Description of the rate‐limiting hydrogen tunneling mechanism for plant 9‐ and 13‐lipoxygenases

Abstract

Soybean lipoxygenase (GmLOX1), a plant 13‐LOX, has long been considered a model protein for non‐trivial, quantum hydrogen tunneling in enzyme catalysis. Hydrogen tunneling mechanisms have also been confirmed for LOXs across mammalian, fungal, and bacterial kingdoms. Notably, GmLOX1 exhibits a low activation energy (E _a ) for catalysis relative to characterized non‐plant LOXs. In the present study, we examined three additional plant isozymes, including representatives of both 9‐ and 13‐LOXs. All plant LOXs show large primary deuterium kinetic isotope effects associated with rate‐limiting hydrogen tunneling during cleavage of a carbon‐hydrogen bond on substrate linoleic acid. The monomeric plant 9‐LOXs studied herein uniquely display kinetic cooperativity in their steady‐state kinetics, while GmLOX1 does not. These data implicate fatty acid‐dependent structural regulation of plant 9‐LOX catalysis in a similar manner to mammalian LOXs. In lieu of experimental structures of these plant LOXs with substrate, we predicted representative 9‐ and 13‐LOX isoforms in AlphaFold3 (AF3) with fatty acids and compared them with their crystal structures to better predict modes of substrate acquisition that could support the alternative kinetic models. Further, we find that two 9‐LOXs have larger E _a values relative to GmLOX1. To understand this property, we also present a visualization of LOX evolutionary conservation across the plant kingdom. The data provide deeper insights into the reaction mechanisms of plant LOXs in the context of the activation barriers for catalysis and the relationships to the thermal activation of hydrogen tunneling.

1. INTRODUCTION

The concept of quantum mechanical barrier penetration (i.e., tunneling) was first described nearly 100 years ago (Hund, ^{1927a, 1927b}). However, it was not until 1989 that the quantum mechanical phenomenon associated with hydrogen atom transfer in enzyme catalysis was reported from the observation of anomalous secondary kinetic isotope effects for the oxidation of benzyl alcohol by yeast alcohol dehydrogenase (Cha et al., 1989). Since the pinnacle study, several case studies have emerged from various laboratories that manifest varying degrees of hydrogen tunneling in enzymatic reactions. Among these systems, the reaction of the isoform‐1 lipoxygenase from soybean seeds has demonstrated some of the most prominent and compelling kinetic features for non‐classical behavior and has thus emerged as a paradigm H‐tunneling enzyme (Glickman et al., 1994; Hu et al., ²⁰¹⁴; Knapp et al., ²⁰⁰²).

Lipoxygenases (LOXs) are well‐represented enzymes in the plant, animal, fungi, and prokaryotic kingdoms that oxidize polyunsaturated fatty acids (PUFAs) to form a diverse set of hydroperoxides (Brash, 1999; Liavonchanka & Feussner, ²⁰⁰⁶). In plants, LOX‐derived hydroperoxides participate in a broad range of physiological roles, including seed germination, plant growth and development, pathogenic defense, and wound healing (Feussner & Wasternack, 2002; Viswanath et al., ²⁰²⁰). The various physiological responses are mediated by different LOX isozymes, resulting from gene duplications and paralog specialization (Camargo et al., 2023). These isoforms are expressed and distributed in specific plant tissues and can produce different oxidized products. For example, five soybean LOX isoforms have been isolated and characterized in vitro, with several more potential LOX genes encoded in the soybean genome (Axelrod et al., 1981; Shin et al., ²⁰⁰⁸; Youn et al., ²⁰⁰⁶).

The initial step in the LOX reaction mechanism is the cleavage of a bis‐allylic carbon‐hydrogen bond by the ferric hydroxide (Scheme 1). Hydrogen abstraction from C‐H bond of substrate leads to a delocalized carbon‐centered radical that is subsequently quenched by molecular oxygen. One of the most common and simplest fatty acid substrates in plants is linoleic acid (LA). The oxidation of LA leads to the formation of four possible hydroperoxide products (9S, 9R, 13S, and 13R‐HpODE), though only one of these products is generally synthesized by a given LOX. As such, LOXs are named and differentiated by their regioselectivity (e.g., 9S‐ and 13S‐LOX).

SCHEME 1.

General reaction mechanism of plant LOX reactions for the oxidation of LA. Oxygen is inserted antarafacially (opposite side of hydrogen abstraction). Plant LOXs mainly produce one of two products, 9S‐ or 13S‐HpODE.

Plant (and animal) LOXs are structurally comprised of two domains in a single polypeptide: an N‐terminal PLAT (Polycystin‐1, Lipoxygenase, Alpha‐Toxin) domain, which can facilitate regulation of the enzyme, and a C‐terminal catalytic domain (Newcomer & Brash, 2014). The C‐terminal domain houses a deeply buried mononuclear, non‐heme iron that mediates fatty acid oxidation. Common only to plant LOXs is the lid domain situated below the π‐helix reported to be dynamic for catalysis (Bradshaw & Gaffney, 2014). This π‐helix is situated in helix alpha‐2 (Ha2) that is also considered to be a gating helix. For example, Ha2 in human 5‐LOX was reported in a “closed” conformation and must undergo a conformational change for substrate to access the active site (Gilbert et al., 2011). Within the catalytic pocket of plant LOXs, an arched helix containing an invariant Leu residue serves to orient the PUFA for precise oxygenation. This positional specificity is further refined by a “regiospecificity switch” eleven amino acids downstream of the invariant Leu, a concept pioneered by Ivo Feussner's group (Hornung et al., 1999). Deep in the substrate pocket of 13‐LOXs lies a conserved threonine (Thr‐556 in GmLOX1) followed by a bulky phenylalanine (Phe‐557), which only allows “tail‐first” substrate entry into the U‐shaped active site (Andreou & Feussner, 2009). Conversely, in 9‐LOXs, the substitution of the bulky phenylalanine with a smaller valine could allow the substrate's larger carboxylate head to penetrate past the valine and potentially interact with an invariant arginine, resulting in “head‐first” (inverse) orientation and 9‐product formation. Thus, the F and V sidechains are often referred to as the “switch residue”.

The most well characterized plant LOX is the isoform‐1 from soybean (designated as GmLOX1 in this report, but also commonly referred to as “SLO‐1”). The GmLOX1 reaction with LA is associated with a very large primary deuterium kinetic isotope effect on the first order rate constant (^D k _cat = k _cat(H)/k _cat(D) ~ 60–80) (Glickman et al., 1994; Hwang & Grissom, ¹⁹⁹⁴). A large ^D k _cat demonstrates that the chemical step, namely the C‐H bond cleavage step, is rate‐limiting for the reaction. Further, the ^D k _cat value is weakly temperature dependent, indicating that the hydrogen transfer occurs quantum mechanically (i.e., via hydrogen tunneling) (Hatcher et al., 2006; Knapp et al., ²⁰⁰²). Kinetic isotope effect studies have also been reported for select mammalian, fungal and prokaryotic LOXs, with ^D k _cat values reported in the range of 20–100. There are also weakly temperature dependent effects, when reported, supporting a conserved tunneling mechanism across the LOX family (Carr & Klinman, 2014; Kostenko et al., ²⁰¹⁹; Seagraves & Holman, ²⁰⁰³).

