Azospirillum brasilense and Azospirillum lipoferum Hydrolyze Conjugates of GA₂₀ and Metabolize the Resultant Aglycones to GA₁ in Seedlings of Rice Dwarf Mutants

Abstract

Azospirillum species are plant growth-promotive bacteria whose beneficial effects have been postulated to be partially due to production of phytohormones, including gibberellins (GAs). In this work, Azospirillum brasilense strain Cd and Azospirillum lipoferum strain USA 5b promoted sheath elongation growth of two single gene GA-deficient dwarf rice (Oryza sativa) mutants, dy and dx, when the inoculated seedlings were supplied with [17,17-²H₂]GA₂₀-glucosyl ester or [17,17- ²H₂]GA₂₀-glucosyl ether. Results of capillary gas chromatography-mass spectrometry analysis show that this growth was due primarily to release of the aglycone [17,17-²H₂]GA₂₀ and its subsequent 3β-hydroxylation to [17,17-²H₂]GA₁ by the microorganism for the dy mutant, and by both the rice plant and microorganism for the dx mutant.

Azospirillum spp. are considered to be important plant growth promotive rhizobacteria that can improve the growth and yield of at least several plant species (Okon and Labandera-González, 1994). However, the mechanism by which Azospirillum spp. and other promotive rhizobacteria promote plant growth has yet to be elucidated (Glick et al., 1999, and literature cited therein). Phytohormone production (Tien et al., 1979; Okon and Kapulnik, 1986), including gibberellins (GAs; Bottini et al., 1989; Fulchieri et al., 1993; Lucangeli and Bottini, 1997), is one mechanism that has been proposed.

GAs are a class of phytohormones with many demonstrated effects on a number of physiological processes (Davies, 1995). Among the 130 GAs identified up to now from plants, fungi, and bacteria are GA₁, GA₃, and GA₄, the three most common directly effective GA shoot elongation promoters. Their levels in plant tissues appear to be regulated by three processes: (a) biosynthesis, (b) reversible conjugation, and (c) catabolism. The activation step in the biosynthesis of growth-promotive GAs is 3β-hydroxylation, i.e. conversion of the 3-deoxy GA₂₀ into the bioactive GA₁ (Kobayashi et al., 1994). Conjugation of GAs is almost exclusively with Glc, either via the carboxyl group forming glucosyl esters or via the hydroxyl group generating glucosyl ethers in a range of isomeric forms. These glucosyl conjugates do not appear to be biologically active per se; rather, their role is proposed to be as reserve, transport, or entry to catabolism forms (Schneider and Schliemann, 1994).

GA production in vitro by Azospirillum spp. and the effects of Azospirillum spp. on infected plants have been studied earlier. Thus, GA₁ and GA₃ were characterized by capillary gas chromatography-mass spectrometry (GC-MS) from gnotobiotic cultures of Azospirillum lipoferum (Bottini et al., 1989) and Azospirillum brasilense (Janzen et al., 1992). In addition, the bacteria have also been shown to metabolize exogenous GAs (Piccoli and Bottini, 1994; Piccoli et al., 1996). External factors such as light quality (Piccoli and Bottini, 1996) or oxygen availability and osmotic strength (Piccoli et al., 1999) may also influence the amount and type of GA produced. It was also found that application of GA₃ in concentrations similar to those produced by the microorganism, or by inoculation with different Azospirillum spp. strains, can promote growth of roots in corn (Zea mays) seedlings (Fulchieri et al., 1990). It is interesting that inoculation with A. lipoferum promoted the appearance of free forms of GA₃ in inoculated plant roots, whereas putative GA conjugates predominated in controls (Fulchieri et al., 1993). The latter results, however, did not establish whether the higher content of GA₃ found in the inoculated roots was due to de novo production, hydrolysis by the microorganism per se, or by the plant under bacterial influence. Regarding this issue, Piccoli et al. (1997) demonstrated that A. lipoferum grown in chemically defined medium has the capacity to hydrolyze in vitro [17,17-²H₂] GA₂₀ glucosyl ester (GA₂₀-GE) and [17,17-²H₂] GA₂₀ glucosyl ether (GA₂₀-G).

