INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Phytic acid (myo-inositol-1,2,3,4,5,6-hexakisphosphate or Ins P₆, Fig. 1A) is the most abundant P-containing compound in mature seeds, typically representing from 65% to 80% of the mature seed's total P (Cosgrove, 1980; Raboy, 1997). In the mature maize (Zea mays) seed, most (>80%) of the phytic acid is found in the germ with the remainder in the aleurone layer (O'Dell et al., 1972). In normal non-mutant seeds, phytic acid P typically represents >95% of total, acid-extractable myo-inositol (Ins) phosphates. Substantial quantitative variation in seed phytic acid P has been observed among genotypes, lines, or cultivars of several crop species. However, in these earlier studies the relationship between seed total P and phytic acid P was not observed to vary greatly, with the correlation between seed phytic acid P and seed total P typically 95% (Raboy, 1990).

View larger version (16K): [in this window] [in a new window]   Figure 1.   Biosynthetic pathways to phytic acid (myo-inositol-1,2,3, 4,5,6-hexakisphosphate or Ins P6) in the eukaryotic cell. A, Structure of phytic acid. B, Structure of Ins. The numbering of the carbon atoms follows the "D-Convention" (Loewus and Murthy, 2000). C, Biochemical pathways: (1), D-Ins(3)-P1 (or L-Ins[1]-P1) synthase; (2), D-Ins 3-phosphatase (or L-Ins 1-phosphatase); (3), D-Ins 3-kinase (or L-Ins 1-kinase); (4), Ins P- or polyP kinases; (5), Ins (1,3,4,5,6) P5 2-kinase or phytic acid-ADP phosphotransferase; (6), PtdIns synthase; (7), PtdIns and PtdIns P kinases, followed by PtdIns P-specific phospholipase C, and Ins P kinases; (8), D-Ins(1,2,3,4,5,6) P6 3-phosphatase; (9) D-Ins(1,2,4,5,6) P5 3-kinase; (10), D-Ins(1,2,3,4,5,6) P6 5-phosphatase; (11), D-Ins(1,2,3,4,6) P5 5-kinase; (12), pyrophosphate-forming Ins P6 kinases; (13), pyrophosphate-containing Ins PolyP-ADP phosphotransferases.

In the context of plant and seed biology, phytic acid has been viewed primarily as a P and mineral storage compound or as an important metabolite in P homeostasis (Strother, 1980; Lott, 1984; Raboy, 1997). Regulation of cellular inorganic phosphate (P_i) concentration may play an important role in starch synthesis and accumulation and in the function of other metabolic pathways (Strother, 1980). Recent studies have shown that phytic acid may be ubiquitous in eukaryotic cells and that phytic acid and certain Ins pentakisphosphates typically represent the most abundant Ins phosphates in cells (Sasakawa et al., 1995; Safrany et al., 1999).

The biosynthetic pathway to phytic acid can be summarized as consisting of two parts: Ins supply and subsequent Ins polyphosphate synthesis (Fig. 1C). The sole synthetic source of the Ins ring (Fig. 1B) is the enzyme Ins(3) P₁ synthase (MIPS), that converts Glc-6-P to Ins(3) P₁ (Fig. 1C, step 1; Loewus and Murthy, 2000). Proximal MIPS activity in the developing seed may provide Ins as Ins(3) P₁ (Yoshida et al., 1999), which then may be converted directly to phytic acid via sequential phosphorylation by two or more kinases (Biswas et al., 1978; Stephens and Irvine, 1990; Fig. 1C, step 4). The Ins backbone for phytic acid may also derive in part from MIPS activity at distal vegetative sites, followed by Ins translocation to the developing seed (Sasaki and Loewus, 1990). The first Ins phosphorylation step would then be catalyzed by the enzyme Ins kinase, which also produces Ins(3) P₁ (English et al., 1966; Loewus et al., 1982; Fig. 1C, step 3). A pathway to phytic acid that begins with Ins as initial substrate and Ins kinase activity and proceeds through sequential phosphorylation steps via defined intermediates, was first described in studies of the cellular slime mold Dictyostelium discoideum (Stephens and Irvine, 1990), and subsequently in studies of the monocot Spirodela polyrhiza (Brearley and Hanke, 1996a, 1996b). The D. discoideum pathway proceeded through the intermediates Ins(3) P₁, Ins(3,6) P₂, Ins(3,4,6) P₃, Ins(1,3,4,6) P₄, and Ins(1,3,4,5,6) P₅. The S. polyrhiza pathway proceeded through the intermediates Ins(3) P₁, Ins(3,4) P₂, Ins(3,4,6) P₃, Ins(3,4,5,6) P₄, and Ins(1,3,4,5,6) P₅. These pathways are similar in their first and last intermediates, and in that these Ins phosphates are not known to function as second messengers. In none of the above studies has the relative contribution, in spatial or temporal terms, of MIPS or Ins kinase activity been determined unequivocally.

Phytic acid synthesis may also proceed in part via pathways typically associated with second messenger metabolism that involve phosphatidylinositol (PtdIns) phosphate intermediates and Ins(1,4,5) P₃ (Fig. 1C, steps 6 and 7; Van der Kayy et al., 1995; York et al., 1999). Also, Ins phosphates more highly phosphorylated than phytic acid, such as Ins P₇ and Ins P₈, have been documented to occur widely in eukaryotic cells (Fig. 1C, steps 12 and 13; Mayr et al., 1992; Menniti et al., 1993; Stephens et al., 1993; Brearley and Hanke, 1996c; Safrany et al., 1999). These compounds contain pyrophosphate moieties and may be involved in ATP regeneration. Phytic acid was originally proposed to play a role in ATP regeneration by Morton and Raison (1963). Therefore in a current view, phytic acid is seen not simply as a P-storage product or end-product for Ins phosphorylation, but as a pool for both P and Ins phosphates, the latter function of importance to signaling and ATP formation (Voglmaíer et al., 1996; Safrany et al., 1999). Recently a role for Ins P₆ in mRNA export in yeast was demonstrated (York et al., 1999).

