Evolutionary Alteration of ALOX15 Specificity Optimizes The

Correction

BIOCHEMISTRY Correction for “Evolutionary alteration of ALOX15 specificity optimizes the biosynthesis of antiinflammatory and proresolving lipoxins,” by Susan Adel, Felix Karst, Àngels González-Lafont, Mária Pekárová, Patricia Saura, Laura Masgrau, José M. Lluch, Sabine Stehling, Thomas Horn, Hartmut Kuhn, and Dagmar Heydeck, which appeared in issue 30, July 26, 2016, of Proc Natl Acad Sci USA (113:E4266–E4275; first published July 13, 2016; 10.1073/pnas.1604029113). The authors note that Table 5 appeared incorrectly. The corrected table appears below. The authors also note that on page E4266, in line 20 of the Abstract, “ratPhe353Ala” should instead appear as “ratLeu353Phe;” and that on page E4270, right column, first full paragraph, line 12, “ratPhe353Leu” should instead appear as “ratLeu353Phe.” These errors do not affect the conclusions of the article.

Table 5. Relative lipoxin synthase activity of mammalian ALOX15 orthologs Relative lipoxin synthase activity, %

5-HETE as 5,6-DiHETE as Species 15-/12-ratio substrate substrate

15-lipoxygenating Human 8.1 100.0 100 Chimpanzee 8.1 118.0 145.8 Orangutan 8.1 172.2 105.6 Rabbit 24.0 39.5 108.6 ratL353F 13.3 197.3 262.5 Mean ± SD 12.3 ± 6.9 125.4 ± 62.1* 144.5 ± 68.4† 12-lipoxygenating Macaca 0.01 25.7 19.9 Mouse 0.03 36.1 1.5 Rat 0.26 8.4 0.0 Pig 0.04 35.4 61.1 humI418A 0.11 29.2 2.1 Mean ± SD 0.09 ± 0.10 27.0 ± 11.2* 17.1 ± 25.9†

The relative lipoxin synthase activity of the ALOX15 orthologs was quantified as described in Materials and Methods. For 5S-HETE oxygenation, lipoxin A and lipoxin B isomers were quantified. During 5S,6(S/R)-DiHETE, only lipoxin A isomers were formed. *P = 0.008 by Student’s t test. †P = 0.005 by Student’s t test.

www.pnas.org/cgi/doi/10.1073/pnas.1618183113

E8006 | PNAS | December 6, 2016 | vol. 113 | no. 49 www.pnas.org Downloaded by guest on September 25, 2021 Evolutionary alteration of ALOX15 specificity optimizes PNAS PLUS the biosynthesis of antiinflammatory and proresolving lipoxins

Susan Adela,1, Felix Karsta,1, Àngels González-Lafontb,c, Mária Pekárováa,d, Patricia Saurab,c, Laura Masgrauc, José M. Lluchb,c, Sabine Stehlinga, Thomas Horna,2, Hartmut Kuhna,3, and Dagmar Heydecka aInstitute of Biochemistry, University Medicine Berlin–Charité, D-10117 Berlin, Germany; bDepartament de Química, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain; cInstitut de Biotecnologia i de Biomedicina, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain; and dDepartment of Cell and Molecular Biology of Drugs, Faculty of Pharmacy, Comenius University, 832 32 Bratislava, Slovakia

Edited by Klaus van Leyen, Harvard Medical School, Charlestown, MA, and accepted by Editorial Board Member Ruslan Medzhitov May 27, 2016 (received for review March 10, 2016) ALOX15 (12/15-lipoxygenase) orthologs have been implicated in Traditionally, mammalian LOXs have been classified according maturational degradation of intracellular organelles and in the to their reaction specificity with arachidonic acid into 5-LOX, 12- biosynthesis of antiinflammatory and proresolving eicosanoids. LOX, and 15-LOX, but this classification has several problems Here we hypothesized that lower mammals (mice, rats, pigs) (18). For those isoforms exhibiting their biological functions via express 12-lipoxygenating ALOX15 orthologs. In contrast, 15- the formation of bioactive mediators, the reaction specificity is of lipoxygenating isoforms are found in higher primates (orangutans, major biological importance. For instance, ALOX5 orthologs men), and these results suggest an evolution of ALOX15 specific- oxygenate arachidonic acid specifically to (6E,8Z,11Z,14Z)-5- ity. To test this hypothesis we first cloned and characterized hydroperoxy-6,8,11,14-eicosatetraenoic acid (5S-HpETE), which ALOX15 orthologs of selected Catarrhini representing different is further converted to various leukotrienes of the 5,6-series (4). stages of late primate evolution and found that higher pri- In contrast, 15-lipoxygenating LOX isoforms are not capable of mates (men, chimpanzees) express 15-lipoxygenating orthologs. In forming leukotrienes of the 5,6-series. Similarly, 12-lipoxygenating contrast, lower primates (baboons, rhesus monkeys) express 12- enzymes have been implicated in hepoxilin biosynthesis (5), but BIOCHEMISTRY lipoxygenating enzymes. Gibbons, which are flanked in evolution ALOX5 orthologs are not involved. by rhesus monkeys (12-lipoxygenating ALOX15) and orangutans Although mammalian ALOX15 orthologs have been known (15-lipoxygenating ALOX15), express an ALOX15 ortholog with for more than 40 y, their biological role is still a matter of dis- pronounced dual specificity. To explore the driving force for this cussion. The enzymes have been implicated in cell differentiation evolutionary alterations, we quantified the lipoxin synthase activity (6, 7) and in the pathogenesis of different diseases (1, 19), but of 12-lipoxygenating (rhesus monkey, mouse, rat, pig, humIle418Ala) there is no unifying concept for their functionality. Compared and 15-lipoxygenating (man, chimpanzee, orangutan, rabbit, with other LOX-isoforms, mammalian ALOX15 orthologs are ratPhe353Ala) ALOX15 variants and found that, when normalized to somewhat special. First, they are capable of oxygenating arachidonic their arachidonic acid oxygenase activities, the lipoxin synthase activities of 15-lipoxygenating ALOX15 variants were more than Significance fivefold higher (P < 0.01). Comparative molecular dynamics simulations and quantum mechanics/molecular mechanics calculations in- Lipoxygenases are lipid-peroxidizing enzymes that have been dicated that, for the 15-lipoxygenating rabbit ALOX15, the energy classified according to their reaction specificity. ALOX15 (12/15- barrier for C13-hydrogen abstraction (15-lipoxygenation) was 17 kJ/mol lipoxygenase) has been implicated in inflammatory resolution via lower than for arachidonic acid 12-lipoxygenation. In contrast, for biosynthesis of antiinflammatory and proresolving lipoxins. We the 12-lipoxygenating Ile418Ala mutant, the energy barrier for 15- found that lower mammals including lower primates express ara- lipoxygenation was 10 kJ/mol higher than for 12-lipoxygenation. chidonic acid 12-lipoxygenating ALOX15 orthologs, whereas higher Taken together, our data suggest an evolution of ALOX15 speci- primates express 15-lipoxygenating enzymes. Gibbons constitute ficity, which is aimed at optimizing the biosynthetic capacity for the missing link interconnecting 12- and 15-lipoxygenating ALOX15 antiinflammatory and proresolving lipoxins. orthologs. To explore the evolutionary driving force for this specificity alteration, we quantified the lipoxin synthase activity of 12- eicosanoids | lipoxygenase | evolution | inflammation | protein design and 15-lipoxygenating ALOX15 orthologs and observed that the lipoxin synthase activities of 15-lipoxygenating enzymes were ipoxygenases (LOXs) form a diverse family of lipid-peroxidizing significantly higher. These results suggest an evolution of ALOX15 Lenzymes that catalyze the specific oxygenation of polyenoic specificity, which optimizes the biosynthetic capacity for antiin- fatty acids to their corresponding hydroperoxides (1, 2). Geno- flammatory and proresolving lipoxins. mic LOX sequences occur in two domains (Bacteria, Eukarya) of terrestrial life (3). In mammals, LOXs have been implicated in Author contributions: S.A., F.K., T.H., H.K., and D.H. designed research; S.A., F.K., À.G.-L., M.P., P.S., L.M., J.M.L., S.S., H.K., and D.H. performed research; S.A., F.K., À.G.-L., M.P., P.S., the biosynthesis of lipid mediators (4, 5), but they may also play a L.M., J.M.L., S.S., H.K., and D.H. analyzed data; and S.A., À.G.-L., T.H., H.K., and D.H. wrote role in cell differentiation (6, 7), apoptosis (8) and pathogenesis the paper. of inflammatory (9), hyperproliferative (10), and neurological (11) The authors declare no conflict of interest. disorders. Targeted gene inactivation in mice provided useful in- This article is a PNAS Direct Submission. K.v.L. is a guest editor invited by the Editorial formation on the functionality of Alox12B (12R-lipoxygenase) Board. 1 (12) and Aloxe3 (epidermal lipoxygenase-3) (13), but Alox5 (5- S.A. and F.K. contributed equally to this work. 2Present address: Institute of Biotechnology, RWTH Aachen University, 52074 Aachen, lipoxygenase)- (14), Alox15- (15), and Alox12-deficient mice (16) Germany. do not show major defects unless challenged otherwise. In the 3To whom correspondence should be addressed. Email: [email protected]. human genome, six functional LOX genes have been identified, This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. which encode for six distinct LOX isoforms (17). 1073/pnas.1604029113/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1604029113 PNAS Early Edition | 1of10 acid to 12-HpETE and 15-HpETE in variable amounts (20–22). logs of men (24), extinct human subspecies (25), various non- The molecular basis for this dual specificity has been explored human primates (22), rabbits (26), mice (27), rats (28), and pigs (23), and three critical amino acids have been identified (i.e., (29). Second, ALOX15 orthologs are capable of oxygenating “triad concept”). Alterations in the side-chain geometry of the phospholipids and cholesterol esters (30, 31), and thus, these LOX triad determinants impacted the specificity of ALOX15 ortho- isoforms can modify biomembranes and lipoproteins. Third, highly

