Soybean Lipoxygenase-1 Oxidizes 3Z-Nonenal

Abstract

In previous work with soybean (Glycine max), it was reported that the initial product of 3Z-nonenal (NON) oxidation is 4-hydroperoxy-2E-nonenal (4-HPNE). 4-HPNE can be converted to 4-hydroxy-2E-nonenal by a hydroperoxide-dependent peroxygenase. In the present work we have attempted to purify the 4-HPNE-producing oxygenase from soybean seed. Chromatography on various supports had shown that O₂ uptake with NON substrate consistently coincided with lipoxygenase (LOX)-1 activity. Compared with oxidation of LOX's preferred substrate, linoleic acid, the activity with NON was about 400- to 1000-fold less. Rather than obtaining the expected 4-HPNE, 4-oxo-2E-nonenal was the principal product of NON oxidation, presumably arising from the enzyme-generated alkoxyl radical of 4-HPNE. In further work a precipitous drop in activity was noted upon dilution of LOX-1 concentration; however, activity could be enhanced by spiking the reaction with 13S-hydroperoxy-9Z,11E-octadecadienoic acid. Under these conditions the principal product of NON oxidation shifted to the expected 4-HPNE. 4-HPNE was demonstrated to be 83% of the 4S-hydroperoxy-stereoisomer. Therefore, LOX-1 is also a 3Z-alkenal oxygenase, and it exerts the same stereospecificity of oxidation as it does with polyunsaturated fatty acids. Two other LOX isozymes of soybean seed were also found to oxidize NON to 4-HPNE with an excess of 4S-hydroperoxy-stereoisomer.

In the past animal researchers have spent considerable effort on 4-HNE research because of its cytotoxicity and mutagenicity (Esterbauer et al., 1991). More recently, 4-HNE has been implicated as a lipid signal because it activates phosphoinositide-specific phospholipase-C (Rossi et al., 1994) and phospholipase-D (Natarajan et al., 1993), and triggers Ca²⁺ influx in hepatocytes (Carini et al., 1996).

Despite the interest in 4-HNE, a biosynthetic pathway in animals has not been found. In plants the biosynthetic route originates with oxidation of linoleic acid by 9-specific LOX leading to NON by hydroperoxide lyase cleavage. Subsequently, NON is oxidized by a “3Z-alkenal oxygenase” and the resultant 4-HPNE is reduced by a hydroperoxide-dependent peroxygenase (Gardner et al., 1991; Gardner and Hamberg, 1993; Takamura and Gardner, 1996). In a second pathway, the higher oxidation state of 4-HPNE is utilized by peroxygenase to oxidize NON to 3,4-epoxynonanal, which subsequently rearranges into 4-HNE (Gardner and Hamberg, 1993). To our knowledge, the specific enzymes involved in 4-HNE formation have not previously been isolated and characterized. In an attempt to isolate the proposed first enzyme, a 3Z-alkenal oxygenase, we found strong evidence that 3Z-alkenal oxidizing activity was identical with soybean (Glycine max) LOX-1 activity. Additional data pointed to the possible existence of other NON oxidizing activity, including the other LOX isozymes of soybean. This research was communicated in part at Plant Biology '97 in Vancouver, British Columbia (Gardner and Grove, 1997).

MATERIALS AND METHODS

LOX-1 from soybean (Glycine max cv Williams) seeds was prepared by the method of Axelrod et al. (1981). The LOX-1 isolates were stored as a suspension in 2.3 m (NH₄)₂SO₄ at 3°C; protein concentration of the isolates used ranged from 12 to 30 mg/mL. NON was prepared from 3Z-nonen-1-ol by the procedure of Ratcliffe and Rodehorst (1970), and the product was purified by silica chromatography (Takamura and Gardner, 1996) followed by open-column chromatography on Sephadex LH-20 (4 × 23 cm) with hexane elution to remove contaminants, principally 3Z-nonenyl 3Z-nonenoate. 4-HNE was prepared as described (Gardner et al., 1992), and 4-ONE was synthesized from 4-HNE by oxidation with pyridinium chlorochromate (Corey and Suggs, 1975), followed by TLC isolation with hexane: diethyl ether (1:1, v/v) development (R_F = 0.68); UV λ_max = 216 nm.