These and other studies (Biswas et al., 1978b; Phillippy et al., 1994; Brearley and Hanke, 1996b) have led to a consensus that, regardless of precursor pathway, Ins(1,3,4,5,6) P₅ represents the penultimate Ins phosphate in the primary synthetic pathway to phytic acid in the eukaryotic cell. In D. discoideum, two additional Ins pentakisphosphates were observed to accumulate, Ins(1,2,4,5,6) P₅ and Ins(1,2,3,4,6) P₅ (Stephens et al., 1991). Although all three compounds serve as substrate for Ins P₅ kinase(s), conversion of these latter compounds to phytic acid was slower than that observed for D-Ins(1,3,4,5,6) P₅, and they accumulate to higher steady-state levels. These two compounds appeared only to interconvert with phytic acid. Both compounds were also observed in the soybean (Glycine max; Phillippy and Bland, 1988), in S. polyrhiza (Brearley and Hanke, 1996a), and in the barley (Hordeum vulgare) aleurone layer (Brearley and Hanke, 1996c).

We sought non-lethal mutants that would greatly alter the basic P and Ins phosphate phenotype of normal seeds and decouple the close relationship between seed total P and phytic acid P. We reasoned that such mutants would represent mutations proximal to phytic acid synthesis in the developing seed and would be valuable in studies of phytic acid biology. The first two non-lethal mutants of this type we found were maize low phytic acid 1-1 (lpa1-1) and lpa2-1 (Raboy and Gerbasi, 1996). Recently similar mutants have also been isolated in barley (Larson et al., 1998; Rasmussen and Hatzack, 1998). Here we describe the origin and inheritance of maize lpa1-1 and lpa2-1, characterize their seed P and Ins phosphate phenotypes, and report an association between reduced seed phytic acid and reduced seed dry weight.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Origin, High-Voltage Paper Electrophoresis (HVPE) Phenotype, and Chromosomal Map Position of lpa1-1 and lpa2-1

The lpa1-1 mutant was first observed segregating in a single M₂ progeny obtained following the self-pollination of the M₁ plant, 90046-13. No phenotypically similar mutant was observed in any of other M₂ descendants of M₁ 90046, nor in the M₂ descendants of other M₁s comprising this first screened population. Therefore this mutation probably occurred in a single ethyl methanesulfanate-treated pollen grain, as expected. The HVPE phenotype of this mutant (Fig. 2A, lane 2) is an approximately 66% reduction in seed phytic acid P as compared with sibling non-mutant seeds (Fig. 2A, lane 1). This reduction in phytic acid P is accompanied by what appears to be a molar-equivalent (in terms of P) increase in P_i. No unusual accumulations of Ins phosphates other than phytic acid are observed. Also, the mutant phenotype of seeds produced by a plant homozygous for lpa1-1 (Fig. 2A, lanes 3-5) is similar if not identical to the mutant phenotype of homozygous lpa1-1 seeds obtained following the self-pollination of a heterozygote. This indicates that the lpa genotype or phenotype of the parent plant does not greatly affect the lpa1-1 seed phenotype.

View larger version (100K): [in this window] [in a new window]   Figure 2.   HVPE of inositol phosphates and Pi in lpa1-1 and lpa2-1 seed. A, HVPE phenotype of lpa1-1: lane S, standards: P6, phytic acid or Ins hexakisphosphate; P2 through P5 are a mixture of Ins bis- through pentakisphosphates produced via the partial hydrolysis of phytic acid; lanes 1 and 2, HVPE tests of sibling normal (+/+ or +/lpa1-1, lane 1) and homozygous mutant (lpa1-1/lpa1-1, lane 2) kernels sampled from an F2 ear produced by the self-pollination of an F1 heterozygote (+/lpa1-1); lanes 3 through 5, HVPE tests of three kernels sampled from an ear produced by the self-pollination of an F2 lpa1-1 homozygote. B, HVPE phenotype of lpa2-1: lane S, standards as in A; lanes 1 and 2, HVPE tests of sibling normal (+/+ or +/lpa2-1, lane 1) and homozygous mutant (lpa2-1/lpa2-1, lane 2) kernels sampled from an F2 ear produced by the self-pollination of an F1 heterozygote (+/lpa2-1); lanes 3 through 5, HVPE tests of three kernels sampled from an ear produced by the self-pollination of an F2 lpa2-1 homozygote.

The HVPE phenotype of lpa2-1 is what appears to be a 50% reduction in seed phytic acid P (Fig. 2B, lane 2) as compared with sibling non-mutant seeds (Fig. 2B, lane 1). This reduction in phytic acid P is accompanied by an increase in P_i and novel accumulations of two P-containing compounds with mobilities similar to Ins P₄ and Ins P₅, the latter being the more abundant of the two. This mutant phenotype was first observed in seeds obtained from not one, as would be expected, but three related M₂ progenies; 90041-1, 90041-4, and 90041-12. The seed that produced these progenies were siblings from a single M₁ ear, M1 90041. Mutants phenotypically similar to lpa2-1 were not observed to segregate in any other M₂ ears of this population. This indicates that the mutation occurred spontaneously in one of the two parent plants used to produce 90041, prior to chemical mutagenesis. If it had occurred at an earlier point in the inheritance of this population, we would have observed it segregating in additional M₂ progenies, descended from other M₁s. All subsequent studies of lpa2-1 were conducted using materials developed from M₂ 90041-4. As with lpa1-1, the mutant phenotype of seeds obtained from a homozygote (Fig. 2B, lanes 3-5) is similar or identical to that observed in mutant seeds obtained from the self-pollination of a heterozygote (Fig. 2B, lane 2). Therefore the mutant seed phenotype is a seed-specific effect.

HVPE tests of seeds produced by the cross-pollination of lpa1-1 and lpa2-1 homozygotes indicated that these seeds contained non-mutant levels of phytic acid P and P_i and no unusual accumulations of Ins phosphates other than phytic acid, demonstrating that these two mutants complement each other and therefore are non-allelic (data not shown). This was confirmed in the following chromosomal-mapping experiments (Fig. 3). We obtained crosses of lpa1-1 homozygotes by 13 different simple and compound B-A translocations, representing portions of 15 different chromosome arms. The TB-1Sb translocation stock, which contains approximately 75% of chromosome 1S arm distal to the centromere (Fig. 3A), was the only translocation that uncovered the lpa1-1 phenotype at a significant frequency (11 of 40 seeds obtained from the cross displayed the mutant phenotype). TB1Sb-2L4464, a compound translocation that uncovers approximately 50% of the same chromosome arm, but not the distal-most portion of the 1S arm (Fig. 3A), did not uncover the lpa1-1 phenotype. This indicates that lpa1-1 maps to the distal region of 1S. In the case of lpa2-1, we obtained crosses by 19 simple and compound translocations, representing significant portions of 19 chromosome arms (all but 8S). As in lpa1-1, TB-1Sb uncovered lpa2-1 (23 of 90 seeds obtained from the cross displayed the mutant phenotype). However, TB1Sb-2L4464 also uncovered the mutant (18 of 63 seeds obtained from the cross displayed the mutant phenotype), indicating that lpa2-1 is located on the proximal half of chromosome 1S.