Fig. 1. Triad determinants of mammalian ALOX15 orthologs including predicted and measured reaction specificity. The triad determinants of ALOX15 orthologs were retrieved from the databases. Only those entries were considered for further evaluation, for which the triad determinants have completely been sequenced. ALOX15 orthologs, for which the reaction specificity has been determined experimentally, are indicated in italic letters. On black background, those ALOX15 orthologs are indicated for which the reaction specificity has been determined for the first time in this study to our knowledge.

2of10 | www.pnas.org/cgi/doi/10.1073/pnas.1604029113 Adel et al. PNAS PLUS

Fig. 2. Simplified schematic view of late primate evolution. The main arachidonic acid oxygenation products are given in the gray shaded area. Unidentified product specificities are indicated by the question marks. developed primates such as modern and extinct humans (25, 32, that the evolutionary change from 12- to 15-lipoxygenation has 33), as well as orangutans (22, 34), express 15-lipoxygenating happened during later primate evolution (Fig. 2). Macaca ALOX15 orthologs, whereas less developed mammals (21, 35– mulatta is the most highly developed nonhuman primate for 38), including lower primates, have 12-lipoxygenating enzymes. which a 12-lipoxygenating ALOX15 has been identified, and Although no functional data are currently available for most Pongo pygmaeus is the lowest nonhuman primate for which a 15- mammalian ALOX15 orthologs, multiple sequence alignments lipoxygenating ALOX15 has been described (22, 34). Unfortunately, suggested that their reaction specificity was altered during late the ALOX15 ortholog of the gibbon (Nomascus leucogenys), which primate evolution from 12- to 15-lipoxygenation (3). is flanked in evolution by M. mulatta and P. pygmaeus, has not been To test this “evolutionary concept of ALOX15 specificity” (3), specified. To test the evolutionary concept of ALOX15 specificity, we cloned the ALOX15 orthologs of selected primates and found we addressed the following problems: (i) What is the reaction that the evolutionary switch from 12- to 15-lipoxygenation oc- specificity of the ALOX15 ortholog of Papio anubis, which precedes BIOCHEMISTRY curred between Cercopithecidae and Hominidae. Moreover, we M. mulatta during primate evolution? The evolutionary concept of observed that 15-lipoxygenating ALOX15 orthologs exhibit a sig- ALOX15 specificity suggests a 12-lipoxygenating ortholog. (ii)What nificantly higher biosynthetic capacity for antiinflammatory and is the reaction specificity of the ALOX15 ortholog of N. leucogenys? proresolving lipoxins, suggesting that the targeted switch in en- Such functional data are of particular interest (Fig. 2) because zyme specificity was aimed at optimizing the immune response. N. leucogenys is flanked in evolution by M. mulatta (12-lipoxygenating Results ALOX15) and P. pygmaeus (15-lipoxygenating ALOX15). (iii)The Pongo abelii Pan troglodytes Pan Prediction of Reaction Specificity of Mammalian ALOX15 Orthologs. ALOX15 orthologs of , ,and paniscus should constitute 15-lipoxygenating enzymes, but their re- The evolutionary concept of alox15 specificity (3) suggests that mammals ranked low in evolution (mice,rats,pigs,cattle),in- action specificities have not yet been determined. cluding lower primates (rhesus monkeys), express 12-lipoxygenating Reaction Specificity of ALOX15 Orthologs During Late Primate Evolution. ALOX15 orthologs. In contrast, higher primates (recent and extinct To characterize the later stages of mammalian ALOX15 evolu- humans, orangutans) express 15-lipoxygenating isoforms (3). To tion, we first compared the amino acid sequences of selected test this concept, to characterize its mechanistic basis, and to explore the possible biological relevance of the evolutionary speci- highly developed mammals with that of the human enzyme (Table ficity alterations we used a research strategy involving multiple 1). The corresponding sequence of the ancient human subspecies Homo neanderthalensis and Homo denisovan are 99.7% and sequence alignments, cloning, and functional characterization of Homo sapiens ALOX15 orthologs and in silico molecular dynamics (MD) sim- 99.5% identical to that of modern humans ( ), and ulations. First, we extracted cDNA sequences of ALOX15 orthologs mutagenesis data suggested that neither of the observed amino from three major genomic databases. From the list of hits, in- acid exchanges altered the reaction specificity (25). Thus, the two complete sequences were eliminated and multiple amino acid extinct human subspecies express a 15-lipoxygenating ALOX15 Pan alignments were carried out. A total of 63 complete ALOX15 ortholog. Chimpanzees ( ) are the closest living relatives of sequences were retrieved for which all triad determinants have been sequenced (Fig. 1). The degree of amino acid conservation to human ALOX15 varied between 60% and 99% (SI Appendix, Table 1. Degree of sequence conservation between ALOX15 Table S1), and all sequences involved the iron liganding amino orthologs of higher primates acids. Next, we compared the triad determinants [Sloane deter- Species Amino acid identity, % Amino acid exchanges minants (SL), Borngräber determinants 1 (BG1), Borngräber de- H. sapiens 100 0 terminants 2 (BG2)] and found that, in lower mammals, at least H. neanderthalensis one of the two major triad determinants (SL, BG1) is occupied by 99.7 2 H. denisovan 99.5 3 a small amino acid (Leu instead of Phe at the BG1 position, Val- P. paniscus Val, Ala-Val, Val-Met, Ala-Met instead of Ile-Met at the SL po- 99.7 2 P. troglodytes 99.7 2 sition). These data suggest that, according to the triad concept (23, P. abelii 39), the ALOX15 orthologs of these mammals exhibit dominant 98.6 8 P. pygmaeus 98.6 8 12-lipoxygenating activities. The only exception was the rabbit. For N. leucogenys this species, two distinct ALOX15 cDNAs can be retrieved, and 96.8 21 M. mulatta 94.6 35 previous studies indicated pronounced differences in the reaction P. anubis specificity of these ALOX15 isoforms (38). 95.0 35 Summarizing the currently available data on the reaction speci- The amino acid sequences of P. paniscus and P. troglodytes among Pan ficity of mammalian ALOX15 orthologs (Fig. 1), one may conclude and of P. abelii and P. pygmaeus among Pongo are identical.

