Biochemical Characterization of 13-Lipoxygenases of Arabidopsis thaliana

13-lipoxygenases (13-LOX) catalyze the dioxygenation of various polyunsaturated fatty acids (PUFAs), of which α-linolenic acid (LeA) is converted to 13-S-hydroperoxyoctadeca-9, 11, 15-trienoic acid (13-HPOT), the precursor for the prostaglandin-like plant hormones cis-(+)-12-oxophytodienoic acid (12-OPDA) and methyl jasmonate (MJ). This study aimed for characterizing the four annotated A. thaliana 13-LOX enzymes (LOX2, LOX3, LOX4, and LOX6) focusing on synthesis of 12-OPDA and 4Z,7Z,10Z)-12-[[-(1S,5S)-4-oxo-5-(2Z)-pent-2-en-1yl] cyclopent-2-en-1yl] dodeca-4,7,10-trienoic acid (OCPD). In addition, we performed interaction studies of 13-LOXs with ions and molecules to advance our understanding of 13-LOX. Cell imaging indicated plastid targeting of fluorescent proteins fused to 13-LOXs-N-terminal extensions, supporting the prediction of 13-LOX localization to plastids. The apparent maximal velocity (Vmax app) values for LOX-catalyzed LeA oxidation were highest for LOX4 (128 nmol·s−1·mg protein−1), with a Km value of 5.8 µM. A. thaliana 13-LOXs, in cascade with 12-OPDA pathway enzymes, synthesized 12-OPDA and OCPD from LeA and docosahexaenoic acid, previously shown only for LOX6. The activities of the four isoforms were differently affected by physiologically relevant chemicals, such as Mg2+, Ca2+, Cu2+ and Cd2+, and by 12-OPDA and MJ. As demonstrated for LOX4, 12-OPDA inhibited enzymatic LeA hydroperoxidation, with half-maximal enzyme inhibition at 48 µM. Biochemical interactions, such as the sensitivity of LOX toward thiol-reactive agents belonging to cyclopentenone prostaglandins, are suggested to occur in human LOX homologs. Furthermore, we conclude that 13-LOXs are isoforms with rather specific functional and regulatory enzymatic features.


In Silico and Subcellular Targeting of Arabidopsis 13-lipoxygenases
Based on annotations in publicly available databases and using bioinformatics tools (see Section 4.1), the amino acids sequences of the four 13-LOX isoforms were dissected into putative chloroplast transit sequences, as well as PLAT and catalytic LOX domains ( Figure 1A). The 13-LOX sequences differed in length, amino acid composition and physicochemical properties. The calculated isoelectric points (pI) of the LOX domains ranged between 5.2 and 6.2. A major difference was seen for the PLAT domain; while LOX6, LOX4 and LOX3 had a basic pI value (9.1-9.5), the respective domain of LOX2 was strongly acidic (4.8). Distinct characteristics of 13-LOXs were seen when comparing their amino acid sequences, revealing that LOX3 and LOX4 are similar, with 85.0% identity, but only 53% identical to LOX6 and 45% to LOX2, whereas LOX2 is 49% identical to LOX6 [20]. Among the conserved features are metal cofactor-coordinating amino acid residues and the determinants for regio-and stereospecificity (for conventions, see [21]). In addition, the absence of the amino acids typical for manganese LOX, such as Phe 332 or Phe 526, supports 13-LOXs annotation as Fe-LOXs [22]. Interestingly, invariant Trp and one or two Cys residues are located in the PLAT domain of 13-LOX (see domain map Figure 1A). In conjunction with the participation of the PLAT domain in LOX stabilization [23][24][25] and binding to LOX inhibitors, this in silico finding served as a starting point for further studies, as addressed below. Taking into account all active site residues identified or suggested before [20], it is noteworthy that a Pro residue in LOX3 and LOX4 and a Thr residue in LOX2 and LOX6 are in proximity to the 13-S determining Phe ( Figure 1B). The 3D models presenting 13-LOXs and their active sites are shown in Figure 1B,C. Arabidopsis 13-LOX enzymes are most probably plastid-localized, due to their Nterminal extension [2,12,15]. To explore their subcellular targeting, we fused yellow fluorescent protein (YFP) downstream of the putative transit peptide sequences of LOXs, and transiently expressed these constructs in protoplasts isolated from A. thaliana leaves in a similar manner, as described for the analysis of rice LOX-1 (Q9FSE5, [26]). As seen in Figure 2, the distribution of the 13-LOX transit peptide:YFP protein fusions matched the pattern of chlorophyll autofluorescence. Merging these images proved co-localization, and indicated that the predicted 13-LOX plastid transit peptides are sufficient to target the reporters to the plastids. cysteinyl and charged amino acid residues. Tertiary structures of 13-LOX models, based on the crystal structure of SLOX (4wfo), are depicted in grey. Basic (K, R, H) and acidic (D, E) amino acid side-chains (AA) that are positively and negatively charged under physiological pH values are depicted as sticks in blue and red. Of the aliphatic index (AI) determining AA, only Cys (yellow) are shown. The amount and percentage of the pI value determining AA in each 13-LOX, together with Cys residues, are also listed. The PLAT and LOX domains exemplified for LOX2 are circled. (C) Active sites (backbone not shown) of 13-LOX are depicted in blue in reference to (A), with 13-S determining and catalytic residues highlighted in black.
Arabidopsis 13-LOX enzymes are most probably plastid-localized, due to their N-terminal extension [2,12,15]. To explore their subcellular targeting, we fused yellow fluorescent protein (YFP) downstream of the putative transit peptide sequences of LOXs, and transiently expressed these constructs in protoplasts isolated from A. thaliana leaves in a similar manner, as described for the analysis of rice LOX-1 (Q9FSE5, [26]). As seen in Figure 2, the distribution of the 13-LOX transit peptide:YFP protein fusions matched the pattern of chlorophyll autofluorescence. Merging these images proved co-localization, and indicated that the predicted 13-LOX plastid transit peptides are sufficient to target the reporters to the plastids. revealed that the indicated sequences target the reporter to the chloroplast, as shown by the overlay (merge) of the YFP fluorescence signal (green) with chloroplast autofluorescence (magenta). The same pattern of co-localization was observed in several protoplasts. Size bars: 10 μm.