Enzyme Assays

NON oxidation activity was measured by two methods. O₂-electrode monitoring measured initial activity, but the method suffered from considerable baseline activity of unknown origin attributable to the presence of the NON substrate. This pseudoactivity of the blank control had to be subtracted from activity of the enzyme assays, making precise measurements difficult. O₂-electrode conditions for routine assay of column fractions was 0.1 mm NON and 0.1 m potassium borate buffer, pH 9.0, at 25°C. The second method of assay employed GC separation with flame-ionization detection of diethyl ether-extracted products after a 5- or 15-min reaction time. The assay reactions were comprised of 1 mm NON, varying amounts of LOX-1, and 0.05 or 0.1 m potassium borate buffer, pH 8.3 or 8.6, in a total of 2 mL. The reaction solutions were incubated on ice and aerated with bubbling, pure O₂, and the reactions were terminated by shaking with 2 mL of diethyl ether. Aliquots of the diethyl ether solution were separated by flame-ionization detection GC.

Conveniently, there were usually three main peaks in the chromatograms, unoxidized NON (6.2 min), 2E-nonenal (7.6 min, from isomerization of NON, generally amounting to less than 5% of the NON peak), and “oxidized nonenal” (9.4 min). By extensive GC-MS of standards, it was found that the “oxidized nonenal” peak was comprised of two closely migrating peaks of 4-ONE and a rearrangement product of 4-HNE. GC-MS of a relatively large sample of underivatized 4-HNE separated into three peaks with mass spectra indicative of pentylfuran (minor peak with retention time = 4.8 min), an unknown rearrangement compound with a molecular ion of m/z 156 (retention time = 8.1 min), and 4-HNE (retention time = 9.6 min). The relative abundance of the three changed as a function of the amount of sample injected and the injector temperature, and the putative 4-HNE peak at 9.6 min disappeared completely when the sample size was decreased sufficiently. It was also ascertained that 4-HPNE decomposes in the GC to mainly 4-ONE plus a lesser amount of rearranged 4-HNE. Data was calculated as the percent of oxidized NON. 2E-Nonenal was included with NON as the unoxidized portion. To determine the composition of all components of the product mixture, selected samples were separated by straight-phase HPLC with a Microsorb silica column (250 × 4.6 mm, 5-μm spherical particle size from Rainin [Woburn, MA]) using hexane:isopropanol (197:3, v/v) monitoring peaks at 226 nm with a Spectra 100 detector (Spectra Physics Corp., San Jose, CA).

GC-MS and Flame-Ionization Detection GC

GC-MS was accomplished with a gas chromatograph (model 5890, Hewlett-Packard) interfaced with a mass selective detector (model 5971, Hewlett-Packard) operating at 70 eV. The capillary column used was an HP-5MS (Hewlett-Packard) cross-linked 5% phenyl methyl silicone, 0.25 mm × 30 m, film thickness, 0.25 μm. The aldehydes were separated by temperature programming from 65 to 260°C at a rate of 10°C/min. (–)-Menthoxycarbonyl derivatives were separated by programming from 160 to 260°C at 5°C/min (He flow rate = 0.67 mL/min).

Flame-ionization detection GC was accomplished with a model 5890 gas chromatograph equipped with a SPB-1 capillary column (30 m × 0.32 mm; film thickness, 0.25 μm) from Supelco (Bellefonte, PA). Underivatized aldehydes were separated using temperature programming from 65 to 165°C at 5°C/min. For separation of (–)-menthoxycarbonyl derivatives of methyl 2R- and 2S-hydroxyheptanoates, temperature was programmed from 160 to 260°C at 3°C/min (He flow rate = 2 mL/min).

