requires a tight coupling or exact stoichiometry, as the amount of N transferred compared with P is very variable and other counterions are probably involved to different extents at different times. Finally, variable amounts of protein have been reported to be associated with polyP in ectomycorrhizal species (Kottke et al., 1995) especially under conditions of high external N (Wallenda & Kottke, 1998), and this may play a role in storage and/or detoxification of excess N. No clear role for this polyP-associated protein in translocation has yet emerged, nor has such a protein been reported in AM fungi, and it is accordingly omitted from the present model. The fungal vacuole contains most of the amino acids that are basic, while acidic amino acids are predominantly cytoplasmic (Pateman & Kinghorn, 1975; Jennings, 1995). For instance, 90% of Orn, Lys, His and Arg is located in the fungal vacuole; only 10% is cytoplasmic (Messenguy et al., 1980). In the case of Arg and Orn, concentrations in the vacuole reach at least 50 times those in the cytoplasm, though it has been reported that a 1000-fold concentration difference is possible (Jennings, 1995). A fungal tonoplast carrier for Arg has been identified (Paek & Weiss, 1989). The vacuolar accumulation of Arg in fungi has generally been seen as a storage mechanism and if the external medium is depleted of Arg the vacuolar fraction acts as an N reserve (Davis, 1986). The proposed translocation of vacuolar Arg may be in addition to a storage role. There are other considerations that make the vacuole an attractive candidate for N translocation. First, the other main translocation flows in AM fungi, P from extraradical to intraradical hyphae and C in the opposite direction, appear to be mediated by the movement of membrane-bound or at least aggregate bodies. Thus, polyPs in AM fungi are mainly large (Solaiman et al., 1999) and found in vacuoles (Rasmussen et al., 2000); and triacylglycerides in lipid bodies seem to be the main form of C moving along hyphae (Bago et al., 1999; Pfeffer et al., 1999; B. Bago et al., unpublished). Indeed in a coenocytic network of fine hyphae that extends many centimeters from the root, simple diffusion of soluble cytosolic metabolites is probably insufficient for efficient bulk transport (even with cytoplasmic streaming). In agreement with this, labelling studies show that trehalose in the intraradical and extraradical mycelia have different metabolic origins and there appears to be little or no transfer of this soluble (presumably cytosolic) carbohydrate by translocation (Pfeffer et al., 1999; Bago et al., 2000). Also, there is good evidence that N, and particularly Arg, is associated with vacuolar polyP in fungi (Dürr et al., 1979; Jennings, 1995), including ectomycorrhizal fungi (Martin et al., 1985; Bücking et al., 1998). A related possibility for compartmentation arises from the finding that long distance translocation of solutes in the mycelium of several filamentous fungi occurs by means of a tubular vacuolar system (Ashford et al., 1994, Smith & Read, 1997). Such a system interconnects vacuoles and other cytoplasmic cisternae, moves by peristaltic pulses independent of the cytoplasmic streaming, and seems to transfer material between different hyphal compartments (Shepherd et al., 1993a,b). Therefore, Fig. 1 shows vacuolar compartmentation without specifying whether vacuolar N moves by translocation of the vacuoles themselves or via a tubular vacuolar network. If N is translocated in vacuoles in association with polyP, then the breakdown of polyP in the intraradical hyphae, before release of Pi to the host, would tend simultaneously to release N to the fungal cytoplasm in the intraradical locations most suitable for transfer. Accordingly, Fig. 1 shows release of N from vacuoles associated with the breakdown of polyP. If N is transferred from the fungus to the host in an organic form such as amino acids, substantial amounts of C must move with it. Since the question of whether meaningful transfer of C occurs from fungus to host is not settled (Watkins et al., 1996; Graves et al., 1997; Fitter et al., 1998), it may be that organic N is in fact transferred. Indeed the transfer of Gln has been postulated in the ectomycorrhizal symbiosis (Martin et al., 1986; Jennings, 1995; Botton & Chalot, 1999). If organic N is transferred, a molecule with high N : C ratio, such as Gln, or still better