In a number of rice (Oryza sativa) cultivars the dwarf phenotype is expressed in several single-gene mutants. One mutant, dx, blocks at the level of kaurene oxidation; a second dwarf, dy, blocks the 3β-hydroxylation step (Takahashi and Kobayashi, 1991; Kobayashi et al., 1994).

We report herein the bacterial hydrolysis and metabolism of GA conjugates in these two dwarf mutants of rice by endophytic Azospirillum spp.

RESULTS

Reversal of the Dwarf (dy) Phenotype in Rice

The responses of the dy rice mutant seedlings treated with [17,17-²H₂]GA₂₀-G and [17,17-²H₂]GA₂₀-GE at several concentrations and/or inoculated with A. brasilense strain Cd and A. lipoferum strain USA 5b are shown in Table I. Both strains were effective in promoting growth of seedlings and reversing dwarfism when the two types of GA₂₀-glucosyl conjugates were added. The promotive effects were similar for each of the two forms of GA₂₀-glucosyl conjugates. Inoculated control plants (but without GA₂₀ conjugate application) also showed dwarfism reversal, having a significantly greater growth than the ethanol-treated controls. Treatments with deuterated GA glucosyl conjugates alone (no Azospirillum spp.) could also slightly reverse dwarfism, relative to ethanol control treatments (Table I). Thus, it seems likely that the plant has some intrinsic capacity to hydrolyze GA-glucosyl conjugates. The difference in growth for inoculated versus noninoculated seedlings can best be explained as hydrolytic plus metabolic activity of the bacteria on the conjugate. Although dwarfism reversal was observed in treatments with each addition of GA-glucosyl conjugates, or with independent inoculation, the greatest growth increases were observed for those seedlings that received the conjugated hormones as well as the microorganism. There were no significant differences between seedlings treated with [17,17-²H₂]GA₂₀-G or [17,17-²H₂]GA₂₀-GE. Thus, no particular Azospirillum spp. specificity for substrate was found.

Table I.

Length of second leaf sheath (in mm, ±p < 0.05 confidence limits) of rice dy seedlings treated with GA A₂₀ glucosyl conjugates and inoculated with Azospirillum spp.

Treatments	Noninoculated	A. lipoferum Strain USA 5b	A. brasilense Strain Cd
[17,17-2H2] GA20-G (μg plant−1)
10−1	12.02 ± 0.63	18.20 ± 1.46	15.60 ± 0.48
10−2	11.10 ± 0.89	13.30 ± 0.47	10.80 ± 0.40
10−3	10.40 ± 0.48	12.00 ± 1.09	10.60 ± 0.80
[17,17-2H2] GA20-GE (μg plant−1)
10−1	11.80 ± 0.40	17.20 ± 1.46	17.40 ± 0.80
10−2	11.60 ± 0.80	13.40 ± 0.48	13.40 ± 1.01
10−3	11.00 ± 0.63	11.75 ± 0.82	11.00 ± 0.89
Ethanol 95% (v/v) + buffer	9.33 ± 0.74	10.60 ± 0.80	10.40 ± 0.48

  

Reversal of the Dwarf (dx) Phenotype in Rice

The responses of the rice dx mutant treated with decreasing concentrations of [17,17-²H₂]GA₂₀-G and [17,17-²H₂]GA₂₀-GE and inoculated with both A. brasilense strain Cd and A. lipoferum strain USA 5b are shown in Table II. Internode length of inoculated seedlings was significantly longer relative to noninoculated treatments for the higher concentration of [17,17-²H₂]GA₂₀-G and [17,17-²H₂]GA₂₀-GE. A. brasilense strain Cd and A. lipoferum strain USA 5b were effective in promoting growth of seedlings and reversing dwarfism when deuterated glucosyl conjugates were added, relative to noninoculated treatments. A. brasilense strain Cd and A. lipoferum strain USA 5b (no GA₂₀ conjugates) showed similar growth-promoting effects. Again, as for the dy mutant, although reversal of dwarfism was observed in treatments with addition of hormones or with independent inoculation, the greatest growth differences were observed in those seedlings that received hormones and microorganism together. Table II also shows that for the dx mutant there was an apparent preference for the glucosyl ester as substrate.

Table II.

Length of second leaf sheath (in mm, ±p < 0.05 confidence limits) of rice dx seedlings treated with GA A₂₀ glucosyl conjugates and inoculated with Azospirillum spp.