Chiral Analysis

For chiral analysis 4-HPNE was produced by incubating 1 mm NON and 0.09 mm 13S-HPODE with 0.48 mg of LOX-1 in 8 mL of 50 mm potassium borate, pH 8.3, for 15 min on ice with an O₂ stream of bubbles. Incubations with the other LOX isozymes were similar, except the reaction volumes were doubled, and additionally, 0.1 m potassium Pipes, pH 6.5, was used. The amount of LOX-2 and LOX-3 used was 25 and 52 mg, respectively; the relatively larger amount of protein required reflects in part the nonhomogeneity of the LOX-2 and -3 fractions. The diethyl ether-extracted product was reduced with 10 mg of KI dissolved in 200 μL of methanol for 1 h in the dark, and then extracted into CHCl₃ by addition of water/CHCl₃. 4-HNE was isolated from the extracted material by TLC (Silica Gel 60 F₂₅₄ precoated plates, Merck, 20 cm × 20 cm × 0.25 mm) using hexane:diethyl ether (1:1, v/v) development. The 4-HNE band was located by UV absorbance and a separately spotted standard (R_F = 0.26–0.31). The scraped material was extracted with diethyl ether, the solvent was evaporated, and residual 4-HNE was derivatized with (–)-menthoxycarbonyl chloride for chiral analysis using a slightly modified procedure of Hamberg (1971). The (–)-methoxycarbonyl derivative of 4-HNE was isolated from Silica Gel 60 F₂₅₄ TLC (hexane:ethyl acetate [95:5, v/v] development) immediately above the by-product menthol band at a R_F of about 0.2. This isolate separated by GC-MS into two peaks (retention times = 14.45 and 14.55 min) giving virtually identical mass spectra: m/z (relative intensity) 200 (4); 199 (1); 156 (10); 139 (40); 138 (60); 123 (20); 109 (8); 95 (52); 83 (100); 81 (65); 69 (38); and 55 (47).

Instead of oxidizing the (–)-methoxycarbonyl derivative of 4-HNE by ozonolysis, 12 mg of KMnO₄ in 0.3 mL of acetic acid was employed (Hamberg et al., 1986). The resultant methyl esterified (–)-menthoxycarbonyl derivatives of methyl 2R- and 2S-hydroxyheptanoate were analyzed by GC-MS giving two peaks (retention time = 12.42 and 12.49 min) with virtually identical mass spectra: m/z (relative intensity) 205 (5); 173 (4); 139 (50); 138 (100); 123 (33); 111 (13); 95 (65); 83 (98); 81 (65); 69 (33); and 55 (48). The relative percentage of each of the (–)-menthoxycarbonyl derivatives of methyl 2R- and 2S-hydroxyheptanoate isomer was determined by flame-ionization detection GC as described above for the flame-ionization equipment. Retention times for the S- and R-isomers were 11.30 and 11.43 min, respectively. The identites of these derivatives were determined to be authentic by subjecting known 13S-HPODE, as its reduced methyl ester, to the same chiral analysis affording mainly the (–)-menthoxycarbonyl derivative of methyl 2S-hydroxyheptanoate.

Preparation of Crude Soybean Enzyme

A crude soybean enzyme was prepared by homogenizing 2 g of hexane-defatted soybean flour (seeds from cv Williams) for 30 s with a Polytron homogenizer after soaking in 20 mL of 0.1 m potassium borate buffer (either pH 8.3 or 9.0) for 10 min. The homogenate was filtered through cheesecloth and centrifuged at 9300g for 15 min. The resultant supernatant was diluted 10-fold with the same buffer giving a protein concentration of 3.4 to 3.5 mg/mL. The diluted supernatant (20 mL) was used to oxidize NON (1 mm) for 5 min on ice while bubbling pure O₂ through the solution. In certain experiments the diluted supernatant was preincubated for 5 min at 25°C with 20 mm H₂O₂ before incubation with NON to inactivate hydroperoxide-dependent peroxygenase (Takamura and Gardner, 1996). Products were extracted into CHCl₃ after adding a 3-fold volume of CHCl₃:CH₃OH (2:1, v/v). Either the product mixture or isolated 4-HPNE was reduced with KI, 4-HNE isolated, and subjected to chiral analysis as described above.

Protein Determination and Other Methods

Protein was determined by the bicinchoninic acid assay (Smith et al., 1985). All other chemical methods, including the preparation of 13S-HPODE, and preparation of derivatives are reviewed (Gardner, 1997).