Adel et al. PNAS Early Edition | 3of10 Fig. 3. Partial sequence alignment of ALOX15 orthologs of selected mammals. The sequence regions of the triad determinants (BG1, SL, BG2) are given, and the critical amino acids are indicated in bold. The amino acid differences with potential impact on the reaction specificity are labeled on gray background. modern humans. When we compared the amino acid sequences of To obtain such data, we cloned the ALOX15 orthologs of P. paniscus ALOX15 with that of P. troglodytes, we did not find any P. anubis, P. abelii,andN. leucogenys, expressed them as recombinant differences. Comparison with human ALOX15 indicated a 99.7% proteins, and performed in vitro activity assays. As predicted, amino acid identity. Here only two different amino acids were ALOX15 of P. anubis converted arachidonic acid mainly to detected (Table 1), and these exchanges were unlikely to be of 12-H(p)ETE (Fig. 4A), whereas the chimpanzee (P. troglodytes) functional relevance. The two orangutan subspecies (P. abelii, ortholog produced mainly 15-H(p)ETE (Fig. 4B). Interestingly, the P. pygmaeus), which have identical amino acid sequences, share a gibbon ALOX15 oxygenated arachidonic acid to a product mixture 98.6% amino acid identity with the H. sapiens ortholog (eight dif- consisting of almost equal amounts of 12- and 15-H(p)ETE (Fig. 4C). ferent amino acids). For gibbons (N. leucogenys), rhesus monkeys Such pronounced dual reaction specificity of arachidonic acid oxy- (M. mulatta), and baboons (P. anubis), significantly lower degrees genation has never been described for any naturally occurring LOX of amino acid conservation (Table 1) were calculated. isoform to our knowledge. For comparison (i.e., positive control), To predict the reaction specificities of the ALOX15 orthologs the specificity of the human ALOX15 was also tested (Fig. 4D), and of these Catarrhini, we compared the amino acid sequences of our data confirmed dominant arachidonic acid 15-lipoxygenation. the triad determinants (Fig. 3) and found that the sequences of Because RP-HPLC does not reliably separate 12- and 8-hydroxy the three amino acid clusters for H. sapiens, H. neanderthalensis, eicosatetraenoic acid (HETE) we carried out additional straight- H. denisovan, P. troglodytes, P. paniscus, P. pygmaeus, and P. abelii phase (SP)-HPLC, which confirmed our conclusions. For more were identical (Fig. 3). Thus, according to the triad concept, detailed quantification, for determination of the relative specific these enzymes should exhibit major 15-lipoxygenating activity, activities, and for statistical evaluation of the interspecies differences, we repeated expression and activity assays several times (Materials which is consistent with the functional data shown previously for and Methods the enzymes of H. sapiens and P. pygmaeus (22). The ALOX15 ), and the results are summarized in Table 2. All enzyme orthologs of M. mulatta and Mus musculus contain at least one preparations exhibit similar relative specific activities (the specific small amino acid at the Sloane positions (i.e., SL), classifying activity of the human enzyme was set at 100%). The ALOX15 orthologs of humans and chimpanzees are major 15-lipoxygenating these enzymes as 12-lipoxygenating ALOX15 variants. More- enzymes. In contrast, gibbons express an ALOX15, which converts over, in mice, the Borngräber 1 position (i.e., BG1) is occupied arachidonic acid to almost similar amounts of 12- and 15-H(p)ETE. by a Leu, the side chain of which is less space-filling than the side To make sure that the products quantified were of enzymatic origin, chain of Phe present in 15-lipoxygenating ALOX15 orthologs. we determined the enantiomer ratio and found that both 12- and 15- Baboons (P. anubis) have almost identical sequences around the H(p)ETE were mainly the S isomer (Table 2). triad determinants as rhesus monkeys (M. mulatta; Fig. 3), and thus, on the basis of sequence comparison, arachidonic acid 12- Mutagenesis Studies on Gibbon ALOX15 Confirm Applicability of N. leucogenys lipoxygenation could be predicted. The gibbon ( ) the Triad Concept. To test the applicability of the triad concept ALOX15 ortholog shows interesting sequence differences at the for gibbon ALOX15 and to quantify the relative contribution of triad determinants (Fig. 3), which made predictions of the re- the amino acid exchanges responsible for the high degree of action specificity risky. At the BG1 position, a bulky Phe is dual positional specificity, we performed a number of site- present, suggesting arachidonic acid 15-lipoxygenation. At the directed mutagenesis studies. For this purpose, we first partly SL positions, Ile418 is conserved, but, instead of Met419 (van der “gibbonized” the human ALOX15 to induce an increased share Waals radius of 124 Å), a Thr (van der Waals radius of 93 Å) is of 12-H(p)ETE formation (Met419Thr, Thr594Val). Here, we found, suggesting a gain of space at the active site. At the BG2 found that, as expected, the Met419Thr exchange increased the position, the Ile593 is conserved, but a Val instead of a Thr share of 12-H(p)ETE formation (Table 3). In fact, for this mutant, occupies the adjacent position. This situation does not allow a 12-H(p)ETE was the major arachidonic acid oxygenation product. precise prediction of the reaction specificity on the basis of the In contrast, Thr594Val exchange did not lead to major alterations triad concept, and thus, direct experimental data are required. in the positional specificity (Table 3). These data suggest that

4of10 | www.pnas.org/cgi/doi/10.1073/pnas.1604029113 Adel et al. Ile418Ala mutants. These data suggest that the Ile418Ala mutant of PNAS PLUS rabbit ALOX15 might be considered a suitable model for exploring the molecular basis for the evolutionary switch in reaction specificity of ALOX15 orthologs.