Purification and Enzymatic Characteristics of Arabidopsis 13-Lipoxygenases
To gain further information on biochemical properties, the sequences encoding the mature At-13-LOXs (LOX2 71-896 , LOX3 71-919 , LOX4 71-926 , LOX6 41-917 ) were placed as Histag fusion constructs under the control of the IPTG-inducible promoter in E. coli expression vectors. This followed the strategies described in previous reports [21,27,28]. Total proteins were extracted from induced expression strains by mild sonication, and the soluble fusion proteins were obtained in the supernatant containing active LOXs. No activities were observed using control homogenates prepared from E. coli cells devoid of LOX plasmids, or when 13-LOX-containing solutions had been boiled (30 s, 85 • C). As shown for LOX6, (see Supplementary Figure S1) this demonstrated that the herein-utilized 13-LOX protein sources were active and specific for PUFA oxidation, containing no impurities that would have caused the unspecific oxidation of PUFA. The theoretical and, thus, expected molecular masses of the recombinant His-tagged 13-LOX were 95,061, 96,791, 97,830 and 100,912 kDa. As judged by SDS-PAGE analysis, dominant bands appeared at the expected mass size (see Figure S2), the proteins were estimated as being of~90% (LOX4),~80% (LOX2 and LOX3) and~50% (LOX6) purity. LOX3, LOX4 and LOX6, unlike LOX2, were enriched by~5-fold when cleared lysates were purified via metal affinity chromatography. This is demonstrated via SDS-PAGE ( Figure S2) and activity analysis (activity analysis is shown for LOX6 in Figure S1). Further attempts to purify 13-LOX were not performed due to our observation of unexpected elution positions during gel-filtration (see Figure S2D) and the instability of 13-LOX [12]. We thus decided to utilize LOX3, LOX4 and LOX6 eluates and LOX2-cleared lysate for subsequent 13-LOX analysis, if not mentioned otherwise.
As chloroplast pH-values range between 7.0 in the dark and >8.0 in the light, we recorded the activities of 13-LOX with their optimal substrate LeA between pH 6.6 and 8.6, to detect the pH-optima of 13-LOX. All members showed high activity around neutral pH and low activity at pH 8.6 ( Figure S2C). LOX2 was most sensitive and LOX4 was least sensitive toward pH changes. At pH 6.6, LOX3 and LOX6 activity were lowered by~30%, in contrast to LOX2 and LOX4, when compared to their optimal pH values at pH 6.6 (3.9 ± 0.4 nmol·s −1 ·mg −1 ) and pH 7.2 (62.3 ± 0.4 nmol·s −1 ·mg −1 ). The apparent affinities of 13-LOX enzymes toward LeA were highest for LOX6 (K m app = 1.2 ± 0.4 µM) and lowest for LOX2 (K m app = 26.3 ± 2 µM). The respective velocities of product formation were highest for LOX3 and LOX4 and lowest for LOX2 and LOX6, the latter two had apparent V max -values~45-fold lower than that of LOX4 (128 ± 18 nmol·s −1 ·mg −1 ), which is partly due to the different degrees of purity. The purities and iron loads below 100% were not considered in the activity calculation or related properties studied herein.
We intentionally tested whether the 22:6 (ω-3) PUFA DHA is a substrate of 13-LOX. As shown in Table 1, 13-LOX enzymes dioxygenated DHA, with rates of~7 to 56% relative to LeA, of which LOX2 was the most active enzyme. In line with the previous data on SLOX and LOX6 [33], a dominant oxylipin derivative of DHA, after incubation of DHA with LOX2, fractioned at Rt~13 min. The MS analysis of this fraction revealed the presence of HPDH, as indicated by the major signal appearing at m/z of 383.2 due to the association of HPDH (MW = 360.5 g·mol −1 ) with sodium [M+Na] + . The molecular ion was subjugated to an MS 2 experiment and the formation of the fragment ion m/z 297 confirmed the position of hydroperoxidation at ω-6, due to the loss of an 86 Da fragment (C 5 H 10 O), in accordance with the findings of [33]. This hypothesis was further endorsed by the confirmation of the sum formulas of the molecular ion and the m/z 297.2 fragment ion ( Figure S5 and Table S2). The specificity of LOX3 and LOX4 was rather poor for producing 17-HPDH (see Figure S5). We observed that 13-LOXs were highly sensitive toward PUFAs or their oxidation products, whereas LOX2 was most robust. As analyzed for their favored substrates, more than 50% inhibition occurred when the enzymes (0.22 µg/µL) were pre-incubated with 29 µM LA (LOX6) or 118 µM LA (LOX3 and LOX4). The effect of LA and LeA on 13-LOX inhibition is depicted in Figure S6, revealing similar trends when pre-incubated with LeA.

Arabidopsis 13-LOX Isoforms Mediate the Synthesis of 12-OPDA and OCPD
13-HPOT is the substrate for the subsequent synthesis of 12-OPDA. As shown previously for LOX6 [33], 12-OPDA was produced by all At-LOX lysates, coupled to AOS and AOC, as revealed by the comparison of retention times with the standard (Figure 4).
The result underscores the involvement of 13-LOXs in 12-OPDA synthesis and the potential mechanism of feedback modulation by 12-OPDA, as investigated below. Due to the aforementioned criteria, LOX2 was, thus, utilized for OCPD synthesis. As revealed by the comparison of retention times with the OCPD standard, LOX2 like SLOX and LOX6, coupled in a one-pot synthesis with AOS and AOC [33], enabled the formation of the promising bioactive compound OCPD from DHA.