RESULTS AND DISCUSSION

LOX-1 Gives 3Z-Alkenal Oxygenase Activity

During preliminary attempts to isolate NON-oxidizing activity by ionic-exchange and gel-filtration methods, it was noted that activity consistently coincided with LOX-1 activity using linoleic acid as a substrate. Utilizing an activity assay at pH 9.0 that was determined to be optimum for NON oxidation (Takamura and Gardner, 1996), coelution of the two activities occurred employing columns of Sephracyl S-200 and DEAE-Sephacel using procedures similar to those reported previously (Salch et al., 1995) (data not shown). Therefore, we resorted to a method used to prepare highly purified soybean LOX-1 (Axelrod et al., 1981). A check by SDS-electrophoresis showed that the isolate was a 90- to 100-kD protein with a minor contaminant of a slightly lower molecular mass, which was probably due to degradation of LOX-1. As seen in Figure 1, the final chromatographic separation employed by this method gave superimposed oxidation activities of NON and linoleic acid (LOX). It can also be seen in Figure 1 that there was an immense difference in the amount of relative activity; that is NON was 400- to 1000-fold less effective as a substrate, compared with linoleic acid. Dependency of pH on NON oxidation activity was also reminiscent of literature data (Christopher et al., 1970) for linoleic acid oxidation by LOX-1 (Fig. 2).

Figure 1.

Coelution of LOX-1 activity (linoleic acid substrate) with NON-oxidizing activity from DEAE-Sephadex. O₂-electrode assay. □, Linoleic acid assay, μmol min⁻¹ mL⁻¹; •, NON, μmol min⁻¹ mL⁻¹.

Figure 2.

pH dependence for oxidation of NON by LOX-1 measured by O₂ electrode. Conditions were: 0.2 m buffer, 0.1 mm NON, and 400 μL of a 1.1 m (NH₄)₂SO₄ solution of LOX-1 (2.4 mg of protein) incubated at 25°C in a total volume of 2.4 mL. Buffers were potassium acetate, pH 5.1; potassium Mes, pH 6.1; potassium Pipes, pH 7.0; potassium Hepes, pH 7.9, potassium borate, pH 8.5 and 8.9; and potassium carbonate, pH 9.2 and 10.0.

Products of NON Oxidation

The oxidation products of NON catalyzed by LOX-1 were separated by HPLC (Fig. 3), collected, and identified by GC-MS before and after preparation of appropriate derivatives. The GC-MS of HPLC-isolated 4-ONE was identical to the 4-ONE standard (retention time, 8.05 min) as follows: m/z (percent abundance, fragment identity) 154 (1, M⁺); 139 (2, M⁺ −CH₃); 125 (100, M⁺ −CHO); 98 (92, M −[CH₂]₃CH₃ + H⁺); 83 (68, M⁺ −[CH₂]₄CH₃); 70 (40); and 43 (45). Isolated 4-ONE was also derivatized with methoxylamine hydrochloride (Aldrich) giving three separable bis methoxime peaks (due to syn and anti isomerism, four isomers are possible). The spectra of the three isomeric peaks were similar, but they differed mainly in fragment ion intensities: m/z (fragment identity) 212 (M⁺); 181 (M⁺ −CH₃O); 166 (M⁺ −CH₃O −CH₃); 156; 154; 141 (M⁺ −[CH₂]₄CH₃); 125; and 110 (M⁺ −[CH₂]₄CH₃ −CH₃O). GC-MS of 4-HPNE isolated from HPLC gave a peak with a mass spectrum identical to 4-ONE on the front side of the peak, and on the back side of the peak the mass spectrum was identical to an unknown rearrangement product of 4-HNE. Selected ion monitoring confirmed this assessment. Since heat decomposition of 4-HPNE in the injector port was anticipated, the isolate was reduced with neutral methanolic KI. For GC-MS, the reduced compound was treated either with OTMS reagent or benzylhydroxylamine followed by OTMS reagent. The GC-MS of the first of the two derivatives proved to be the OTMS of 4-HNE (retention time, 10.1 min) as follows: m/z (relative intensity and ion structure) 228 (1, M⁺); 213 (5, M⁺ −CH₃); 199 (17, M⁺ −CHO); 184 (5, M⁺ −CHO −CH₃); 157 (100, M⁺ −[CH₂]₄CH₃); 143 (5); 129 (23); and 73 (73, TMS⁺). The second derivative, the syn and anti benzyloxime-OTMS of 4-HNE (retention times, 12.2 and 13.1 min), gave essentially the same mass spectrum as reported previously (Gardner and Hamberg, 1993). The last-eluting HPLC peak was identified as 4-HNE by making its benzyloxime-OTMS derivative for GC-MS of the syn and anti isomers, as was completed above for KI-reduced 4-HPNE.

