Author for correspondence: Carroll P. Vance Tel: +1 612 6255715 Fax: +1 651 6495058 Email: email@example.com
I. Introduction 424
II. The phosphorus conundrum 424
III. Adaptations to low P 424
IV. Uptake of P 424
V. P deficiency alters root development and function 426
VI. P deficiency modifies carbon metabolism 431
VII. Acid phosphatase 436
VIII. Genetic regulation of P responsive genes 437
IX. Improving P acquisition 439
X. Synopsis 440
Phosphorus (P) is limiting for crop yield on > 30% of the world's arable land and, by some estimates, world resources of inexpensive P may be depleted by 2050. Improvement of P acquisition and use by plants is critical for economic, humanitarian and environmental reasons. Plants have evolved a diverse array of strategies to obtain adequate P under limiting conditions, including modifications to root architecture, carbon metabolism and membrane structure, exudation of low molecular weight organic acids, protons and enzymes, and enhanced expression of the numerous genes involved in low-P adaptation. These adaptations may be less pronounced in mycorrhizal-associated plants. The formation of cluster roots under P-stress by the nonmycorrhizal species white lupin (Lupinus albus), and the accompanying biochemical changes exemplify many of the plant adaptations that enhance P acquisition and use. Physiological, biochemical, and molecular studies of white lupin and other species response to P-deficiency have identified targets that may be useful for plant improvement. Genomic approaches involving identification of expressed sequence tags (ESTs) found under low-P stress may also yield target sites for plant improvement. Interdisciplinary studies uniting plant breeding, biochemistry, soil science, and genetics under the large umbrella of genomics are prerequisite for rapid progress in improving nutrient acquisition and use in plants.
Phosphorus (P) is one of 17 essential elements required for plant growth (Bieleski, 1973; Raghothama, 1999). The P concentration in plants ranges from 0.05 to 0.50% dry weight. This element plays a role in an array of processes, including energy generation, nucleic acid synthesis, photosynthesis, glycolysis, respiration, membrane synthesis and stability, enzyme activation/inactivation, redox reactions, signaling, carbohydrate metabolism, and nitrogen (N) fixation. The concentration gradient from the soil solution P to plant cells exceeds 2000-fold, with an average inorganic phosphate (Pi) concentration of 1 µm in the soil solution (Bieleski, 1973; Schachtman et al., 1998). This concentration is well below the Km for plant uptake. Thus, although bound P is quite abundant in many soils, it is largely unavailable for uptake. As such, P is frequently the most limiting element for plant growth and development. Crop yield on 30–40% of the world's arable land is limited by P availability (Runge-Metzger, 1995; von Uexküll & Mutert, 1995). Phosphorus is unavailable because it rapidly forms insoluble complexes with cations, particularly aluminum and iron under acid conditions. The acid-weathered soils of the tropics and subtropics are particularly prone to P deficiency (Sanchez & Uehara, 1980). Application of P-containing fertilizers is usually the recommended treatment for enhancing soil P availability and stimulating crop yields.
II. The phosphorus conundrum
Application of P fertilizer, however, is problematic for both the intensive and extensive agriculture of the developed and developing worlds, respectively. In intensive agriculture, a maize crop yield of 6–9 t ha−1 requires crop uptake of 30–50 kg P ha−1 (Ellington, 1999; Vance, 2001; Johnston, 2002) with about two-thirds of that removed in the harvested portion of the crop. Small grains yielding 3 t ha−1 take up 15–22 kg P ha−1, again with a 70% removal rate. Soybean takes up 20–25 kg P ha−1, with 80–100% removed in the harvested portion (Johnston, 2002). However, even under adequate P fertilization, only 20% or less of that applied is removed in the first year's growth because of retention by the soil (Russell, 1973). This results in P loading of prime agricultural land. Run-off from P-loaded soils is a primary factor in eutrophication and hypoxia of lakes and marine estuaries of the developed world (Runge-Metzger, 1995; Bumb & Baanante, 1996). Another reason for alarm is that by some estimates inexpensive rock phosphate reserves could be depleted in as little as 60–80 yr (CAST, 1988; Vance, 2001). Phosphorus fertilizer use increased four- to five-fold between 1960 and 2000 and is projected to increase further by 20 Tg yr−1 by 2030 (Table 1). As noted by Abelson (1999), a potential phosphate crisis looms for agriculture in the twenty-first century. An even greater concern than the excessive use of P fertilizers by intensive agriculture, is the lack of available P fertilizers for extensive agriculture in the tropics and subtropics where the majority of Earth's people live. Lack of fertilizer infrastructure, money for purchase, and transportation make P fertilization unattainable for these areas. Sustainable management of P in agriculture requires that plant biologists discover mechanisms that enhance P acquisition and exploit these adaptations to make plants more efficient at acquiring P, develop P-efficient germplasm, and advance crop management schemes that increase soil P availability.
Table 1. Agriculture production and resource use in the recent past to the near future
Data adapted from Vance (2001). Mt, metric tons; Tg, 1012 g or million metric tons.
Phosphorus is taken up by plants in the orthophosphate (Pi) forms H2PO4− and HPO42–, which occur in soil solutions at very low concentrations (0.1–10 µm; Hinsinger, 2001). A pH optimum for Pi uptake of 4.5–5.0 indicates preferential plant uptake of Hi2PO4− over HPO42– (Raghothama, 1999). Although total soil P content typically varies from 500 to 2000 p.p.m., total bioavailable P, as measured by soil extractants may be only a few p.p.m. (Sanyal & DeDatta, 1991). Up to half the soil P can be organic, derived from plant residues and soil organisms. Organic P must be mineralized before it can be taken up by plants (Horst et al., 2001).
The reactions controlling the amounts of Pi in solution include dissolution–precipitation of P-bearing minerals, adsorption–desorption of phosphate on soil surfaces, and the hydrolysis of organic matter (Matar et al., 1992; Comerford, 1998; Hinsinger, 2001). It can be difficult to distinguish adsorbed and precipitated forms of P, but retention or ‘fixation’ of P by soil components is greatest in the presence of Fe- and Al-hydroxylated surfaces (from Fe and Al oxides and clay minerals) and, at higher pH, calcium carbonate (Matar et al., 1992; Comerford, 1998). Along with the types and amounts of clay and metal oxides, P availability is also controlled by soil solution pH, ionic strength, concentrations of P and metals (Fe, Al and Ca) and the presence of competing anions, including organic acids (Sanyal & DeDatta, 1991; Hinsinger, 2001). Thus, plant roots (and microbes) can alter solution Pi availability by acidification of the rhizosphere, exudation of organic acids, and secretion of extracellular phosphatases (Comerford, 1998; Hinsinger, 2001).
Because of its strong reactions with soil components, Pi is principally supplied to plant roots by diffusion rather than mass flow (Comerford, 1998; Hinsinger, 2001). At the root surface, Pi is rapidly taken up, resulting in a Pi depletion shell of 0.2–1.0 mm around the root (Barber et al., 1963; Holford, 1997). Although the soil solution Pi concentration rarely exceeds 2 µm, that in plant cells is much higher, 2–20 mm (Bieleski, 1973; Schachtman et al., 1998). For the plant to surmount this concentration difference as well as the negative membrane potential, active transport across the plasmalemma is required. Physiological research over the last few decades, including both inhibitor and kinetic studies, laid the foundation for the molecular investigation of transport processes (Epstein, 1953; Smith, 2001). The striking reduction in Pi accumulation of tissues treated with inhibitors reflects the energy requirement for uptake (Bieleski, 1973; Raghothama, 1999). Moreover, kinetic analysis of Pi uptake shows that plants have both a low- and high-affinity uptake system (Bieleski, 1973; Smith et al., 2000). The high-affinity system operating at low Pi concentrations has an apparent Km ranging from 3 to 10 µm; the low-affinity system operating at high Pi concentrations has a Km ranging from 50 to 300 µM (Nandi et al., 1987; Furihata et al., 1992). The high-affinity uptake process is induced when Pi is deficient whereas the low-affinity system appears to be constitutive in plants (Raghothama, 1999). Phosphorus moves symplastically from the root surface to the xylem at a rate of about 2 mm h−1 (Bieleski, 1973). Xylem flow then transports Pi to the above-ground organs, where symplastic transport carries it to cells within individual tissues. The movement of Pi from the xylem to the cell cytoplasm and from cytoplasm to vacuole is also against a steep electrochemical potential gradient which requires energized transport (Ullrich & Novacky, 1990). Because most Pi is transported as H2PO4−, cotransport involves a cation. Acidification of the cytoplasm which occurs upon Pi addition to P-deficient cells suggests H+ is the cotransport product for the vast majority of plants (Schachtman et al., 1998). However, authors of a recent study on internodal cells of the giant alga Chara have reported a Na+-coupled Pi uptake system that was induced by Pi deficiency (Reid et al., 2000).