Effect of Metal Ions and LOX-Associated Phytohormones on Arabidopsis 13-LOX Activities
Ca 2+ and Mg 2+ play important roles in enzyme regulation. Stromal Mg 2+ increases in the light phase, when the pH is >8.0 [34]. In addition, the activities of mammalian and plant lipoxygenases are affected by divalent cations [23,35,36]. Therefore, we investigated whether the same result occurs for 13-LOXs and included activity studies with Cu 2+ and Cd 2+ . At a tested concentration of 1 mM, 13-LOX enzymes were differentially stimulated by Ca 2+ and Mg 2+ . In comparison to control (mean values n ≥ 3), LOX2 activity increased more 7 of 23 in the presence of 1 mM Ca 2+ (590%) than 1 mM Mg 2+ (70%). LOX4 and LOX3 were slightly activated (6% and 35%) by Mg 2+ , whereas Ca 2+ stimulated LOX3 activity. Ca 2+ and Mg 2+ enhanced LOX6-catalyzed oxygen consumption by 21% and 42%, respectively (Table 2). dation products, whereas LOX2 was most robust. As analyzed for their favored substrates, more than 50% inhibition occurred when the enzymes (0.22 µ g/µ L) were pre-incubated with 29 µ M LA (LOX6) or 118 µ M LA (LOX3 and LOX4). The effect of LA and LeA on 13-LOX inhibition is depicted in Figure S6, revealing similar trends when preincubated with LeA.