15-Lipoxygenating ALOX15 Orthologs Exhibit Significantly Higher Lipoxin-Synthesizing Capacities. Our sequence and specificity data suggest an evolution-dependent alteration in reaction specificity, but the evolutionary driving forces for this process remained unclear. ALOX15 orthologs have been implicated in the biosynthesis of antiinflammatory and proresolving mediators (41– 43), but it has not been explored whether the lipoxin synthase activities of 12- and 15-lipoxygenating ALOX15 orthologs are different. To address this point, we expressed five different 12- lipoxygenating (rhesus monkey, mouse, rat, pig, humIle418Ala) and five different 15-lipoxygenating (man, chimpanzee, orangutan, rabbit, ratPhe353Leu) ALOX15 variants and confirmed their reaction specificity (Table 5). Next, we tested the lipoxin synthase activity of these ALOX15 variants with two different substrates [5S-HETE, 5S,6R/S-dihydroxy eicosatetraenoic acid (DiHETE)]. Lipoxins can be biosynthesized via a number of different mechanisms, and these metabolic routes involve a concerted action of ALOX5, ALOX15, and/or ALOX12 (44, 45). One of the transcellular sce- narios suggests that arachidonic acid is oxygenated by ALOX5, yielding 5S-H(p)ETE, which is subsequently further oxygenated by ALOX12 and/or ALOX15 isoforms to a mixture of LXA4 and LXB4 isomers. To compare the lipoxin-biosynthesizing activity of 12- and 15-lipoxygenating ALOX15 orthologs, we incubated our BIOCHEMISTRY enzyme preparations with 5S-HETE (30 μM) for 10 min and quantified the formation of LXA4 and LXB4 isomers by RP-HPLC. The chemical identity of the lipoxin isomers was concluded from cochromatography with authentic standards and by comparison of their UV spectra (SI Appendix,Fig.S1). The experimental raw data were then normalized to the arachidonic acid oxygenase activity of Fig. 4. Positional specificity of arachidonic acid oxygenation by ALOX15 the different enzyme preparations, and the relative lipoxin synthase orthologs representing late primate evolution. Arachidonate oxygenase activity activity of the human ALOX15 ortholog was set at 100%. From assays and RP-HPLC analysis were carried out as described in Materials and Table 5 it can be seen that 15-lipoxygenating ALOX15 orthologs Methods. Each chromatogram was scaled for the highest HETE peak. (A)Prod- exhibit a fivefold higher lipoxin synthase activity (P < 0.01) com- B ucts formed by baboon ALOX15. ( ) Products formed by chimpanzee ALOX15. pared with the 12-lipoxygenating enzymes. The lower LX-synthase (C) Products formed by gibbon ALOX15. (D) Products formed by human ALOX15. activity of the 12-lipoxygenating ALOX15 orthologs is consistent with the reaction mechanism. These ALOX15 isoforms convert 5- the unusual positional specificity of the gibbon enzyme is mainly HETE mainly to 5-OH,12-OOH-ETE. This double oxygenation a result of the Met419Thr exchange. Next, we inserted at the compound constitutes a dead-end product, which cannot be further major SL position a small Ala (Ile418Ala). An identical mutation transferred to lipoxin isomers. In contrast, the major product of in rabbit ALOX15 converted this enzyme almost completely to a 5-HETE oxygenation by 15-lipoxygenating ALOX15 orthologs was 12-lipoxygenating variant (23). When we performed this muta- 5-OH,15-OOH-ETE, which can subsequently be further converted tion with the ALOX15 ortholog of gibbon (Fig. 5 and Table 3), to lipoxin isomers. orangutans, and chimpanzee (Table 4), we also observed dominant In parallel, we tested a distinct metabolic route of lipoxin 12-lipoxygenation. Finally, we introduced a bulky Phe at this po- biosynthesis. ALOX5 converts arachidonic acid to leukotriene sition (Ile418Phe). As expected, we found major 15-lipoxygenation (Fig. 5). Taken together, these mutagenesis data indicate that the Table 2. Product pattern of primate ALOX15 orthologs triad concept is fully applicable for gibbon ALOX15. Product share, % A Single Amino Acid Exchange Mimics the Evolutionary Alterations in Relative specific Reaction Specificity. Multiple mutagenesis studies (23, 24, 27, 29, Species activity, %* 15-HETE 12-HETE 40) on a number of mammalian ALOX15 orthologs have pre- H. sapiens 100 ± 41 78.7 ± 1.6 (99:1) 21.3 ± 1.6 (98:2) viously confirmed the triad concept, and an Ala-scan of the triad P. troglodytes 135 ± 45 80.3 ± 2.1 (98:2) 19.7 ± 2.1 (98:2) determinants of rabbit ALOX15 indicated that a single Ile418Ala N. leucogenys 76 ± 14 50.5 ± 7.7 (98:2) 50.0 ± 7.7 (99:1) exchange is sufficient to completely convert the reaction specificity of P. anubis 37 ± 5 21.9 ± 2.5 (ND) 78.1 ± 2.5 (ND) this enzyme from 15- to 12-lipoxygenation (23). Thus, according to these data, a single point mutation is able to mimic the evolutionary ALOX15 preparation and HPLC analysis of the reaction products were switch in reaction specificity. To test whether similar dramatic changes carried out as described in Materials and Methods. Four expression experi- ± in reaction specificity can be induced in other ALOX15 orthologs, we ments were performed, and each sample was HPLC-quantified twice. Mean SD of the relative shares of the two major reaction products are given. expressed human, orangutan, and chimpanzee ALOX15 and their Enantiomer compositions are given in parentheses. ND, not determined. corresponding Ile418Ala mutants. As indicated in Table 4, all WT *The relative specific activity was estimated by quantification of the product enzymes were dominantly 15-lipoxygenating, whereas 12-H(p)ETE formation (RP-HPLC), correcting this value for the amount of LOX protein was the almost exclusive arachidonic acid oxygenation product of the that was determined by immunoblotting.

Adel et al. PNAS Early Edition | 5of10 Table 3. Product pattern of ALOX15 mutants alignment of the substrate fatty acid at the active site of ALOX15 Relative product share, % orthologs (2, 24, 46). For 15-lipoxygenating ALOX15 orthologs, the proS-hydrogen bound at carbon 13 (C13) was predicted to be Species (mutant) 15-HETE 12-HETE located in close proximity to the iron-bound hydroxyl group ± ± functioning as hydrogen abstracting group. In contrast, for the Human (WT) 78.7 1.6 21.3 1.6 12-lipoxygenating ALOX15 orthologs, the fatty acid substrates Human (Met419Thr) 34.7 ± 8.7 65.3 ± 8.7 ± ± may slide in deeper into the substrate-binding pocket approaching Human (Thr594Val) 81.5 1.2 18.6 0.9 the proS-hydrogen at C to the enzyme-bound iron–hydroxyl Gibbon (WT) 46.1 ± 5.2 53.9 ± 5.2 10 complex (2). Because no X-ray structures are currently available Gibbon (Ile418Ala) <199± 14 for ALOX15–substrate complexes and because the geometry- Gibbon (Ile418Phe) 83.6 ± 3.5 16.4 ± 3.5 based model of the triad concept does not consider energetic Human and gibbon ALOX15 orthologs were expressed in E. coli. The prod- details, we performed MD simulations and quantum mechanics/ uct pattern was analyzed by consecutive RP-HPLC, SP-HPLC, and chiral-phase molecular mechanics (QM/MM) calculations to explore the struc- HPLC. Four independent expression experiments were performed, and each ture of enzyme–substrate complex for the 15-lipoxygenating rabbit ± sample was quantified twice. Means SD of the relative shares of the two ALOX15 and its 12-lipoxygenating Ile418Ala mutant. major reaction products are given. Because of the structural flexibility of the fatty acid backbone, a large number of energetically similar arachidonic acid con- A4 (5,6 epoxy arachidonic acid), which can further be trans- formers are bound at the active site of WT rabbit ALOX15 (47), formed to leukotriene B4 or C4. Alternatively, it may undergo and similar results were obtained here for the 12-lipoxygenating nonenzymatic epoxide hydrolysis to yield 5S,6(R/S)-DiHETE. Ile418Ala mutant. Next we screened the arachidonic acid con- The hydrolysis products can subsequently be oxygenated to lip- formers at the active site of the 12-lipoxygenating Ile418Ala oxin isomers by 12- and 15-lipoxygenating ALOX isoforms. When mutant for catalytically productive structures using the follow- i d d ≤ we incubated the recombinant ALOX15 orthologs with a 1:1 mix- ing filtering criteria: ( ) (H13-OH) or (H10-OH) 3.0 Å and ii d > d d > d ture of 5S,6S- and 5S,6R-DiHETE (30 μM) for 10 min and ( ) (C13-OH) (H13-OH) or (C10-OH) (H10-OH). These quantified the formation of LXA4 isomers, we found that, when criteria ensure that the C-H bond to be split during hydrogen normalized to identical arachidonic acid oxygenase activities, 15- abstraction is properly aligned in relation to the iron-bound hy- lipoxygenating ALOX15 orthologs exhibit an almost 10-fold higher droxyl. Applying this strategy, two sets of potentially productive SI lipoxin A4 synthesizing capacity than 12-lipoxygenating ortho- conformers were extracted from our MD trajectories ( Appendix i logs (Table 5). , Table S5): ( ) Structures suitable for proS-H abstrac- ii Taken together, these data indicate that 15-lipoxygenating tion from C13 (15-lipoxygenation) and ( ) structures suitable for ALOX15 variants exhibit a higher lipoxin synthase activity than proS-hydrogen abstraction from C10 (12-lipoxygenations). These the 12-lipoxygenating counterparts when normalized to an iden- data indicate that at the active site of the Ile418Ala mutant ar- tical arachidonic acid oxygenase activity. It should be stressed at achidonic acid conformers are present, which allow both 12- and this point that our lipoxin synthase activity measurements do not 15-lipoxygenation. Similar results were previously reported for involve detailed kinetic studies, and thus, direct comparison of the 15-lipoxygenating WT enzyme (47). These data indicate that, basic kinetic parameters (KM, Vmax, catalytic efficiency) for the on the basis of the original geometry-based triad concept, which different ALOX15 orthologs was not possible. Moreover, our in stresses the binding distances, it may not be possible to reliably vitro data might not adequately mirror the lipoxin synthase activity discriminate between enzyme–substrate complexes suitable for of the ALOX15 orthologs in vivo. To address this topic, experi- 12- or 15-lipoxygenation. ments should be carried out with humanized knock-in mice, which Next, we performed clustering analysis for the 12-lipoxygenating express a 15-lipoxygenating mutant of murine Alox15. We are Ile418Ala mutant within the two sets of arachidonic acid con- currently in the process of creating such animals. formers (12-lipoxygenating vs. 15-lipoxygenating), which was based on the rmsd of the heavy atoms of the substrate fatty acid. In Fig. MD Simulations and Quantum Mechanics/Molecular Mechanics 6A, a representative arachidonic acid conformer, which is suitably Calculations. As suggested by the triad concept, the evolutionary aligned at the active site of the 15-lipoxygenating WT enzyme, is switch in reaction specificity might be related to a modified shown. This complex is overlaid with a conformer, which is aligned