One of the distinctive features of enzymatic hydrogen tunneling is the inherent temperature independent nature of wave function overlap as measured by the temperature invariance of the kinetic isotope effect (Klinman & Offenbacher, 2018; Sutcliffe & Scrutton, ²⁰⁰⁹). The observation of temperature dependent enzyme rates (i.e., E _a >0 kcal/mol) reflect the involvement of protein thermal motions that are required to help bring the donor (carbon‐11 of substrate) and acceptor (metal‐bound hydroxide) into the proper distance(s) and geometries for effective hydrogen wave function overlap (Klinman et al., 2017). The GmLOX1 reaction is associated with a relatively small E _a of ca. 2 kcal/mol (Knapp et al., 2002), whereas several examples of non‐plant LOX reactions exhibit elevated E _a values ranging from 8 to 13 kcal/mol (see Table S1). While plant LOXs have been studied previously in vitro (Boeglin et al., 2008; Lorenzi et al., ²⁰⁰⁶; Park et al., ²⁰¹⁰), to our knowledge, GmLOX1 is the only plant LOX to have been well characterized for its tunneling properties. Thus, it is hitherto undetermined if the kinetic properties ascribed to GmLOX1 are common across the diverse family of plant LOXs.

In this study, we present an investigation of the hydrogen tunneling properties for select LOX orthologs from various plant species. Enzyme kinetic assays, including temperature and isotope effects, were performed for three plant LOXs to determine insights into the reaction mechanism. The kinetic isotope effect data indicate that the plant LOX reactions are associated with an initial, rate‐limiting hydrogen tunneling mechanism, consistent with GmLOX1. Other kinetic properties, including activation barriers and kinetic cooperativity effects, were surprisingly varied. AlphaFold3 (AF3) models of representative 9‐LOX and 13‐LOX isoforms, complexed with fatty acids, successfully recapitulate the established head‐ or tail‐first binding orientations for each respective isoform. Furthermore, these predicted structures reveal distinct substrate access portals. Spatial mapping of evolutionary sequence analyses onto GmLOX1 shows conserved and variable residues within this active site. Altogether, these results provide deeper insights into the relationship between non‐trivial quantum mechanics and the role of the LOX structure and its corresponding protein thermal motions to optimize these features.

2. RESULTS

Table 1 presents the series of plant LOXs that were targeted for this study. Our goal was to obtain a library of plant LOXs that represented both 9‐ and 13‐LOXs and sequence variability across the diverse plant kingdom. The work was initiated as a Course‐based Undergraduate Research Experience (CURE) laboratory sequence at East Carolina University, involving upper‐level undergraduate laboratories in the Departments of Biology (Biotechnology II; BIOL 3110) and Chemistry (Biological Chemistry; CHEM 3771). In the Biotechnology II CURE, students sub‐cloned plant LOX genes into bacterial expression vectors and screened protein expression using SDS‐PAGE. While several genes were sub‐cloned, only the vector encoding OsLOX1 successfully produced substantive protein levels from these tests. Students in the Biological Chemistry CURE performed large‐scale expressions and protein isolation methods that were followed by in vitro kinetic assays including kinetic isotope effects to determine the rate‐limiting step(s) of catalysis. During these CUREs, students were able to screen the activity and test preliminary mechanisms of these enzymes. For example, ZmLOX5 was characterized with poor enzyme activity and was not pursued further. Importantly, the collective aim of this CURE was to make quantum mechanical phenomena in the context of biological processes accessible and relevant to undergraduate students. The assessment of student outcomes is currently underway. Two other plant LOXs (AtLOX1 and SlLOXA), as well as more thorough biochemical analyses, were supplemented from parallel efforts outside of the CURE labs.

TABLE 1.

Information on select plant LOXs ^a .

Plant LOX	No. amino acids	Product	“Switch residue”	Identity (%) b	Cotyledon c	Accession (UniProt)
GmLOX1 Soybean	839	13	F557	—	D	P08170
AtLOX1 Arabidopsis	859	9	V577	57	D	Q06327
SlLOXA Tomato	860	9	V579	59	D	P38415
OsLOX1 Rice	863	9/13	V576	54	M	Q76I22
GhLOX2 Cotton	877	9	V591	57	D	A0A1U8LI71
NtLOX1 Tobacco	862	9	V581	60	D	Q43800
StLOX1 Potato	861	9	V580	58	D	P37831
TaLOX3 Wheat	878	9	V587	53	M	A0A3B6IJT8
ZmLOX5 Corn	887	9	V595	53	M	A1XCI0
ZmLOX10 Corn	905	13	F622	43	M	A1XCI5

^a Boldface represents isoforms examined in this study. ^bRelative to GmLOX1. ^cD: dicotyledon, M: monocotyledon.

AtLOX1, GmLOX1, OsLOX1, and SlLOXA were expressed in and successfully isolated to near homogeneity from E. coli bacterial cultures (see Figure S1a for SDS‐PAGE). AtLOX1 and SlLOXA isoforms are 9‐LOXs (Boeglin et al., 2008). OsLOX1 is predicted to be a 9‐LOX (Mizuno et al., 2003), consistent with its switch residue (Table 1); however, there is biochemical evidence that this plant LOX produces 13‐hydroperoxides with LA (Wang, Shen, Liu, Jiang, Liu et al. ²⁰⁰⁸). Consistent with the latter, OsLOX1 has been considered a 9/13‐LOX or a “type‐II” LOX (Wang, Shen, Liu, Jian, Zhai et al. ²⁰⁰⁸). Circular dichroism (CD) spectroscopic experiments indicated that these proteins all exhibited secondary structure similar to that of the well‐characterized GmLOX1 protein (see Figure S1b), and the ferrozine assay (see Methods) showed 0.6–0.85 iron atoms per protein monomer for these LOXs. Size exclusion chromatography (SEC) confirmed that all four LOXs are monomeric, based on the SEC calibration curve using protein standards (Figure S1c). Despite many efforts (including various cell lines and expression media tested), other plant LOX isoforms, such as GhLOX2, NtLOX1, StLOX1, TaLOX3, and ZmLOX5, failed to produce sufficient protein (determined by kinetic screens following purification). ZmLOX10 was isolated as a dimer and had minimal LOX activity. These latter LOXs were excluded from the remainder of the studies.

Steady‐state kinetic assays were performed to monitor LA oxidation by the LOX reaction. These experiments were performed with single wavelength kinetics mode (λ = 234 nm). Initial studies focused on the pH profiles of the reactions, in which the activity was examined from pH 5 to 9. We selected a common pH 7.0 buffer to analyze the kinetics of AtLOX1, OsLOX1, and SlLOXA. At this pH, the activities of these three enzymes are at (or nearly at) their peak activity; further, AtLOX1 and SlLOXA are confirmed as 9‐LOXs under these conditions (Bild et al., 1977; Boeglin et al., ²⁰⁰⁸). Note that the GmLOX1 kinetics were performed at pH 9, where the first‐ and second‐order rate constants and 13‐HpODE production are maximal (Gardner, 1989; Glickman & Klinman, ¹⁹⁹⁵).

The kinetic parameters for LA oxidation by the plant LOXs at 25°C are listed in Table 2. The first‐order rate constant, k _cat, and Michaelis constant, K_m, of GmLOX1 are determined as 253 s⁻¹ and 22 μM. AtLOX1, OsLOX1, and SlLOXA exhibit k _cat values in the range of 230–400 s⁻¹ (Table 2). The plots of the initial velocity versus LA substrate concentrations for GmLOX1 reaction produced a hyperbolic trace consistent with Michaelis–Menten kinetic behavior (Figure 2a). Conversely, the reactions of the other three plant LOXs, AtLOX1, OsLOX1, and SlLOXA, give rise to a distinct, sigmoidal kinetic response, which is modeled well by the Hill equation (Figure 2b–d). This kinetic model for these plant LOXs was validated using Akaike Information Criterion (AIC) analysis. The corresponding Hill coefficients (n) were 2.3 for OsLOX1 to nearly 4 for AtLOX1 and SlLOXA (Table 2). Values of n > 1 are associated with positive cooperativity. Similarly, monomeric forms of humanStable‐5‐LOX have also been reported with Hill coefficient between 2 and 3 for the oxidation of arachidonic acid (Gallegos et al., 2022).