Within recent years, functional complementation of yeast mutants defective in Pi transport has been used to isolate and characterize Pi transporters from a diverse array of plants (Muchhal et al., 1996; Leggewie et al., 1997; Daram et al., 1998; Liu et al., 1998; Liu et al., 1998). Molecular characterization of Pi transporters coupled to the discovery of as many as 16 Pi transport genes within the genome of Arabidopsis confirms that plants have a multiplicity of Pi transporters functional in specific organs and tissues (Raghothama, 1999; Gilroy & Jones, 2000). The deduced amino acid sequence of plant Pi transporters indicate they have an Mr of ≅ 58 kDa and are comprised of 525–550 amino acid residues. All characterized Pi transporters have 12 transmembrane domains that occur as two groups (6 + 6) connected by a hydrophobic domain of 60 amino acids (Schachtman et al., 1998; Smith et al., 2000). Computer modeling predicts that both N- and C-terminal domains of high-affinity Pi transporters are found on the inner membrane surface (Fig. 1), while in the low-affinity transporters these domains are on the exterior surface of the membrane. Initial Pi uptake kinetic studies with a yeast mutant complemented with a Pi transporter disappointingly showed an apparent uptake Km to be greater than 100 µm, suggesting that it was not a high-affinity transporter (Leggewie et al., 1997). This reduced affinity may also reflect that expression was in a heterologous system. However, more recent studies have yielded a Km of 31 µm for the tomato LePT1 Pi transporter (Daram et al., 1998). An even higher affinity, apparent Km of 3.1 µm, was shown when the Arabidopsis PHT1 Pi transporter gene was expressed in cultured tobacco cells (Mitsukawa et al., 1997). In situ hybridization and immunolocalization of high-affinity Pi transporters transcripts and protein in root epidermal and root hair cells (Daram et al., 1998; Liu et al., 1998; Chiou et al., 2001) suggests their importance for Pi uptake from soil solution. Localization of plant uptake systems in the youngest regions of roots and in the root hairs is most effective for recovery of nutrients from undepleted soil (Smith et al., 2001). As a depletion zone develops, effective uptake of Pi behind the root hair zone requires either mycorrhizal uptake (through hyphae that bridge the depletion zone) or plant adaptations, such as cluster roots, that solubilize and extract additional Pi from localized soil patches.
When placed in +P nutrient solution, uptake of Pi on a gram fresh weight basis by P-deficient cluster roots of white lupin is much greater than that of P-sufficient plants (Keerthisinghe et al., 1998; Neumann et al., 1999). Moreover, the apparent Km for Pi uptake of cluster roots is 8.6 µm compared with a Km of 30.7 µm for P-sufficient controls (Neumann et al., 1999). These results suggest that a high-affinity Pi uptake system is induced in cluster roots of P-deficient plants. Liu et al. (2001) have recently characterized a high-affinity type Pi uptake gene (LaPT1) from cluster roots of white lupin (Lupinus albus) that shows highly intensified expression in P-deficient plants. Similar to other plant Pi transporters, the white lupin cluster root Pi transporter has a 1620 bp open reading frame that encodes a protein of 540 amino acids with a Mr of 59 kDa organized into 12 transmembrane domains (Fig. 1). The deduced amino acid sequence of LaPT1 is 85% similar to previously reported high-affinity type Pi transporters but only 75% similar to low affinity type Pi transporters. Transcripts of LaPT1 are highly expressed in Pi-deficient cluster roots, normal roots, and stems with little to no expression in P-sufficient plants (Fig. 2). Thus, the enhanced uptake of Pi displayed by P-deficient cluster roots may be directly related to increased expression of a high-affinity type Pi transporter.
V. P deficiency alters root development and function
Coordinated expression of plant genes within a given environment trigger biochemical events that give rise to plant organs and tissues (Torrey, 1986; Boerjan et al., 1992; Celenza et al., 1995; Lynch, 1995; Scheres & Wolkenfelt, 1998; Gilroy & Jones, 2000). Changes in either gene expression or the environment can alter the developmental fate of any given organ. Understanding the genetic and biochemical mechanisms that regulate the development of plant organs are cardinal questions in plant biology. Studies of Arabidopsis, common bean (Phaseolus vulgaris) and white lupin have proven particularly valuable for revealing genetic and biochemical factors that mediate plant adaptations to P deficiency.
Root architecture (Lynch, 1995; Lynch & Brown, 2001) refers to the complexity of root system spatial configurations that arise in response to soil conditions. It includes root morphology, topology and distribution patterns. Soil P limitation is a primary effector of root architecture (Charlton, 1996; Johnson et al., 1996b; Borch et al., 1999; Williamson et al., 2001) and is known to impact on all aspects of root growth and development. The classic studies of Drew (1975) and Jackson et al. (1990) demonstrated the effect of localized supply of soil P on root proliferation in grass species. Elegant experiments with common bean coupled to simulation modeling have shown genotypic adaptations to P deficiency involve changes in root architecture that facilitate acquisition of P from the topsoil (Ge et al., 2000; Liao et al., 2001; Lynch & Brown, 2001). Adaptations that enhance acquisition of P from topsoil are important because of the relative immobility of P in soil, with the highest concentrations usually found in the topsoil and little movement of P into the lower soil profiles. Lynch and Brown (2001) refer to P-deficiency induced modifications in bean root architecture as adaptations for topsoil foraging. Root characteristics associated with improved topsoil foraging for P are a more horizontal basal-root growth angle, resulting in more shallow roots, increased adventitious root formation, enhanced lateral root formation and increased root hair density and length.
Williamson et al. (2001) have shown that P availability exerts a marked effect on the root system architecture of Arabidopsis. Growth under P-deficient conditions resulted in a redistribution of root growth from the primary root to lateral roots. Reduced primary root elongation under low P conditions was accompanied by increased lateral root density and elongation. Similar to common bean, Arabidopsis root biomass was concentrated near the soil surface, suggesting topsoil foraging. Moreover, Arabidopsis accessions with enhanced P acquisition have the root architecture modifications mentioned previously and greater root penetration capacity (Narang et al., 2000).
The abundant development of lateral roots associated with P-deficiency induced alterations in root architecture is almost invariably accompanied by increased root hair density and length. Root hairs are tubular-shaped cells specialized for nutrient uptake (Gilroy & Jones, 2000). They arise from root epidermal cells known as trichoblasts and undergo tip growth, thereby extending the root surface area in contact with the soil matrix (Ridge, 1995; Peterson & Farquhar, 1996). Root hairs can form as much as 77% of the root surface area of field crops (Parker et al., 2000). For plants lacking mycorrhizae they are the primary site of nutrient uptake (Jungk, 2001; Schmidt, 2001). Root hair formation and growth is regulated largely by the supply of mineral nutrients, particularly NO3− and P (Gilroy & Jones, 2000; Jungk, 2001). In rape, spinach, and tomato, root hair length and number are inversely related to the P concentration in the plant (Jungk, 2001). Similarly, in legumes, P-deficiency results in both increased root hair density and length (Reid, 1981; Jungk et al., 1990). Recent results from our laboratory show that Medicago truncatula responds quickly to P-deficiency with increased numbers and length of root hairs (S. Miller and C. P. Vance unpubl. data). Phosphorus uptake in barley cultivars is closely related to variation in root hair abundance and length (Gahoonia & Nielsen, 1997). In Arabidopsis, root hair density (Ma et al., 2001) and elongation (Bates & Lynch, 2000) are regulated by P availability. Root hair density was fivefold greater in low-P than in high P-media. Ma et al. (2001) found that low P stimulated trichoblast number and the likelihood that trichoblasts would form root hairs. The average length of root hairs on P-deficient Arabidopsis was threefold greater than that on P-sufficient plants. Grierson et al. (2001) report that at least 40 genes in Arabidopsis affect root hair initiation and development. Many of these may be responsive to P-deficiency.
Cluster roots (proteoid roots)
Although much can be garnered from Arabidopsis as a model for analysis of the fundamental underpinnings of root development, alternative models must be invoked for determining how P deficiency affects root growth of crop and native species. Native species of Western Australia and South Africa are adapted to grow on the most heavily leached soils in the world (Pate et al., 2001). Species of these regions have evolved specialized structures, cluster roots, for nutrient acquisition from impoverished soils. Beside mycorrhizal associations, cluster roots are regarded as one of the major adaptations for P acquisition (Skene, 1998; Pate & Watt, 2001). Cluster roots are formed in most members of the Proteaceae and in several other plant species adapted to habitats of extremely low soil fertility, including members of the Betulaceae, Casuarinaceae, Cucurbitaceae, Cyperaceae, Eleagnaceae, Leguminosae, Moraceae, Myricaceae and Restionaceae (Louis et al., 1990; Dinkelaker et al., 1995; Skene, 2000; Adams & Pate, 2002). In native habitats, many plant species that form cluster roots are slow-growing, sclerophyllous shrubs and trees that grow on severely P-deficient soils, such as highly leached sands, sandstones and laterites (Dinkelaker et al., 1995; Pate et al., 2001; Lambers et al., 2002). Plant species that form cluster roots usually do not form mycorrhizal associations (Skene, 1998). The cluster roots of Proteaceae occur in close association with decomposing litter. This is especially true for Banksia, a genus including trees and shrubs that are dominant in the northern sandplains of south-west Western Australia. (Pate & Watt, 2001). Banksia typically forms dense mats of cluster roots beneath the litter layer, exploiting the relatively P-rich A horizon, while vertically descending tap roots provide access to groundwater (Jeschke & Pate, 1995; Pate & Watt, 2001). Growth of these root clusters is seasonal, starting after the onset of winter rain for nutrient uptake from the newly acquired litter during winter and spring (Lamont, 1982; Grierson & Attiwill, 1989; Jeschke & Pate, 1995; Pate & Watt, 2001). During winter, increased concentrations of P and other nutrients were found in xylem sap from cluster roots of Banksia prionotes, compared with the remainder of the lateral root system (Jeschke & Pate, 1995). This period of nutrient uptake is accompanied by nutrient storage in the trunk and leaves (Jeschke & Pate, 1995). It is followed by senescence of root clusters when the soil surface dries out during summer. At this stage, minerals stored in trunks and foliage are released for renewed shoot growth during summer (Jeschke & Pate, 1995).