Figure 3.

HPLC separation of products of NON oxidation. LOX-1 added, left to right: none, 0.12 mg, 0.30 mg, and 0.59 mg. NON (1 mm) was incubated for 15 min on ice under pure O₂ in 2 mL of 50 mm potassium borate buffer, pH 8.3. UV detection occurred at 226 nm. Samples of equivalent size were analyzed; values assigned to peaks are relative peak areas. Abbreviations and retention times are: 4-ONE, 9.4 min; 4-HPNE, 24.8 min; and 4-HNE, 53 min.

Threshold Requirement and Product Shift

There were two intriguing aspects of NON activity. First, there was a requirement for a threshold concentration of LOX-1 to trigger the reaction (Fig. 4). This threshold phenomenon was ascertained by determining initial rates of O₂ uptake with an O₂ electrode (Fig. 5). Second, when sufficient LOX-1 was present, the product was principally 4-ONE, not the expected 4-HPNE (Fig. 3). The observation of a threshold enzyme requirement was believed to be due to a “lag phase” observed when LOX is incubated with substrate in the absence of hydroperoxide. That is, native LOX, an Fe²⁺ species, needs to be oxidized to Fe³⁺ to start the catalytic cycle. The length of the lag is inversely dependent on the amount of both LOX and hydroperoxides present (Smith and Lands, 1972). This need for a threshold amount of LOX is undoubtedly the reason soybean LOX-1 was previously found to be inactive with NON (Gardner and Hamberg, 1993).

Figure 4.

Nonlinearity of NON-oxidizing activity with amount of LOX-1 added illustrating the threshold requirement (▪), compared with the same conditions plus 55 μg (0.09 mm) of 13S-HPODE (○). NON (1 mm) in 0.1 m potassium borate buffer (pH 8.6) was incubated for 5 min on ice in the presence of pure O₂; reaction volume was 2 mL.  

Figure 5.

O₂-electrode measurement of nonlinearity of NON-oxidizing activity versus amount of LOX-1 added illustrating the threshold requirement. Initial rate was measured. Conditions: 0.1 m potassium borate buffer, pH 8.6, 1 mm NON, and varying amounts of LOX-1 suspended in 2.3 m (NH₄)₂SO₄ (amount added was kept constant at 90 μL) incubated at 25°C in a total volume of 2.4 mL.

The second observation is similar to the O₂-starved reaction of LOX with linoleic acid, where 13-oxooctadecadienoic acid is produced (Garssen et al., 1971). O₂-starved reactions also can be triggered by excessive amounts of LOX (H.W. Gardner, personal observation). Therefore, we spiked the reaction with a small amount of a LOX-1 product, 13S-HPODE, to oxidize the enzyme to the active Fe³⁺ state. This spiked activity was compared with oxidation of NON in the absence of 13S-HPODE. Under these conditions NON-oxidizing activity was enhanced and the threshold phenomenon disappeared (Fig. 4). In the presence of 13S-HPODE at relatively low LOX-1 concentration, the main product was 4-HPNE instead of 4-ONE (Fig. 6). Apparently, NON, being a comparatively poor substrate, is inefficient in electron cycling the Fe active site by substrate and O₂ and, consequently, 4-HPNE is consumed in the process of keeping the active site oxidized. This tendency to further oxidize to the ketone is reminiscent of LOX-1 oxidation of another α,β-unsaturated carbonyl, 12-oxo-9Z-octadecenoic acid, to afford 9,12-dioxo-10E-octadecenoic acid (Kühn et al., 1991).

Figure 6.

HPLC separation of NON oxidation products after addition of a below-threshold amount of LOX-1 (left), and an identical treatment spiked with 55 μg of 13S-HPODE (right). Conditions, methods, and abbreviations are the same as in Figure 3.

Chiral Analysis of 4-HPNE

The 4-HPNE obtained by spiking the reaction with 13S-HPODE was reduced to 4-HNE with neutral KI, and the TLC-isolated 4-HNE was analyzed for its stereoconfiguration. As seen in Figure 7 and Table I, the 4-hydroxyl was 83.1% of the S-stereoisomer. A duplicate experiment gave exactly the same result. This S-stereopreference is the same as found in the product of linoleic acid oxidation by LOX-1 (Hamberg, 1971). The structural similarities of the two substrates, and how they might fit in the active site of LOX-1, is illustrated in Figure 8 using a model suggested for the oxidation of linoleic acid (de Groot et al., 1975).