TABLE 2.

Kinetic parameters of LOXs and their reactions with LA ^a .

LOX	LA b	n	k cat (s−1)	Km (μM)	D k cat	E a (kcal/mol)
GmLOX1	H	1	253 ± 8	22 ± 2	72 ± 3	2.2 ± 0.4
	D	1	3.5 ± 0.1	7.3 ± 0.8		3.1 ± 0.3
OsLOX1	H	2.3 ± 0.2	388 ± 13	21 ± 1	37 ± 2	1.2 ± 0.2
	D	4.1 ± 0.7	10.5 + 0.4	35 + 2 c		N.M. d
SlLOXA	H	3.7 ± 0.8	362 ± 15	19 ± 1	62 ± 7	6.0 ± 0.3
	D	3.7 ± 1	5.8 ± 0.6	43 ± 4 c		N.M. d
AtLOX1	H	3.9 ± 0.5	228 ± 5	12.1 ± 0.4	50 ± 1	7.2 ± 0.6
	D	3.3 ± 0.3	4.59 ± 0.08	17.8 ± 0.4		6.9 ± 0.1

^a The temperature was 25°C.

^b H represents natural abundance substrate, D represents 11,11‐D₂‐LA.

^c Value is an estimate as the K_m for the deuterated substrate (D₂‐LA) approaches or exceeds the critical micelle concentration of substrate.

^d N.M., not measured because of the elevated K_m value. The buffer for GmLOX1 was 100 mM sodium borate, pH 9.0, while all other plant LOXs were assayed in 50 mM HEPES, 150 mM NaCl, pH 7.0 buffer. The data represent the average ± s.e.m. from at least three separate measurements.

FIGURE 2.

Kinetic saturation plots, with the rates corrected for enzyme concentration, for the reactions involving (a) GmLOX1, (b) OsLOX1, (c) AtLOX1, and (d) SlLOXA with varying LA concentrations. The solid lines represent fit to (a) Michaelis–Menten and (b‐d) Hill equations. For reference, the dashed lines in (b–d) represent the data fit to the Michaelis–Menten equation. The reactions were collected at 25°C in (a) 100 mM sodium borate (pH 9.0) or (b–d) 50 mM HEPES (pH 7.0) and 150 mM sodium chloride buffer. The data are averages from three independent measurements with the error bars representing ±s.e.m.

To better understand this sigmoidal behavior of 9‐LOX catalysis, we investigated the impact of pre‐activating the enzyme, using AtLOX1 as an example. As isolated, the catalytically essential non‐heme iron center of LOXs is in the resting 2+ oxidation state but becomes activated during catalysis. We pre‐activated the iron center of AtLOX1 to the ferric (Fe³⁺) state and then examined the kinetic properties upon removal of HPODE (see Methods for details); this sample was associated with a slight decrease in the k _cat (Figure S2a). Metal analysis of the pre‐activated AtLOX1 revealed 0.62 iron atoms per LOX compared to 0.84 iron atoms for AtLOX1 as isolated. Thus, this reduced activity of pre‐activated AtLOX1 sample is due at least in part to a slight loss of the metal during the activation procedure, similar to our previous reports for GmLOX1 (Hill et al., 2025). Importantly, the reaction was still sigmoidal with a Hill coefficient of ca. 3.0 ± 0.4. We also explored the effect of detergents on AtLOX1 activity. The K_m values for AtLOX1 and SlLOXA (12 and 19 μM, respectively; see Table 2) are lower than the critical micelle concentration (CMC) of LA at pH 7, which was determined as ca. 32 μM by the fluorescence‐based assay using 1,6‐diphenyl‐1,3,5‐hexatriene dye (Chattopadhyay & London, 1989). In the presence of 0.01% Triton X‐100 (Figure S2b), the k _cat (i.e., maximal velocity) was found to be unaffected by the presence of detergent, though the Hill coefficient decreased to near unity (n = 1.4 ± 0.2). However, we chose to avoid the use of detergents for the remainder of the kinetic experiments to prevent any potential complications from co‐contaminants (e.g., peroxides and aldehydes) that often accompany commercial detergents and the non‐physiological nature of Triton X‐100 in plant systems.

We next explored the primary deuterium kinetic isotope effects of these reactions using a commercially available per‐deuterated substrate (D₃₁‐LA) and a synthetic di‐deuterated (11,11‐D₂‐LA) substrate (Figure S2a). The ^D k _cat was determined by dividing the first‐order rate constants for the natural abundance substrate by that of the deuterated substrate. For reference, the GmLOX1 reaction revealed a ^D k _cat of ~70 at 25°C and pH 9, which is in good agreement with reported literature values (Glickman et al., 1994; Hwang & Grissom, ¹⁹⁹⁴). The ^D k _cat values for the other plant LOXs follow a similar trend, in the range of ~40–60. These values indicate that the chemical step is rate‐limiting for the first‐order rate constant. It is important to note that the K_m for the deuterated substrate was elevated in the 9‐LOXs. For example, the K_m of D₃₁‐LA for the AtLOX1 reaction was found to be ca. 20 μM (Figure S2c). The use of the di‐deuterated substrate at the reactive carbon (11,11‐D₂‐LA) resulted in nearly identical kinetic properties to the perdeuterated substrate (cf. Figure S2c and Table 2). Thus, the additional deuterium labels did not greatly influence the kinetic parameters. This overall behavior is distinct from the GmLOX1 reaction, in which the K_m value of the deuterated substrate (7 μM at 25°C; see Table 2) is reduced from the natural abundance (22 μM).

Next, we determined the first‐order rate constants for these reactions across a temperature range of 10–40°C. Differential scanning calorimetry (DSC) was employed to validate that the proteins were structurally stable across this temperature range. The corresponding melting temperatures (T_m) for this group of plant LOXs were determined from these DSC experiments to be in the range of 59–71°C (Table S2), which is at least 20°C above the temperature range explored in the kinetic experiments herein. Arrhenius plots were generated from the temperature‐dependent kinetic data to extract the activation energies (E _a) from the linear slopes (Figure 3 and Table 2). For the protium substrate, the E _a value of the GmLOX1 reaction was determined as 2.2 ± 0.4 kcal/mol for LA. OsLOX1, which represents our (“type II”) 9/13‐LOX, also exhibited a low E _a value of ~1 kcal/mol. In contrast, both AtLOX1 and SlLOXA reactions gave rise to elevated E _a values of 6–7 kcal/mol.

FIGURE 3.

Arrhenius plots for the first‐order rate constants, k _cat, of A) GmLOX1 and B) AtLOX1 reactions with natural abundance (H‐LA) and perdeuterated (D₃₁‐LA) substrates.

