Leukotrienes regulate complex inflammatory responses by interacting with the host and external stimuli as paracrine mediators1,2,3. Induction of these pathways involves the processing of arachidonic acid to biologically active leukotrienes1,4. Leukotriene A4 hydrolase (LTA4H) is a zinc metalloenzyme that stands out with unique dual functionalities that include peptidase activity in addition to lipid metabolism. As an epoxide hydrolase (EH), LTA4H converts leukotriene A4 (LTA4) to leukotriene B4 (LTB4). Increased production of LTB4 induces neutrophilic inflammation found in chronic obstructive pulmonary disease (COPD), sepsis, and acute lung injury5,6,7,8. As an aminopeptidase (AP), LTA4H hydrolyzes natural ligands such as dynorphin, enkephalin, and the tripeptide proline-glycine-proline (PGP), of which PGP has garnered significant interest1,9,10,11,12,13,14,15. PGP is an extracellular matrix-driven neutrophil chemoattractant, which is generated from collagen through a multistep proteolytic pathway involving matrix metalloproteases 8 and 9 (MMP-8 and MMP-9) and prolyl endopeptidase (PE)7,38,52. The current understanding is that LTA4H AP activity resolves neutrophilic inflammation specifically by digesting and clearing PGP since accumulation of PGP activates the CXCR2 receptors on neutrophils5,16,17,18. Therefore, bi-functional EH and AP activities undergird the intricate pro- and anti-inflammatory regulation, respectively, exerted by LTA4H.

Snelgrove reported that lipopolysaccharide (LPS) exposure causes bioproduction of PGP and acute lung injury5. Unlike cigarette smoke exposure, LPS exposure does not suppress endogenous LTA4H AP activity, and therefore, active LTA4H AP clears PGP after exposure to LPS in murine lung5. Numao reports additional observations, which show that PGP does not induce inflammatory responses at micromolar concentrations found in the murine air-pouch model of inflammation19. This report on PGP is puzzling since several independently corroborating reports demonstrate that the in vivo levels of PGP correlate with host inflammatory responses and outcomes5,6,20. These contradicting reports also suggest that cigarette smoke causes lung injury by an accumulation of acrolein and acidification of the airways, which result in an accumulation of PGP in the lungs via suppression of LTA4H AP activity6,8,21. In murine models of lung injury, reducing PGP levels in the lung by 4-methoxydiphenylmethane (4MDM) correlates with the resolution of neutrophilic inflammation and other beneficial therapeutic effects6,20. 4MDM selectively augments the LTA4H AP activity and does not affect the EH activity6,20. 4MDM has minimal off-targeting effects reported by our group as demonstrated in an LTA4H knockout murine model6. We have demonstrated that treatment with 4MDM restores the LTA4H AP activity and prevents PGP accumulation by cigarette smoke exposure6. To clarify the contradicting reports on PGP, we set forth to conduct studies using 4MDM as a pharmaceutical tool to interrogate the biochemistry of the LTA4H AP activity.

First, we characterized the effects of selectively augmenting LTA4H AP activity on pulmonary inflammation induced by LPS. We have taken advantage of this unique aspect with the LPS murine model to determine the effect of selectively augmenting LTA4H AP activity with 4MDM in a PGP-independent model for pulmonary inflammation with 4MDM. Since PGP interaction with LTA4H is not completely characterized biochemically, our study is mainly focused on the AP activity of LTA4H in the presence of 4MDM. In addition to PGP, dynorphin and enkephalin are peptide-based substrates for the LTA4H AP activity9,10,11,12. As reported by Orning and co-workers, LTA4H hydrolyzes tripeptides containing an N-terminal arginine more efficiently than other peptide substrates such as dipeptides, tetrapeptides, pentapeptides, and tripeptides with non-arginine N-termini22. Therefore, LTA4H can be considered an arginine N-aminotripeptidase22. In order to gain more insight into the substrate-specific properties of LTA4H AP activity, we determined the effect of 4MDM on the kinetic mechanisms for hydrolysis of PGP, Ala-pNA, Arg-pNA, and Pro-pNA. Lastly, we reported the first X-ray crystal structures of LTA4H in complex with 4MDM and LTA4H bound to 4MDM and N-(4-oxo-4-pyrrolidinyl-butanoyl)-proline (OPB-Pro), a non-hydrolyzable analogue of PGP.