View larger version (42K): [in this window] [in a new window]   Figure 3.   Chromosomal mapping of maize lpa1 and lpa2. A, Approximate map positions of lpa loci and markers on chromosome 1S and their relation to two chromosome 1S B-A translocations. Approximate distance (cM) of lpa1 to umc157 and lpa2 to umc167 is shown. For B-A translocations TB-1Sb and TB-1SB-2L4464, B (dashed line) indicates B chromosome component and A-1S or A-2L (solid lines) indicate relative position and composition (to chromosome 1S sequence) of indicated A chromosome component. B and C, RFLP mapping of lpa loci using bulked segregant analyses. A genotypic bulk DNA was prepared to represent the three lpa1 or lpa2 F2-mapping population segregant classes: +/+, homozygous normal (or Lpa/Lpa); +/, heterozygous (+/lpa or Lpa/lpa); /, homozygous mutant (lpa/lpa). DNAs isolated from each of the individuals representing each class were combined so that each individual contributed equally to the bulk. Bulk DNA was digested with EcoRV, fractionated, and probed with the indicated RFLP marker. P and E are the parental Pioneer Hi-Bred inbred and Early-ACR RFLP alleles, respectively.

These approximate chromosome arm positions for lpa1-1 and lpa2-1 were confirmed with RFLP mapping (Fig. 3, B and C). Bulk-segregant analysis of the 50 segregating F₂s first identified linkage of lpa1-1 to the RFLP marker umc157 (Fig. 3B), which maps to the distal portion of chromosome 1S (Davis et al., 1999). A readily scorable EcoRV polymorphism was detected. Based on observed differences in signal, the parental linkage "E;" and "P;+" was observed in approximately 90% of the chromosomes assayed, with the "E;+" and "P;" crossover types observed in approximately 10% of the chromosomes assayed, indicating linkage of approximately 10 centiMorgans (cM). A follow-up study of the individual F₂s, using umc157 and a second marker that maps to the distal region of chromosome 1S, bnl5.62, confirmed the bulk-segregant result and further defined lpa11 map position. Two "E;+" and two "P;" recombinants between lpa11 and umc157 were found in the 28 homozygous F₂ individuals, and 16 recombinants between lpa11 and bnl5.62 were found in these 28 F₂s. These data place lpa11 approximately 7.7 cM proximal to umc157 (Fig. 3A). Bulk-segregant analysis detected linkage of lpa21 to umc167 (Fig. 3C), which maps to the centromere-proximal portion of chromosome 1S (Davis et al., 1999), with the RFLP marker at a position proximal to the TB1-Sb breakpoint of chromosome 1S. The relative amount of signal observed in the "E" and "P" alleles in the three F₂ genotypic bulks was similar to that observed in the lpa1-1 bulk segregant test (approximately 90% parental linkage in the chromosomes assayed). These data place lpa2-1 approximately 10 cM distal to umc167 on chromosome 1S (Fig. 3A).

Quantitative Analyses of Seed P Fractions

The ferric-precipitation method yields an accurate and reproducible assay of phytic acid P in non-mutant and lpa1-1 seeds where phytic acid P represents >95% of total Ins phosphate (Fig. 2A). However, HVPE indicated that lpa2-1 seeds may contain more substantial amounts (>5% of total Ins phosphate) of Ins phosphates other than phytic acid (Fig. 2B). These would be precipitated in the ferric salt along with phytic acid P and incorrectly measured as "phytic acid P." Therefore we will refer to the value for the P-fraction obtained using the ferric-precipitation method as "total Ins phosphate." This assay indicated that in mature lpa1-1 seeds, total Ins phosphate is reduced approximately two-thirds, compared with non-mutant seeds (Table I). This is accompanied by a molar-equivalent (in terms of P) increase in P_i, with no net change in seed total P. This represents approximately a 5- to 10-fold greater level of P_i as compared with levels typical of mature, non-mutant seeds. The total Ins phosphate in mature lpa2-1 seeds is reduced by approximately one-third, as compared with non-mutant seeds (Table I). As in lpa1-1 seeds, this reduction is accompanied by a molar-equivalent (in terms of P) increase in P_i, with no net change in seed total P. The level of P_i in mature lpa2-1 seeds represents approximately a 3- to 4-fold increase over that observed in mature non-mutant seeds. Thus in both lpa1-1 and lpa2-1 seeds the sum of total Ins phosphate and P_i is constant and similar to that of non-mutant seeds.

Heterozygosity for either mutant had little observable effect on mature seed total P, phytic acid P, and in the case of lpa2-1, P_i (Table I). P_i appeared to be increased approximately 2-fold in lpa1-1 heterozygotes as compared with normal seeds. This increase in P_i was confirmed in an additional analysis of lpa1-1 heterozygotes obtained by the reciprocal pollination of mutant and non-mutant homozygotes (data not shown) and has also been observed in numerous studies of lpa1-1 inheritance. Thus, whereas studies to date indicate that the lpa2-1 mutant allele is recessive to non-mutant, the lpa1-1 mutant allele clearly is not strictly recessive. This first quantitative analysis also indicated a trend for reduced seed dry weight in lpa genotypes as compared with non-mutant (Table I).

Analyses of P fractions during the development of normal, lpa1-1, or lpa2-1 seed revealed that at any given point in development the three genotypes had similar levels of seed total P (Fig. 4). Seed total P concentrations remained relatively constant throughout the development of each genotype (4-5 mg total P g¹), indicating that P uptake closely paralleled dry weight accumulation (Table II). By 30 d after pollination (DAP) reductions in seed dry weight were observed in both mutants as compared with the non-mutant control, typically ranging from 10% to 20%. In normal seeds total Ins phosphate concentration increased, and P_i concentration decreased, throughout development, maintaining a relatively constant sum of total Ins phosphate and P_i. In contrast, total Ins phosphate accumulation was perturbed throughout seed development in both mutants such that clear differences between mutant and non-mutant were observed by 30 DAP (Fig. 4). The reductions in total Ins phosphate concentration observed in lpa1-1 and lpa2-1 during development, as compared with normal seed, were in both cases closely matched by increases in P_i. Thus the sum of total Ins phosphate and P_i remained relatively constant throughout development of normal and mutant seed, representing approximately 75% of seed total P concentration (Fig. 4).