Adaptation to P deficiency via formation of cluster roots is the result of a highly coordinated modification of root development and biochemistry. Cluster roots can comprise single clusters of very densely packed determinate lateral rootlets formed on a parent axis, as found in such Proteaceae as Hakeae spp., Leucadendron laureolum, Grevillea robusta as well as in the legume white lupin (Gardner et al., 1981; Lamont, 1993; Dinkelaker et al., 1995; Skene et al., 1996) (Fig. 3). Banksia species, however, form more complex compound clusters (Dinkelaker et al., 1995; Pate & Watt, 2001; Lambers et al., 2002). Several features of cluster root development and morphology are distinguished from that of typical dicot lateral roots. First, lateral roots are randomly initiated from the pericycle of primary roots near the zone of metaxylem differentiation (Torrey, 1986; Charlton, 1996), while cluster roots are initiated in waves (Skene, 2000). Second, lateral roots are initiated singularly opposite a protoxylem point, unlike cluster roots which are in multiples opposite every protoxylem point within the wave of differentiation (Fig. 4). Third, in typical lateral roots, root hair development is highly regulated and occurs from a discrete number of epidermal cells (Ridge, 1995; Malamy & Benfey, 1997; Dolan, 2001), while cluster roots produce a superabundance of root hairs. The accompanying increase in root hair density of clustered rootlets results in an increase of surface area of greater than 100-fold compared with normal roots. Finally, contrasting with the indeterminate growth of lateral roots, cluster root growth is determinate, ceasing shortly after emergence (Fig. 3a). This highly synchronous developmental pattern indicates that cluster root formation is a finely tuned process. Moreover, because root pericycle cells are arrested in the G2 phase of the cell cycle (Skene, 1998, 2000), cluster root initiation must involve concerted release of multiple pericycle cells from the G2 phase in a wave-like pattern along second-order lateral roots.
While mycorrhizal hyphae increase the soil volume that is exploited by roots, cluster roots explore a comparatively small soil volume. The hairy and densely packed lateral rootlets bind tightly to trapped sand and particles of organic matter (Lamont, 1973; Grierson & Attiwill, 1989; Pate & Watt, 2001). The dense aggregates of cluster roots are thought to mobilize sparingly soluble Pi more effectively by concentrating root exudates in localized patches (Gardner et al., 1983; Grierson & Attiwill, 1989). Excretion of organic acids and acid phosphatase from cluster roots has been shown for species of the Proteaceae (Grierson, 1992; Dinkelaker et al., 1995; Lambers et al., 2002). Detailed studies of the carboxylate exudation in B. grandis showed exudation of significant amounts of a range of carboxylates when plants were grown on Fe-phosphate or Al-phosphate (Lambers et al., 2002). Interestingly, different carboxylate patterns were recovered from the rhizospheres in Fe-phosphate and Al-phosphate treatments, indicating that the plants perceived a difference in the chemistry of the rhizosphere environment (Lambers et al., 2002). Development of cluster roots was suppressed when plants were grown with higher P supply (K-phosphate) (Lambers et al., 2002).
Although the development and function of cluster roots in Proteaceae have received growing attention, most of the detailed physiological and biochemical work on the functioning of root clusters has been performed with white lupin. This well-characterized legume has proven an illuminating model system for understanding plant adaptations to low P in species lacking mycorrhizal associations (Avio et al., 1990; Watt & Evans, 1999; Skene, 2000; Neumann & Martinoia, 2002). Biochemical changes that can increase the acquisition of scarce P occur within the cluster root zones of P-stressed white lupin (Neumann & Martinoia, 2002) (Fig. 3b). Copious quantities of organic acids, H+ and acid phosphatase (APase) are exuded to solubilize bound P from inorganic and organic complexes (Marschner et al., 1986; Gerke et al., 1994; Johnson et al., 1994; Dinkelaker et al., 1995; Neumann et al., 1999; Miller et al., 2001). In addition, Pi uptake within cluster root zones is much greater on a gram fresh weight basis than that of normal roots (Keerthisinghe et al., 1998; Neumann et al., 1999). Molecular events that give rise to the metabolic changes which increase P acquisition by cluster roots are mediated in part by enhanced expression and accumulation of transcripts encoding Pi transporters (Liu et al., 2001), exuded acid phosphatase (Miller et al., 2001) and enzymes of carbon metabolism (Uhde-Stone et al., 2003a,b). Recently, we have taken a functional genomics approach to investigating adaptation to P deficiency in white lupin by sequencing some 2000 expressed sequence tags (ESTs) from P-deficiency induced cluster roots (Uhde-Stone et al., 2003a,b). This approach has provided us with access to several white lupin genes important for adaptation and acclimatization to low P conditions (noted in later sections).
Hormones and root architecture
As might be expected, the internal balance of the plant growth regulators auxin, ethylene, and cytokinin is thought to play a role in P-deficiency induced alterations in lateral root development and architecture, root hair formation and cluster root development. Substantial support for the role of auxin in lateral root development is derived from evidence that: (1) exogenous auxin applications stimulate lateral root formation in many species (Torrey, 1986; Blakely et al., 1988; Muday & Haworth, 1994; Charlton, 1996); (2) auxin transport inhibitors block lateral root formation and this block can be alleviated by exogenous auxin (Torrey, 1986; Muday & Haworth, 1994; Reed et al., 1998); (3) Arabidopsis mutants that overproduce auxin have enhanced lateral root formation (Boerjan et al., 1995; Celenza et al., 1995; King et al., 1995; Scheres & Wolkenfelt, 1998), while mutants insensitive to auxin have impaired lateral root development (Celenza et al., 1995; Hobbie & Estelle, 1995; Rogg et al., 2001; Fukaki et al., 2002); and (4) mutants in auxin transport have altered lateral root formation (Ruegger et al., 1997; Casimiro et al., 2001; Marchant et al., 2002). Because cluster root formation during P deficiency reflects a striking change in root development, auxins have been implicated in their formation (Lamont et al., 1984; Dinkelaker et al., 1995). We (Gilbert et al., 2000) and others (Skene & James, 2000) have shown that exogenous addition of auxin to P-sufficient white lupin mimics cluster root formation as seen under P-deficient conditions. Moreover, Gilbert et al. (2000) demonstrated that auxin transport inhibitors added to P-deficient plants dramatically reduced the formation of cluster roots. The data suggest the cluster root response to P deficiency in white lupin is directly under control of auxin transport. Conversely, auxin insensitive mutants of Arabidopsis (aux1, axr1 and axr4) which have reduced lateral root formation, showed typically enhanced production of laterals on low P media (Williamson et al., 2001). The authors interpret the data to mean that auxin is not directly involved in the induction of lateral roots under low P. Unfortunately, auxin transport inhibitors were not added to the P-stressed mutants, neither was auxin measured; thus interpretation of this work is open to question. An auxin transport inhibitor experiment with aux1 under P-stress would be particularly informative since aux1 may act as an auxin transporter (Scheres & Wolkenfelt, 1998).
Although the role of ethylene in lateral root formation is less clear there is convincing evidence that ethylene plays a role in root hair formation and abundance (Kieber, 1997; Michael, 2001). Treatments that inhibit ethylene synthesis inhibit root hair formation and growth, while the stimulation of ethylene synthesis results in an increase in root hair density and length (Tanimoto et al., 1995; Masucci & Schiefelbein, 1996; Schmidt, 2001). Of the more than 40 genes affecting root hair development (Grierson et al., 2001) at least eight act directly or indirectly on ethylene biosynthesis. Mutations in the ethylene signaling pathway involving the constitutive triple response ctr1 gene cause plants to respond as though ethylene is continually present and root hair formation as well as density is much greater than wild type. Ethylene overproduction mutants eto1, 2, 3 have much longer and more root hairs than normal, while the ethylene receptor mutant etr1 has much shorter than normal root hairs. The ethylene-overproducing mutant eto3 exhibits a root hair phenotype reminiscent of P-deficiency induced conditions (Schmidt, 2001). Data from auxin insensitive mutants axr2, iaa7, and aux1 show root hairs that are reduced in numbers or absent, suggesting that ethylene and auxin are both involved in regulating root hair numbers.