Regulation of LTA4H bifunctionality in the murine model of acute lung inflammation and injury induced by intra-nasal LPS

The experimental design of the murine model is shown in Fig. 1A. Mice were treated daily from Days 0 to 4 with intranasal (IN) 4MDM after being exposed to IN LPS on Day 0. LTA4H AP activity was significantly elevated after five days of 4MDM treatment over that of the vehicle (Fig. 1B). Levels of LTB4 in the bronchoalveolar lavage fluid (BALF) were comparable between the 4MDM- and vehicle-treated animals (Fig. 1C). These studies indicated that 4MDM treatment selectively enhanced the LTA4H AP activity without affecting the EH activity in the murine model of LPS-induced acute lung injury. PGP concentration was assessed in the whole lung BALF collected from two treatment cohorts after LPS exposure (vehicle vs. 4MDM) on Days 1 and 5. While the PGP levels in the BALF were above the level of detection, they were well below the lower limit of quantification except in the BALF samples from the cohort treated with 4MDM for five days. In this model, the concentration of PGP was at a level where its biological contribution was believed to be trivial (Tables S1 and S2).

Figure 1
figure 1

(A) Experimental design of the murine model of acute lung injury induced by intranasal lipopolysaccharide (LPS). (B) LTA4H aminopeptidase activity in the bronchoalveolar lavage fluid. (C) LTA4H epoxide hydrolase activity in the bronchoalveolar lavage fluid reflected by the levels of LTB4. Veh = cyclodextrin vehicle in phosphate-buffered saline. 4MDM = drug in the vehicle. *Significant p-values by Holm-Sidak comparison. n.s., not significant.

4MDM treatment consistently reduced the number of CD45+ leukocytes and CD45+CD11b+Ly6G+ neutrophils in the LPS-exposed lungs on Days 1 and 5 (Fig. 2A,B), while LTA4H activity in BALF was significantly augmented on Day 5 (Fig. 1B). LTA4H activity measured in the BALF may not fully reflect that of activity in lung tissue. The severity of acute lung injury was assessed by wet-to-dry lung weight ratio and Sireq Flexivent pressure–volume loop. 4MDM treatment maintained the lung to be more compliant by the premortem pressure–volume loop, consistent with less water content and less severe acute lung injury (Fig. 2C,D). Representative H&E staining shows significantly fewer leukocytes infiltrating the peri-bronchial areas (arrow) with 4MDM treatment than with vehicle alone (Fig. 2E).

Figure 2
figure 2

(A) CD45+ leukocytes in the bronchoalveolar lavage fluid by flow cytometry. (B) CD45+CD11b+Ly6G+ leukocytes in the bronchoalveolar lavage fluid by flow cytometry. (C) Premortem Pressure–Volume loop measured by Sireq Flexivent on Day 5. (D) Lung wet-to-dry weights ratio on Days 1 and 5. (E) H&E staining of the bronchovesicular bundle with vehicle or 4MDM treatment. Arrow indicates leukocytes infiltrating the peri-bronchial vascular bundles. Veh = cyclodextrin vehicle in phosphate-buffered saline. 4MDM = drug in the vehicle. *Significant p-values by Holm-Sidak comparison. n.s., not significant.

Enzyme kinetics of LTA4H for the hydrolysis of PGP, Arg-pNA, Ala-pNA, and Pro-pNA in the presence of 4MDM

The reaction velocity plots of PGP hydrolysis were determined at escalating concentrations of 4MDM (Fig. 3). The data is consistent with substrate-induced inhibition at increasing substrate concentrations. Our data suggested that a substrate-enzyme–substrate (SES) complex formed, which would involve the addition of a second PGP molecule to the LTA4H-PGP complex23,24. 4MDM accentuated PGP-induced inhibition in a dose-dependent manner.

Figure 3
figure 3

Reaction velocity plot of LTA4H-mediated hydrolysis of PGP. Reaction velocity plot showing the modulation of PGP-induced (substrate-induced) inhibition in the presence of 4MDM. Each sample was prepared in twenty replicates (n = 20), and the coefficient of variation is within 5%. Data shown in the figure is the average of twenty replicates.