View larger version (41K): [in this window] [in a new window]   Figure 4.   Seed phosphorus fractions in non-mutant (white bars), lpa1-1 (gray bars), and lpa2-1 (hatched bars) homozygotes during development. Seed of the three genotypes were harvested from field-grown plants at three dates during development (15, 30, and 40 DAP) and at maturity, and assayed for seed total P, total inositol P, and Pi. These fractions are expressed as P concentrations (atomic weight = 31) to facilitate comparisons. The data represent the mean of duplicate analyses of three individuals of each genotype at each date and are expressed on a dry weight basis.

HPLC analysis confirmed that in lpa1-1 seeds the reduction in total Ins phosphate is primarily accounted for by a reduction in phytic acid P (Fig. 5). HPLC analysis of normal seeds reproducibly detects a small peak with a mobility similar to an Ins P₅ (Fig. 5B), representing 0.12 mg P g¹ or 4% of total Ins phosphate. This peak is reduced to the extent that it is not detectable in HPLC assays of lpa1-1 seeds (Fig. 5C). HPLC also confirmed that no unusual accumulations of other Ins phosphates are observed in lpa1-1 seeds. Similar findings were reported in an independent analysis of the same non-mutant and lpa1-1 materials tested here (Mendoza et al., 1998). HPLC analysis (Fig. 5D) also confirmed the Ins phosphate phenotype of lpa2-1 seeds observed with HVPE: phytic acid P is reduced approximately 50% as compared with normal seeds, and represents approximately 75% of lpa2-1's reduced levels of total Ins phosphate. The remaining 25% consists primarily of what appears to be an Ins P₅, representing 0.45 mg P g¹, or 22% of total Ins phosphate, and trace levels of the less abundant Ins P₄.

View larger version (21K): [in this window] [in a new window]   Figure 5.   HPLC of acid-soluble Ins phosphates in non-mutant and lpa seed. A, Na Ins P6 or phytic acid standard. Shown is a typical result obtained from the elution of 99.5 nmol of phytic acid. B through D, HPLC tests of extracts prepared from homozygous non-mutant (or Lpa) seed (B), homozygous lpa1-1 seed (C), and homozygous lpa2-1 seed (D). To allow for direct comparison, equal amounts of flour and equal aliquot sizes were tested. Ins P4, Ins P5, and Ins P6 are Ins tetrakis-, pentakis-, and hexkisphosphates, respectively. These identities were obtained and confirmed via comparisons with known standards in HPLC and HVPE, comparison with results of quantitative analyses following ferric-precipitation, and with subsequent NMR.

Purification and Structural Identification of Ins Phosphates in lpa2-1 Seeds

The two putative novel Ins phosphates that accumulate in lpa2-1 seeds to an extent sufficient for reproducible detection with the HVPE and HPLC methods used here were obtained as individual, purified free acids (data not shown). In addition, one-third less abundant P-containing compound was obtained from the same bulk ferric-precipitate. ¹H-NMR revealed that the most abundant novel Ins phosphate in lpa2-1 seeds is an isomer of Ins P₅ (Fig. 6A). The relative up-field (approximately  3.5 ppm) position of a doublet of doublets (J = 10 and 3.0 Hz) compared to the other resonances was clearly evident and this indicates that dephosphorylation had occurred at the H-3, or the enantiomeric H-1, position. Enantiomeric protons cannot be distinguished by NMR spectroscopy so the structure is D-Ins(1,2,4,5,6) P₅ and/or D-Ins(2,3,4,5,6) P₅. Additional information obtained by ³¹P-decoupling and J-resolved NMR experiments provided confirmation of the structure (data not shown).

View larger version (26K): [in this window] [in a new window]   Figure 6.   Determination of structure of Ins tris-, tetrakis-, and pentakisphosphates that accumulate in homozygous lpa2-1 seed. Putative Ins phosphates were purified to homogeneity, and one-dimensional-NMR spectra were obtained. In descending order the most abundant Ins Ps were found to be D-Ins(1,2,4,5,6) P5 or its enantiomer D-Ins(2,3,4,5,6) P5, D-Ins(1,4,5,6) P4 or its enantiomer D-Ins(3,4,5,6) P4, and D-Ins(1,2,6) P3 or its enantiomer D-Ins(2,3,6) P3.

¹H-NMR revealed that the second most abundant Ins phosphate in lpa2-1 seeds is an Ins P₄ (Fig. 6B). The appearance of a triplet (J = 2.9 Hz) at approximately  4.16 in addition to the resonance from H-3 (or H-1) on non-phosphorylated carbon (mentioned above) indicated that the additional dephosphorylation had occurred at H-2. Additional experiments (homonuclear decoupling and J-resolved experiments) were conducted and the structure consistent with all the NMR data was D-Ins(1,4,5,6) P₄ and/or its enantiomer D-Ins(3,4,5,6) P₄. The third and least abundant Ins phosphate obtained from our purification of lpa2-1 seed Ins phosphates was identified as an Ins P₃, D-Ins(1,2,6) P₃, and/or its enantiomer D-Ins(2,3,4) P₃ (Fig. 6C). The ¹H-NMR of this compound showed three sets of "up-field" resonances (relative to other resonances) thus suggesting three protons geminal to hydroxyl groups. The presence of triplets at the  3.4 (J = 9 Hz) and at  3.66 (J = 9.5 Hz) are due to H-5 and H-4 (or H-6) respectively, and a doublet of doublets (J = 10 and 3.0 Hz) is due to H-3 (or H-1). The structure consistent with these resonances, the rest of the NMR spectrum, and additional J-resolved and two-dimensional-DQCOSY experiments was D-Ins(1,2,6) P₃, or its enantiomer D-Ins(2,3,4) P₃. The presence of small concentrations of additional Ins phosphates are evident in the spectra (Fig. 6C), however the concentrations were insufficient for unequivocal identification.