Fig. 5. Positional specificity of arachidonic acid oxygenation by gibbon ALOX15 mutants. Arachidonate oxygenase activity assay of gibbon ALOX15 and RP- HPLC analysis of the oxygenation products were carried out as described in Materials and Methods. Each chromatogram was scaled for the highest HETE peak.

6of10 | www.pnas.org/cgi/doi/10.1073/pnas.1604029113 Adel et al. Table 4. Reaction specificity of mammalian ALOX15 orthologs Summarizing the major finding of our in silico calculation, one PNAS PLUS and their Ile418Ala mutants may conclude that, for the rabbit ALOX15, the exponential av- Product share, % erage potential energy barrier for the 12-lipoxygenation is 17 kJ/mol higher than that for 15-lipoxygenation. On the contrary, for the Species/variant 12-H(p)ETE 15-H(p)ETE Ile418Ala mutant, the exponential average potential energy Oryctolagus cuniculus barrier for 12-lipoxygenation is 10 kJ/mol lower than that for (rabbits) 15-lipoxygenation, and this difference explains preferential WT 3 97 12-lipoxygenation. Ile418Ala 92 8 H. sapiens (humans) Discussion WT 15 85 The conventional scenario of LOX classification, which stresses Ile417Ala* 94 6 P. pygmaeus the positional specificity of the enzymes, is misleading because (orangutans) orthologous enzymes are frequently classified in different cate- WT 14 86 gories. This problem is particularly obvious for ALOX15 Ile417Ala* 82 18 orthologs. Before this study was initiated, it was well established P. troglodytes (chimpanzee) that some mammals (rhesus monkeys, mice, rats, pigs) express WT 20 80 12-lipoxygenating ALOX15 isoforms, whereas the orthologs of Ile419Ala 99 1 other mammals (H. sapiens, H. neanderthalensis, H. denisovan, ALOX15 expression and product analysis were carried out as described in P. pygmaeus) express 15-lipoxygenating enzymes. A systematic search Materials and Methods. of the currently available mammalian ALOX15 sequences (Fig. 1) *In the primary structure of human and orangutan ALOX15 orthologs, a Glu, suggested that lower mammals have small amino acids at either of which is present in the rabbit enzyme in the linker peptide interconnecting the triad positions and thus, should function as 12-lipoxygenating the N-terminal and the C-terminal domains, is lacking. Thus, amino acid enzymes (3, 22, 34). In contrast, highly developed mammals numbering is shifted by one residue in comparison with the rabbit enzyme. (H. sapiens, H. neanderthalensis, H. denisovan, P. pygmaeus)express 15-lipoxygenating enzymes as the triad positions are occupied for 12-lipoxygenation at the active site of the 12-lipoxygenating by more space-filling amino acid residues. Unfortunately, the Ile418Ala mutant. Consistent with the triad concept, the arachidonic experimental basis for this evolutionary concept of ALOX15 BIOCHEMISTRY acid molecule has moved in deeper into the substrate-binding specificity (3, 22, 34) was rather narrow because key ALOX15 pocket to fill the vacancy, which is provided by the Ile418Ala orthologs have not been characterized. The aim of the present exchange (Fig. 6 B and C and Movie S1). study was to express and characterize the ALOX15 orthologs of The most populated arachidonic acid conformers observed at selected Catarrhini to confirm or disconfirm this concept and to the active site of the 12-lipoxygenating Ile418Ala mutant were define the evolutionary switching point at which 12-lipoxygenating energetically optimized, and the corresponding reactant minima enzymes have turned into 15-lipoxygenating orthologs. Because were located in the QM/MM potential energy surface. From those the reaction specificity of LOXs is important for LOX classifica- minima, potential energy profiles for the abstraction of the proS- tion and their biological roles, answers to these questions appear hydrogen from C10 (H10-abstraction, structures I–X) and of the to be of general interest. By applying a multiple research strategy involving detailed sequence comparison, experimental character- proS-hydrogen at C13 (H13-abstraction, structures XI–XX) were calculated. In the enzyme–substrate complexes suitable for 12- ization of recombinant ALOX15 orthologs, multiple mutagenesis lipoxygenation (structures I–X), the distances between the proS- studies, and in silico simulations of the enzyme–substrate complex. hydrogen at C10 and the iron-bound hydroxyl group [d(H10-OH)] range from 2.43 to 3.17 Å. Similarly, the distances between C10 and the iron bound hydroxyl [d(C -OH)] vary between 3.28 to Table 5. Relative lipoxin synthase activity of mammalian 10 ALOX15 orthologs 3.67 Å (SI Appendix,TableS5). For structures XI to XX, which represent arachidonic acid conformers for 15-lipoxygenation, C13 Relative lipoxin synthase activity, % is also located at reactive distances (SI Appendix,TableS5). These data indicate that, if one considers only the geometric distances, 5-HETE as 5,6-DiHETE as Species 15-/12-ratio substrate substrate hydrogen abstraction from both C10 and C13 is possible. Finally, we calculated the potential energy barriers for the 15-lipoxygenating transition states derived from the Michaelis complexes for Human 8.1 100.0 100 structures I–X (suitable for 12-lipoxygenation) and XI–XX Chimpanzee 8.1 118.0 145.8 (suitable for 15-lipoxygenation). The energy barriers calculated Orangutan 8.1 172.2 105.6 for C13-hydrogen abstraction (15-lipoxygenation) ranged from Rabbit 24.0 39.5 108.6 64 to 91 kJ/mol (SI Appendix, Table S6). In contrast, for ratI353F 13.3 197.3 262.5 † C10-hydrogen abstraction (12-lipoxygenation), the energy barriers Mean ± SD 12.3 ± 6.9 125.4 ± 62.1* 144.5 ± 68.4 varied between 53 and 125 kJ/mol. To explore whether 12- or 15- 12-lipoxygenating lipoxygenation is preferentially catalyzed by the Ile418Ala mutant, Macaca 0.01 25.7 19.9 the exponential average potential energy barriers (SI Appendix, Mouse 0.03 36.1 1.5 Computational Methods)forH10-andH13-abstractions were cal- Rat 0.26 8.4 0.0 culated from the individual values given in SI Appendix, Table S6. Pig 0.04 35.4 61.1 Here, we found an exponential average potential energy barrier of humI418A 0.11 29.2 2.1 † 59 kJ/mol for H10-abstraction but 69 kJ/mol for H13-abstraction at Mean ± SD 0.09 ± 0.10 27.0 ± 11.2* 17.1 ± 25.9 300 K. The 10-kJ/mol higher exponential average energy barrier for the transition states of 15-lipoxygenation is consistent with the The relative lipoxin synthase activity of the ALOX15 orthologs was quantified as described in Materials and Methods. For 5S-HETE oxygenation, lip- experimental observation of preferential 12-lipoxygenation of the oxin A and lipoxin B isomers were quantified. During 5S,6(S/R)-DiHETE, only Ile418Ala mutant. For WT rabbit ALOX15, a 17-kJ/mol higher ex- lipoxin A isomers were formed. ponential average energy barrier was calculated for C10-hydrogen *P = 0.008 by Student’s t test. † abstraction (47). P = 0.005 by Student’s t test.