Enzyme kinetics studies were performed to elucidate the kinetic mechanism of LTA4H-mediated hydrolysis of Arg-pNA, Ala-pNA, and Pro-pNA at escalating concentrations of 4MDM (Figs. 45, 6). 4MDM induced hyperbolic predominantly specific inhibition in the presence of Arg-pNA, hyperbolic mixed predominantly catalytic activation with Ala-pNA, and hyperbolic catalytic activation with Pro-pNA.

Figure 4
figure 4

The kinetic scheme of non-essential mixed type activation and rate constants (Baici, 2015). The k1, k2, k3, k4, and k5 are rate constants, and the unit with k2, k3, and k5 in s−1, and the unit of k1 and k4 in M−1 s−1. The corresponding dissociation constants KS and KX, in M, equal k2/k1 and k5/k4, respectively. The Michaelis–Menten constant Km equals (k2 + k3)/k1 with units of M. The catalytic constant kcat (k3) is in s−1. α and β are dimensionless positive coefficients. KSp and KCa refer to the specific (competitive) and catalytic (uncompetitive) modifications, respectively. The catalytic efficiency kcat/Km of the enzyme is used for comparing the relative rates of enzyme activity acting on the substrate with or without modifiers.

Figure 5
figure 5

The AC50 and IC50 curves for 4MDM were determined using four different substrates Ala-pNA (orange), Arg-pNA (blue), and Pro-pNA (red). Hydrolysis of Arg-pNA by LTA4H was inhibited in the presence of 4MDM with an IC50 of 328.10 μM. Hydrolysis of Ala-pNA and Pro-pNA was activated by 4MDM with an AC50 of 4.83 μM and 462.81 μM, respectively. All assays were run under the same buffer conditions with 4MDM concentrations between 0 and 8000 μM. The bars represent ± standard deviation (n = 6).

Figure 6
figure 6

Enzyme kinetic data represented in the Michaelis–Menten plot (A, B, C) and Lineweaver–Burk plot (D, E, F). In the presence of 4MDM, the hyperbolic mixed predominantly specific inhibition is shown with Arg-pNA (A, D), the hyperbolic mixed predominantly catalytic activation with Ala-pNA (B, E), and the hyperbolic catalytic activation with Pro-pNA (C, F). The bars represent ± standard deviation from six replicates (n = 6).

These studies revealed the effect of 4MDM on LTA4H kinetics by perturbing the equilibrium coupling constant α and the enzyme kinetic catalytic constant β. Decreasing values for α signify increasing stabilization of the enzyme-modifier-substrate (EXS) complex where X is 4MDM, E is LTA4H, and S is amino acid-pNA substrate (Fig. 4). Therefore, 4MDM is likely an unsuitable activator for potential R-X-X substrates, because it inhibits Arg-pNA binding competitively by stabilizing the formation of the EX complex. Given the high α value for hydrolysis of Arg-pNA in the presence of 4MDM, 4MDM and Arg-pNA cannot simultaneously bind to LTA4H, presumably due to the large size of the Arg side chain. However, 4MDM predominantly activates Ala-pNA hydrolysis by increasing catalytic turnover (β > 1) for the EXS complex25. In contrast to Arg- and Ala-pNA mechanisms, 4MDM "uncompetitively" activates Pro-pNA hydrolysis where the equilibrium coupling constant α equals the kinetic catalytic constant β (α = β). The catalytic specificities and kinetic parameters for hyperbolic enzyme modifications of Arg-, Ala-, Pro-pNA are shown in Tables 1 and 223. These studies suggested that for activation, 4MDM does not universally activate the LTA4H enzyme and its ability to activate or inhibit hydrolysis highly depends on the nature of the substrate. Notably, 4MDM affects the ES complex by contributing to partial predominantly catalytic activation (Ala-pNA) or exclusive catalytic activation (Pro-pNA).

Table 1 Catalytic efficiency (kcat/KM) of 4MDM with Ala-pNA, Arg-pNA, and Pro-pNA.
Table 2 Calculated kinetic parameters of each substrate mechanism by LTA4H AP activity.