Arabidopsis 13-LOX Isoforms Mediate the Synthesis of 12-OPDA and OCPD
13-HPOT is the substrate for the subsequent synthesis of 12-OPDA. As shown previously for LOX6 [33], 12-OPDA was produced by all At-LOX lysates, coupled to AOS and AOC, as revealed by the comparison of retention times with the standard (Figure 4). The result underscores the involvement of 13-LOXs in 12-OPDA synthesis and the potential mechanism of feedback modulation by 12-OPDA, as investigated below. Due to the aforementioned criteria, LOX2 was, thus, utilized for OCPD synthesis. As revealed by the comparison of retention times with the OCPD standard, LOX2 like SLOX and LOX6, coupled in a one-pot synthesis with AOS and AOC [33], enabled the formation of the promising bioactive compound OCPD from DHA.   The activity of 13-LOXs was abolished (LOX3, LOX4, LOX6) or strongly decreased (LOX2) by the addition of 100 µM Cu 2+ .
In contrast to the inhibition of LOX3, 4 and 6 in the presence of 1 mM CdCl 2 , the rate of oxygen consumption increased when LOX2 was injected into LeA supplemented with 1 mM Cd 2+ ( Table 2). The concentration dependency of Cd 2+ on LOX2 activity was analyzed spectrophotometrically ( Figure S6B), revealing the following values (nmol·s −1 ·mg −1 , mean ± SD of n ≥ 3): 0.8 ± 0.2 for control, 2.3 ± 0.3 in the presence of 40 µM Cd 2+ , 5.2 ± 0.9 in 150 µM Cd 2+ and 5.7 ± 0.7 in 1180 µM Cd 2+ . With regard to applications, e.g., for the optimization of 13-LOX activity in 12-OPDA synthesis, we were interested in exploring whether 13-LOXs are also activated when lysed in buffer without supplements, as utilized in 12-OPDA synthesis (see Section 4.8). As suggested from the dialyzed LOX2 lysate, all 13-LOXs except LOX2 were inhibited by Cd 2+ , whereas Ca 2+ or Mg 2+ stimulated LOX3 and LOX6 activity, in line with Table 2 (see Table S3).
The next assay addressed the hypothesis that 13-LOX may be inhibited by its product 12-OPDA, as one possible mechanism of feedback regulation. We incubated 12-OPDA with 13-LOXs and analyzed the results for the remaining LOX activity. The non-physiological Michael acceptor N-ethylmaleimide (NEM) was also included as a control to support any observations made with MJ and 12-OPDA. As seen in Table 3, a decrease in LOX activity occurred in the presence of 12-OPDA, which was less pronounced with the 12-OPDA-derivative MJ. MJ was also tested to reveal insights into the selectivity and feedback modulatory aspects: (i) the cyclopentanone-like-prostaglandin is a model jasmonatelike compound devoid of α,β-unsaturated carbonyl system, unique to Michael acceptors (for more details, see [37]); (ii) MJ was revealed to enhance the transcription of LOX2 and LOX3 (see Section 3.5). In several systems, endogenous JA accumulation is inhibited by the application of salicylic acid (SA) to plants [19]. Therefore, we also explored whether this phytohormone, in addition to the LOX-HPL-derived TA, affects 13-LOX activity in vitro. The first tests did not result in any significant effects of SA and TA on LOX activity. To augment these observations, LOX activity was spectrophotometrically assayed in presence of reactive molecules. Due to the 10-fold higher sensitivity of detection [38], we used less enzyme but kept the same compound concentrations of 280 µM and 4.8 µM LeA. Here, NEM, 12-OPDA and MJ (except for LOX4) diminished LOX activity more intense, whereas the effects of the other tested chemicals on 13-LOX were in the same range as recorded polarographically (see Table 3).   We concluded that of the three compounds derived from 13-LOX, only MJ and 12-OPDA were inhibitory on LOX activity, supporting a model of feedback inhibition. To gain additional biochemical information, we aimed at determining enzyme inhibitory constants. Furthermore, we analyzed whether MJ and 12-OPDA alter the LOX protein conformation, which is often accompanied by protein functional changes [33]. Incubation of LOX4 (0.22 µg/µL) with indicated amounts of chemicals revealed the inhibition of LOX4 by 12-OPDA, with an IC 50 value of 48 ± 3 µM ( Figure 5A). Incubation of LOX2, LOX3 and LOX6 with 12-OPDA revealed that a level of 108 µM significantly (p ≤ 0.05) decreased their activities in comparison to the control (see Figure S7A). An IC 50 value of 130 ± 31 µM was estimated for LOX3. Due to the effects of pH, reliable IC 50 values could not be derived for LOX2 and LOX6; however, as inferred from the dose-response curves and data shown in Table 3, these IC 50 values likely range between 280 and 338 µM. Intrinsic fluorescence studies of LOX4 (see Figure 5B) revealed a decrease in fluorescence emission in the presence of 156 µ M 12-OPDA. At the LOX4 fluorescence emission maxima of 333 nm (λExc = 280 nm), which was not altered by the indicated chemicals, the relative fluorescence emission mean values (± SD of n = 4) were as follows: 98.1 ± 3.9 (control), 86.2 ± 3.9 (12-OPDA), 98.0 ± 2.4 (MJ). Thus, a change in the intrinsic fluorescence of LOX4 was not observed when incubated with MJ. Unlike LOX4, MJ altered the intrinsic fluorescence of LOX2 and LOX3 and, as observed by photometry, the activity decreased by more than 40% at 280 µ M MJ (see Table 3 and Figure S7B) in comparison to 7%, as observed for LOX4 at 338 µ M (see Figure 5A). To further evaluate the mechanism of inhibition of 13-LOXs by MJ and 12-OPDA on 13-LOXs, kinetic studies were undertaken. Michaelis-Menten plots and the corresponding Lineweaver-Burk double reciprocal plots at increasing inhibitor and substrate concentrations are presented in Supplementary Figures S8-S10. The Lineweaver-Burk plot of 12-OPDA against LOX2 revealed a decrease in the velocity of the reaction when the enzyme was saturated by substrate at apparent Vmax-values. Together with the change in the apparent Michaelis-Menten-constant, 12-OPDA tested at 226 µ M was consistent with a non-competitive mode of LOX2-inhibition. As for MJ, the intersecting lines on the y-axis were only slightly changed, whereas the slope was increased in comparison to control and 12-OPDA incubates. The result is in line with a competitive effect of the cyclopentanone on LOX2.
As for LOX6, the identified inhibitor 12-OPDA was non-competitive at concentrations of 226 µ M (see Figure S8). As for LOX3, MJ only minimally affected the apparent Km-values, ranging between 10 µ M and 11 µ M, in comparison to the control (no effector present) value of 11 µ M, observed with a Lineweaver-Burk plot, and 9 µ M with Michaelis-Menten plot. The apparent Vmax values declined from 66 µ M (control) to 39 µ M, with Intrinsic fluorescence studies of LOX4 (see Figure 5B) revealed a decrease in fluorescence emission in the presence of 156 µM 12-OPDA. At the LOX4 fluorescence emission maxima of 333 nm (λ Exc = 280 nm), which was not altered by the indicated chemicals, the relative fluorescence emission mean values (± SD of n = 4) were as follows: 98.1 ± 3.9 (control), 86.2 ± 3.9 (12-OPDA), 98.0 ± 2.4 (MJ). Thus, a change in the intrinsic fluorescence of LOX4 was not observed when incubated with MJ. Unlike LOX4, MJ altered the intrinsic fluorescence of LOX2 and LOX3 and, as observed by photometry, the activity decreased by more than 40% at 280 µM MJ (see Table 3 and Figure S7B) in comparison to 7%, as observed for LOX4 at 338 µM (see Figure 5A). To further evaluate the mechanism of inhibition of 13-LOXs by MJ and 12-OPDA on 13-LOXs, kinetic studies were undertaken. Michaelis-Menten plots and the corresponding Lineweaver-Burk double reciprocal plots at increasing inhibitor and substrate concentrations are presented in Supplementary Figures S8-S10. The Lineweaver-Burk plot of 12-OPDA against LOX2 revealed a decrease in the velocity of the reaction when the enzyme was saturated by substrate at apparent Vmax-values. Together with the change in the apparent Michaelis-Menten-constant, 12-OPDA tested at 226 µM was consistent with a non-competitive mode of LOX2-inhibition. As for MJ, the intersecting lines on the y-axis were only slightly changed, whereas the slope was increased in comparison to control and 12-OPDA incubates. The result is in line with a competitive effect of the cyclopentanone on LOX2.
As for LOX6, the identified inhibitor 12-OPDA was non-competitive at concentrations of 226 µM (see Figure S8). As for LOX3, MJ only minimally affected the apparent Km-values, ranging between 10 µM and 11 µM, in comparison to the control (no effector present) value of 11 µM, observed with a Lineweaver-Burk plot, and 9 µM with Michaelis-Menten plot. The apparent Vmax values declined from 66 µM (control) to 39 µM, with increasing MJ. The Dixon plot for MJ produced intersecting lines on the x-axis, confirming the non-competitive mode of inhibition and providing a Ki value of 295 µM ( Figure 6A). In short, 12-OPDA acted in a non-competitive or mixed mode. As seen in the respective Michaelis-Menten and Lineweaver-Burk-plots, both the apparent Vmax and Km-values were affected. The Dixon plot ( Figure 6B) was in line with a noncompetitive mode of inhibition, with an estimated Ki-value of 13 µM, which is in disagreement with the titrated IC 50 -value of 130 µM. As for LOX4, the Dixon plot revealed a Ki-value of 10 µM, similar to that of the observed IC 50 value ( Figure 6C). By considering the Michaelis-Menten curves ( Figure S10A), where I = Ki would reveal half-maximal activity for noncompetitive inhibition, the value of 10 µ M seems overestimated. The Lineweaver-Burk plot revealing a decrease in apparent Vmax-and an increase in apparent Km-values, together with the Dixon plot, supported the conclusion that 12-OPDA was either a non-competitive or mixed LOX4 inhibitor. Slight decreases in apparent Vmax values might be due to minor variations in enzyme quality or the pre-incubation procedure at 30 °C, as the activities of 13-LOXs decrease at elevated temperatures. LOX6 was most sensitive, as depicted by the half-life values in Figure S11.
To translate our findings on the impacts of 12-OPDA on 13-LOX to in vivo conditions, we assayed root and leaf extracts from A. thaliana for LOX activity with no positive results, as was in line with previous reports [39]. An alternative source of 13-LOX is the dicot vegetable pea [40]. We detected pea LOX protein by antibody staining and activity in enzyme tests in young roots; that is, the organ depleted of LOX inhibitors, in contrast to leaves [41] ( Figure S12). The oxygen-electrode recording with 36 µ g pea root extract, pretreated with 1 mM 12-OPDA, revealed an inhibition of activity by about 43% (2.0 ± 0.5 By considering the Michaelis-Menten curves ( Figure S10A), where I = Ki would reveal half-maximal activity for noncompetitive inhibition, the value of 10 µM seems overestimated. The Lineweaver-Burk plot revealing a decrease in apparent V max -and an increase in apparent Km-values, together with the Dixon plot, supported the conclusion that 12-OPDA was either a non-competitive or mixed LOX4 inhibitor. Slight decreases in apparent V max values might be due to minor variations in enzyme quality or the pre-incubation procedure at 30 • C, as the activities of 13-LOXs decrease at elevated temperatures. LOX6 was most sensitive, as depicted by the half-life values in Figure S11.
To translate our findings on the impacts of 12-OPDA on 13-LOX to in vivo conditions, we assayed root and leaf extracts from A. thaliana for LOX activity with no positive results, as was in line with previous reports [39]. An alternative source of 13-LOX is the dicot vegetable pea [40]. We detected pea LOX protein by antibody staining and activity in enzyme tests in young roots; that is, the organ depleted of LOX inhibitors, in contrast to leaves [41] ( Figure S12). The oxygen-electrode recording with 36 µg pea root extract, pretreated with 1 mM 12-OPDA, revealed an inhibition of activity by about 43% (2.0 ± 0.5 nmol·mL −1 ·min −1 , n = 4) relative to the control (4.6 ± 0.5 nmol·mL −1 ·min −1 , n = 4). Thus, 12-OPDA may also inhibit 13-LOX in vivo.
The PLAT domain is involved in LOX activity regulation. In human hLOX5, the PLAT domain is the reaction site of inhibiting Michael systems [42,43]. Sequence alignment of the mentioned LOXs and annotated plant 13-LOXs revealed that two cysteine residues in the respective PLAT domain partially align ( Figure 8A). The thiol-bearing F-C-W motif is involved in the stability and catalysis of LOXs [24,25] and is absent in LOX6. Furthermore, the N-terminal Cys203 of LOX4 is also present in hLOX12 in other plant LOX, e.g., Zea mays, and in PLAT 1-3. Among the At-13-LOXs, LOX4 displays the highest sequence identity with hLOXs, this being 26.43% and 27.34% to LOX5 and LOX12 (supplementary File S1). Structural alignment of LOX4 with LOX12 predicts that the Cys residues of the PLAT domain are located at similar positions. Furthermore, blind docking studies employing the CB-DOCK and Swiss Dock online tools revealed interdomain interaction with 12-OPDA ( Figure 8B), similar to that reported for hLOX5 with 3-acetyl-11-keto-beta-boswellic acid [43]. Based on these results with LOX4, namely, its sensitivity to thiol modifications, the intrinsic fluorescence change, and in silico analyses, it appeared that C203 might be involved in activity regulation. Therefore, C203 of LOX4 was replaced with Ser by sitedirected mutagenesis. Unfortunately, we were unsuccessful in obtaining the site-directed variant of 13-LOX as immunoreactive heterologously expressed protein (data provided on request).  Structural alignment of LOX4 with LOX12 predicts that the Cys residues of the PLAT domain are located at similar positions. Furthermore, blind docking studies employing the CB-DOCK and Swiss Dock online tools revealed interdomain interaction with 12-OPDA ( Figure 8B), similar to that reported for hLOX5 with 3-acetyl-11-keto-betaboswellic acid [43]. Based on these results with LOX4, namely, its sensitivity to thiol modifications, the intrinsic fluorescence change, and in silico analyses, it appeared that C203 might be involved in activity regulation. Therefore, C203 of LOX4 was replaced with Ser by site-directed mutagenesis. Unfortunately, we were unsuccessful in obtaining the site-directed variant of 13-LOX as immunoreactive heterologously expressed protein (data provided on request).