Treatments	Noninoculated	A. lipoferum Strain USA 5b	A. brasilense Strain Cd
[17,17-2H2] GA20-G (μg plant−1)
10−1	17.60 ± 1.85	21.80 ± 1.32	23.60 ± 1.35
10−2	12.60 ± 0.48	14.20 ± 0.97	13.40 ± 0.80
10−3	10.00 ± 0.63	10.40 ± 0.48	12.00 ± 0.63
[17,17-2H2] GA20-GE (μg plant−1)
10−1	21.50 ± 0.50	29.80 ± 1.32	30.80 ± 1.46
10−2	15.60 ± 0.48	17.40 ± 0.48	17.00 ± 0.89
10−3	10.40 ± 1.01	11.20 ± 0.74	11.40 ± 1.35
Ethanol 95% (v/v) + buffer	12.30 ± 0.50	13.60 ± 1.01	13.40 ± 0.74

The dx mutant was much more efficient than the dy mutant in utilizing the glucosyl conjugates of [17,17²H₂]GA₂₀ for sheath elongation growth. It is likely due to a higher inherent capacity of the dx mutant to 3β-hydroxylate the aglycone, [17,17²H₂]GA₂₀, to the bioactive [17,17²H₂]GA₁, i.e. the genetic lesion in the dy mutant results in a very much reduced ability to 3β-hydroxylate 3-deoxy GAs (Takahashi and Kobayashi, 1991; Kobayashi et al., 1994).

Endophytic Bacteria

Endophytic presence of the bacteria in roots and stems of inoculated plants was shown for both cultivars of all experiments (summarized in Table III).

Table III.

Bacterial counts 72 h after inoculation

Strain Inoculated	Fraction	dx, Colony Forming Units (CFU) g plant−1	dy, CFU g plant−1
A. lipoferum strain USA 5b	Root	6.106	3.106
A. lipoferum strain USA 5b	Stem + leaves	3.104	5.104
A. brasilense strain CD	Root	4.107	2.107
A. brasilense strain CD	Stem + leaves	5.105	7.105
Buffer phosphate	Root + stem + leaves	Nda	Nd

^aNd, Not detected. 

Identification of GAs and Their Quantitation by GC-MS

Identification of [17,17-²H₂]GA₂₀ (the aglycone of the conjugates applied) and [17,17-²H₂]GA₁ (the 3β-hydroxylated metabolite of [17,17-²H₂]GA₂₀) was based on the peak area counts of eight diagnostic ions (Table IV). Both [17,17-²H₂]GA₂₀ and its metabolite [17,17-²H₂]GA₁ were identified for the two rice mutants.

Table IV.

Gibberellin identification by GC-MS based on relative abundance (in %) of characteristic ions

Compound	Abundance in % of Characteristic Ions
[17,172H2]GA20 standard	420 (100), 405 (13), 377 (66), 361 (18), 347 (5), 303 (20), 237 (9), and 209 (43)
[17,172H2]GA20 putative	420 (100), 405 (14), 377 (62), 361 (17), 347 (7), 303 (21), 237 (12), and 209 (57)
[17,172H2]GA1 standard	508 (100), 493 (9), 450 (21), 378 (20), 315 (11), 239 (12), 225 (11), and 209 (44)
[17,172H2]GA1 putative	508 (100), 493 (10), 450 (21), 378 (17), 315 (14), 239 (13), 225 (10), and 209 (58)

Rice seedlings had been treated with GA-A₂₀-glucosyl conjugates and inoculated with Azospirillum spp.