Correspondence between Reduced Phytic Acid, Increased P_i, and Reduced Seed Weight

Since normal mature maize seeds contain consistently low levels (0.3-0.5 mg g¹) of P_i, the high-P_i (HIP) phenotype of lpa seeds (Table I, Figs. 2 and 4) should provide a quick and inherently sensitive assay for lpa genotype. A survey of maize defective kernel (dek) mutants revealed that mutations that perturb germ or aleurone development, the tissues that accumulate phytic acid in maize and other cereals, result in substantial reductions in phytic acid P, and these are always accompanied by equivalent increases in P_i (Raboy et al., 1990). However all such dek mutants are lethal as homozygotes. If care is taken to inspect for the presence of normal germ and aleurone tissues, the HIP phenotype (Fig. 7) should accurately and consistently predict homozygosity for lpa1-1 or lpa2-1. The following inheritance experiments tested the correspondence between the "low phytic acid," "high P_i," and reduced seed weight phenotypes of lpa seeds. F₁ heterozygotes were either self-pollinated to produce F₂s, or used both as males and females in pollinations with the appropriate homozygous mutant testers. In the case of lpa1-1, all seeds from a total of six F₂ ears and 12 test-cross ears were individually inspected, weighed, and tested for P_i (using the assay illustrated in Fig. 7). Approximately 5% of the seed extracts were also tested with HVPE to confirm correspondence between "low phytic acid" and "high P_i." In the case of lpa2-1, all seeds from a total of six F₂s and five test-cross ears were similarly analyzed for P_i, and in addition all seed extracts were also tested with HVPE for the distinctive lpa2-1 HVPE phenotype. Of the six lpa2-1 F₂s, only three showed segregation for a consistent and stable lpa2-1 HVPE phenotype that could be reliably scored, and these were included in the analysis below. The remaining three showed no clear segregation for an lpa2-1-like HVPE phenotype that could be reliably scored, even though tests showed that the sibling M3 lpa2-1 parents used to make the F₁s appeared homozygous for the lpa2-1 allele. Since inheritance of lpa2-1, or expression of its HVPE phenotype, could not be detected in these three F₂ progenies, they could not be included in the subsequent analyses. The cause of this reduced penetrance or instability of inheritance is not known, and such instability was not observed with lpa1-1.

View larger version (110K): [in this window] [in a new window]   Figure 7.   The HIP phenotype of lpa seeds. Twenty seeds from a given ear were individually crushed, extracted, and assayed for Pi using a microtitre plate-based colorimetric assay. To allow for direct comparison, all seeds were extracted in 10 volumes on a single-seed basis, and equal aliquot volumes were tested. A and B, Twenty seeds from a non-mutant (Lpa) homozygote; C and D, 20 seeds from a lpa1-1 homozygote; E and F, 20 sibling F2 seeds sampled from an ear obtained following the self-pollination of an F1 +/lpa1-1 (or Lpa1/lpa1-1) heterozygote; G and H, 20 seeds from a lpa2-1 homozygote; I and J, 20 sibling F2 seeds sampled from an ear obtained following the pollination of an F1 +/lpa2-1 (or Lpa2/lpa2-1) heterozygote. S, Standards; five standards contained 0.0, 0.15, 0.46, 0.93, and 1.39 µg of P.

There was a strict correspondence between reduced seed phytic acid and increased P_I in all seeds tested. In every ear tested, the mean dry weight of the lpa mutant class of seeds was reduced as compared with its sibling non-mutant seed class (Table III). This reduction in seed dry weight approached being twice as great in the case of lpa1-1, ranging from 8% to 23%, as compared with lpa2-1, where the reductions ranged from 4% to 16%. The results also confirm the monogeneic inheritance of both lpa1-1 and lpa2-1 (Table III).

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

These results indicate that lpa1-1 and lpa2-1 represent reduced-function or loss-of-function alleles at two loci on chromosome 1S in maize. It is unlikely that either mutant represents a gain-of-function mutation such as a novel increase in phytase activity. Such gain of function mutations are rare events typically found once in 10⁵ individuals in a mutated population, rather than once in 10³ individuals as observed here, typical of loss-of-function mutations. Also, gain of function mutations usually are additive or dominant, whereas both lpa1-1 and lpa2-1 appear recessive or nearly so. When homozygous these mutants are viable and result in substantial reductions in phytic acid P accumulation during seed development but have little or no effect on seed total P. Therefore, the reduction in seed phytic acid P is not due to reduced uptake or translocation of P to the developing seed. The alteration of a biochemical or genetic function in lpa1-1 and lpa2-1 seed is sufficient to condition the mutant seed phenotype, independent of parent plant genotype. Homozygosity for these alleles may also alter some function throughout the plant, but if so it does not appear to contribute to the seed phenotype. We have isolated a number of additional alleles at these two loci. Studies of these additional alleles will determine if homozygosity for one or more conditions a plant or seed phenotype more extremely than that observed in the initial alleles.

In lpa1-1, seed reductions in all soluble Ins phosphate species typically observed in normal seeds contribute to total Ins phosphate reduction. In lpa2-1 seed total Ins phosphate is reduced as compared with normal seed, but this reduction is accompanied by increases in novel Ins phosphates not observed to accumulate in normal seeds. Based on these phenotypes and the observation that these Ins phosphate reductions occur in the presence of normal levels of total P, we hypothesize that lpa1-1 is a mutation in the first part of the phytic acid synthesis pathway, Ins supply, and lpa2-1 is a mutation in the later part, Ins phosphate metabolism. The maize genome contains a number of MIPS-homologous sequences (possibly as many as seven), and one maps in the proximity of lpa1-1 on chromosome 1S (Fig. 3; Larson and Raboy, 1999). Studies are under way to determine if lpa1-1 is in fact a lesion in the chromosome 1S MIPS or in some other function in this part of the pathway.

The correspondence between the reduction in phytic acid and increase in D-Ins(1,2,4,5,6) P₅ (and/or its enantiomer) observed in lpa2-1 seeds indicates that this later compound plays some significant role in phytic acid metabolism in the maize seed. Maize lpa2-1 may be a lesion in a gene encoding a D-Ins(1,2,4,5,6) P₅ 3-kinase. Such a lesion might also account for the accompanying accumulations of D-Ins(1,4,5,6) P₄ and D-Ins(1,2,6) P₃ (and/or their respective enantiomers) in lpa2-1 seed. The presence of these apparent breakdown products of D-Ins(1,3,4,5,6) P₅ indicates that this later compound does not simply interconvert with phytic acid but can be further metabolized in the developing maize seed.