Nutrient stress, particularly P and Fe, are known to stimulate ethylene production (Romera & Alcantara, 1994; Borch et al., 1999; Waters & Blevins, 2000; Lynch & Brown, 2001). Schmidt (2001) has proposed that the mechanism of P-deficiency induced changes in root hair density and length involves ethylene through two functionally redundant pathways. Similarly, Cho & Cosgrove (2002) present evidence that root hair initiation and growth are modulated by convergent developmental and environmental pathways, with ethylene/auxin signalling a predominant feature of the environmental stress pathway.
The changes in root development and root hair formation in cluster roots of white lupin are postulated to occur through ethylene signaling (Gilbert et al., 1997; Watt & Evans, 1999). Gilbert et al. (2000) found a significant increase in ethylene synthesis in white lupin cluster roots. Surprisingly, application of ethylene synthesis inhibitors had little to no effect on cluster roots, but root hair number and length were not evaluated. In our EST analysis of genes expressed in P-deficient cluster roots we found several involved in ethylene biosynthesis that were highly redundant, including ACC (1-aminocyclopropane-1-carboxylate) oxidase, methionine synthase and S-adenosylmethionine synthetase (Uhde Stone et al., 2003b), but whether ethylene is a primary factor in cluster root formation and function remains to be established.
The role of cytokinins in root growth and architecture, like that of ethylene, is not resolved. Traditionally, cytokinins are thought to inhibit root growth while stimulating shoot growth (Skoog & Miller, 1957). Moreover, low cytokinin/auxin ratios favor root initiation and development in plant tissue culture experiments (Christianson & Warnick, 1985). Phosphate starvation and N deficiency appear to result in decreased cytokinin content. In addition, exogenously applied cytokinin can counteract lateral root growth stimulation induced by nutrient deficiency (Salama & Wareing, 1979; Horgan & Wareing, 1980; Kuiper et al., 1988) and can repress expression of several Pi-starvation responsive genes in Arabidopsis (Martin et al., 2000). Cytokinin inhibition of root growth may also be coupled to ethylene action (Cary et al., 1995). In recent years, Arabidopsis mutants altered in cytokinin sensitivity (Baskin et al., 1995; Faure et al., 1998; Cary et al., 2001) have been isolated but these mutants have no clear-cut root phenotype. Neumann et al. (2000) have shown that exogenous application of kinetin inhibits cluster root formation in P-deficient white lupin. Also, kinetin content was increased in cluster roots compared with normal roots. While we have not measured cytokinin profiles, it is revealing that an abundant EST from cluster roots showing enhanced expression (three- to five-fold) encoded a deduced protein having high similarity to cytokinin oxidase (E = 10−20). Cytokinin oxidase is the key enzyme implicated in cytokinin degradation (Morris et al., 1999). Enhanced degradation of cytokinins in cluster root formation and/or development might be expected because low cytokinin levels favor root growth, and P deficiency reportedly results in reduced xylem sap cytokinin levels (Wagner & Beck, 1993; Binns, 1994; Martin et al., 2000). Alternatively, in planta regulation of the potentially large quantities of cytokinins that could be released by the mass induction of cluster root meristems may require cytokinin oxidase (Morris et al., 1999; Mok & Mok, 2001).
VI. P deficiency modifies carbon metabolism
Organic acid efflux
Under normal growth and development conditions, plant roots exude a wide variety of organic compounds including: simple sugars, organic acids, amino acids, phenolics, quinones (iso)-flavonoids, growth hormones, proteins, and polysaccharides (Curl & Truelove, 1986; Marschner, 1995). Exudation of organic compounds from roots can alter rhizosphere chemistry, soil microbial populations, competition, and plant growth. Exuded compounds are functionally diverse and can be involved in a wide array of processes ranging from signaling in plant–microbe interactions, to allelopathy and nutrient acquisition (Curl & Truelove, 1986; Marschner et al., 1986; Harrison, 1997).
Under the influence of environmental stress the complement of compounds found in exudates can be significantly altered in either or both quality and quantity (Marschner, 1995; Ryan et al., 2001; Neumann & Martinoia, 2002). For example, when plants are attacked by soil pathogens, the amount and type of phenolics and phytoalexins are altered, depending upon the pathogen. Similarly, rhizobia bacteria and mycorrhizal fungi induce changes in amount and types of compounds released from roots. During nutritional stress (i.e. P limitation), Al toxicity, low Fe availability, and exposure to heavy metals, roots show enhanced synthesis and exudation of several organic acids (anions) (Dinkelaker et al., 1989; Delhaize et al., 1993; Delhaize & Ryan, 1995; Ryan et al., 1995a,b, 1997; Larsen et al., 1998; Neumann et al., 2000). Of the changes in response to plant nutrition, those related to low P and excess Al are among those that have been most thoroughly documented. Convincing evidence now exists for exudation of malate and citrate as a principal mechanism in alleviating the edaphic stress of P-deficiency and Al-toxicity. The release of organic acids allows for the chelation of Al3+, Fe3+ and Ca2+, and subsequent displacement of Pi from bound or precipitated forms (Gerke, 1994; Jones, 1998; Hinsinger, 2001; Ryan et al., 2001), and may also cause organic P to become more susceptible to hydrolysis by acid phosphatases (Gerke, 1994; Braum & Helmke, 1995) (Fig. 3B). In Al-toxic soils root exudation of organic acids protects plants by chelating Al3+ ions in the rhizosphere, thus excluding their entry into the root. Because several recent excellent reviews (Jones, 1998; Lopez-Bucio et al., 2000b; Hinsinger, 2001; Ryan et al., 2001) have critically evaluated organic acid synthesis and exudation from roots, we will focus on the salient features related to the role of organic acids in alleviating P-deficiency in cluster roots of white lupin, including recent evidence for a potential efflux mechanism.
When white lupin is P deficient, cluster roots synthesize and exude striking amounts of malate and citrate (Gardner et al., 1982; Dinkelaker et al., 1989; Johnson et al., 1994; Keerthisinghe et al., 1998; Neumann et al., 1999). Grown under P-deficiency, cluster roots exude 20- to 40-fold more citrate and malate than P-sufficient roots (Fig. 3b). The amount of carbon exuded in these two compounds can range from 10% to greater than 25% of the total plant dry weight. Surprisingly, P-deficient plants do not appear to suffer any loss in either dry matter accumulation or N fixation until the reproductive stage of growth (Dinkelaker et al., 1989; J. Schulze, G. Gilbert, C. Vance, unpubl. data). Although the internal concentrations of organic acids increase in P-deficient roots, it is not proportional to the amount released as exudates. This suggests specific selective synthesis and exudation of malate and citrate in response to P-stress.
Measurements of CO2 fixation and pulse-chase labeling with 14CO2 show the carbon required for enhanced organic acid exudation from cluster roots of P-deficient white lupin is derived from both photosynthetic CO2 fixation and dark CO2 fixation by cluster roots (Johnson et al., 1994, 1996a). Photosynthetic CO2 fixation provides about 65% of the carbon exuded, while cluster root dark CO2 fixation provides about 35%. Rates of dark CO2 fixation in cluster roots are similar to those occurring in legume root nodules (Johnson et al., 1996a). The increased cluster root in vivo CO2 fixation is accompanied by enhanced specific activity of phosphoenolpyruvate carboxylase (PEPC), malate dehydrogenase (MDH), and citrate synthase (CS) (Johnson et al., 1996a,b; Gilbert et al., 1999; Uhde-Stone et al., 2003a).
Enhanced exudation of organic acids may not only occur because of upregulation of selected reactions leading to greater synthesis but also may be due to either reduced degradation or utilization of citrate. The activity of aconitase (AC) is reduced in cluster roots (Neumann & Römheld, 1999; Neumann et al., 1999). Similarly, respiration rates are also reduced in P-deficient cluster roots (Johnson et al., 1994; Neumann et al., 1999). Both phenomena could result in more citrate available for exudation.
Organic acid anion (OA−) exudation from cluster roots of white lupin coincides with a marked acidification of the rhizosphere (Marschner et al., 1987; Dinkelaker et al., 1989; Neumann & Martinoia, 2002). This is consistent with reports of a correlation of OA− exudation and the release of protons from white lupin and chickpea roots (Neumann & Römheld, 1999; Sas et al., 2001). Sas et al. (2001) have shown that extrusion of protons and organic acids in white lupin was highly dependent upon P supply. On an equimolar basis, H+ extrusion in P-deficient plants was two- to three-fold greater than organic acid exudation. Similarly, a recent report by Yan et al. (2002) showed substantially enhanced proton release from cluster roots of P-deficient white lupin. Proton release occurred primarily from young clusters and decreased significantly as cluster roots aged. During prolonged P deficiency in white lupin plants, OA− exudation from cluster roots increased while the release of protons decreased, indicating that different mechanisms are involved in proton release and OA− exudation (Sas et al., 2001). A comparison with tomato (Lycopersicon esculentum) showed an increase of the net release of protons from roots of P-starved tomato while OA− excretion from tomato roots decreased under P deficiency (Neumann & Römheld, 1999). Wheat (Triticum aestivum) displayed a decrease in OA− exudation under P deficiency and no proton extrusion (Neumann & Römheld, 1999).