Synthesis of Functional 13-LOX Isoforms
In this study, Arabidopsis 13-LOXs could be obtained from an E. coli protein expression system, previously derived from an insect expression system used by Bannenberg et al. [12]. Despite the disadvantage of utilizing partially purified proteins for biochemical characterizations, the data presented herein extend the molecular characteristics of At-LOX, first provided by Bannenberg et al. [12]. Difficulties in the synthesis of functional 13-LOX or hLOXs were observed in previous studies and the utilization of enzyme-containing lysates or partially purified LOXs is not unusual [12,14,44,45] when gaining the biochemical characteristics of LOXs. The iron loads and enzyme purities of 13-LOXs were not taken into consideration in the determination of activity characteristics; thus, the activities are apparent values that might alter if isolated with iron loads of 100% and increased purities, especially for LOX6. Their apparent K m and V max activity values with the favored substrates, LA or LeA, are similar when compared to the 13-LOXs of other plant sources, for example, tomato, banana, melon and soybean, with values between 1.4 and 200 µM [21,46,47] and hLOXs [48].

Subcellular Localization of 13-LOX Isoforms
Arabidopsis 13-LOXs and orthologs are predicted to be located in plastids [2,12,15]. This was previously indicated in proteomics studies for LOX2 and LOX6 [49]. Accordingly, the putative transit peptide ( Figure 1A) of each A. thaliana 13-LOX isoform was sufficient to target YFP to the chloroplasts in transfected protoplasts ( Figure 2); thus, 13-LOXs are likely to reside in plastids. However, it is likely that 13-LOXs locate to non-photosynthetic organelles as well, as shown for the LOX6 and LOX2 that were also identified in root plastids [17] and nucleus [50].