Previous studies uniformly show that the most likely synthetic pathway to phytic acid begins with the synthesis of D-Ins(3) P₁ and ends with the conversion of D-Ins(1,3,4,5,6) P₅ to phytic acid (Biswas et al., 1978a; Stephens and Irvine, 1990; Phillippy et al., 1994; Van der Kayy et al., 1995; Brearley and Hanke, 1996b). Therefore, an alternative is that D-Ins(1,2,4,5,6) P₅ accumulates in lpa2-1 seed indirectly as a result of a lesion in gene encoding something other than a 3-kinase, such as a gene encoding a D-Ins(1,3,4,5,6) P₅ 2-kinase. A study of fruitfly (Drosophila melanogaster) Ins polyP 1-phosphatase (ipp) mutants demonstrated that flies homozygous for an ipp allele cannot metabolize Ins(1,4) P₂, a critical component of the Ins(1,4,5) P₃ signaling pathway (Majerus, 1992), yet several cellular processes dependent on Ins P₃ signaling pathways functioned normally (Acharya et al., 1998). Apparently ipp homozygotes adjust in vivo via compensatory up-regulation of an alternative Ins P₃ pathway involving Ins(1,3,4) P₃. A study of D. discoideum PtdIns P-specific phospholipase C nulls, incapable of synthesizing Ins(1,4,5) P₃ via the PtdIns intermediate pathway (Fig. 1C, step 7), revealed that Ins(1,4,5) P₃ pools and the signaling processes using it were maintained via a PtdIns P-independent, alternative pathway involving breakdown of Ins P₆ (Van Dijken et al., 1995). These studies illustrate the metabolic adjustment and balancing that the Ins polyP and PtdIns P pathways are capable of in vivo. In the case of maize lpa2-1, the novel accumulation of D-Ins(1,2,4,5,6) P₅ and/or its enantiomer may occur in compensation for a block in Ins P₆ synthesis. Perhaps D-Ins(1,2,4,5,6) P₅ along with D-Ins(1,2,3,4,6) P₅ and phytic acid together represent a "buffer pathway," functioning as a complex pool for Ins phosphate (Fig. 1C).

As one approach to the nutritional and environmental problems attributed to seed-derived dietary phytic acid (Erdman, 1981; Cromwell and Coffey, 1991), efforts are under way to breed "low phytic acid" crops using lpa mutants. The initial efforts to breed elite maize "low phytic acid" inbreds and hybrids used lpa1-1 and simple backcrossing methods (Ertl et al., 1998). The HIP phenotype of this mutant provided a quick, inexpensive, and accurate test for its inheritance, greatly facilitating introgression of the trait into numerous breeding lines. Fourteen "near-isogeneic" hybrid pairs were produced, each consisting of sibling non-mutant and lpa1-1 variants. Field studies of these found little or no effect of homozygosity for the lpa1-1 allele on germination, or on stalk strength, grain moisture at harvest, and flowering date. However, yield reductions were observed in eight of the 14 hybrid pairs. When meaned across the 14 pairs, a yield reduction of 5.5% was observed.

This yield loss in lpa1-1 hybrids may be due in part to the inheritance of deleterious factors inherited from the "Early-ACR" parent, closely linked in coupling to lpa1-1 ("linkage drag"). However, in the present study the seed dry weight loss was observed for both mutants between sibling seed classes on individual ears and within the Early-ACR genetic background. Linkage drag is therefore probably not the major cause of this seed-specific effect. Blocks in either Ins supply (lpa1-1), or Ins phosphate metabolism (lpa2-1), may contribute in part to this dry weight loss. The seed dry weight loss may also in part be a direct outcome of the increase in P_i concentration that results from each mutant's block in phytic acid synthesis. For example, the rate-limiting step in starch synthesis in the cereal seed is catalyzed by the enzyme ADP-Glc pyrophosphorylase, and this enzyme is allosterically inhibited by P_i (Plaxton and Preiss, 1987). This hypothesis is supported by the fact that the dry-weight loss was inversely proportional to the increase in P_i in lpa1-1 and lpa2-1 seed. The level of yield reduction observed in the study of lpa1-1 hybrids and its variability closely reflects the seed dry weight reduction associated with homozygosity for lpa1-1 observed in the present study. It is therefore also most likely a direct outcome of the genetic lesion and its mutant phenotype. Studies to address this phenomenon and breeding efforts to overcome it are currently under way.

Previous studies have observed substantial variation in seed total P concentration among different non-mutant lines of a given species grown in the same environment (for review, see Raboy, 1997). Variation in seed total P concentration can also result from varying levels of nutrient P supply to the developing plant. During the development of normal seeds total P content (net total P) typically increases in a linear fashion (Raboy, 1997). In each of these three cases phytic acid P accumulation varies in turn to maintain a relatively constant non-phytic acid P, or "cellular P," level (defined as all P necessary for basic cellular metabolism). In this context phytic acid P is seen as excess P or storage P (all seed P over and above that needed for cellular metabolism). However the present studies of lpa1-1 and lpa2-1 indicate that it is not solely phytic acid P but the sum of phytic acid P and P_i that represents excess or storage P. It is this sum that remains relatively constant across the genotypes and developmental stages studied here. It remains to be determined how the relative contributions of phytic acid P and P_i to their sum might vary in response to variation in the supply of P to the developing seed of these genotypes.