Adel et al. PNAS Early Edition | 7of10 a TTA triplet encoding for Leu353, and mutagenesis studies indicated 12-lipoxygenation for this enzyme variant. The mechanistic details for the tissue-specific recoding process remains to be explored. It might well be that, in the future, additional exceptions from the evolutionary concept of ALOX15 specificity will be discovered, but, based on the currently available data, the likeli- hood of such discoveries is rather low (1.6%). In addition to ALOX15, there is a dramatic difference in the reaction specificity of ALOX15B when mouse and human orthologs are compared. Human ALOX15B converts arachidonic acid to 15-H(p)ETE, whereas the mouse ortholog forms 8-H(p)ETE (48). Here, we explored whether this difference in reaction specificity of ALOX15B is also a consequence of an evolutionary process. Comparing the amino acid sequences of various mammalian ALOX15B orthologs, we found that there is no systematic development of ALOX15B reaction specificity. In fact, the retrieved sequence data suggest that most mammalian ALOX15B orthologs are arachidonic acid 15-lipoxygenating isoforms (SI Appendix,Ta- ble S2). The only exception we found was mouse Alox15b, which carries different amino acids at the critical positions (Jisaka determinants). Even the rat ALOX15B, which is most closely related to the mouse enzyme, should be a 15-lipoxygenating enzyme, as concluded from the sequence data. To address the question about the driving force behind this evolutionary concept of ALOX15 specificity, we first compared the fatty acid specificity and membrane oxygenase activity of 12- and 15-lipoxygenating ALOX15 orthologs, but did not find conserved differences. However, when we tested the biosynthetic capacity of 12- and 15-lipoxygenating ALOX15 orthologs for antiinflammatory and proresolving lipoxins (Table 4), we found that 15-lipoxygenating ALOX15 orthologs exhibit a higher lipoxin synthase activity. Putting these results into a phylogenetic context, one may conclude that the evolutionary switch from 12- to 15-lipoxygenation was aimed at optimizing the lipoxin syn- Fig. 6. Overlay (A) of WT ALOX15-AA (orange) and Ile418Ala ALOX15-AA thase activity. The 12-lipoxygenating ALOX15 orthologs of less (blue) active sites from two snapshots of the MD trajectories of both in silico highly developed mammals are capable of synthesizing proresolv- models. The location of the arachidonic acid molecule with respect to the ing lipoxins, but the 15-lipoxygenating orthologs of higher B C bottom of the cavity is shown in and . Arachidonic acid, the iron-bound mammals can do so better. In other words, mammals expressing hydroxyl ion (OH), and Fe are represented in sticks, and C10, Ala-418, and Ile418 are represented in balls and sticks. 15-lipoxygenating ALOX15 orthologs show a more efficient inflammatory resolution and this might confer them an evolutionary advantage. Because resolution actively concludes inflammation as we confirmed the evolutionary concept of ALOX15 specificity and immune response, the evolutionary switch in ALOX15 reaction identified Hylobatidae as the evolutionary switching point (Scheme 1). specificity may be considered as part of a more complex de- When we searched the major genomic databases for mamma- velopmental concept optimizing the immune system. lian ALOX15 orthologs, we retrieved a total 63 complete entries The initial version of the triad concept was based exclusively (Fig. 1). Applying the triad concept of ALOX15 specificity (23, 39) on geometric parameters (23, 39) stressing binding distances and to these sequences, one can conclude the reaction specificity of alterations in the volume of the substrate-binding pocket. Pre- ALOX15 orthologs from their primary structures. Currently, vious MD simulations indicated that, at the active site of rabbit specificity data are available for 15 mammalian ALOX15 orthologs ALOX15, arachidonic acid conformers are present, which are (Fig. 1), and their primary structures indicate that all of them follow suitable for both 12- and 15-lipoxygenation. In other words, binding distances and formal volume calculations cannot reliably the triad concept. When we used the triad concept to predict the discriminate between 12- or 15-lipoxygenation. However, calcu- reaction specificity of the ALOX15 orthologs listed in Fig. 1, we lations of the energy landscapes for the enzyme–substrate com- found that, among the 63 primary structures, 62 follow the evolu- plexes indicated that, for WT rabbit ALOX15, the average energy tionary concept of ALOX15 specificity, which suggests that higher barrier for the 15-lipoxygenating complexes is by 17 kJ/mol lower primates express 15-lipoxygenating ALOX15 orthologs whereas compared with the corresponding value for the 12-lipoxygenating lower primates and other mammals express 12-lipoxygenating complexes (47). These in silico results are consistent with experi- enzymes. The only exception from this rule is the rabbit. In mental in vitro data (20). Here, we performed similar calculations this species, a 12- and a 15-lipoxygenating ALOX15 isoform are for the 12-lipoxygenating Ile418Ala mutant of rabbit ALOX15 expressed in a tissue-specific manner. Originally, the existence of two and found that the exponential average energy barrier of enzyme– separate ALOX15 genes has been suggested (38), but completion of substrate complexes suitable for 15-lipoxygenation was 10 kJ/mol therabbitgenomedidnotconfirmthis suggestion. In fact, the major higher than for 12-lipoxygenation, explaining the dominant 12- publically available genome databases [ENSEMBL (www.ensembl. lipoxygenation of the mutant enzyme. Although these energy cal- org) and the National Center for Biotechnology Information culations do not prove the triad concept, they are consistent with (NCBI; www.ncbi.nlm.nih.gov/)] suggest the existence of a single the experimental data and move the concept a step forward. The ALOX15 gene, which encodes for a 15-lipoxygenating ALOX15 data advance the simple distance-based hypothesis to an energy- ortholog. However, in rabbit monocytes, a mRNA encoding for a based model, which broadens the mechanistic basis of the entire 12-lipoxygenating enzyme is expressed (38). This mRNA involves concept by including kinetic aspects.

8of10 | www.pnas.org/cgi/doi/10.1073/pnas.1604029113 Adel et al. PNAS PLUS

Scheme 1. Evolutionary concept of ALOX15 development. ALOX15 orthologs of lower mammals are 12-lipoxygenating enzymes, whereas higher mammals including humans express 15-lipoxygenating orthologs. Hylobatidae (gibbons) appear to be the evolutionary switching point because they express an ALOX15 ortholog with pronounced dual reaction specificity.