Proton release into the rhizosphere is a common response of plant cells to counter intracellular acidity (Raven & Smith, 1976). Generally, proton release results from the activity of a plasma membrane H+ ATPase. This enzyme uses ATP to pump protons out of the cell, thereby creating pH and electrical potential differences across the plasmalemma. Yan et al. (2002) reported enhanced catalytic activity of a plasma membrane H+ ATPase of P-deficient cluster roots in white lupin that might be responsible for the increase in H+ extrusion. Characterization of the H+ ATPase from P-deficient cluster roots showed a threefold increase in Vmax and a fourfold increase in H+ ATPase enzyme protein compared with P-sufficient roots. Only the youngest P-deficient cluster roots had significant proton extrusion and H+ ATPase activity. It is noteworthy that Sakano (2001) addresses the potentially problematic generation of protons during OA− synthesis via glycolysis in his revised pH-stat hypothesis of the plant cell. By contrast to the ‘feed-forward’ fashion of the nonplant system, plant glycolysis is regulated by a feedback process. This feedback regulation only permits glycolytic activity when the cytoplasm is alkaline enough to stimulate PEPC, an enzyme with an alkaline pH optimum (Sakano, 2001). As Sakano points out, active H+ extrusion is a possible response of the plant cell to prevent the inhibiting effect of cytosolic acidification.
While neither the specific cell types for nor the mechanism of organic acid exudation have been characterized for cluster roots of white lupin, studies on Al tolerance in wheat (Ryan et al., 1997) and maize (Zea mays) (Kollmeier et al., 2001; Pineros & Kochian, 2001) have identified the root apex as the zone of organic acid exudation. Moreover, these patch-clamp and inhibitor studies indicate that Al may activate a malate/citrate-permeable anion channel to facilitate the exudation of large quantities of organic acids. Zhang et al. (2001) showed that anion channel frequency and rapidity of activation was greater in an Al3+-tolerant wheat line compared with a near-isogenic Al33+-sensitive line. Neumann et al. (1999) showed that exposure of P deficiency induced cluster roots to the anion channel-blockers ethacrynic- and anthracene-9-carboxylic acids inhibited citrate exudation by 40–60%. Moreover, we have recently shown through in situ hybridization (Uhde-Stone et al., 2003a) that PEPC and MDH transcripts are localized to cluster root apices and elongation zones, indicating the enzymes necessary for malate and perhaps citrate synthesis occur in the root tips (Fig. 5). Our data coupled with that of Neumann et al. (1999) suggest that P deficiency induced rapid exudation of organic acids, similar to Al3+-tolerance, may involve selective anion channel proteins at sites of exudation. Since malate and citrate are fully dissociated in the cytosol and because cell membranes are in essence impermeable to ions, it is not surprising that the release of organic acids from root tips in either P deficiency or Al toxicity might involve some type of channel protein.
Although a large number of cation channels and transporters belonging to a multiplicity of gene families have been identified over the past decade (Cao et al., 1995; Maathuis & Sanders, 1999; Barbier-Brygoo et al., 2000; Howitt & Udvardi, 2000; Theodoulou, 2000; Maser et al., 2001), the molecular events involved in organic acid (anion) excretion remain elusive. While anion channels are probably ubiquitous in plants, few have been characterized (Schroeder, 1995; Barbier-Brygoo et al., 2000). Mechanisms of anion transport potentially could include chloride channels (CLC), aquaporins, and multidrug extrusion proteins (MATE). To date, only putative anion channel genes related to the CLC family have been characterized (Hechenberger et al., 1996; Lurin et al., 1996; Geelen et al., 2000). These genes encode messages of approximately 2.4 kb giving rise to proteins of approximately 785 amino acids with Mr of c. 85 kDa. The Arabidopsis CLC-d gene could functionally substitute for a yeast CLC protein involved in iron-limited growth, while the CLC-a gene appeared to be related to nitrate acquisition. Anion channel genes encoding transmembrane proteins involved in organic acid efflux have yet to be characterized.
During our annotation of ESTs from cluster roots we found that several proteins with putative transport-related functions were highly expressed in P-stressed conditions. One of these ESTs was redundant (found 10 times) and showed a ninefold induction in -P cluster roots (Uhde-Stone et al., 2003b). This EST displayed high homology to a putative integral membrane protein in Arabidopsis belonging to the MATE protein family. MATE proteins are a large family of putative antiporters that are thought to be involved in excretion of a variety of drugs and toxins (Brown et al., 1999; Debeaujon et al., 2001; Diener et al., 2001). Around 55 putative members of the MATE family have been predicted in Arabidopsis, two distantly related MATE proteins have been predicted in yeast. To date, only a few MATE proteins have been cloned (Debeujon et al., 2001; Diener et al., 2001; Rogers & Guerinot, 2002). An Arabidopsis MATE has been shown to control response to iron deficiency. It represents either an Fe sensor or transporter of small organic molecules (Rogers & Guerinot, 2002). Analysis of the white lupin P-stress induced MATE revealed a 2.1-kb cDNA encoding a deduced amino acid sequence of 531 amino acids having 10–12 transmembrane helices, depending upon the program used for analysis. These findings make this putative MATE EST an extremely interesting candidate for a role in organic acid excretion. Functional studies in yeast and Xenopus oocytes will allow us to examine the role of the MATE type protein in organic acid excretion.
Alternative glycolytic pathways
Besides increased acquisition of soil P, conservation of internal pools of acquired Pi is considered an important adaptation for growth on low P. As several enzymes of the glycolytic pathway depend on Pi or adenylates as cosubstrates, metabolic processes may be impaired by severe P deprivation. Yet the generation of energy and the production of carbon skeletons continues during conditions of P limitation, as evident in the excretion of significant amounts of organic acids. In response to limited P, many plants show remarkable flexibility in adjusting metabolic rates and utilizing alternative metabolic pathways. Alternative glycolytic reactions can bypass Pi- or ATP-requiring steps of glycolysis under environmental stress conditions such as Pi starvation or anoxia (Duff et al., 1989b; Mertens, 1991; Theodorou et al., 1992; Theodorou & Plaxton, 1996). While ATP and ADP-levels have been reported to decline under P deficiency (Ashihara et al., 1988; Duff et al., 1989b), pyrophosphate (PPi) concentrations appear to be buffered during Pi stress (Duff et al., 1989b; Dancer et al., 1990). Despite an earlier assumption of cellular bioenergetics that PPi anhydride bonds were not utilized in cellular metabolism, studies have shown that under certain environmental stresses PPi can serve as an energy donor in the plant cytosol (Duff et al., 1989b; Plaxton, 1996; Plaxton & Carswell, 1999). A well-documented alternative glycolytic pathway is catalysed by a PPi-dependent phosphofructokinase (PFP) that, under P deficiency, can bypass the ATP-dependent phosphofructokinase (PFK), generating fructose 1,6-biphosphate (Theodorou et al., 1992; Theodorou & Plaxton, 1996; Plaxton & Carswell, 1999). Other processes that might use PPi are the cleavage of sucrose by a PPi-dependent sucrose synthase pathway and the active transport of protons into the vacuole by a PPi-dependent H+ pump in the tonoplast (Plaxton & Carswell, 1999). PPi has been assumed to be a byproduct of secondary metabolism and anabolism, therefore use of PPi as an energy donor helps to conserve limited ATP pools (Duff et al., 1989b; Theodorou et al., 1992; Plaxton, 1996; Theodorou & Plaxton, 1996; Plaxton & Carswell, 1999).
Another alternative glycolytic pathway known in plants is catalysed by the action of a nonphosphorylating NADP-dependent glyceraldehyde-3P dehydrogenase (NADP-G3PDH) that bypasses Pi-dependent NAD-G3PDH and phosphoglycerate kinase (Duff et al., 1989b; Theodorou et al., 1992; Plaxton & Carswell, 1999). A third bypass of the glycolytic pathway can be catalysed by the combined activities of PEPC, MDH and NAD-malic enzyme (Theodorou & Plaxton, 1993). Phosphorus stress can severely limit the activity of pyruvate kinase (PK), an enzyme requiring Pi and ADP. The PEPC, MDH and NAD-malic enzymes can bypass PK and thus maintain the flow of carbon from glycolysis to the TCA cycle by avoiding the use of ADP but generating free Pi (Theodorou et al., 1992; Plaxton & Carswell, 1999).
The described metabolic adaptations to Pi stress appear to be plant specific. P-deficiency induced PFP activity was found in plants that are adapted to infertile soils and contain very low PFP activity when grown under sufficient P, e.g. Arabidopsis, rape (Brassica napus), sugar beet (Beta vulgaris), and buckwheat (Fagopyrum esculentum) (reviewed in Plaxton & Carswell, 1999). Murley et al. (1998) reported that P-stress inducible PFP activity is not found in plants that form symbiotic associations with mycorrhizal fungi. In plants that typically contain abundant PFP activity when grown under sufficient P, PFP activity stayed unchanged or declined under P deficiency (in tobacco (Nicotiana tabacum) seedlings, bean seedlings and tomato root cultures; Plaxton & Carswell, 1999). Analysis of ESTs from P-deficient cluster roots of white lupin indicated the induction of genes encoding for PFP and the PEPC/MDH bypass of PK, but did not reveal evidence for the involvement of a nonphosphorylating NADP-dependent G3PDH (Fig. 6).