That both mutants result in seed dry weight loss suggests that phytic acid metabolism is at least in part a component of P homeostasis during seed development. However, both mutants are viable as homozygotes and at least in the case of lpa1-1 have little effect on seed function other than a relatively minor loss in dry weight accumulation. Therefore if P homeostasis is critical to seed function, some second mechanism not involving phytic acid metabolism, such as a combination of localization and compartmentalization of P, must play the major role. In light of the other possible functions for phytic acid metabolism, such as an Ins phosphate pool important for signaling pathways and possibly ATP regeneration (Menniti et al., 1993; Van der Kayy et al., 1995) or as an anti-oxidant (Graf et al., 1987), it is surprising that lpa1-1 and lpa2-1 seeds are in fact viable and are essentially normal in phenotype other than in their seed P chemistry. Perhaps the major function for phytic acid accumulation in seeds is as an efficient P-storage metabolite. Under cultivation, long-term sequestering of P in seeds may not be essential. However, efficient storage of P may be essential in the natural environment where plants evolved, where seeds must survive in soils for extended periods. The impact of the change in seed storage P chemistry (phytic acid P to P_i) in lpa mutants, on this long-term P storage function, remains to be determined.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION MATERIALS AND METHODS LITERATURE CITED

Plant Materials

A population of ethyl methanesulfanate-induced mutants was generated using the pollen-treatment method (Neuffer and Coe, 1978). The main maize (Zea mays) stock used for these studies, a synthetic population referred to as "Early-ACR," was kindly provided by Dr. M.G. Neuffer (University of Missouri, Columbia). In addition, an F₂ obtained from the cross of the public inbred lines A632 and Mo17 was also used as a pollen parent for some of the mutagenesis treatments. Treated pollen was applied to silks of 54 untreated Early-ACR plants, producing M₁ seeds heterozygous for induced mutations. These were planted and self-pollinated to produce 872 M₂ progenies each consisting of sibling seeds on a single M₂ ear. We screened for M₂s segregating for seeds with reduced phytic acid P or unusual increases of other Ins phosphates or P_i, as compared with that typical of non-mutant seeds. Five or more seeds that appeared phenotypically normal or non-mutant to the unaided eye were sampled from each M₂ ear, individually crushed with a hammer blow, and incubated over-night at 4°C in 0.4 M HCl (10 µL per mg seed weight). The extracts were then briefly vortexed and allowed to settle for a minimum of one-half h. Aliquots were fractionated using a HVPE assay for acid-extractable P-containing compounds (Raboy et al., 1990). Standards were Na phytate (Sigma, St. Louis) and a mixture of Ins phosphates and P_i produced by the chemical hydrolysis of phytic acid (Raboy et al., 1990).

Remnant seed from M₂s containing putative mutants were planted in a field nursery. The resulting plants were self-pollinated to produce M3 ears and cross-pollinated onto non-mutant Early-ACR lines to produce F₁ ears. To provide materials for quantitative analyses and to test allelic relationships, M3 and F3 homozygotes were identified, seeds were planted, and the resulting plants were self- or sib-pollinated and intercrossed. For analyses of P fractions during seed development, immature ears representing each genotype were harvested at 15, 30, and 40 DAP, frozen in liquid N₂, and stored at 80°C. Ears harvested at maturity were dried at 40°C for 48 h and stored at 4°C. To provide materials for inheritance studies, F₁ heterozygote seeds were planted in field nurseries, and the resulting plants were either self-pollinated to produce F₂ progenies or used in test-crosses to respective mutant homozygote testers.

Chromosomal-Mapping Experiments

B-A translocation stocks were used to map the first two mutants, lpa1-1 and lpa2-1, to chromosome arm (Beckett, 1978). B-A translocations undergo non-disjunction during the developing microspore's second mitotic division, producing male gametes containing two sperm nuclei. One of the sperm nuclei is hyperploid (containing two copies of the A chromosome segment contained in the translocation) and one is hypoploid (containing no copies of the translocated segment; Beckett, 1978). Preferential fertilization by the hyperploid sperm typically occurs in approximately 66% of zygotes. Therefore, if the frequency of non-disjunction approaches 100%, the frequency of fertilization by a hypoploid sperm will approach 33%. The bulk of seed phytic acid P is localized in the diploid embryo. Fertilization of an egg produced by an lpa homozygote by a sperm hypoploid for the corresponding chromosome segment will result in a germ hemizygous for the mutant allele, "uncovering" the mutant phenotype. This indicates that the mutant locus was contained on the A chromosome fragment contained in the translocation. We crossed lpa1-1 and lpa2-1 homozygotes by a collection of B-A translocations, and analyzed the resulting seeds for their respective mutant phenotypes.

For these and other genetic analyses, we followed the inheritance of lpa1-1 or lpa2-1 via testing for the HIP phenotype associated with homozygosity for either mutant. Single seeds were weighed, crushed, and extracted overnight in 10 (v/w) 0.4 M HCl at 4°C and 10 µL of extract were assayed for P_i using the method of Chen et al. (1956), modified to be conducted in microtitre plates. To each microtitre plate well were added 10 µL of extract, 90 µL distilled, deionized water, and 100 µL of colorimetric reagent consisting of a 1:1:1:2 mixture of 10% (w/v) ascorbic acid:6 N H₂SO₄:2.5% (w/v) ammonium molybdate:distilled, deionized water. Each microtitre plate also contained five wells prepared to contain the following P standards: 0.0 µg P; 0.15 µg P; 0.46 µg P; 0.93 µg P; 1.39 µg P. Following development for 2 h at ambient temperature, results were obtained either via visual inspection of the plates or quantified via use of a microtitre-plate spectrophotometer. Depending on the study, selected extracts were also tested with HVPE to confirm the correspondence of HIP with the HVPE phenotype of either lpa1-1 or lpa2-1.

The mutants were then mapped in segregating F₂-mapping populations using RFLPs. F₂ seed were obtained from a cross of a homozygous lpa1-1 plant (Early-ACR or "E" background) and the inbred PHP38 (Pioneer or "P" background) and from a cross of a homozygous lpa2-1 plant ("E" background) and the inbred PHN46 ("P" background), and planted in a field nursery. The inbred lines and initial crosses were kindly provided by Pioneer Hi-Bred International (Des Moines, IA). DNAs were prepared from leaf samples obtained from each individual in the F₂ populations (Dellaporta et al., 1983). F₂ plants were then self-pollinated to produce F3 progeny ears. These F3 progenies were then tested to determine parent F₂ plant lpa genotype: homozygous normal (+/+); heterozygous (+/lpa or +/); homozygous mutant (lpa/lpa, or /). A bulk-segregant analysis was first conducted to identify linkage to RFLP markers (Michelmore et al., 1991). Three "bulk" DNAs were prepared to represent each of the three lpa F₂ genotypic classes by combining aliquots of DNA from all of the individuals representing a given class. These bulk DNAs were digested with restriction endonucleases, fractionated on agarose gels (Southern, 1975), and probed with RFLP markers kindly provided by the Maize RFLP Lab (Dr. Edward Coe, University of Missouri, Columbia). If a scorable polymorphism at a given RFLP locus exists between the "E" and "P" parental backgrounds, producing "E" and "P" alleles, and if this RFLP locus is linked to an lpa locus, then as the proximity of linkage increases an increase in signal in the "E" allele relative to the "P" allele will be observed in the / bulk, the reverse will be observed in the +/+ class, and similar levels of signal in "E" and "P" alleles will be observed in the heterozygous +/ class. If there is no linkage between an RFLP locus and the lpa locus then similar amounts of signal in both the "E" and "P" alleles will be observed in tests of the three class bulks. In the case of lpa1-1, F₂ DNAs representing the individuals comprising the two homozygous segregant classes were individually subjected to analysis. The data obtained were analyzed with MAPMAKER 3 (Lander et al., 1987).