Quantification of [17,17-²H₂]GA₂₀ (the aglycone) and [17,17-²H₂]GA₁ (its 3β-hydroxylated metabolite) were made using the absolute areas of the molecular ions (M⁺) at m/z 420 and 508 by GC-MS for the dy seedlings, with or without Azospirillum spp. inoculation (Table V). There is a high correlation between the application of deuterated GA conjugates and the growth response of the rice mutant seedlings in the presence of Azospirillum spp. It was a coincidence that [17,17-²H₂]GA₂₀ and [17,17-²H₂]GA₁ M⁺ peak areas were higher for those treatments where inoculation with Azospirillum spp. was a cotreatment (Tables V and VI). Thus, it is apparent that inoculation with the microorganism allows for deconjugation (release of the aglycone [17,17-²H₂]GA₂₀) and its metabolism to [17,17-²H₂]GA₁ in both dx and dy mutants of rice seedlings. These results also prove that the bacteria can 3β-hydroxylate GA₂₀ because in noninoculated [17,17-²H₂]GA₂₀ glucosyl conjugate treatments of dy, [17,17-²H₂]GA₁ was not identified (Table V) and the dy mutant seedlings showed very little growth (Table I). However, dx seedlings without Azospirillum spp. inoculation showed fair uptake and good conversion to [17,17-²H₂]GA₁ (Table VI), as well as good growth (Table II). It is noteworthy that GA₁ levels are higher in the Azospirillum-inoculated dx mutant after treating with GA₂₀-G than after treating with GA₂₀-GE (Table VI). This seems to be inconsistent with the internode length results obtained (Table II); i.e. GA₂₀-GE was more effective than GA₂₀-G to induce the second leaf sheath growth in the Azospirillum-inoculated dx mutant. However, this could be because the GA measurements were done using whole seedlings.

Table V.

Absolute areas of peak abundance for the M⁺ of [17,17²H₂]GA₂₀ (m/z 420) and [17,17-²H₂]GA₁ (m/z 508) obtained by GC-MS for dy seedlings (equal sample aliquots were injected; see “Materials and Methods” for details)

Treatment (10−1 μg plant−1)	M+ 420 Area × 103 ([2H2]GA20)	M+ 508 Area × 103 ([2H2]GA1)	Total Peak Areas × 103 of ([2H2]GA20) + ([2H2]GA1)
[2H2]GA20-G without Azospirillum spp.	73	Nda	73
[2H2]GA20-GE without Azospirillum spp.	67	Nd	67
[2H2]GA20-G +A. lipoferum strain USA5b	570	140	710
[2H2]GA20-GE +A. lipoferum strain USA5b	480	360	840
[2H2]GA20-G +A. brasilense strain Cd	1,100	160	1,260
[2H2]GA20-GE +A. brasilense strain Cd	400	130	530

Results represent a typical subsample from one replicate, injected several times without noticeable differences.

^aNd, Not detected. 

Table VI.

Absolute areas of peak abundance for the M⁺ of [17,17²H₂]GA₂₀ (m/z 420) and [17,17-²H₂]GA₁ (m/z 508) obtained by GC-MS for dx seedlings (equal sample aliquots were injected; see “Materials and Methods” for details)

Treatment (10−1 μg plant−1)	M+ 420 Area × 103 ([2H2]GA20)	M+ 508 Area × 103 ([2H2]GA1)	Total Peak Areas × 103 of ([2H2]GA20) + ([2H2]GA1)
[2H2]GA20-G without Azospirillum spp.	7	45	52
[2H2]GA20-GE without Azospirillum spp.	12	260	272
[2H2]GA20-G +A. lipoferum strain USA5b	2,100	410	2,510
[2H2]GA20-GE +A. lipoferum strain USA5b	310	78	388
[2H2]GA20-G +A. brasilense strain Cd	530	110	640
[2H2]GA20-GE +A. brasilense strain Cd	240	38	278

Results represent a typical subsample from one replicate, injected several times without noticeable differences.

DISCUSSION

Sheath elongation growth of two single gene dwarf rice mutants, dy and dx, can be promoted when seedlings inoculated with A. brasilense strain Cd and A. lipoferum strain USA 5b are also supplied with [17,17-²H₂]GA₂₀-GE or [17,17-²H₂]GA₂₀-G. This growth is due primarily to release of the aglycone, [17,17-²H₂]GA₂₀, and its subsequent 3β-hydroxylation to [17,17-²H₂]GA₁ by the microorganism for the dy mutant, and by both the plant and microorganism for the dx mutant. This mutant retained the ability to 3β-hydroxylate GA₂₀. Hence, Azospirillum spp. can effectively hydrolyze GA-glucosyl conjugates in vivo. This confirms earlier work by Piccoli et al. (1997) where in vitro cultures of A. lipoferum effectively released the GA aglycone in chemically defined medium.