Figure 7.

Chiral analysis by GC separation of R- and S-isomers of (-)-menthoxycarbonyl derivatives of methyl 2-hydroxyheptanoates obtained from 4-HNE by chemical modification. Top, Analysis of synthetic 4-HNE; bottom, analysis of 4-HNE derived from LOX-1 oxidation of NON.

Table I.

Analyses of stereoconfiguration of 4-HNE and 4-HPNE

Sample	pH	Total
		4S	4R
		%
4-HPNE from LOX-1 oxidation	8.3	83.1	16.9
4-HPNE from LOX-2aoxidation	8.6	60.8	39.2
	6.5	86.5	13.5
4-HPNE from LOX-3aoxidation	8.6	55.5	44.5
	6.5	71.4	28.6
4-HPNE + 4-HNE from crudeb	8.3	44.0	56.0
	9.0	42.9	57.1
4-HPNE + 4-HNE from H2O2inactivated crudec	8.3	52.4	47.6
	9.0	52.2	47.8
4-HPNE from crudeb	9.0	56.3	43.7
4-HNE from crudeb	9.0	38.4	61.6
4-HPNE from H2O2-inactivated crudec	9.0	52.5	47.5
4-HNE from H2O2-inactivated crudec	9.0	50.4	49.6

^aTentative identification of LOX isozyme. 

^bLow-speed supernatant (9300g for 15 min) of hexane-defatted soybean homogenate. Protein concentration of 3.4 to 3.5 mg/mL. 

^cLow-speed supernatant was preincubated for 5 min at 25°C with 20 mm H₂O₂ to inactivate peroxygenase. 

Figure 8.

Mechanism of LOX oxidation of linoleic acid (top) compared with oxidation of NON (bottom) showing the redox cycling of the Fe active site and the hydrophobic pocket (hatched marks) that accommodates the ω6 tail-end of linoleic acid or NON.

Because 4-HNE and 4-HPNE were readily formed in crude soybean preparations (Takamura and Gardner, 1996), it was necessary to ascertain the stereochemistry of the KI-reduced product of the crude preparation to determine if there was evidence for a significant contribution of LOX-1 in the biogenesis of 4-HNE. As seen in Table I, the 4R-isomer was unexpectedly predominant at pH 8.3 (56.0% 4R-isomer) and at pH 9.0 (57.1% 4R-isomer).

Two other LOX isozymes, LOX-2 and LOX-3, exist in soybeans with pH optima near neutrality (Christopher et al., 1972). Using the procedure of Axelrod et al. (1981), two peaks of activity eluted early from the first of two DEAE-Sephadex columns used. These peaks showed LOX activity at pH 6.5, but negligible activity at pH 9.0. The material from the two peaks were tested for activity with 0.09 mm 13S-HPODE at both pH 8.6 and 6.5, giving oxidation ranging from 27 to 54% of total NON. 4-HNE was isolated from the KI-reduced product to complete stereochemical analyses. The first eluting isozyme peak (presumably LOX-3 [Axelrod et al., 1981]) gave 55.5 and 71.4% 4S-enantiomeric excess at pH 8.6 and 6.5, respectively (Table I). The second isozyme peak (presumably LOX-2 [Axelrod et al., 1981]) furnished 60.8 and 86.5% 4S-enantiomeric excess at pH 8.6 and 6.5, respectively (Table I). As might be expected, stereochemical purity was improved at lower pHs where these isozymes have their optima. However, even at pH 8.6 these isozymes afforded a slight excess of the 4S-isomer. Thus, the 56 to 57% enantiomeric excess of 4R-isomer obtained from crude preparations was still difficult to attribute solely to the combined action of LOX isozymes.