Substrate Specificities and Evolutionary Relations
As reported in [12], LOX3, LOX4 and LOX6 were maximally active at neutral pH (pH 7.2). In slight contrast to our data, [12] reported LOX2 activities below 50% with LA (~40%) and ARA (~15%) and revealed no pH optimum. This might be due to different methods for protein and substrate preparation. Our observation of LOX2, as being an enzyme with good peroxidation activity toward ARA and LA, fits into its evolutionary relatedness to potato LOX2 (O24370) and the rice leaf pathogen-inducible lipoxygenase (OsLOX7, P38419), shown to have similar activities toward ARA and LA [31,51]. The identification of ω-6 DHA peroxidation might indicate that LOX2 is closer to hLOX15 than to hLOX12, which is also corroborated by their sequence similarities of 26.99 vs. 25.55% (See Supplementary File S1). LOX15 acts as ω-6 LOX toward both LeA and DHA. The 3D modeling and the protein alignment of LOXs suggest that Pro residues (P769 and P796 in LOX3 and 4) are present in the active site and, together with Cys 203 (according to LOX4), could play decisive roles in differentiating LOX3/4 from LOX2/6 functional characteristics in plants. The lack of these particular prolyl and cysteinyl residues assigns these 13-LOX to a group different from LOX3 or LOX4 (see Figure 1), e.g., AtLOX2 and StLOX2, in contrast to AtLOX4, with StLOX3 (O24371), as explored in Supplementary File S1.

Differential Impact of Divalent Cations on 13-LOX Isoforms
Apart from the above outlined enzymatic characteristics, clear differences were also observed for 13-LOX sensitivity to alkaline earth metals, as previously hypothesized by [52]. LOX2 and LOX6 were strongly activated by Ca 2+ and Mg 2+ , whereas LOX4 and LOX3 were only moderately or not activated by these ions. Whether this is due to the high number of acidic residues ( Figure 1B) that often bind Ca 2+ and Mg 2+ [53] or owing to Ca 2+ /Mg 2+ -binding motifs, as observed for human LOXs [25], awaits elucidation (see Supplementary File S1). In planta, Cd 2+ and Cu 2+ affect the oxylipin blend [54] and abundance of 13-LOX transcripts [55,56]. We report that 13-LOXs are all inhibited by Cu 2+ , while Cd 2+ stimulated LOX2 activity, similar to Ca 2+ ( Table 2). Ca 2+ and Cd 2+ have partly similar properties [53]. These in vitro data may provide an explanation for the earlier findings of Montillet et al. [57], who observed that Cd 2+ increased the content of 13-HPOT in A. thaliana and might relate to the specific activation of LOX2 by Cd 2+ in vivo.

Suicide Inhibition and Feedback Regulation
Wu [58] and Conrad [59] reported some sort of catalytic self-inhibition of SLOX and hLOX12/15 by their substrates. Likewise, we noted the enzyme inhibition of LOX3, LOX4 and LOX6 when incubated with LA or LeA, at concentrations enabling high LOX2 activity (see Figure S6A). This might imply the PUFA-dependent regulation of the 9-LOX (LAspecific, [12]) and/or 13-LOX (LeA-specific, [12]) pathways in vivo. The results support the view that all 13-LOXs potentially channel 13-HPOT in 12-OPDA biosynthesis, as anticipated for LOX6 in leaves and roots, LOX2 in leaves, and LOX3 and LOX4 (shown for JA) in inflorescences and wounded leaves [17,20,60].
Similar to hLOX5, hLOX12, pea and sLOXs [58,[61][62][63], we observed that cysteinyl residues are involved in the activity of 13-LOXs, as demonstrated by inhibition with NEM. As listed in Table 3, the physiological Michael acceptor 12-OPDA likewise decreased LOX activities. This is in accordance with a model of thiol-sensitive feedback inhibition. From the dose-response curves shown in Figure 5 and Figure S7, the IC 50 values were estimated both below and above 50 µM 12-OPDA for 13-LOXs. The higher concentration of 12-OPDA required to inhibit LOX2, LOX3 and LOX6, in comparison to LOX4, might be due to differential 12-OPDA sensitivities.
The gene expression of LOX3, unlike LOX4, is under the control of jasmonates [20] and, like LOX2, LOX3 is induced by MJ [64,65]. It is interesting to note that MJ, a 12-OPDA derivative (for the structure, see Figure 5), more actively interacted with LOX2 and LOX3, and that LOX4 and LOX6 were insensitive to MJ. As inferred from the photometric and polarographic activity determinations (Table 3), high substrate concentrations (30 µM) decreased the MJ's inhibitory action, which is suggestive of a competitive inhibition mode that differs from 12-OPDA. This was confirmed by kinetic studies revealing that, except for LOX3, MJ was identified as a competitive inhibitor. It is of note that high concentrations of substrate eliminated MJ's activity on LOX3 when probed at inhibitor concentrations of 113 µM. This competitive characteristic was also the case for LOX6 with 12-OPDA (see Figures S8B and S9A), suggestive of a mixed mode of enzyme inhibition [66]. The types of inhibition tentatively assessed herein might be complicated and not uniform [67], but in-depth physical chemistry analysis, as performed in Mogul et al. [67], was beyond the scope of this manuscript. Next to substrate and inhibitor concentration-dependent LOX inhibition, 12-OPDA inhibited in a time-dependent manner, as shown for LOX4 (see Figure S10B). Furthermore, it was demonstrated that 12-OPDA modifies LOX4 cysteinyl thiols and inhibits LOX4, due to its electrophilic character ( Figure S13B). The interaction sites and the extent of feedback inhibition by MJ and 12-OPDA on 13-LOX isoforms await clarification.
The co-expression of HPL with LOX2 and LOX3 (see ATTED II (http://atted.jp/) entries, accessed on 22 May 2021) and network association with remaining LOXs (see STRING entries (https://string-db.org/, accessed on 22 May 2021) might relate to additional feedback regulatory mechanisms on oxylipin synthesis, as mentioned in [19]. Our data reveal that TA did not influence the activities of LOX2 and its isoforms. Therefore, feedback inhibition from the HPL branch-derived TA to 13-LOX seems unlikely to occur in vivo.