Alternative mitochondrial respiration
Phosphorus limitation impairs mitochondrial respiration in many plants. Considering the adaptive flexibility of plant responses to P deficiency that is displayed in the glycolytic pathway, adaptations which allow alternative mitochondrial electron transport under P starvation have been hypothesized (Rychter & Mikulska, 1990; Theodorou & Plaxton, 1993).
The significant reduction in cellular ADP and Pi levels in plants undergoing P limitation is thought to inhibit the cytochrome (Cyt) pathway of electron transport. Low ADP and Pi can result in a high ATP : ADP ratio which inhibits respiration, a phenomenon well known as adenylate control of respiration (Bryce et al., 1990). Nonphosphorylative pathways that can bypass energy-requiring sites include rotenone-insensitive NADH dehydrogenase (Rasmusson & Moller, 1991) and the alternative oxidase (AOX) (Lambers, 1985; Lance et al., 1985; Day & Wiskich, 1995). The cyanide-resistant AOX catalyses the oxidation of ubiquinol and the reduction of oxygen to water but bypasses the last two energy-conserving sites normally associated with the Cyt pathway. Alternative oxidase activity is subject to regulation by a sulfhydryl–disulfide redox system (Siedow & Umbach, 2000) and by allosteric activation via α-keto acids, especially pyruvate (Millar et al., 1993; Hoefnagel et al., 1995). It was proposed that induction of AOX during P deficiency might be a consequence of elevated PEPC activity.
In roots of P. vulgaris, P deficiency led to decreased activity of the respiratory cytochrome pathway, while alternative, cyanide-resistant respiration and the NADH : NAD ratio increased (Rychter & Mikulska, 1990; Rychter et al., 1992). Indication for a possible link between AOX respiration and P starvation has also been found in Catharanthus roseus (Hoefnagel et al., 1993), Chlamydomonas reinhardtii (Weger & Dasgupta, 1993) and tobacco cell cultures (Parsons et al., 1999). However, oxidative respiration rates under P deficiency can vary greatly among different plants. Respiratory O2 consumption in the green alga Scenedesmus obtusiusculus was stimulated during the first 96 h of P starvation (Tillberg & Rowley, 1989), while pea (Pisum sativum) roots displayed no effect of P starvation on respiration (reviewed by Theodorou & Plaxton, 1996). The respiration rates of P. vulgaris, Selenastrum minutum, Lemna gibba and suspension-cultured C. roseus, however, declined under P deprivation (Theodorou & Plaxton, 1996). In P-deficient cluster roots of white lupin, up to 60% reduction in mitochondrial respiration, as measured by O2 consumption, has been reported (Johnson et al., 1994; Neumann et al., 1999), accompanied by an increased NADH : NAD ratio. These results indicate that in some plants inhibition of the CN-sensitive cytochrome pathway of respiration might not be fully compensated by AOX activity.
Enhanced glycolysis in P-deficient white lupin was accompanied by significant induction of genes encoding for an alcohol dehydrogenase (Massonneau et al., 2001) and a formate dehydrogenase (fdh) (Uhde-Stone et al., 2003b), indicating a possible role of fermentative processes during P deficiency. In the green algae C. reinhardtii, phosphate starvation resulted in the accumulation of an mRNA with homology to pyruvate formate-lyase, another enzyme of fermentative metabolism (Dumont et al., 1993). Taken together, these findings indicate that the increase of glycolytic activity and the resulting organic acid synthesis under P deprivation in white lupin could be, in part, a consequence of severely impaired oxidative respiration.
A switch from primary to secondary metabolism is a common response of plants undergoing nutrient limitation. Phosphorus deficiency typically results in the accumulation of secondary metabolites such as flavonoids and indole alkaloids (reviewed in Plaxton & Carswell, 1999). In rice culture, accumulation of the polyamine putrescine is thought to mediate growth inhibition in response to P depletion (Shih & Kao, 1996). Yamakawa et al. (1983) reported that P depletion induced anthocyanin accumulation in Vitis cell culture, while P supplementation induced primary metabolism and inhibited anthocyanin synthesis. Increased synthesis of anthocyanins is a frequent response of plants to P deficiency and presumably functions to ameliorate photoinhibitory damage to chloroplasts (Takahashi et al., 1991).
In addition to the possible involvement of phenolics in senescence, phenolic compounds can be exuded into the rhizosphere in response to P starvation and might act as chelators and/or reductants increasing the release of bound Pi. Examples of phenolics that are exuded into the rhizosphere are piscidic acid from roots of pigeon pea (Cajanus cajan) (Ae et al., 1990), alfafuran from alfalfa (Medicago sativa) roots (Masaoka et al., 1993), and isoflavonoids from P-deficient cluster roots of white lupin (Neumann et al., 2000).
Sclerophylly, characteristic of many of the Proteaceae, is also thought to be an adaptation involving enhanced phenolic metabolism by many species that occupy low P habitats. Sclerophylls are typically considered to be hard and stiff leaves that are tough (Turner, 1994). They are characterized by a thick cuticle and outer epidermal cell wall with abundant sclerification (Loveless, 1961, 1962). Bundle sheaths and leaf margins in particular show sclerification. The secondary wall of sclerids are heavily lignified, hence the relationship to phenolic secondary metabolism. Loveless (1962) postulated that sclerophylly is an adaptation to conserve nutrients particularly P and N. Turner (1994) concluded that sclerophylly is also related to protection against herbivory.
Enzymes in the phenolic and flavonoid pathways have been reported to be upregulated in Pi-deficient plants (Plaxton & Carswell, 1999). Several white lupin ESTs with high similarity to enzymes involved in secondary metabolism were represented in our cDNA library derived from mature cluster roots of P-deficient plants, including DAHP (3-deoxy-d-arabino-heptulosonate-7-phosphate synthase), phenylalanine ammonia-lyase, chalcone synthase, chalcone isomerase, isoliquiritigenin 2′-O-methyltransferase, 4-coumarate-CoA ligase, caffeoyl-CoA O-methyltransferase, cinnamoyl-CoA reductase and cinnamyl alcohol dehydrogenase (http://www.home.earthlink.net/~whitelupinacelimation). Whether these enzymes are involved in either anthocyanin or lignin biosynthesis remains to be established. By contrast to primary metabolism, secondary metabolism in general does not consume as much Pi but can serve to recycle significant amounts of Pi from phosphate esters (Plaxton & Carswell, 1999; Sakano, 2001). However, secondary metabolism produces an excess of reducing equivalents. The resulting cytosolic acidification could lead to the activation of AOX and other pathways that can relieve the accumulation of reducing equivalents (Sakano, 2001).
Modified thylakoid membranes
Adaptations to P deficiency related to carbon metabolism also include modification of the lipid composition of photosynthetic membranes (Essigmann et al., 1998; Yu et al., 2002). The lipid composition of thylakoids is comprised primarily of: monogalactosyl diacylglycerol, digalactosyl diacylglycerol, phosphaidyl-glycerol and sulfoquinovosyl diacylglycerol. Under P-deficiency, thylakoid membranes display decreased amounts of phospholipids accompanied by large increases in the sulfolipid, sulfiquinovosyl diacyylglycerol (Benning, 1998; Yu et al., 2002). Arabidopsis genes encoding the two critical steps in sulfolipid synthesis have recently been isolated and characterized (Essigmann et al., 1998; Yu et al., 2002). The first step is catalysed by SQD, a protein that converts UDP-glucose and sulfite to UDP-sulfoquinovose. The second enzyme required for sulfolipid synthesis is SQD2 which catalyses the condensation of UDP-sulfoquinovose and diacylglycerol to sulfoquinovosyl diacylglycerol (Benning, 1998). Both SQD1 and SQD2 have greatly enhanced expression under P-deficient conditions. Moreover, an Arabidopsis line with a mutation in SQD2 is impaired in phosphate-limited growth (Yu et al., 2002). Essigmann et al. (1998) proposed that the function of the changes in thylakoid membrane sulfolipid : phospholipid ratio would be to conserve P while maintaining membrane integrity. Anionic sulfolipid could substitute for anionic phosphatidylglycerol, thus reducing the need for membrane-bound P, yet allowing photosynthesis to continue under P-limited conditions.