The activation energies for the reactions with the deuterated substrate are also informative. The so‐called temperature dependence of the kinetic isotope effect, ΔE _a (ΔE _a = E _a(D) − E _a(H)), serves as a powerful kinetic probe for hydrogen tunneling (Klinman & Offenbacher, 2018). Reactions involving naturally evolved enzymes dominated by hydrogen tunneling are associated with near‐zero ΔE _a values, while enzymes with impaired or modest tunneling are associated with larger ΔE _a values. For example, one impairing variant of GmLOX1, I553G, was associated with a modest impact on k _cat, but exhibited a ΔE _a value of 5 kcal/mol (Meyer et al., 2008). Thus, the ΔE _a can serve as a ruler for tunneling efficiency in enzymes. The GmLOX1 reaction has a weak temperature dependence for the kinetic isotope effect, ΔE _a = 0.9 ± 0.5 kcal/mol (Figure 3a). The ΔE _a for AtLOX1 with LA is found to be −0.3 ± 0.6 kcal/mol (Figure 3b), which is within uncertainty to zero. The near‐zero ΔE _a value for AtLOX1 corroborates that tunneling is dominant for its hydrogen transfer reaction, despite being associated with a larger E _a for the protium substrate. Because the K_m values for the reactions of SlLOXA and OsLOX1 with D‐LA exceeded the CMC of LA at pH 7, we did not attempt to determine the ΔE _a for these reactions.

It is important to note that the LOX reactions presented in this work have focused on using LA as the primary substrate. LA was used due to its accessible deuteration to follow kinetic isotope effects for mechanistic studies. In addition, LA is typically the most predominant fatty acid in many plants, including soybean (45%–70%), rice (35%–40%), and tomatoes (48%–54%) (see Table S3 and references [Taira, ¹⁹⁸³, Weber, 1983, O'Neill et al., ²⁰⁰³, Ivanov et al., 2010, Yoshida et al., 2011, Chaudhary et al., 2014, Teh et al., 2019, Li et al., 2020]). Notably, alpha‐linolenic acid, ALA, is highly abundant (20%–30%) in A. thaliana. ALA comprises about 5%–12% of fatty acid content in soybeans. Note that the relative composition of LA and ALA varies among plant tissues (e.g., seeds vs. vegetative tissues) where individual LOX isoenzymes are active. We therefore examined the steady‐state kinetics for the GmLOX1 and AtLOX1 reactions with ALA to determine any impact from a different FA substrate. As listed in Table 3, both the reaction rates and activation energies for GmLOX1 and AtLOX1 were found to be comparable for both LA and ALA substrates. Thus, we can conclude that the kinetic properties are comparable for both ALA and LA in AtLOX1 and GmLOX1.

TABLE 3.

Steady‐state kinetic parameters for LA and ALA substrates.

Plant LOX	Substrate	k cat (s−1)	Km (μM)	E a (kcal/mol)
GmLOX1	LA	253 ± 8	22 ± 2	2.2 ± 0.4
	αLA	263 ± 21	24 ± 5	1.0 ± 0.2
AtLOX1	LA	228 ± 5	12.1 ± 0.4	7.2 ± 0.6
	αLA	272 ± 20	24 ± 3	7.7 ± 0.4

To gain some molecular insight into the interactions of these LOXs with their fatty acids, we utilized AlphaFold 3 (AF3) to predict structural models for AtLOX1 and GmLOX1 in the presence of zero, one, or three oleic acid (OA) molecules. These predictions were performed without structural templates to allow the diffusion model to explore novel or less common conformational states. For GmLOX1, template‐based modeling yielded highly consistent results (without pruning any amino acids, RMSD 1.1 Å across five predictions to the crystal structure PDB: 1YGE); however, omitting templates increased the RMSD to 2.2–2.3 Å. The regions of greatest divergence were localized to the loops of the PLAT domain and the catalytic domain lid, which shifts the N‐terminus of the gating π‐helix. The addition of fatty acids (OA) did not significantly alter the global architecture of GmLOX1 (Figure 4a,b), with RMSDs of 2.1–2.3 Å for single‐OA predictions and 2.6–3.0 Å for triple‐OA predictions. The primary conformational fluctuations remained centered in the PLAT domain loops and the catalytic lid (not shown). In all single‐OA predictions for GmLOX1, the fatty acid was positioned within the U‐shaped channel in a tail‐first orientation (Figure 4a,b). This supports the hypothesis that the bulky F557 residue, part of the “regiospecificity switch,” sterically blocks the carboxylate head from deep penetration into the channel. Notably, predictions of the GmLOX1 F557V variant resulted in three out of 10 models adopting a head‐first carboxylate orientation. This structural finding closely mirrors the kinetic data reported by Hershelman et al., who observed 33% 9‐LOX product formation for this variant (Hershelman et al., 2019). Predictions involving three OA molecules did not identify a second highly reproducible binding site; the OA molecule with the highest pLDDT (a measure of local accuracy) was consistently located in the U‐shaped channel. However, AF3 predicted a hydrophobic cleft between the π‐helix and the parallel helix alpha 21 (residues 741–754). This potential substrate access portal represents a novel entry point for soluble LOXs.

FIGURE 4.

AF3 predicted models of plant LOXs with OA. (a,b) Five predictions of GmLOX1 (cartoon loop) with 1 OA or 3 OA (sticks) colored by pLDDT (60 red, 80 white, 100 blue) superimposed on GmLOX1 x‐ray structure (green, cartoon loops). Fe is shown as orange sphere with Lid domain removed for clarity. Key helices are highlighted. OA is predicted to bind “tail‐first” for GmLOX1 in the U‐shaped channel with additional OAs binding between and below Hα2 and Hα21. (c,d) Similar AF3 predictions were made for AtLOX1 superimposed on x‐ray structure (green, cartoon loops). OA was predicted both “head and tail‐first” for AtLOX1 in U‐shaped channel with additional OAs binding above Hα2 and under the “arched helix.” OAs in U‐shaped channel had higher predictive values (colored white to light red) as compared to 3 OA predictions (dark red, sticks) illustrating the confidence values.

Parallel predictions were performed for the 9‐LOX AtLOX1, for which a crystal structure has become recently available for comparison (PDB: 8I6Y). To promote conformational flexibility, these predictions were also performed without templates. RMSD values for apo‐AtLOX1, single‐OA, and triple‐OA models were 1.5–1.6 Å, 2.5–2.8 Å, and 2.7–3.3 Å, respectively (Figure 4c,d). Similar to GmLOX1, the most mobile regions were the PLAT domain loops and the catalytic domain lid; these regions are frequently disordered or absent in plant LOX crystal structures due to high B‐factors. In AtLOX1, the OA orientation was more promiscuous, with both head‐first and tail‐first orientations predicted. The smaller V557 residue in AtLOX1 appears to allow the carboxylate head of the fatty acid to bypass the isopropyl side chain, whereas the bulkier benzyl group of F557 in GmLOX1 prevents this entry. Unexpectedly, the typically bent π‐helix elongated into a straight conformation in the prediction with OA (Figure 4c,d), resembling the structure of human 5‐LOX in an “open” conformation. This elongation facilitated the formation of a substrate access portal above the gating helix, adjacent to the arched helix. This architecture is reminiscent of the crystal structures of coral (Plexaura homomalla) 8R‐LOX bound to substrate and human 15‐LOX‐2 bound to the substrate mimetic C₈E₄ (Kobe et al., 2014; Neau et al., ²⁰¹⁴).