Possible Origin of the 4R-Isomer of 4-HNE in Crude Preparations

It was suggested previously that a portion of 4-HNE was produced by rearrangement of 3,4-epoxynonanal in broad bean seed preparations (Gardner and Hamberg, 1993). 3,4-Epoxynonanal originates from oxidation of NON by peroxygenase utilizing the oxidation potential derived from 4-HPNE in its reduction to 4-HNE. According to Blée and Schuber (1990), linoleic acid is oxidized by soybean peroxygenase to predominantly the 9R,10S- and 12R,13S-epoxides. Based on these data, one might predict that 3R,4S-epoxynonanal could be the preferred isomer formed. Soybean epoxide hydrolase hydrolyzed 9R,10S-epoxystearic acid to 9R,10R-dihydroxystearic acid (Blée and Schuber, 1992), thus inverting the stereoconfiguration of carbon-10. Applied to 4-HNE formation by the peroxygenase/hydrolase route, one might predict the preferential formation of the 4R-hydroxyl with loss of the 3R-hydroxyl by rearrangement/dehydration. It is known that peroxygenase is inactivated by a 5-min preincubation with H₂O₂ (Hamberg and Fahlstadius, 1992). Therefore, a crude preparation was treated with 20 mm H₂O₂ for 5 min, then incubated with 1 mm NON as above. The 4-HNE obtained after KI reduction of the products was 52.4 and 52.2% of the 4S-isomer at pH 8.3 and 9.0, respectively (Table I).

To further investigate the hypothesis that peroxygenase/hydrolase was responsible for formation of the 4R-isomer, 4-HNE and 4-HPNE were individually isolated by TLC (Takamura and Gardner, 1996) for chiral analysis to determine if there was any segregation of configuration. Without H₂O₂ preincubation, the crude preparation afforded 4-HPNE that was 56.3% of the 4S-isomer, whereas 4-HNE was 61.6% of the 4R-isomer (Table I). With H₂O₂ preincubation, 4-HPNE was 52.5% 4S-isomer, and the small amount of 4-HNE isolated (2% yield compared with the absence of H₂O₂ preincubation) was analyzed at 50.4% of the 4S-isomer (Table I). In the absence of H₂O₂, the prevalence of the 4R-isomer in 4-HNE, compared with the 4S-isomer excess in 4-HPNE, further suggested the participation of a 4R preference of the peroxygenase/hydrolase pathway. Since H₂O₂ preincubation inactivates peroxygenase, the stereoconfigurations of 4-HPNE and 4-HNE isolated from the H₂O₂ treatment should reflect the stereoconfiguration of 4-HPNE from the absence of H₂O₂ treatment, which was largely the result obtained. However, it is noted that the 52 to 56% 4S-isomeric excess in 4-HPNE obtained with the crude preparations does not seem to compare with the 55 to 83% 4S-isomeric excess obtained with LOX isozymes at pH 8.3 to 8.6. Since LOX-1 is the most active LOX isozyme at pH 8.0 to 9.0, one would expect a result approaching 83% 4S-isomer. Therefore, the origin of a significant percentage of the 4R-isomer seems to be unresolved. In this regard, we have not yet ruled out the existence of a membrane-bound 3Z-alkenal oxygenase as originally proposed (Gardner and Hamberg, 1993; Takamura and Gardner, 1996). In these investigations significant 3Z-alkenal-oxidizing activity occurred with washed microsomal fractions. In the present work we found that enzyme extraction of defatted soybean flour with 0.5% Triton X-100 enhanced NON oxidation activity 3-fold, but had little influence on LOX activity with linoleic acid. This suggests that 4R-oxidizing activity may reside in membranes.