VII. Acid phosphatase
A universal response by plants to P-deficiency is the synthesis of acid phosphatase enzymes (APases). These enzymes are ubiquitous in plant organs and APase activity can be detected throughout development (Duff et al., 1994). APases are implicated in: providing P during seed germination from stored phytate (Biswas & Cundiff, 1991; Brinch-Pedersen et al., 2002); internal remobilization of P (Duff et al., 1991; del Pozo et al., 1999; Baldwin et al., 2001); release of P from soil organic P-esters by exudation of enzymes into the rhizosphere (Lefebvre et al., 1990; Goldstein et al., 1988; Miller et al., 2001); and the synthesis of glycolate from P-glycolate (Christeller & Tolbert, 1978) as well as glycerate from 3-PGA during photorespiration (Randall et al., 1971). Characteristically, phytases and root-secreted APase have little substrate specificity, while APases involved in carbon metabolism (i.e. phosphoglycolate phosphatase, 3-PGA phosphatase and phosphoenolpyruvate (PEP) phosphatase) have much stricter substrate specificities (Duff et al., 1991; del Pozo et al., 1999; Miller et al., 2001). Both intra- and extra-cellular APases are prominent in plants, and their activities have traditionally been used as markers for P-deficiency (Duff et al., 1994; Baldwin et al., 2001). Intracellular forms are found in the vacuole or cytoplasm. Although extracellular APases occur in the root apoplast and are frequently released from cell suspension cultures, extracellular types are usually localized in the cell wall, outer surface of root epidermal cells, and the root apical meristem. Intracellular APases appear to be much less stable than extracellular forms, which remain stable for hours to days (Goldstein et al., 1988; Duff et al., 1989a, 1991; Miller et al., 2001). Increased stability would, of course, be a critical characteristic of any APase excreted into the rhizosphere. While intracellular APases induced during P deficiency function primarily in the release of P from senescent tissue for remobilization and in bypassing the P-requiring steps in C metabolism (Duff et al., 1989b; Plaxton & Carswell, 1999), root extracellular APases are more likely to be involved in acquisition from soil (Marschner et al., 1986; Tarafdar & Claasen, 1988). Although the acquisition of P from soil organic matter is equivocal, Miller et al. (2001) postulated that secretion of even low-affinity P-cleaving enzymes into soils with high organic P is an effective mechanism to provide additional sources of P for plant growth. Here, we focus on recent progress in understanding biochemical and molecular aspects of APases with special emphasis on exuded APases from cluster roots of P-deficient white lupin.
The literature abounds with reports of APases secreted from roots of P-deficient plants (Goldstein et al., 1988; Li et al., 1997a,b; Gilbert et al., 1999; Hayes et al., 1999; Zhang & McManus, 2000; Hunter & Leung, 2001). However, only recently have APase genes and proteins from Arabidopsis, tomato, and lupin cluster roots been fully characterized (del Pozo et al., 1999; Wasaki et al., 1999, 2002; Haran et al., 2000; Baldwin et al., 2001; Miller et al., 2001). del Pozo et al. (1999) isolated and characterized a type 5 purple acid phosphatase, AtACP5, gene from Arabidopsis. The gene contained a 1014-bp open reading frame (ORF) encoding 338 amino acid residues. The protein had a 31 amino acid N-terminal extension with characteristics of a membrane targeting signal peptide. Both roots and shoots showed highly induced expression of AtACP5 under P deficiency and addition of P could reverse induction. Induction of AtACP5 also occurred in senescent tissues and in response to salt stress. Although AtACP5 was highly expressed in roots under P-stress, the authors could find no evidence for secretion of AtACP5 protein. They concluded that the protein was tightly anchored to the cell wall or plasmalemma. The AtACP5 promoter, when fused to GUS and used as a reporter, showed high activity in response to P-deficiency, senescence, and salt stress. They concluded that AtACP5 was involved more in internal P remobilization than P acquisition. Another APase (LePS2) implicated in internal remobilization of P has been characterized from tomato (Baldwin et al., 2001). LePS2 contains an 810-bp ORF encoding a 269 amino acid protein. The protein contains no 5′-signal peptide and is postulated to be in the cytosol. LePS2 was highly induced in all tissues in P-deficient plants and expression was not induced by other nutrient stresses. Induction of LePS2 transcripts could also be reversed by addition of P to stressed plants. Both AtACP5 and LePS2 were rapidly induced under P-deficient conditions, suggesting that plant molecular responses to P are tightly controlled.
The first truly secreted APase (sAPase) gene to be characterized was from Arabidopsis (AtsAPase; Haran et al. 2000). The AtsAPase transcript is 1380 bp corresponding to a 46 kDa protein of approximately 450 amino acids. This protein contains a 34 amino acid N-terminal extension predicted to target the mature protein to the endomembrane system. The gene was highly induced in roots of P-deficient plants. When the AtsAPase transit sequence was fused to green fluorescent protein (GFP) as a reporter, GFP was detected in root exudates of P-deficient plants. The authors proposed that the APase transit sequence would be useful in strategies to produce and obtain relatively pure proteins of industrial importance from plant roots.
Root APase activity increases dramatically in P-deficient white lupin plants (Ozawa et al., 1995; Gilbert et al., 1999). This large increase in APase activity is associated with the appearance of a novel isoform (Wasaki et al., 1997; Gilbert et al., 1999). Moreover, this novel form induced by P-stress is also secreted into the rhizosphere of cluster roots (Gilbert et al., 1999; Miller et al., 2001; Wasaki et al., 2002). When cluster roots from P-deficient plants are placed on agar containing APase substrate, exudation of APase activity into the rhizosphere can be detected within minutes (Fig. 7). White lupin sAPase is synthesized as a preprotein having an Mr of 52 kDa. During secretion the protein's 31-amino acid presequence is cleaved, giving rise to a 49 kDa processed protein. Similar to other APases, the sAPase is a glycoprotein. Protein blots probed with antibodies for the sAPase showed rapid accumulation of the protein in P-deficient roots accompanied by secretion into the rhizosphere (Miller et al., 2001). The cDNA for white lupin sAPase contained a 1380 bp open reading frame capable of coding a 460-amino acid protein. Expression of sAPase mRNAs was quite specific (Fig. 2) and found predominantly under P-deficiency in cluster roots (Miller et al., 2001; Wasaki et al., 2002). Slight induction was also found in Al stressed roots. Miller et al. (2001) isolated the promoter for sAPase and showed that the gene was responsive to internal P concentration. In addition, the promoter was functional in alfalfa, showing that the molecular events controlling plant response to P deficiency are conserved between plants. The fact that a 50 bp region within the promoter of white lupin sAPase is 72% identical to a region in the Arabidopsis P-induced APase supports this interpretation. White lupin's ability to access P unavailable to other species (Braum & Helmke, 1995) may not only reside in its exudation of organic acids but may also be a consequence of exuded APase activity (Fig. 3b).
VIII. Genetic regulation of P responsive genes
As mentioned previously, numerous molecular events are set in motion in response to P deficiency resulting in biochemical changes that facilitate P conservation and/or acquisition. Raghothama (1999) has suggested that more than 100 genes may be involved in plant response to low-P stress. The fact that many of the molecular and biochemical changes in response to P deficiency occur in synchrony suggests that the genes involved are coordinately expressed and share a common regulatory system. Raghothama (1999) and others (Goldstein, 1992; Schachtman et al., 1998) have suggested that regulation of plant response to P stress may reside in a phosphate (PHO) regulatory system similar to that found in yeast and bacteria. The PHO regulon in bacteria and yeast represents a complex multigene system having both structural and regulatory components (Vogel & Hinnen, 1990; Wanner, 1993; Lenburg & O’Shea, 1996; Oshima, 1997). The Pi signaling in these organisms involves a classic two-component system which includes a sensor protein and a response regulator protein. The sensor element is ascribed to a transmembrane protein comprised of an extracellular module, involved in sensing environmental cues, fused to a protein kinase module, capable of autophosphorylation. The response regulator is cytosolic and also has two modules, one of which has a phosphorylation site, and the other acts as a transcription (trans) factor. Induction of -P response genes, in simplest terms, involves the sensor protein receiving the environmental cue and catalysing autophosphorylation. The kinase module of the sensor then phosphorylates the receiver module of the response regulator. In turn, the trans-factor module of the response regulator is then rendered capable of binding to specific cis-elements in the promoters of P responsive genes. Those specific promoter cis-elements in P responsive genes that bind with the PHO response regulator trans-factor are called PHO-boxes.
In recent years, plant two-component signaling systems have been identified that are involved in perception of and response to ethylene, cytokinin and osmoticum (for reviews see Sakakibara et al., 1999; Urao et al., 2000; Hwang & Sheen, 2001; Lohrmann & Harter, 2002). By contrast to the traditional prokaryotic two-component signaling system which comprises two proteins, the two-component system of plants appears to be more complex, involving a multistep process and at least three proteins (Sakakibara et al., 2000; Urao et al., 2000; Chang & Stadler, 2001; Gilroy & Trewavas, 2001; Lohrmann & Harter, 2002). The fundamental system in plants (Fig. 8) involves perception by the input domain of a membrane bound sensor histidine (His) kinase which autophosphorylates an aspartate (Asp) residue in the transmitter domain. The transmitter domain then phosphorylates a His residue on a phosphotransfer (HPt) protein which shuttles into the nucleus and transfers the P to an Asp residue of the response regulator (RR). Upon phosphorylation, the RR undergoes conformational change enabling it to bind to cis-elements within the promoter of genes to be activated by the signal molecule. More than 15 His-Asp kinases, five HPt, and 16 RR proteins have been identified in plants (Urao et al., 2000; Lohrmann & Harter, 2002). Although this multistep two-component signaling mechanism has yet to be linked to signaling low and/or high P, it has been implicated in signaling N (Sakakibara et al., 1999) and osmotic (Urao et al., 1999) stress.