3. DISCUSSION

In this study, we have undertaken a kinetic investigation into the tunneling properties for three plant LOXs and compared them to the well‐characterized model tunneling enzyme from soybean (GmLOX1). While the sampling size was limited due to moderate success in isolation of purified protein, the results reveal three new key features. First, GmLOX1, the canonical hydrogen tunneling enzyme, exhibits Michaelis–Menten kinetic behavior as expected, while the kinetic traces of AtLOX1, SlLOXA, and OsLOX1 all follow sigmoidal behavior that is consistent with positive kinetic cooperativity. A recently published kinetic and structural study of a grape LOXA, which was determined to be a 13‐LOX, similarly showed sigmoidal kinetics (Pilati et al., 2024). The grape LOXA enzyme was isolated as a functional dimer, and the positive cooperativity was suggested to be controlled by this oligomerization. The plant LOXs reported herein, however, are different from the grape LOXA as they were isolated as functional monomers (Figure 1c). The enzyme activity of the grape LOX was further shown to be influenced by the presence of detergents that mimic a lipid surface (Pilati et al., 2024). Similarly, in our hands, we observed an effect from detergent on the Hill coefficient for the AtLOX1 reaction that decreased to near unity. Note that the inclusion of detergents did not influence the maximal velocity of the enzyme at elevated LA concentrations. The most notable effect was therefore only on the lower concentrations of substrate that were in the solubility range in the absence of detergent. Detergent is likely influencing the protein structure in a way that eliminates the apparent kinetic cooperativity observed with LA. Thus, the interpretation of LOX reactions in the presence of detergents could be complicated and needs to be carefully investigated.

FIGURE 1.

Model of plant LOX structure. (a) Global architecture of plant LOX includes two domains: An N‐terminal, C2‐like PLAT domain and primarily alpha‐helical catalytic domain. The arched helix and helix a2 are labeled for reference. (b) Active site of plant LOXs is displayed with a C18 fatty acid (modeled for reference). Panel (b) was rotated by 150° along the x axis, relative to (a). The representative structure is of soybean LOX‐1 (PDB: 3PZW).

To explore the possibility that fatty acids might potentially regulate an open‐to‐close (allosteric) conformational transition in the plant LOXs tested here, we carried out a series of AF3 predictions by varying the number of fatty acids included in the predictions (Figure 4). Predictions with AF3 did not reveal a reproducible fatty acid binding site with high predictive values outside of the U‐shaped channel. However, another interpretation for positive cooperativity was first addressed with glucokinase, a monomeric, single substrate binding enzyme. The sigmoidal kinetics are thought to arise from shifting of ensemble populations at different substrate concentrations (Larion & Miller, 2012). More simply put for positive cooperativity with a Hill coefficient above one, a significant population of enzymes at low substrate concentrations will adopt a conformation less favorable for turnover. However, as substrate concentration is increased, there is a shift of the ensemble population to the more catalytically competent conformation resulting in the sigmoidal curve. This kinetic cooperativity (mnemonic) occurs in monomeric enzymes due to slow, ligand‐induced conformational changes (hysteretic) that create the population shift at higher substrate concentrations. A stable form of the human 5‐LOX exhibits this type of sigmoidal kinetic traces for the oxidation of arachidonic acid. Cooperativity in 5‐LOX was hypothesized to be modulated by a protein conformation of Ha2 (Gallegos et al., 2022). Mutations to residues along the gating helix to shift the ensemble populations from the “closed” kinetically sluggish conformation to “open” conformations not only resulted in lowering K_half, the ligand concentration at half‐maximal saturation, but also increased k _cat. Positive cooperativity is maintained for these variants reinforcing the theory that the ensemble population of protein conformers (open vs. closed) are just shifting with changes in substrate concentration. This behavior is also notable in the AF3 models of AtLOX1 (Figure 4c,d), in which addition of fatty acids does not change the global architecture of the enzyme, but is predicted to shift the ensemble populations of the gating helix (π‐helix) from bent to fully elongated forming a fully accessible substrate portal as the amount of fatty acid is increased. These structural fluctuations of the gating helix may be responsible for the positive cooperativity observed in the kinetics of plant 9‐LOXs, by a similar mechanism previously detailed for human 5‐LOX.

Another plausible explanation for the apparent kinetic cooperativity observed mainly for the 9‐LOXs and not the 13‐LOXs is that multiple substrate‐binding modes could occur for the 9‐LOXs at different substrate concentrations. Structural and kinetic studies on LOXs have shown that substrate can bind in distinct productive and nonproductive orientations within the active‐site cavity, and that ligand binding can alter substrate inhibition or product regio‐specificity(reviewed in Brash & Oliw, ²⁰²⁶). Removal of anchoring residues such as arginine for binding of the carboxylate of the substrate can minorly alter product specificity but increase substrate inhibition as observed for 8R‐LOX (Neau et al., 2014).

The second kinetic feature to emerge from this study is the observation of K_m values for the 9‐LOX reactions with deuterated substrates that are larger than the K_m values for the natural abundant substrates. In comparison, the K_m of GmLOX1 reaction is reduced from 22 to 7 μM for the deuterated substrate. The increased K_m(D) values for the 9‐LOX are attributed to kinetic complexity of the LOX reaction coordinate, as previously described for GmLOX1 as follows (Offenbacher et al., 2018):

Note that the LOX reaction is considered to be virtually irreversible (Glickman & Klinman, 1996); thus, the apparent isotope effect on K_m is not caused by an equilibrium effect from a reverse reaction (Cleland, 2003). From Scheme 2, when k ₂ (i.e., k _cat) is reduced upon isotopic substitution, the commitment to catalysis also becomes small and, as a consequence, the kinetic isotope effect of the second‐order rate constant (^D k _cat/K_m = k _cat/K_m(D)/k _cat/K_m(H)) approaches ^D k _cat. For example, at 25°C, the ^D k _cat/K_m for GmLOX1 is ca. 24 (from data in Table 2). This value increases with increasing temperature to nearly match ^D k _cat at elevated temperatures (Glickman & Klinman, 1995). K_m will not only reflect on and off rates, which would be regulated in part by the conformational change included in the Hill coefficient, but is also a composite of the partitioning of the enzyme from an initially bound, unproductive enzyme‐substrate complex (ES in Scheme 2) to a productive complex (ES′). This pre‐chemical reorganization of the substrate, denoted by K_eq, is large for wild‐type GmLOX1, though can be modulated by select active site mutations or the presence of an effector (Hu et al., 2019; Mogul et al., ²⁰⁰⁰; Offenbacher et al., ²⁰¹⁸). The K_eq for plant 9‐LOXs, based on the ^D k _cat/K_m, is expected to be smaller, and sensitive to isotopic substitution on substrate. From a structural perspective, the head‐first binding orientation of the substrate in 9‐LOXs may place the pentadiene scaffold in a binding pose that is less optimal for hydrogen abstraction. In this model, the 9‐LOX enzyme would sample multiple substrate conformations before reaching the optimal geometry for catalysis (i.e., ES'), and the deuterated substrate would affect this conformational search, resulting in a larger apparent K_m for this substrate without influencing the cooperative behavior of the enzyme. Consistent with this, the Hill coefficient (n) is retained with the deuterated substrate, indicating that the 9‐LOXs maintain sigmoidal kinetics despite the use of elevated enzyme concentrations (90‐to‐120 nM for D‐LA compared to 2–4 nM for H‐LA) necessary for the kinetic measurements.

SCHEME 2.

Reaction coordinate for GmLOX1. E and S represent the enzyme (LOX) and substrate (LA), respectively. K_eq represents a pre‐chemical internal reorganization from the initial enzyme‐substrate (ES) complex to a productive (ES′) complex.