Quantitative Analyses of Seed P and Inositol P Fractions

Samples of immature seeds were lyophilized. Samples of mature seeds were dried for 48 h at 60°C. These were then milled to pass through a 2-mm screen and stored in a desiccator until analysis. Seed total P was determined following wet-ashing of aliquots of tissue (typically 150 mg) and colorimetric assay of digest P (Chen et al., 1956). The ferric-precipitation method was used to determine total, acid-soluble Ins phosphates (Raboy et al., 1990). Aliquots of tissue (typically 0.5-1.0 gm) were extracted in 0.4 M HCl:0.7 M Na₂SO₄. Acid-soluble Ins phosphates were obtained as a ferric precipitate, wet-ashed, and assayed for P as in the total P analysis. Phytic acid or Ins phosphates are expressed as their P (atomic weight = 31) content to facilitate comparisons between seed P fractions. Seed P_i was determined colorimetrically following extraction of tissue samples (typically 0.5 g in non-mutant seeds and 0.15 g in mutant seeds) in 12.5% (w/v) trichloroacetic acid:25 mM MgCl₂.

Anion-exchange HPLC analyses of seed Ins phosphates were performed using a modification of the method as described (Phillippy and Bland, 1988; Rounds and Nielsen, 1993). Samples of seeds were dried and milled as described above, and extracted in 40 volumes 0.4 M HCl overnight. Following centrifugation (10,000g, 10 min), supernatants were filtered through number 1 filter paper (Whatman, Clifton, NJ), and passed through HV 0.45-µm filters (Millipore, Bedford, MA). Two hundred-microliter aliquots were then fractionated on an IonPac AS7 anion-exchange column (Dionex, Sunnyvale, CA), equipped with an IonPac AG7 guard column (Dionex), which had been equilibrated with 10 mM methyl piperazine, pH 4.0 (buffer A). The Ins phosphates were then eluted with the following gradient system at a flow rate of 0.5 mL min¹: 0 to 1 min 100% (v/v) buffer A; 1 to 26 min a concave gradient from 0% to 15% 1 M NaNO₃, pH 4.0 (buffer B); 21 to 41 min a linear gradient from 15% to 100% (v/v) buffer B. The column elutent was mixed with colorimetric reagent (0.015% [w/v] FeCl₃:0.15% [w/v] sulfosalicylic acid) at a flow rate of 0.5 mL min¹, using a PEEK tee and a Lazar pulseless pump (Alltech, Deerfield, IL), and the mixture passed through a 290-cm reaction coil prior to peak detection via A₅₅₀. Ins phosphate in a sample peak was calculated using the following standard curve, obtained via the analysis of four Na Ins P₆ standards containing 24.9, 49.7, 74.6, and 99.5 nM Na Ins P₆; nM Ins P = 1.66 × 10⁵ (peak area)  3.85; R² = 0.99.

Purification of Inositol Phosphates in lpa2-1 Seeds and Structural Identification Using NMR

The objective was to purify to homogeneity the most abundant Ins phosphates, other than phytic acid P, found in maize lpa21 seed, and then to determine their structures using NMR. One hundred grams of seed homozygous for lpa2-1 was ground with a coffee grinder and extracted in 1 L of 0.4 M HCl overnight. Extracts were centrifuged (10,000g, 10 min), and Ins phosphates were obtained as a ferric precipitate with a modification of the method as described above. Ferric Ins phosphate precipitates were converted to soluble Na Ins phosphate salts by treatment with NaOH, and the insoluble ferric hydroxide was removed via centrifugation. To obtain individual Ins phosphates in pure form, the supernatants containing the Na Ins phosphates were neutralized with HCl and loaded onto preparative Dowex (Sigma) 1X2-400 anion-exchange columns (packed volume 5 mL). These were eluted with a 400-mL 0.0 to 0.4 M HCl linear gradient or a 400-mL 0.4 M HCl isocratic gradient and collected in 5-mL fractions. Fractions containing Ins phosphates were identified following acid digestion of fraction aliquots, and colorimetric assay for P in the digests. Ins phosphates in peak fractions were precipitated as barium salts, and then converted to free acids via passage through AG 50W-X8 cation exchange columns. The purity of a given sample was confirmed with HVPE and HPLC (data not shown) and subsequently NMR. Aliquots of these free acids were then dehydrated in a Speed-Vac Concentrator (Savant Instruments, Holbrook, NY).

The structures of these Ins phosphates were determined by a combination of one- and two-dimensional NMR spectroscopy. NMR characteristics that are particularly useful for structure determination of Ins phosphates have been previously described (Barrientos et al., 1994; Johnson et al., 1995; Barrientos and Murthy, 1996). NMR spectra were recorded on a 400-MHz Unity Inova-400 spectrometer (Varian, Palo Alto, CA). The dehydrated samples (0.002-0.2 g) were dissolved in D₂O (0.8 mL), and the pH adjusted to 5.0 by addition of NaOH (1 M) or perdeuterated acetic acid, as necessary. The pH values reported in this paper are readings of the glass electrode, uncorrected for deuterium effects. One-dimensional ¹H-NMR spectra were obtained at 399.943 MHz. ¹H-Chemical shifts were referenced to the residual proton absorption of the solvent, D₂O ( 4.67). The acquisition conditions were as follows: spectral windows 6,738 Hz, pulse width 90° tipping angle. Typically, 16 to 32 scans with recycle delays of 4 to 6 s between acquisitions were collected. The residual H₂O resonance was suppressed by a 1.5-s selective presaturation pulse. ³¹P-decoupled spectra were decoupled continuously with Waltz decoupling. TOCSY, DQCOSY, and J-resolved spectra were obtained as described previously (Barrientos et al., 1994; Johnson et al., 1995; Barrientos and Murthy, 1996).