Materials and Methods as enzyme activator (total assay volume, 0.1 mL). After 10 min of incubation μ Chemicals. The sources of the chemicals used are specified in the SI Appendix. at room temperature, the hydroperoxy derivatives were reduced (5 Lofan SnCl2 solution of 10 mg/mL). Then, 0.1 mL of ice cold MeOH was added, protein precipitate was spun down, and aliquots of the clear supernatant Database Analysis. The following genomic databases were used for searches were injected to HPLC analysis. The lipoxin isomers formed were quantified, of ALOX15 sequences: NCBI (www.ncbi.nlm.nih.gov/), ENSEMBL (www.ensembl. recording the absorbance at 300 nm by using a molar extinction coefficient org/index.html), University of California, Santa Cruz, Genome Bioinformatics of 53,000 (M cm)−1. The lipoxin synthase activity of the different enzyme (https://genome.ucsc.edu/). The nucleotide sequences of the ALOX15 orthologs preparations was first normalized for the arachidonic acid oxygenase activity of different Catarrhini were extracted from the NCBI genome database with of the enzymes, and then the corresponding value of the human ALOX15 the following accession numbers: M. mulatta (XM_001094627), N. leucogenys was set 100% for each of the two substrates. (XM_012501351.1), P. abelii (NM_001133877), P. troglodytes (XM_003315313.2), P. paniscus (XM_008962574.1), and P. anubis (XM_009189398.1). HPLC Analysis. HPLC analysis of the LOX products was performed on a Shi- madzu instrument equipped with a Hewlett-Packard diode array detector BIOCHEMISTRY Cloning of the ALOX15 Ortholog from Pan troglodytes and Papio anubis. The 1040 A by recording the absorbance at 235 nm. RP-HPLC was carried out on a ALOX15 cDNAs from P. troglodytes and P. anubis were prepared by RT-PCR Nucleosil C18 column (KS-system, 250 × 4mm,5-μm particle size; Marcherey- cloning from the blood of two individuals from each species. The method- Nagel) coupled with a guard column (30 × 4 mm, 5-μm particle size). A ological details of the cloning process are provided in SI Appendix. Because solvent system of methanol/water/acetic acid (85/15/0.1 by volume) was used ALOX15 of P. paniscus has the identical amino acid sequence as P. troglo- at a flow rate of 1 mL/min. SP-HPLC was performed on a Zorbax-SIL column dytes, separate cloning of the P. paniscus ALOX15 cDNA was not necessary. (250 × 4 mm, 5-μm particle size) with the solvent system n-hexan/2-propanol/ acetic acid (100/2/0.1 by volume) and a flow rate of 1 mL/min. Hydroxy fatty Cloning of the ALOX15 Ortholog from N. leucogenys (Gibbon). Because we were acid enantiomers were separated by chiral-phase HPLC. The 15-HETE enan- not able to obtain blood from this primate species, the complete ALOX15 tiomers were separated as free fatty acids on a Chiralcel OD column (Daicel) cDNA was chemically synthesized (Biomatik), which includes a SalI restriction by using a solvent system consisting of hexane/2-propanol/acetic acid (100/5/ site immediately in front of the starting methionine and a HindIII site behind 0.1 by vol.) and a flow rate of 1 mL/min. The 12-HETE enantiomers were the stop codon. The synthesis product was inserted into a pME vector (Bio- separated as methyl ester on a Chiralcel OB column (Daicel) with a solvent matik) for bacterial amplification. After SalI-HindIII digestion, the 2,000-bp system consisting of n-hexane/2-popanol/acetic acid (100/4/0.1 by volume) restriction fragment was ligated into the linearized bacterial expression and a flow rate of 1 mL/min. Lipoxin isomers were analyzed by RP-HPLC plasmid pET28b. This construct was completely sequenced. using a Nucleosil C18 column (KS-system, 250 × 4 mm, 5-μm particle size; Marcherey-Nagel) coupled with a guard column (30 × 4 mm, 5-μm particle Bacterial and Eukaryotic Expression of ALOX15 Isoforms. WT and mutant size) and the solvent system acetonitrile/water/acetic acid (38/62/0.1 by vol- ALOX15 isoforms of different mammalian species (man, chimpanzee, orang- ume) at a flow rate of 1 mL/min. utan, gibbon, rhesus monkey, mouse, rat, pig) were expressed as N-terminal Escherichia coli his-tag fusion proteins in or in N2a cells, and the experimental Assay Repetitions and Statistical Evaluation. For activity assays and for quan- SI Appendix details of expression are provided in the . tification of the product pattern, two to four different expression samples (50 mL each) were grown, and at least two activity assays were carried out per Site-Directed Mutagenesis. Site-directed mutagenesis was carried out by using sample. Each activity assay was quantified twice by RP-HPLC, and the obtained × Pfu UltraII Hot Start 2 PCR Master Mix (Agilent Technologies), followed by raw data were statistically evaluated by using Microsoft Excel 2008 (12.8). DpnI digestion of the parent DNA. For each mutant, as many as five clones were Means ± SDs are given. screened for the LOX insert (digestion with a suitable combination of different restriction enzymes), and one clone was sequenced to confirm mutagenesis. MD Simulations and QM/MM Calculations. MD and QM/MM simulations were carried out for the Ile418Ala mutant of rabbit ALOX15 as described before Fatty Acid Oxygenase Activity. For the activity assay, the sequenced clone was (47). For this, we have generated molecular configurations of the Michaelis replated, four well-separated colonies were picked, and protein expression complex of the 12-lipoxygenating Ile418Ala mutant of rabbit ALOX15 by was carried out as described earlier. For routine activity assay, different means of MD simulations. The system was fully solvated in an orthorhombic – μ amounts of the stroma-free bacterial lysis supernatant (2 20 L) were added box of preequilibrated TIP3P water molecules, with dimensions of 125 Å × μ μ to 250 L PBS solution, and arachidonic acid (100 M final concentration) 79 Å × 85 Å. Seven sodium ions were added to neutralize the total charge of – was added. The mixture was incubated for different time periods (3 15 min the system. The resulting model contains nearly 75,500 atoms. Three dif- at room temperature), the hydroperoxy compounds were reduced with ferent trajectories of 5 ns of production at 300 K under periodic boundary SnCl2, the mixture was acidified to pH 3, and 0.25 mL of ice-cold methanol conditions at constant pressure and temperature were run for the ALOX15 was added. The protein precipitate was spun down, and aliquots of the clear mutant. All MD simulations were run with CHARMM version c35-b1. More supernatant were injected directly to RP-HPLC. detailed information on methodological details, sequence comparison, and results obtained by MD simulations and QM/MM calculations are provided in Lipoxin Synthase Activity. Lipoxin synthase activity was assayed by HPLC the SI Appendix, Tables S3 and S4 and Figs. S2–S9. quantification of lipoxin A4 and B4 isomers formed during the incubation of the enzymes with different substrates. Bacterial lysate supernatants of ACKNOWLEDGMENTS. We thank the German Primate Center (Göttingen, mammalian ALOX15 orthologs were incubated in PBS solution with 5S-HETE Germany) and the Max Planck Institute of Evolutionary Anthropology Leip- or 5S,6(R/S)-DiHETE (30 μM) in the presence of 3 μM linoleic acid that served zig for providing primate blood. This work was supported in part by Slovak

Adel et al. PNAS Early Edition | 9of10 Grant Agency (Bratislava, Slovakia) Research Grant 1/0392/14 (to M.P.), Economía y Competitividad Grant CTQ2014-53144-P, and “Programa Banco Deutsche Forschungsgemeinschaft Ku961/11-1, Spanish Ministerio de Santander-Universitat Autònoma de Barcelona.”