The third feature to emerge is that the activation energies for the two established 9‐LOXs, AtLOX1 and SlLOXA, are elevated with respect to GmLOX1. Because hydrogen wave function overlap is inherently temperature independent, the weak temperature dependence of hydrogen transfer in GmLOX1 (i.e., E _a(H) of 2 kcal/mol) corresponds to a functional role for thermal motions of the protein to modulate the donor and acceptor distances to promote efficient tunneling. GmLOX1 was previously characterized with a defined thermal network (Figure 5a) that originates at the ferric hydroxide cofactor and the invariant Leu clamp residue (L546 in GmLOX1) and terminates at a loop (residues 317–334); the latter was identified to exhibit activation energies in hydrogen deuterium exchange (HDX) experiments that correlated with E _a values for hydrogen tunneling for a series of activity‐altered protein variants (Offenbacher et al., 2017). Room temperature x‐ray analysis identified several bulky aliphatic residues, mostly Ile, Leu, Phe, and Val, connecting the active site to the thermal loop in a cone‐like arrangement (Figure 5a) (Zaragoza et al., 2023). Clusters of leucine, isoleucine, and valine (ILV) sidechains have been supported to play an important role in protein stability for high‐energy protein states (Kathuria et al., 2016), and have been implicated in the thermal activation of enzyme catalysis (Offenbacher & Guy, 2024; Zaragoza et al., ²⁰²³). In GmLOX1, functional role of this network has been proposed to facilitate the anisotropic transmission of thermal energy from the solvent at the protein surface to the active site via a productive conformational restructuring of the protein promoting effective wave function overlap for tunneling. This thermally‐initiated restructuring of the protein likely represents the origin of the E _a for catalysis in GmLOX1 (Zaragoza et al., 2023).

FIGURE 5.

Analysis of the thermal network of GmLOX1 and the structural homology to AtLOX1. The thermal loop of GmLOX1 (residues 317–334) is colored in firebrick; AtLOX1 structure has a similar loop feature (residues 337–354; salmon). The green and cyan spheres represent sidechains in the putative thermal networks of GmLOX1 and AtLOX1, respectively.

The observation of elevated E _a values for AtLOX1 and SlLOXA, presented herein, raises the question into the detailed mechanism(s) of thermal activation of hydrogen tunneling for plant 9‐LOXs. A structure of AtLOX1 was recently deposited in the Protein Data Bank (PDB: 8I6Y; 3.26 Å). Similar to GmLOX1, this structure of AtLOX1 has a loop‐like feature in a homologous positioning to the “thermal loop” of GmLOX1 (Figure 5b). Further, analysis of the amino acids reveals a high propensity of bulky and/or aliphatic sidechains, including Ile, Leu, Phe, and Val, lining the interface between the thermal loop and the active site.

Evolutionary conserved residues typically indicate they are critical to an enzyme's structure, function, and/or regulation. Phylogenetic analyses and sequence conservation have been explored previously for plant LOXs, albeit at a smaller scale (Bannenberg et al., 2009; Brash, ¹⁹⁹⁹; Feussner & Wasternack, ²⁰⁰²). Here, we investigated whether surface mapping of phylogenetic sequence conservation for the reference GmLOX1 protein may reveal new insights into this thermal network at a broader scale than possible by deep biochemical analyses of many individual LOXs. To this extent, we used BLASTP analysis of Arabidopsis LOX1 against 200+ publicly available genomes and reference transcriptomes that revealed 3582 putative LOX homologs (see Data S2). Not surprisingly, residues closer to the catalytic site were much more conserved than surface residues among all phylogenetic groups (Figure S3). This is particularly true for the side chains that line the arched helix (535–557 of GmLOX1; Figure S3c) that have considerable conservation, with a clear exception of the “switch residue” that is predominantly a Phe (>95%) for putative 13‐LOXs and a Val (>90%) for putative 9‐LOXs and defines the C‐terminus of the arched helix. Important sequence differences in the thermal network and active site regions were revealed (Figure 6, and Figure S3) by using inferred phylogeny to help group putative 9S vs. 13S LOXs. After clustering these based on phylogenetic group (Figure S44) it was apparent that key conserved (or non‐conserved) residues in the active site (relative to GmLOX1) are relatively consistent whether we are analyzing LOX from closely related species (e.g., dicots) or more diverged species (e.g., algae).

FIGURE 6.

Surface mapping of phylogenetic sequence conservation on GmLOX1. (a) Sequence identity conservation is mapped on GmLOX1 model (PDH: 1f8n modeled with LA and Fe) with (top) all LOX sequences from 200+ plant genomes/transcriptomes, (middle) putative 9S LOX proteins, and (bottom) putative 13S‐LOX proteins. (b) Sequence conservation mapped on only thermal loop 317–334 (all structural features shown in Figure S3). (c) Sequence logos for thermal loop. Details of genomes/transcriptomes analyzed, individual protein IDs, sequences, and conservation analyses are provided in Data S2.

Figure 6 and Figure S3 present a detailed phylogenic analysis of the GmLOX1‐detected thermal network. It is worth highlighting the conservation of invariant Leu546 (across the >3500 putative LOX sequences) that positions the reactive carbon for catalysis in LOXs and is expected to be the critical mediator of long‐range thermal energy transfer from the protein surface (Offenbacher et al., 2017). In these phylogenic models, the conservation of the thermal loop (317–334 in GmLOX1) is moderately low. One notable exception is the high occurrence of a bulky hydrophobic residue at position 317. Previous mutational analysis of Y317 in GmLOX1 to smaller aliphatic residues resulted in altered activation energies and a disturbance of the substrate binding (Offenbacher et al., 2017; Offenbacher et al., ²⁰²⁰). Interestingly, we found an invariant proline sidechain at position 330 and a highly conserved proline at 309 in GmLOX1 that bookend the loop feature, presumably to disrupt the flanking helical segments. These proline residues could serve as a target for future studies aimed at investigating loop structure and dynamics as it relates to tunneling proficiency in plant LOXs.

Moving from the surface loop to the active site, the network passes through two major helices: α2 and α21 (Figure 5 and Figure S3). There is considerable sequence variation in helix α2 and helix α21 (residues 741–754 in GmLOX1), which runs parallel to the arched helix at the active site. However, the residues that are expected and/or predicted to play a role in productive energy transfer (i.e., 262, 266, 270 in helix α2 and 742, 746, and 750 in helix α21) are predominantly Ile, Leu, or Val (and sometimes Phe) side chains (Figure S3). A recent molecular dynamics (MD) simulation study incorporating a correlated‐motion protocol revealed that there could be two prominent pathways for thermal energy transfer from the loop to L546 in GmLOX1 (Jiang et al., 2026). Both pathways proceed through I552. Further, potential alternative routes all predicted to converge to L266 in helix α2 of GmLOX1, prior to accessing the common pathway to L546. Both the L266 and I252 sidechains are represented by high occurrences of Leu, Ile, or Val residues across the plant LOXs (Figure S3c,d).

The existence of a primary thermal conduit proceeding through Ha2 in GmLOX1 could provide one potential rationale for the altered catalytic behavior seen in AtLOX1. Our AF3 models of AtLOX1 predict a fatty acid‐mediated regulation of the ensemble of helix α2 conformations (Figure 4b), which is supported by the observation of sigmoidal kinetics (Figure 2c). The kinetic consequence of this dynamic sampling of Ha2 conformers could translate to a larger apparent enthalpic barrier for catalysis. Because both the thermal activation network and substrate binding seem to intersect at the same helix (namely, helix α2), the dynamics of this helix may be balanced to serve as a vital regulatory mechanism in vivo for controlling proper levels of 9‐hydroperoxide output. While the mechanism(s) of the thermal activation of the plant 9‐LOXs (e.g., AtLOX1) remain to be experimentally resolved, we can conclude from our results that the presence of a loop, homologous to the thermal loop in GmLOX1, and a conserved network enriched in bulky, aliphatic residues to mediate thermal energy transfer from the solvent to the active site is not correlated with low activation barriers for hydrogen tunneling across all plant LOXs, as previously hypothesized (Offenbacher et al., 2017; Ohler et al., ²⁰²⁴).