Cyclopentenone Prostaglandins as Potential LOX Inhibitors
LOX inhibitors were reported to enhance the proteolysis of hLOX into its two domains [43], supporting the predicted docking of 12-OPDA to the inter-domain and the potential involvement of C203 in LOX4 stabilization (Figure 8). Whether 12-OPDA targets C203 or other cysteinyl thiols in PLAT domain proteins should be investigated in future studies to elucidate the specificity of protein activity regulation by 12-OPDA, as demonstrated in this study for At-13-LOXs. Effective 13-LOX, hLOX12 and hLOX5 inhibitors are reported to exhibit IC 50 values below 5 µM [66], qualifying 12-OPDA as a weak LOX inhibitor. However, the potential interaction of 12-OPDA or cyclopentenones related to 12-OPDA, for example, PGJ 2 or 15d-PGJ 2 , with proteins of LOX activity, such as the LOX4-related hLOX12 (see Figure 8), provide novel insights and promising perspectives. Together with the extended 12-OPDA and OCPD synthesis protocols provided herein and in the 12-OPDA derivative synthesis protocol [68], these studies are also potential starting points for future investigations. Active site residues were identified by the alignment of 13-LOXs with LOX3, based on [20]. The finalization and depiction of structures were performed with Pymol (the PyMOL molecular graphics system, version 1.2r3pre, Schrödinger, LLC) or UCSF Chimera [70].

Subcellular Localization Studies
The LOX2, LOX3, LOX4 and LOX6 encoding cDNAs of putative chloroplast transit peptides (see the respective TAIR (https://www.arabidopsis.org/ accessed on 18 March 2019) entries At3g45140, At1g17420, At1g72520 and At1g67560) were fused to the yellow fluorescent protein (YFP). These were cloned into the 35S-YFP-NosT vector using BamHI and AgeI restriction sites, which were added to the coding sequences by PCR (for primer sequences, see Supplementary Table S1). The protoplast preparation and transfection were performed as described in [76]. The subcellular distribution of the fluorescent protein was examined by confocal laser scanning microscopy (LSM 5 Exciter, Zeiss) with a C-Apochromat 40×/1.2 W autocorr objective. YFP was excited at 488 nm and a 2% line of the argon-ion laser, and the emission was recorded with the BP 505-600-nm filter. The chlorophyll autofluorescence was detected with the LP 650-nm filter and excited at 488 nm. The pinhole was 90 µm and the pixel dwell time was 4.58 µs, with a line average of 4, and images were encoded by 12-bit. Fluorescence images and spectra were analyzed with the ZEN Digital Imaging Software. The magenta color was used to code the chlorophyll fluorescence and green color for YFP.

Cloning, Expression and Preparation of Lipoxygenase Wild-Type and Variant LOX2, LOX3, LOX4, LOX6 and LOX4 C203S
The cDNAs encoding 13-LOX proteins were amplified from the RAFL cDNA clones obtained from the RIKEN Bioresource Center (https://web.brc.riken.jp/en/accessed on 14 May 2018) by PCR, using gene-specific primers (Supplementary Table S1). The cDNAs were cloned into the Invitrogen pET15b vector (Merck, Darmstadt, Germany) or Invitrogen pEXP5-NT-TOPO vector (Fisher, Schwerte, Germany; LOX6) in frame with the N-terminal His-tag and transformed into Nico21(DE3) cells (NEB) for heterologous expression in E. coli. LOX4C203S was generated with LOX4 as a template and in vitro mutagenesis primers (Supplementary Table S1). The correctness of the constructs was verified by DNA sequencing. Then, 600 mL Luria-Bertani medium, containing 600 µg/mL ampicillin, was inoculated with 50 mL of a non-induced overnight bacteria culture and incubated at 37 • C and 150 rpm to an OD 600 of 0.7-0.8. Protein expression was then induced by the addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 0.35 mM and 0.1 mM of NH 4 Fe-(III)-citrate. Induced cultures were grown at 25 • C, with orbital shaking at 120 rpm for 20 h. The cultures were centrifuged at 6000 rpm for 30 min at 4 • C and the harvested cell pellets were stored at −80 • C.

Lipoxygenase Activity
The LOX-catalyzed hydroperoxidation of PUFAs was assayed using two different methods. Firstly, oxygen electrode measurements were performed at 25 • C (Oxygraph + PC-operated oxygen electrode control unit with USB 2.0 connectivity, Hansatech, Norfolk UK) with a stirring speed of 50 rpm. The electrode was calibrated as described in the manufacturer's manual. Air-saturated substrate medium (1.8 mL, 245 ± 15 nmol O 2 mL −1 ) was equilibrated to a constant baseline for 5 min prior to injection of the sample via a Hamilton syringe. The LOX-initiated oxygen consumption was monitored as a function of time, and the linear slope after enzyme injection was used for activity calculations. No oxygen consumption occurred when the enzyme was omitted from the injection medium. Secondly, in the spectrophotometric approach, the LOX-catalyzed epoxide formation was monitored at 234 nm (ε = 25,000 L·mol −1 ·cm −1 ; [77]) (Shimadzu 2401 spectrophotometer) at 25 • C. Activities were calculated from the initial linear rates. All buffers were saturated with air. Kinetic parameters were determined by adding fixed LOX amounts to 6-10 different PUFA concentrations and plotting the LOX activity (nmol hydroperoxide·s −1 ·mg −1 ) against substrate concentration, applying the Michaelis-Menten equation using the Microsoft Excel-Solver. All LOX incubates and buffers supplemented with metal salts were checked for unchanged pH values in comparison to control or salt-free buffers.
As for the DHA-LOX product analysis, major peaks fractionating at an elution time of 13 min were collected and analyzed with a Q-IMS-TOF mass spectrometer Synapt G2Si (Waters GmbH, Manchester, UK) operated in resolution mode and interfaced to a nano-ESI ion source. N 2 served both as the nebulizer gas and the dry gas. N 2 was generated with the nitrogen generator NGM 11. Argon served as the collision gas for MS 2 experiments by CID (collision-induced dissociation). Samples from the LC-separation were introduced by static nano-ESI using in-house pulled glass emitters. The mass axis was externally calibrated with fragment ions of Glu-1-fibrinopeptide B as the calibration standard. In MS experiments, the protonated signal of leucine-enkephalin was used as an internal mass standard. Scan accumulation and data processing was performed with MassLynx 4.1 (Waters GmbH, Manchester, UK) on a PC workstation. The spectra were generated by accumulating and averaging 50 single spectra. Determinations of the exact masses were performed using centroided data.