The identification of [17,17-²H₂]GA₁ only in inoculated treatments of dy seedlings (Table II) further demonstrates that Azospirillum spp. has the in vivo capacity to 3β-hydroxylate [17,17-²H₂]GA₂₀ to [17,17-²H₂]GA₁ because the dy mutant is blocked in this biosynthetic step. As with the dx mutant, the current results confirm, in vivo, the 3β-hydroxylation abilities of A. lipoferum previously shown in vitro (Piccoli and Bottini, 1994; Piccoli et al., 1996). The finding that only [17,17-²H₂]GA₁ was a metabolic product of [17,17-²H₂]GA₂₀ (the deconjugated aglycone) strongly implies that GA₁ and GA₃ have different precursors in the route of synthesis in Azospirillum spp. (Piccoli and Bottini, 1996; Piccoli et al., 1996).

As expected, dx seedlings, with or without inoculation, that were treated with [17,17-²H₂]GA₂₀-GE and [17,17-²H₂]GA₂₀-G showed greater growth when compared with similarly treated dy seedlings (Tables I and II). Seedlings of dx also produced [17,17-²H₂]GA₁ from [17,17-²H₂]GA₂₀ in noninoculated treatments.

These responses by the dx seedlings are consistent with earlier studies (Takahashi and Kobayashi, 1991; Kobayashi et al., 1994) that place the dwarfing lesion of the dx mutation early in the GA biosynthesis pathway. In a similar manner, the inability of the dy seedlings to grow or to produce [17,17-²H₂]GA₁ without inoculation again confirm that this dwarf has its genetic lesion at the later, 3β-hydroxylation step of GA biosynthesis.

In conclusion, the beneficial effect of Azospirillum spp. on growth and yield (Okon and Labandera-González, 1994) or water stress alleviation (Creus et al., 1997) of graminaceous plants can likely be explained, at least in part, by: (a) GA production by the bacteria (Kucey, 1988; Bottini et al., 1989; Fulchieri et al., 1990; Janzen et al., 1992; Fulchieri et al., 1993; Lucangeli and Bottini 1997; Piccoli et al., 1999), (b) deconjugation of GA-glucosyl conjugates (Piccoli et al., 1997), and (c) 3β-hydroxylation of inactive 3-deoxy GAs present in roots to active forms (Kobayashi et al., 1994; Piccoli and Bottini, 1994; Piccoli et al., 1996).

MATERIALS AND METHODS

Biological Material

The following bacteria were used: Azospirillum lipoferum strain USA5b (kindly provided by Dr. Vera Baldani, EMBRAPA, Itajai, Brazil) and Azospirillum brasilense strain Cd (ATCC accession no. 29710); and the dwarf rice (Oryza sativa) mutants dy (rice cv Waito C) and dx (rice cv Tan-ginbozu; mutants were gifts from Dr. Masaji Koshioka, National Research Institute of Vegetables, Ornamental Plants and Tea, Mie, Japan).

Bacterial Growth

Bacterial strains were grown in liquid nitrogen-free biotin-based (NFb) medium with malic acid (5 g L⁻¹) and NH₄Cl (1.25 g L⁻¹; Piccoli et al., 1997) on an orbital shaker at 30°C and 80 rpm, until an optical density at 540 nm of 1.0 was reached. This corresponds with a concentration of approximately 10⁸ CFU mL⁻¹. Bacteria were harvested by centrifugation at 8,000 rpm and 4°C for 15 min. The cellular pellet was washed twice with 0.85% (w/v) NaCl solution and resuspended in 0.05 m phosphate buffer to obtain a titer of 3 × 10⁶ CFU mL⁻¹ for later inoculation.

Deuterated GA Conjugates

The deuterated GA conjugates [17,17-²H₂] GA₂₀-glucosyl ester ([17,17-²H₂] GA₂₀-GE) and [17,17-²H₂] GA₂₀-glucosyl ether ([17,17-²H₂] GA₂₀-G) were prepared from [17,17-²H₂] GA₂₀ (96% [v/v] ²H₂) as previously described (Schneider et al., 1989, 1990).

Seedling Growth

Seeds of both rice mutants were surface sterilized with ethanol (70% [v/v]) for 20 s, then with sodium hypochlorite (2% [v/v]) for 20 min, finally being washed with sterile distilled water. They were pregerminated in 80 μm Uniconazole (S-3307D, Sumimoto Chem. Co., Nagoya, Japan), an early-stage GA biosynthesis inhibitor, at 30°C for 48 h to deplete their endogenous GA levels. Then after again washing with sterile distilled water, five seeds were sown in glass beakers containing 2 mL of agar (0.8% [w/v]) and incubated for 48 to 72 h at 30°C under continuous fluorescent light (20 μmol m⁻² s⁻¹) in a saturated humidity.