Finally, it is worth noting that OsLOX1 exhibits a reduced E _a of ca. 1 kcal/mol that is comparable to the 13‐LOX of GmLOX1. However, the activity of OsLOX1 followed sigmoidal behavior consistent with positive cooperativity. A 13‐LOX from Arabidopsis (isoform 2) was also characterized with Hill coefficients greater than 1 (Kaur et al., 2024). It is unclear if the sigmoidal behavior for AtLOX2 is attributed to its oligomeric status or if it is due to pH/buffer conditions explored, as the Hill coefficient approached unity with increasing pH. Further biochemical and structural studies may be required to resolve these kinetic properties.

4. CONCLUSIONS

Sigmoidal kinetics coupled with protein modeling by AF3 predict that the monomeric plant LOXs, investigated in the current work, are likely regulated by fatty acids entailing the restructuring of the gating helix α2. These findings present a new model for substrate acquisition in plant LOX catalysis. In addition, the two 9‐LOX representatives exhibited activation energies that were elevated from that of GmLOX1. Broad phylogenic analysis of LOX sequences across the plant kingdom was spatially mapped onto available x‐ray crystal structures. This analysis points toward a central basis for the mechanism of thermal activation of H‐tunneling in plant LOXs that includes a conserved loop‐like structure and an intermediary network enriched in Ile, Leu, and Val sidechains. This work builds on the growing body of evidence that spotlights a role for classical thermal motions of the protein in mediating and enhancing quantum mechanical phenomena in biological function.

5. MATERIALS AND METHODS

5.1. Materials

Genes encoding ZmLOX5 and ZmLOX10 were synthesized by GenScript (Piscataway, NJ). The vector encoding SlLOXA was a generous gift from Prof. Alan Brash. All other plant LOX (see Table 1) genes were subcloned by PCR into a pET28 bacterial expression vector in line with an N‐terminal hexa‐histidine (His₆) tag. The perdeuterated substrate (D₃₁LA) was obtained as the methyl ester from Cambridge Isotope Labs (Tewksbury, MA). The 11,11‐D₂‐LA substrate was synthesized (Offenbacher et al., 2016).

5.2. Plant LOX expression and purification

GmLOX1 was expressed and purified as described previously (Jakobowski et al., 2024). For all other plant LOXs, the LOX vectors were transformed in E. coli BL21(DE3) Codon Plus RIL competent cells and added to overnight starter cultures supplemented with 0.4% (w/v) glucose and the appropriate antibiotic. The starter culture was incubated at 30°C overnight with shaking. After 16–20 h, the starter cultures were used to inoculate large scale ZYM‐5052 autoinduction media (supplemented with antibiotic) in 2.8 L Fernbach flasks. The cultures were incubated at 37°C and 180 rpm, until the optical density (OD₆₀₀) reached 1.0, at which point the temperature was decreased to 18°C for continual shaking for 24–40 h. Cell pellets were harvested and resuspended in 100 mL lysis buffer (50 mM sodium phosphate, 100 mM NaCl, pH 8, 8% (w/v) glycerol, and 2 mM magnesium sulfate supplemented with 10 mg lysozyme, 4 μL DNAse I, and 20 mg AEBSF protease inhibitor). The cell suspensions were lysed by sonication and the cell debris was removed by centrifugation. From the supernatant, the LOX proteins were captured and purified by Ni‐NTA using gravity. Briefly, after loading the protein onto the column (10 mL), the material was washed with 10 column volumes (CVs) of buffer A (20 mM TRIS, pH 8, 500 mM NaCl, and 20 mM imidazole) and then eluted with 32–5 CVs of buffer B (20 mM TRIS, pH 8, 500 mM NaCl, and 200 mM imidazole). The eluted protein was collected using a fraction collector. The presence of protein was confirmed with enzyme activity and SDS‐PAGE. LOXs were further purified using SEC with a HiPrep Sephacryl™ S‐200 HR preparative column on an ÄKTA go fast‐protein liquid chromatography (FPLC). The protein was eluted with the running buffer (50 mM HEPES, 150 mM NaCl, pH 7.5). Aliquots were frozen in liquid N₂, and stored at −80°C until further use. The iron content in the LOX samples was determined by the ferrozine (spectrophotometric) assay as described elsewhere (Jakobowski et al., 2024).

5.3. Steady‐state enzyme kinetics

Steady‐state kinetics for the LOX reactions with LA were monitored continuously at a single wavelength for the detection of product formation (λ = 234 nm) using a UV–vis spectrometer equipped with a temperature‐controlled, circulating water bath. Initial experiments were performed at 25°C across a series of pH values from 6.0 (50 mM MES), 7.0 and 8.0 (50 mM HEPES), and 9.0 (50 mM sodium borate), supplemented with 150 mM sodium chloride. For a typical LOX reaction, six to eight LA concentrations were tested, ranging from 2 to 100 μM, at a constant temperature, using the natural abundance substrate. The Michaelis‐Menton kinetic parameters (k _cat and K_m) were determined and reported in Table 2. Steady‐state kinetics of the LOX reactions were performed across a temperature range (10, 15, 20, 25, 30, 35, and 40°C). The activation energy, E_a, was determined from the slope of the corresponding Arrhenius equation (in the format of y = mx + b):

$$\ln k_{\mathit{cat}}=-\frac{E_a}{R}\frac{1}{T}+\ln A$$ (1)

where inverse temperature is the x‐axis, R is the ideal gas law constant (in kcal/mol K), and A is the Arrhenius pre‐factor.

Primary deuterium kinetic isotope effects were determined for the first‐order (^D k _cat) and second‐order (^D k _cat/K_M) rate constants, using the comparison of the natural abundance and deuterium‐labeled (D₃₁ or 11,11‐D₂) LA substrates. For reactions with D‐LA, the substrate concentrations were the same as for natural abundance LA but the enzyme concentration was increased by 10‐ to 20‐fold over the unlabeled substrate to ensure the reaction was completed in a timely manner. The primary deuterium kinetic isotope effects were calculated by the ratio of the rate constant of the light isotope over the heavy isotope:

(2)

Igor Prov (v5.0) was used to plot kinetic data and perform fitting analyses. All kinetic parameters are reported as the mean ± standard error (s.e.). AIC analysis was performed with GraphPad Prism (GraphPad software).

5.4. Circular dichroism (CD) spectroscopy

The structural integrity of the isolated plant LOXs was assessed using a Jasco model J‐815 CD spectrometer at room temperature with bandwidths of 2 nm and with a Starna cell (pathlength of 0.1 cm). The samples were diluted to about 2 μM using 25 mM potassium phosphate, pH 7.0, so that the PMT high voltage (HT) remained ≤600 V in the range of 190–260 nm. Samples were recorded at 25°C.

5.5. Differential scanning calorimetry

LOX protein thermodynamics were measured using a TA‐instruments Nano‐DSC microcalorimeter. The proteins were prepared at 30 μM concentrations in 50 mM HEPES, 150 mM NaCl, pH 7.5 buffer. For comparison, GmLOX1 was prepared in 50 mM borate, pH 9.0 buffer. The DSC experiments were conducted in “heat” only mode over a temperature range of 30–90°C at a rate of 1°C/min and constant pressure of 3 atm. The raw data were converted to molar heat capacity and analyzed using TA‐instruments NanoAnalyze software. The baseline was integrated using a 4th order polynomial to determine the melting temperature (T_m) and enthalpy of unfolding (ΔH°). Experiments were run in duplicate.