A growing body of evidence derived from studies of P-starvation response defective mutants of Chlamydomonas (Wykoff et al., 1999) and Arabidopsis (Rubio et al. 2001) as well as soybean (Glycine max) P response genes (Tang et al., 2001) indicates that nuclear localized transcription factors are involved in regulation of plant response to P. Genetic analysis of the Chlamydomonas and Arabidopsis P response mutants showed that the mutations had occurred in MYB coiled-coil transcription factors designated PSR1 and PHR1, respectively. Evaluation of the Arabidopsis genome shows the presence of 15 MYB coiled-coil proteins. Although PSR1 and PHR1 transcripts are detected independently of the P status of the plant, both increase under P-deficiency. Both MYB genes are proposed to act downstream from the initial P signaling recognition event. The Arabidopsis PHR1 MYB protein was shown to bind to an imperfect palindromic consensus sequence, 5′-GNATATNC-3′, in the promoters of several P starvation-responsive genes including white lupin sAPase, LaPT1, and LaMYB1 (Liu et al., 2001; Miller et al., 2001). By comparison, the soybean vegetative storage protein (VSPB) which is downregulated in response to added P was shown to be regulated in part by homeodomain leucine zipper protein (HD-ZIP) which binds to the CATTAATTAG element in the VSPB promoter.
Raghothama (1999) has noted that PHO-box like elements occur in some P-deficiency induced genes and has recently identified a cis-element 5′-CACGT(G/T)-3′ in the promoter of several P-response genes. This element is identical to the Pho4 binding domain in yeast P-response genes (Ogawa et al., 1995). We have also recently identified this motif in the promoters of P-deficiency induced APase (Miller et al., 2001) and a MYB transcription factor (J-Q Liu & C. P. Vance unpubl. data), genes which are highly expressed in cluster roots. Interestingly, CACGTG is known as a G-box in plant genes which are regulated by various biotic and abiotic stresses. Also, relevant to identifying genes affecting nutrient acquisition and root development, Zhang and Forde (1998, 2000) have isolated and functionally characterized a MADs box transcription factor, ANR1, that controls nitrate-induced changes in lateral root development. Thus, there are likely many transcription factors and cis-elements that interact to control nutrient acquisition and root architecture. Many of these may be active during the development of white lupin cluster roots as evidenced by our sequencing of ESTs from these organs revealing TUBBY-, GRAB/NAM-, RAS-, DIMINUTO-, and SCARECROW-like genes (http://www.home.earthlink.net/~whitelupinacelimation).
Roots have been notoriously recalcitrant to study and to improve through genetic selection due to the fact that their development is affected by a multitude of genes and their subterranean nature. However, as outlined here and elsewhere in this review, advances in understanding the genetic control of roots, improved methods for root evaluation, and reaffirmation of their importance for improving nutrient use efficiency have led to a renaissance of discovery in root biology that should result in significant improvements to crop plants.
Genetic variability has been documented in a wide range of species (and probably exists in all species) for tap root elongation, basal root growth angle (i.e. rooting depth, lateral and adventitious root development, and root hair numbers; Klepper, 1992; Charlton, 1996). Inclusively these traits comprise root architecture and each has been identified as important to P acquisition (Lynch, 1995; Eshel et al., 1996; Gahoonia & Nielsen, 1997; Care & Caradus, 1999; Yan, 1999; Narang et al., 2000). Phaseolus vulgaris has been the prototype for selection of root architecture ideotypes that result in improved P acquisition (Lynch, 1995; Yan, 1999; Lynch & Brown, 2001). Genetic selection as well as root simulation models with bean have shown enhanced P uptake in genotypes with shallow roots, accompanied by heightened numbers of lateral roots with abundant root hairs. Selection for increased root hair abundance has led to enhanced P acquisition in bean, clover, barley and Arabidopsis. Wissuwa & Ae (2001) have identified four quantitative trait loci (QTL) in rice (Oryza sativa) associated with increased P uptake under P-deficiency, two of which were related to maintenance of root growth under low P. Similarly, these root architecture traits are also important in the acquisition of N and other nutrients (Klepper, 1992; Zhang & Forde, 1998; Lamb et al., 2000; Vance & Lamb, 2001).
Phosphorus acquisition has also been improved through approaches aimed at increasing organic acid synthesis in and/or exudation from plant roots. This approach is based upon the large body of evidence showing that exudation of citrate and malate from roots effectively solubilizes unavailable sources of P (Marschner et al., 1986). Release of organic acids into the rhizosphere in exudates leads to metal chelation and subsequent desorption of Pi from the soil matrix, with a concomitant increase in availability. Rhizosphere acidification by protons and/or organic acids may be an accompanying trait when root architecture is modified through selection as shown in bean (Yan, 1999), Arabidopsis (Narang et al., 2000), rice (Yang et al., 2000) and pigeon pea (Ae et al., 1990). The acidification of the rhizosphere of white lupin cluster roots also exemplifies the concurrent exudation of organic acids accompanying a change in root development. Overexpression of genes involved in organic acid synthesis appears to be a useful strategy to stimulate P acquisition by enhancing exudation of citrate and malate. Koyama et al. (1999) developed carrot (Dacus carota) cell lines that overexpressed mitochondrial citrate synthase. The overexpressing cell lines exuded more citrate and had improved growth on AlPO4− medium compared with untransformed controls. In a similar approach but using a bacterial citrate synthase gene driven by the CaMV35S promoter, Lopez-Bucio et al. (2000a) demonstrated that overexpression of citrate synthase resulted in increased secretion of citrate into the rhizosphere and enhanced P accumulation. Our laboratory recently overexpressed a novel form of MDH in alfalfa (Tesfaye et al., 2001). Transgenic plants containing the novel MDH driven by the CaMV35S promoter had increased organic acid synthesis and exudation, which resulted in increased P accumulation compared with either transformed or untransformed controls. Interestingly, in each of the above cases where overexpression of genes involved in citrate and malate production enhanced organic acid synthesis and P acquisition, the transformed plants had greater Al3+ tolerance than controls (Lopez-Bucio et al., 2000b; Tesfaye et al., 2001). It is well established that Al3+ tolerance in plants is related to organic acid accumulation and/or exudation by roots (Kochian, 1995; Ryan et al., 2001). Moreover, increased organic acid synthesis and exudation has been implicated in tolerance to Pb and other metals (Mathys, 1977; Godbold et al., 1984; Kochian, 1995; Yang et al., 2000). In addition, accumulation and exudation of citrate and malate by roots has been implicated in alleviating Fe deficiency (de Vos et al., 1986; Jones, 1998). Thus, genetic engineering of plants for enhanced organic acid synthesis may have far-reaching effects on plant nutrition and stress tolerance.
Other targets which have potential to improve Pi acquisition include high-affinity Pi transporters (Mitsukawa et al., 1997) and root exuded phytase (Richardson et al., 2001). Overexpression of the Arabidopsis high-affinity Pi transporter gene (PHT1) in tobacco resulted in cell lines with enhanced growth under Pi-limited conditions (Mitsukawa et al., 1997). Transgenic tobacco lines overexpressing PHT1 had threefold greater Pi uptake than controls and 50% greater growth under low Pi conditions. Richardson et al. (2001) generated transgenic Arabidopsis plants that expressed an Aspergillus phytase gene and exuded phytase into the rooting medium. The transgenic plants could grow on phytate as the sole source of P whereas control plants could not. We have recently developed alfalfa plants that can overexpress several diverse proteins in roots in efforts to metabolically engineer improved nutrient use acquisition (APase, MDH and metallothioneins). These transgenic plants are currently awaiting characterization.
Evidence is accumulating that selection for and overexpression of selected genes involved in either root development or P acquisition can improve nutrient acquisition. As the arsenal of plant genes involved in these processes expands through whole-genome approaches, we will undoubtedly see more successful examples of plants with better nutrient acquisition and utilization created through traditional breeding and transgenic technology. Because of the urgent need for plant germplasm with improved P-use efficiency, it is imperative that research programs mount multidisciplinary teams that combine traditional plant breeding, biotechnology and plant physiology.
The world is on the brink of a new agriculture, one that involves the marriage of plant biology and agroecology under the umbrella of biotechnology and germplasm improvement. Although P fertilizers will continue to play a major role in intensive agriculture, depletion of natural resources, loss of biodiversity, and long-term unsustainability require that alternative strategies be investigated and implemented to buffer against food insecurity and environmental degradation. Furthermore, because improved P acquisition and use by plants has immediate and direct benefit in extensive agriculture in developing countries where access to fertilizers is limited, funding for research at international centers should be a high priority. The following recommendations deserve attention: (1) expand research in root biology; (2) continue to isolate, characterize and develop a fundamental understanding of individual genes holding promise for application to improving P nutrition; (3) enhance the expression of genes and increase the synthesis of gene products, such as those involved in transport of nutrients and exudation of organic acids, through transgenic technology and incorporate these traits into adapted germplasm; (4) continue and expand collaborations between plant breeders and physiologists to develop germplasm with enhanced P-use efficiency; and (5) reemphasize applied research programs directed toward sustainable approaches to enhancing soil P through crop management (such as intercropping, rotations and incorporation of legumes).
This work was funded in part by the United States Department of Agriculture, National Research Initiative Grant No. USDA-CSREES/98-35100-6098 and USDA-CSREES/2002-35100-12206, and United States Department of Agriculture, Agricultural Research Service CRIS no. 3640-21000-019-00D.