Treatments

After 72 h roots, of seedlings of both rice mutants were inoculated with the Azospirillum spp. strains using a titer of 3 × 10⁶ CFU plant⁻¹. Control plants received an equivalent amount of buffer. After another 72 h, [17,17-²H₂]GA₂₀-GE or [17,17-²H₂]GA₂₀-G, dissolved in 1 μL of ethanol (95% [v/v]), were carefully placed on the surface of the seedling roots with the aid of a microsyringe. Treatments are summarized as follows: treatment 1, 1 μL ethanol (95% [v/v]) + 0.05 m buffer phosphate; treatments 2 through 6, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-GE; treatments 7 through 11, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-GE + A. lipoferum strain USA 5b; treatments 12 through 16, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-GE + A. brasilense strain Cd; treatments 17 through 21, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-G; treatments 22 through 26, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-G + A. lipoferum strain USA 5b; treatments 27 through 31, 10⁻¹ to 10⁻⁵ μg plant⁻¹ of [17,17-²H₂] GA₂₀-G + A. brasilense strain Cd; treatment 32, A. lipoferum strain USA 5b; and treatment 33, A. brasilense strain Cd.

Measurements and Statistical Analysis

After 72 h in the presence of the deuterated GA conjugates, the length of the first internode was measured. Seedlings (roots + shoots) were frozen in liquid N₂ and stored at −30°C until processed. Each experiment was carried out in a random design, with five repetitions per treatment. Data were analyzed by analysis of variance (ANOVA) followed by Tukey's t test at P ≤ 0.05.

Bacterial Counts in Stems and Roots

Bacterial counts were made for both shoots and roots of control and treated rice seedlings using plates of NFb agar. Plant tissue was macerated using a mortar and pestle with 0.05 m phosphate buffer. Serially diluted inoculations were applied to plates containing agar NFb. After incubating for 72 h at 30°C, the number of CFU was assessed.

Evaluation of [17,17-²H₂] GAs

Seedlings (including roots) treated with 10⁻¹ μg plant⁻¹ of GA₂₀-G and GA₂₀-GE were homogenized 72 h after treatment in 500 mL of methanol:water (4:1) and kept at 4°C overnight. After filtration the solid residue was extracted again and filtered. Filtrates were combined and methanol evaporated under reduced atmospheric pressure. The aqueous phase was adjusted to a pH of 2.5 with acetic acid (1% [v/v]) and partitioned four times with the same volume of water-saturated (1% [v/v] acetic acid) ethyl acetate. Acidic ethyl acetate was evaporated and diluted with acetic acid:methanol:water (1:10:89), then filtered through 0.45-μm membranes and injected into an HPLC system (KNK-500, Konic Inc., Barcelona) with a C₁₈ reverse phase column (μBondapack, 300 × 3.9 mm, Waters Associates, Milford, MA). Elution was at 2 mL min⁻¹ using a gradient 10% to 73% (v/v) methanol in 1% (v/v) acetic acid over 30 min until the 73% (v/v) methanol concentration was reached. The fractions corresponding with the retention times of [17,17-²H₂]GA₂₀ and [17,17-²H₂]GA₁ were subjected to a second HPLC using a Nucleosil 5[N(CH₃)₂] column (15 cm × 4.6 mm, Waters Associates). Elution was carried out with acetic acid:methanol (0.1:99.9) at 1 mL min⁻¹. The fractions that would contain [17,17-²H₂]GA₂₀ and [17,17-²H₂]GA₁ were again collected, evaporated, and processed as described by Volmaro et al. (1998) prior to assessment by capillary GC-MS. The only change was that a Hewlett-Packard-1 cross-linked methyl silicone capillary column (25-m length × 0.25-mm i.d. × 0.22-μm film thickness) with a GC temperature program of 100°C to 195°C at 15°C min⁻¹, then to 260°C at 4°C min⁻¹, was used. For each analysis, full-scan spectra of [17,17-²H₂]GA₂₀ and [17,17-²H₂]GA₁ were monitored at the retention times of authentic [17,17-²H₂]GA₂₀ and [17,17-²H₂]GA₁ standards.