Effect of pH and Various Compounds on LOX Activities
The pH profiles of LOX activities were determined spectrophotometrically with 9.6 µM LeA, dissolved in K-P i , pH 6.6, or Tris-HCl, pH 7.2-8.6. The effect of chemicals on LOX activity was determined using both techniques: (i) with an O 2 -electrode by injecting preincubated (5 min at 35 • C and 5 min at 25 • C) enzyme to 30 µM LeA (in 50 mM Tris-HCl, pH 7.2); and (ii) with a spectrophotometer by incubating 13.5 µL (0.25 µg/µL) 13-LOX with 1.5 µL 2.8 mM chemical stocks or solvent (10 min, 30 • C) and the injection of 3 µL incubate into 123 µL LeA (4.8 µM in 50 mM Tris-HCl, pH 7.2). Inhibitory studies were performed spectrophotometrically by the pre-incubation of 15 µL enzymes (0.25 µg/µL) with 2 µL chemical stocks (120-2880 µM) or solvent (0 µM) for 10 min at 30 • C, and the successive injection of incubate (3 µL) into 123 µL 4.8 µM LeA (in 50 mM Tris-HCl, pH 7.2). This approach was also used for kinetic inhibition studies, using four different substrate concentrations for the determination of inhibition type and inhibitor constant (Ki)-values, with Michaelis-Menten, Lineweaver-Burk and Dixon plots.

Effect of Divalent Cations on LOX-Catalyzed LeA Dioxygenation
The dependence of LOX activity on CaCl 2 , MgCl 2 , CdSO 4 and CuCl 2 dissolved in 20 mM HEPES, pH 8.0, was determined O 2 -polarographically [78]. Briefly, 55.0 µL of protein (0.5 µg/µL) was injected into an O 2 -electrode cuvette filled with 30 µM LeA (±1.0 mM salts). For CuCl 2 , 100 µM was used, due to the strong metal-catalyzed oxidation of LeA. The Cd 2+ effect was studied spectrophotometrically, using 4 µg LOX2 at indicated concentrations, preincubated for 10 min at 25 • C. The residual LOX activity was measured by injecting 8 µL incubate (0.2 µg/µL) into 110 µL LeA (9 µM in 20 mM HEPES, pH 8.0). The effect of metals on LOX was studied with E. coli-expressed protein pellets dissolved in TRIS-HCl, pH 8.0. The protocol followed the details outlined for the determination of the regiospecificity of LOX-mediated DHA and ARA oxygenation (Section 4.5) and used synthesized 12-OPDA and OCPD as described in Section 4.8. Synthesis of 12-OPDA and OCPD with 13-LOX-, AOS-and AOC-lysates was performed as described in [33], here expanded to the testing of LOX2, LOX3 and LOX4. The RP-HPLC analysis of oxylipin extracts was performed as described (see Section 4.5) using an injection volume of 20 µL. Successful formation of 12-OPDA and OCPD was confirmed by comparing the elution times with standards as synthesized in [33]. The time course of LA, ARA and LeA peroxide formation by LOX2, and consumption by AOS, was demonstrated spectrophotometrically at 234 nm [77] by adding 3 µg LOX2 to 12 µM PUFA (in 50 mM Tris-HCl, pH 7.2), followed by the addition of 1.5 µg AOS in a total volume of 129 µL.

Ex Vivo Evaluation of 12-OPDA as a LOX Inhibitor
To evaluate 12-OPDA as a LOX inhibitor under native conditions, we employed pea roots as a plant LOX source. The reliability was assessed with the criterion that an increase in A 234nm and, ideally, oxygen consumption occurred via the injection of plant tissue extract (0.5-2 µg/µL protein in 50 mM Tris-HCl or K-P i , pH 6.6-8.0) in LeA (tested at 4-100 µM). Furthermore, boiling of plant tissue extract (5 min, 95 • C) resulted in decreased activities, determined as described in the following. Pea seeds (Pisum sativum "Kleine Rheinländerin") were placed in darkness for 4 days on moist paper. The emerged roots were cut, frozen in liquid N 2 , and stored at −80 • C. Frozen tissue (~50 mg/mL) was pulverized in a pre-chilled mortar and immediately ground with 50 mM Tris-HCl, pH 7.2. After centrifugation (14,000 rpm, 4 • C), the clear supernatant was used as a protein source (pea protein).

Conclusions
This report provides novel insights into the regulatory effects of physiologically relevant compounds on 13-LOX isoforms from A. thaliana. We confirmed that all 13-LOXs carry chloroplast signal peptides and are potential sources of 12-OPDA. Our hypothesis of LOX inhibition by 12-OPDA was challenged via two methods, revealing that among recombinantly expressed proteins, LOX3 and LOX4 were the most sensitive to inhibition by 12-OPDA. Thus, feedback modulation of these oxygenases by 12-OPDA might be relevant in vivo, to decrease the potential accumulation of potentially toxic LOX products under conditions of stress. In addition, we found that LOX3 and LOX2 were most sensitive toward inhibition by MJ. Insight into selective interactions is also provided for divalent metals; of which, the highly toxic cadmium metal ion activated LOX2 and inactivated LOX3, LOX4 and LOX6. Furthermore, we suggest similarities between A. thaliana 13-LOXs and human LOXs, known to be involved in inflammation and diseases.