CONCLUSIONS

There are two reasons that one might question the physiological significance of LOX-1 oxidation of NON. One is the 400- to 1000-fold less activity of NON oxidation, compared with the preferred substrate, linoleic acid. Another is the threshold requirement and formation of mainly 4-ONE with the highly purified LOX-1. However, this contrasts with the facile production of 4-HPNE and 4-HNE from NON in crude preparations of soybean (Takamura and Gardner, 1996). In addition, 9S-hydroperoxy-10E,9Z-octadecadienoic acid is readily converted in vitro to 4-HNE via hydroperoxide lyase cleavage to give 4-HNE through the intermediate NON (Gardner et al., 1991). This implies that activation of the LOX pathway would certainly lead to 4-HPNE and 4-HNE in soybean. Since the 9-hydroperoxide of linoleic acid is a precursor, catalytic amounts of hydroperoxide needed to overcome the threshold of reaction would always be present in crude preparations, but hydroperoxides would be initially very low in pure LOX-1 systems in the absence of trace amounts of polyunsaturated fatty acids. In the case of a wounded or stressed plant, it seems certain that facile oxidation of polyunsaturated fatty acid would “prime the pump” to form the active Fe³⁺-LOX to make it active in NON oxidation. Thus, fatty acid hydroperoxide stimulated oxidation of NON to 4-HPNE and subsequent conversion by peroxygenase would yield 4-HNE. Soybean seed is one of the richest sources of LOX known, and LOX-1 is the most active of the soybean LOXs. From the data of Axelrod et al. (1981), one can calculate that LOX-1 amounts to about 2.5% of the protein extractable from defatted soybean flour. This suggests that fatty acid hydroperoxide-aided oxidation of NON by this relatively large quantity of LOX is certainly possible, if not probable. For example, even in the absence of catalytic 13S-HPODE, the threshold of NON oxidation was 0.125 mg LOX-1/mL (Fig. 4), whereas the crude preparation contained 3.4 to 3.5 mg protein/mL (3.5 mg/mL × 2.5% = about 0.09 mg/mL LOX-1). In view of the oxidation of NON by all three soybean LOX isozymes, 3Z-alkenal oxidation may be a general reaction for all LOXs, including those of animal origin. 3Z-Alkenals might be oxidized by LOX particularly under conditions of oxidative stress necessary to overcome the threshold requirement.

However, our data indicate that other enzymes also may be involved in NON oxidation in crude preparations. It seems certain from our data that part of the 4R-hydroxy stereoisomer of 4-HNE is derived from peroxygenase action. Nevertheless, this does not account for the larger than expected amount of 4R-hydroperoxy stereoisomer found in 4-HPNE isolates (Table I). When crude enzyme was used at pH 9.0, the 4-HPNE obtained was 56.3% 4S-hydroperoxy stereoisomer. Since LOX-1 is, by orders of magnitude, the predominant enzyme at this pH, one would expect a result close to the 83.1% 4S-hydroperoxy stereoisomer obtained with pure LOX-1. The result with crude enzyme suggests that either the 4S-hydroperoxy stereoisomer is preferentially utilized by peroxygenase to form 4-HNE, or the majority of 4R-hydroperoxide is formed by another oxygenase. Since the percentage of the 4S-hydroperoxide was nearly the same with crude enzyme treated with H₂O₂ to inactivate peroxygenase (Table I), the latter choice was implied. If this putative 4R-specific oxygenase is membrane bound, it would not be expected to be separated by chromatography without detergent-assisted solubilization.

It is also noted that 3Z-alkenals other than NON, that is 3Z-hexenal and 3Z,6Z-nonadienal, are directly biosynthesized from linoleic and linolenic acid hydroperoxides (Gardner, 1995). Although 3Z-hexenal and 3Z,6Z-nonadienal were not tested with the LOX isozymes, 3Z-hexenal was found to be oxidized to 4-hydroxy-2E-hexenal by soybean preparations (Takamura and Gardner, 1996).

Since 4-HNE appears to be a lipid signal in animals, a role for 4-HNE in plants is suggested. It is now known that one branch of the LOX pathway leads to 12-oxo-phytodienoic acid and jasmonic acid, which are potent lipid signals with broad physiological effects (Creelman and Mullet, 1997). The hydroperoxide lyase branch of the LOX pathway, leading to aldehydes and alcohols from aldehyde reduction, is often thought to only give generalized protection by inhibiting the growth of pathogenic microbes and fungi (Gardner, 1995). However, by extending the pathway to include 4-oxo-, 4-hydroperoxy-, and 4-hydroxy-2-nonenals and hexenals, consideration should be given to their possible effects on both the physiology of plants and pathogenic organisms. One such study showed that 4-HNE inhibited the growth of soybean pathogens (Vaughn and Gardner, 1993).

Abbreviations:

4-HNE

4-hydroxy-2E-nonenal

4-HPNE

4-hydroperoxy-2E-nonenal

4-ONE

4-oxo-2E-nonenal

13S-HPODE

13S-hydroperoxy-9Z,11E-octadecadienoic acid

LOX

lipoxygenase

LOX-1

-2, and -3, soybean seed lipoxygenase-1, -2, and -3 isozymes

NON

3Z-nonenal

OTMS

trimethylsilyloxy derivative of hydroxyls