5.6. Structural model predictions

AlphaFold3 (AF3) was used to predict the structures of GmLOX1 and AtLOX1. For both proteins, predictions were conducted with and without structural templates to assess possible variations in conformation. When templates were included for GmLOX1, the predicted structures closely matched existing crystal structures, with root mean square deviations (RMSDs) of approximately 1.0 Å. In contrast, predictions without templates resulted in higher RMSDs of 2.2–2.3 Å, indicating greater conformational flexibility. Five structural models were predicted for each plant LOX with the catalytic iron (Fe), Fe and one OA, and Fe and 3 OAs. The predicted template modeling (pTM) score for the protein has confidence values consistently scored above 90% for all model protein predictions. The interface predicted template modeling score (ipTM) for predictions of LOXs with Fe³⁺ ion is 0.99. In single OA co‐predictions, the ipTM scores are close to 0.9 for both plant LOXs. However, the three OAs co‐predictions were near 0.8 for all models. The two additionally fatty acids had lower ipTMs with little reproducibility of binding sites. From the AF3 server website: “Values higher than 0.8 represent confident high‐quality predictions, while values below 0.6 suggest likely a failed prediction.” Models were analyzed in PyMOL 3.1.6.1 with RMSDs calculated by alignment without pruning amino acids (cycles = 0). GmLOX1 was aligned to the x‐ray model of Soybean LOX‐1 (PDB ID: 3PZW), and AtLOX1 to the x‐ray model of Arabidopsis LOX1 (PDB ID: 8I6Y).

5.7. Sequence conservation analysis

LOX homologs of AtLOX1 were identified from predicted protein sequences from over 227 plant genomes and reference transcriptomes (see Data S2) using BLASTP and structural‐based sequence conservation analyses as described previously (Cannon & Horn, 2024). Very similar BLASTP datasets (~98% overlap) were determined using other Arabidopsis LOX as queries (only LOX1 homologs were used for further analysis). From the initial BLASTP search, unique predicted protein sequences between 600 and 1200 amino acids long and an (query vs. subject) E‐Value ≤1 × 10⁻¹⁰ were retained for further analysis (total of 3582 sequences) to filter out low‐quality assembled sequences. Sequences were initially binned into one of five phylogenetic groups: (1) eudicots, (2) monocots, (3) Archaeplastida and Chlorophyta, (4) Bryophyta, Lycopodiophyta, and Marchantiophyta, and (5) gymnosperms (see Data S2). To aid global sequence alignments via MUSCLE (Edgar, 2004) and sequence conservation analysis to a single reference protein structure GmLOX1 across diverse phylogenetic groupings, Arabidopsis and Soybean LOX isoenzymes were initially included for all alignments. A MAFFT global sequence alignment (Katoh et al., 2017) was performed using default parameters (FFT‐NS‐2). After alignments for non‐eudicot groups, all soybean and Arabidopsis LOX were removed other than GmLOX1. The phylogeny was inferred using the Neighbor‐Joining method with bootstrap analysis performed in MEGA12 (Figure S5 with tree data in Data S2) (Felsenstein, 1985; Kumar et al., ²⁰²⁴; Saitou & Nei, ¹⁹⁸⁷; Zuckerkandl & Pauling, ¹⁹⁶⁵). 9S‐ and 13S‐LOX protein groups were predicted based on the clustering around Arabidopsis 9S‐LOX (LOX1, LOX5) and 13S‐LOX (LOX2‐4, LOX6, GmLOX1) and used for additional sequence alignment filtering. A sequence identity, BLOSUM62, and hydrophobicity score for each amino acid in alignment were calculated using the GmLOX1 as a reference. These values were used as PDB “beta‐factors” and overlaid on the GmLOX1 PDB 1f8n, with sequence conservation mapping visualized using PyMOL. Sequence conservation logos were created using WebLogo (Crooks et al., 2004).

AUTHOR CONTRIBUTIONS

Patrick Horn: Investigation; data curation; formal analysis; writing – original draft. Jordan King: Investigation. Katherine Ray: Writing – review and editing; investigation. Adam R. Offenbacher: Conceptualization; supervision; project administration; writing – original draft; funding acquisition. Alex Kockler: Investigation; methodology; formal analysis; writing – review and editing. Joi Walker: Data curation; formal analysis. Eric Anderson: Investigation. Nathaniel Gilbert: Investigation; formal analysis; writing – original draft.

Supporting information

Figure S1. Biochemical characterization of the isolated plant LOXs: panel (a) SDS‐PAGE and panel (b) CD spectra of the plant LOXs, normalized based on their protein concentration and residue number (MRE, mean residue ellipticity). The buffer was 25 mM potassium phosphate (pH 7.5). Panel C: calibration curve for analytical Superdex 200 increase 10/300 with the peak elution volumes plotted for plant LOXs examined in this report. Protein size standards are represented as black circles. Raw analytical SEC traces are presented in Figure S6.

Figure S2. Kinetic saturation plots for the reaction of AtLOX1 with (a) LA following pre‐activation of AtLOX1 to Fe³⁺ form, (b) LA in buffer supplemented with 0.01% Triton X‐100, and (c) deuterium‐labeled substrates (D₂LA: 11,11‐D₂‐LA; and D₃₁LA). The experiments represent reactions collected at 25°C in 50 mM HEPES (pH 7.0) and 150 mM sodium chloride buffer. The data represent the average from three to four measurements, with each data set collected on three to four separate days.

Figure S3. Extended surface mapping of phylogenetic sequence conservation on GmLOX1. (a) Similar to Figure 6, sequence identity conservation is mapped on entire GmLOX1 model (PDB: 1f8n modeled with LA and Fe). Figure S3 is continued on the next page. (b) Zoomed in on active site with (top) all LOX proteins, (middle) putative 9S LOX proteins, and (bottom) putative 13S‐LOX proteins. (c‐g) Sequence conservation mapped on individual structure features from (b) with positions shown and associated sequence logos. Details of genomes/transcriptomes analyzed, individual protein IDs, sequences, and conservation analyses are provided in Data S2. Figure S4. Surface mapping of phylogenetic sequence conservation on GmLOX1 active site. Similar to Figure 6, sequence identity conservation is mapped on entire GmLOX1 model (PDB: 1f8n modeled with LA and Fe). Zoomed in on active site but in this case scored based on phylogenetic groups (a) All LOX (N = 3582 LOX sequences), (b) Dicots (N = 2592), (c) Monocots (N = 664), (d) Gymnosperms (N = 62), (e) Bryophytes (N = 97) and Algae (N = 21). Details of genomes/transcriptomes analyzed, individual protein IDs, sequences, and conservation analyses are provided in Data S2.

Figure S5. Phylogenetic Tree Analysis: Evolutionary relationships of 3582 taxa.

Figure S6. SEC elution profiles of (a) GmLOX1, (b) OsLOX1, (c) SlLOXA, and (d) AtLOX1. The running buffer was 50 mM HEPES (pH 7.5) buffer with 150 mM sodium chloride.

Table S1. Literature reported Activation energies for various LOX reactions.

Table S2. Protein stability thermodynamic parameters of plant LOXs by DSC ^a .

Table S3. Fatty acid content (in percentage of total fats) in select plants.

Data S1. Supporting information.

Figure S4. Supporting information.

Data S2. Supporting information.

Kockler A, Ray K, King J, Anderson E, Walker J, Gilbert N, et al. Description of the rate‐limiting hydrogen tunneling mechanism for plant 9‐ and 13‐lipoxygenases. Protein Science. 2026;35(7):e70647. 10.1002/pro.70647

DATA AVAILABILITY STATEMENT

The data that supports the findings of this study are available in the supplementary material of this article.