1. Haeggström JZ, Funk CD (2011) Lipoxygenase and leukotriene pathways: Bio- 27. Borngräber S, Kuban RJ, Anton M, Kühn H (1996) Phenylalanine 353 is a primary chemistry, biology, and roles in disease. Chem Rev 111(10):5866–5898. determinant for the positional specificity of mammalian 15-lipoxygenases. J Mol Biol 2. Ivanov I, et al. (2010) Molecular enzymology of lipoxygenases. Arch Biochem Biophys 264(5):1145–1153. 503(2):161–174. 28. Pekárová M, Kuhn H, Bezáková L, Ufer C, Heydeck D (2015) Mutagenesis of triad 3. Horn T, et al. (2015) Evolutionary aspects of lipoxygenases and genetic diversity of determinants of rat Alox15 alters the specificity of fatty acid and phospholipid oxy- Prog Lipid Res – human leukotriene signaling. 57:13 39. genation. Arch Biochem Biophys 571:50–57. 4. Di Gennaro A, Haeggström JZ (2012) The leukotrienes: Immune-modulating lipid 29. Suzuki H, et al. (1994) Site-directed mutagenesis studies on the iron-binding domain mediators of disease. Adv Immunol 116:51–92. and the determinant for the substrate oxygenation site of porcine leukocyte arach- 5. Pace-Asciak CR (2015) Pathophysiology of the hepoxilins. Biochim Biophys Acta idonate 12-lipoxygenase. Biochim Biophys Acta 1210(3):308–316. 1851(4):383–396. 30. Kuhn H, Belkner J, Wiesner R, Brash AR (1990) Oxygenation of biological membranes 6. Rapoport SM, Schewe T (1986) The maturational breakdown of mitochondria in re- by the pure reticulocyte lipoxygenase. J Biol Chem 265(30):18351–18361. ticulocytes. Biochim Biophys Acta 864(3-4):471–495. 31. Belkner J, Stender H, Kühn H (1998) The rabbit 15-lipoxygenase preferentially oxy- 7. van Leyen K, Duvoisin RM, Engelhardt H, Wiedmann M (1998) A function for lip- J Biol oxygenase in programmed organelle degradation. Nature 395(6700):392–395. genates LDL cholesterol esters, and this reaction does not require vitamin E. Chem – 8. Clària J (2006) Regulation of cell proliferation and apoptosis by bioactive lipid me- 273(36):23225 23232. diators. Recent Patents Anticancer Drug Discov 1(3):369–382. 32. Sloane DL, Dixon RA, Craik CS, Sigal E (1991) Expression of cloned human 15-lip- 9. Kühn H, O’Donnell VB (2006) Inflammation and immune regulation by 12/15-lipoxygenase in eukaryotic and prokaryotic systems. Adv Prostaglandin Thromboxane oxygenases. Prog Lipid Res 45(4):334–356. Leukot Res 21A:25–28. 10. Schneider C, Pozzi A (2011) Cyclooxygenases and lipoxygenases in cancer. Cancer 33. Chaitidis P, et al. (2013) Lipoxygenase pathways in Homo neanderthalensis: Func- Metastasis Rev 30(3-4):277–294. tional comparison with Homo sapiens isoforms. J Lipid Res 54(5):1397–1409. 11. Palacios-Pelaez R, Lukiw WJ, Bazan NG (2010) Omega-3 essential fatty acids modulate 34. Johannesson M, Backman L, Claesson HE, Forsell PK (2010) Cloning, purification and initiation and progression of neurodegenerative disease. Mol Neurobiol 41(2-3): characterization of non-human primate 12/15-lipoxygenases. Prostaglandins Leukot 367–374. Essent Fatty Acids 82(2-3):121–129. 12. Epp N, et al. (2007) 12R-lipoxygenase deficiency disrupts epidermal barrier function. 35. Yokoyama C, et al. (1986) Arachidonate 12-lipoxygenase purified from porcine leukocytes J Cell Biol – 177(1):173 182. by immunoaffinity chromatography and its reactivity with hydroperoxyeicosatetraenoic 13. Krieg P, et al. (2013) Aloxe3 knockout mice reveal a function of epidermal lip- acids. J Biol Chem 261(35):16714–16721. J Invest oxygenase-3 as hepoxilin synthase and its pivotal role in barrier formation. 36. Freire-Moar J, Alavi-Nassab A, Ng M, Mulkins M, Sigal E (1995) Cloning and charac- Dermatol 133(1):172–180. terization of a murine macrophage lipoxygenase. Biochim Biophys Acta 1254(1): 14. Chen XS, Sheller JR, Johnson EN, Funk CD (1994) Role of leukotrienes revealed by 112–116. targeted disruption of the 5-lipoxygenase gene. Nature 372(6502):179–182. 37. De Marzo N, Sloane DL, Dicharry S, Highland E, Sigal E (1992) Cloning and expression 15. Sun D, Funk CD (1996) Disruption of 12/15-lipoxygenase expression in peritoneal Am J Physiol – macrophages. Enhanced utilization of the 5-lipoxygenase pathway and diminished of an airway epithelial 12-lipoxygenase. 262(2 pt 1):L198 L207. oxidation of low density lipoprotein. J Biol Chem 271(39):24055–24062. 38. Berger M, et al. (1998) Simultaneous expression of leukocyte-type 12-lipoxygenase J Mol Biol – 16. Johnson EN, Brass LF, Funk CD (1998) Increased platelet sensitivity to ADP in mice and reticulocyte-type 15-lipoxygenase in rabbits. 278(5):935 948. lacking platelet-type 12-lipoxygenase. Proc Natl Acad Sci USA 95(6):3100–3105. 39. Ivanov I, Kuhn H, Heydeck D (2015) Structural and functional biology of arachidonic 17. Funk CD, Chen XS, Johnson EN, Zhao L (2002) Lipoxygenase genes and their targeted acid 15-lipoxygenase-1 (ALOX15). Gene 573(1):1–32. disruption. Prostaglandins Other Lipid Mediat 68-69:303–312. 40. Sloane DL, Leung R, Craik CS, Sigal E (1991) A primary determinant for lipoxygenase 18. Kuhn H, Thiele BJ (1999) The diversity of the lipoxygenase family. Many sequence positional specificity. Nature 354(6349):149–152. data but little information on biological significance. FEBS Lett 449(1):7–11. 41. Romano M (2010) Lipoxin and aspirin-triggered lipoxins. Sci World J 10:1048–1064. 19. Kuhn H, Banthiya S, van Leyen K (2015) Mammalian lipoxygenases and their bi- 42. Lee HN, Surh YJ (2012) Therapeutic potential of resolvins in the prevention and ological relevance. Biochim Biophys Acta 1851(4):308–330. treatment of inflammatory disorders. Biochem Pharmacol 84(10):1340–1350. 20. Bryant RW, Bailey JM, Schewe T, Rapoport SM (1982) Positional specificity of a re- 43. Serhan CN, Dalli J, Colas RA, Winkler JW, Chiang N (2015) Protectins and maresins: ticulocyte lipoxygenase. Conversion of arachidonic acid to 15-S-hydroperoxy-eicosa- New pro-resolving families of mediators in acute inflammation and resolution bio- tetraenoic acid. J Biol Chem 257(11):6050–6055. active metabolome. Biochim Biophys Acta 1851(4):397–413. 21. Watanabe T, Medina JF, Haeggstrom JZ, Radmark O, Samuelsson B (1993) Molecular 44. Schwab JM, Chiang N, Arita M, Serhan CN (2007) Resolvin E1 and protectin D1 acti- Eur J Biochem – cloning of a 12-lipoxygenase cDNA from rat brain. 212(2):605 612. vate inflammation-resolution programmes. Nature 447(7146):869–874. 22. Vogel R, et al. (2010) Applicability of the triad concept for the positional specificity of 45. Serhan CN, et al. (2003) Reduced inflammation and tissue damage in transgenic J Biol Chem – mammalian lipoxygenases. 285(8):5369 5376. rabbits overexpressing 15-lipoxygenase and endogenous anti-inflammatory lipid 23. Borngräber S, et al. (1999) Shape and specificity in mammalian 15-lipoxygenase active mediators. J Immunol 171(12):6856–6865. site. The functional interplay of sequence determinants for the reaction specificity. 46. Kühn H, Schewe T, Rapoport SM (1986) The stereochemistry of the reactions of lip- J Biol Chem 274(52):37345–37350. oxygenases and their metabolites. Proposed nomenclature of lipoxygenases and re- 24. Sloane DL, Leung R, Barnett J, Craik CS, Sigal E (1995) Conversion of human 15-lip- lated enzymes. Adv Enzymol Relat Areas Mol Biol 58:273–311. oxygenase to an efficient 12-lipoxygenase: The side-chain geometry of amino acids 47. Saura P, Suardiaz R, Masgrau L, Lluch JM, Gonzalez-Lafont A (2014) Unraveling how 417 and 418 determine positional specificity. Protein Eng 8(3):275–282. 25. Adel S, et al. (2015) Leukotriene signaling in the extinct human subspecies Homo enzymes can use bulky residues to drive site-selective C-H activation: The case of ACS Catal denisovan and Homo neanderthalensis. Structural and functional comparison with mammalian lipoxygenases catalyzing arachidonic acid oxidation. 4(12): Homo sapiens. Arch Biochem Biophys 565:17–24. 4351–4363. 26. Kuhn H, et al. (2005) Structural biology of mammalian lipoxygenases: Enzymatic 48. Jisaka M, Kim RB, Boeglin WE, Brash AR (2000) Identification of amino acid deter- consequences of targeted alterations of the protein structure. Biochem Biophys Res minants of the positional specificity of mouse 8S-lipoxygenase and human 15S- Commun 338(1):93–101. lipoxygenase-2. J Biol Chem 275(2):1287–1293.

10 of 10 | www.pnas.org/cgi/doi/10.1073/pnas.1604029113 Adel et al.