Arg, which would be more efficient in this respect, might be involved. An amino acid permease has recently been identified in an ectomycorrhizal fungus (Nehls et al., 1999) and this or other such transporters might mediate the efflux of organic N at the AM symbiotic interface. However, recent findings indicate that fungal C does not move beyond the mycorrhizal root into host shoots (Fitter et al., 1998) (although this may not hold for the special case of achlorophyllous phycobionts). It follows that any C transferred as organic N would need to be returned very efficiently to the fungus or sequestered within the root without entering other host metabolic pools. We believe that the transfer of inorganic N, which would require the breakdown of organic N by the fungus before transfer, would be a less elaborate mechanism and would also be more metabolically efficient for both host and fungus. The usual Arg catabolic pathway in fungi involves, successively, the hydrolysis of Arg to urea and Orn (via arginase), the breakdown of urea to ammonia and CO2 (via urease) and the conversion of Orn to Glu-γ-semialdehyde (via ornithine-oxoacid transaminase) and then to Glu (Pateman & Kinghorn, 1975; Davis, 1986; Jennings, 1995). Arginase genes have been identified in yeast, Neurospora and Aspergillus (Sumrada & Cooper, 1984; van Huffel et al., 1994; Pan et al., 1995; Marathe et al., 1998). Urease is also found in fungi (Yu et al., 1997; Cox et al., 2000), and the observation that N supplied as 15N-urea to the extraradical mycelium results in labelling of host and fungal amino acids indicates the existence of urease activity in G. intraradices (Shachar-Hill et al., 1997; D. B. Rolin et al., unpublished). Moreover, Orn has been detected as a significant amino acid pool of Glomus mycorrhizal roots (Johansen et al., 1996). These findings are all compatible with the idea that the catabolic arm of the urea cycle, together with urease, could operate in the intraradical mycelium to release ammonium within the root. The presence of ammonium transport proteins in the symbiosome membrane of an N-fixing symbiosis (Tyerman et al., 1995; Kaiser et al., 1998) suggests that in that situation at least, N transfer is in the form of NH4+. The breakdown of Arg (or indeed of Gln via GDH or GOGAT, for example) releasing ammonium could, therefore, allow transfer of N while allowing the fungus to retain the fixed C. The C skeletons released in the breakdown of organic N would probably be in the form of organic acids or amino acids of lower N content that could re-enter fungal metabolism in a number of ways. Pathways include entry into the TCA cycle and thence to respiration, and/or incorporation into lipids or proteins, and/or return as organic acids to the extraradical mycelium. However, there appears to be little flux through gluconeogenesis in the intraradical mycelium (Pfeffer et al., 1999; Bago et al., 1999). No matter the route for re-use of C, the fungus would retain the benefit of this fixed C and the model does not specify the fate of the organic C released from Arg breakdown. The model proposed here is speculative, but we believe it is plausible in the light of what is known of N movement in the AM symbiosis and related systems. The value of any such model is in making testable predictions and highlighting areas of ignorance. This model requires the activity of various enzymes and transporters at specific places in the arbuscular mycorrhiza, in particular, finding an arginine transporter in the vacuolar membrane as well as arginase and urease within the intraradical mycelium would provide support for the hypothesis. Apart from enzyme and transport assays, there are a number of pulse-chase experiments that could test the model, particularly ones to assess the rates of turnover of C and N in different positions of the Arg molecule. Thus, if Arg were used by the fungus only as a storage molecule it would not turn over rapidly. Animals use the urea cycle to translocate nitrogen from source to sink without losing excessive quantities of carbon. Might it be that a similar mechanism, with a similar overall purpose, operates in AM fungi using the same metabolic steps? The evidence is far from complete, but we hope that this proposal will inspire further studies. We thank C. Azcón-Aguilar, A. Fitter and F. Martin for critically reading the manuscript and for their expert opinions on the subject of this article.