Crystal structure of LTA4H in complex with 4MDM and 4MDM:OPB-Pro

Human recombinant LTA4H was co-crystallized with 4MDM (Fig. 7A) and 4MDM:OPB-Pro (Fig. 7B), where OPB-Pro is a non-hydrolyzable analog of PGP. The full-length structure of LTA4H in complex with 4MDM was refined at 2.9 Å to R-work of 18% and R-free of 21%. The LTA4H-4MDM complex crystallized in space group P32, with three molecules in the asymmetric unit. As shown in the M1 metallopeptidase HEXXH-(X)18E motif, the Zn2+ ion was coordinated with H295, E296 H299, and the carboxylic acid side chain oxygen atom of E31826. The 2.8 Å X-ray crystal structure of LTA4H in complex with 4MDM and OPB-Pro was refined to R-work of 19.9% and R-free of 22.9%. OPB-Pro and 4MDM were both bound in each of the three molecules of LTA4H in the asymmetric unit. The structure revealed that the OPB-Pro and 4MDM bound within the LTA4H active site, but at a significant distance of > 4.0 Å away from each other. The OPB-Pro was bound in the AP active site with its prolyl carbonyl group interacting with the zinc atom. 4MDM bound to the hydrophobic pocket of LTA4H in a similar orientation as the LTA4H:4MDM complex structure, and was stabilized by van der Waals interactions.

Figure 7
figure 7

Superimposed crystal structures of LTA4H:4MDM and LTA4H:4MDM:OPB-Pro complexes (top). (A) A close-up view of the LTA4H binding pocket with 4MDM (yellow) bound within the short end of the L-shaped hydrophobic pocket of LTA4H (salmon) is shown. (B) A close-up view of the LTA4H binding pocket with both 4MDM (yellow) and OPB-Pro bound (green) is shown. Active site Zn2+ and coordinating His residues are indicated. The direction of the Gln-136 side chain points toward the hydrophilic binding site with respect to water. The direction of the methoxy group in 4MDM is rotated in the tri-complex. Background residues were deleted to enhance the visibility of the binding site except for residues from Ser-133 to Arg-141, from Asn-291 to Ser-300, and from Thr-310 to Val-376 are shown in line rendering.

The methoxy group of 4MDM in the LTA4H:4MDM complex was oriented toward the Zn2+ cation, but is oriented toward Q136 in the LTA4H:4MDM:OPB-Pro complex. At the same time, the oxygen atoms of the methoxy group of 4MDM in the LTA4H:4MDM complex is shifted by 1.3 Å in the LTA4H:4MDM:OPB-Pro complex. 4MDM in the LTA4H:4MDM:OPB-Pro complex bound slightly deeper into the hydrophobic binding pocket, and shifted by 0.5 Å toward F362. The position of the side chain of Q136 marked a notable difference between the LTA4H:4MDM complex and the published LTA4H:OPB-Pro structure (PDB ID: 4MS6)27. The polar side chain of Q136 was oriented toward the aminopeptidase active site in the LTA4H:4MDM complex, whereas it was oriented toward the hydrophobic cavity in the LTA4H:OPB-Pro structure (PDB ID: 4MS6)27. Hydrogen bonding interactions were present between the Oε1 atom of Q136 and the Cγ atom of OPB-Pro. In the published LTA4H:ARM1:OPB-Pro structure (PDB ID: 4MKT)27, two rotamers were observed for the side chain of Q136, while the side chain of Q136 in the LTA4H:4MDM:OPB-Pro structure showed only one orientation, where the side chain orients toward the catalytic Zn2+ cation. Also, the crystal structure of LTA4H:4MDM showed that Q136 was oriented towards the Zn2+ cation, which has the same orientation shown in the structure of LTA4H:4-OMe-ARM1 complex (PDB ID: 6O5H)25. This additional structural information provided by the LTA4H:4MDM and LTA4H:4MDM:OPB-Pro complexes indicates that Q136 could have a role in tripeptide substrate recognition and catalytic turnover by maintaining rotational freedom28. In both 4MDM bound structures, the catalytic water molecule was presented near the Zn2+ ion and interacted with E296. Our interpretation is that the Q136 side chain maintains the catalytic water molecule interaction with E296 while simultaneously engaging in multiple hydrophobic interactions via its rotational freedom. These aspects are critical determinants of peptidase potentiation by 4MDM. Also, two non-catalytic water molecules form hydrogen bond interactions with Q136, N317 and E318 in the LTA4H:ARM1:OPB-Pro (PDB ID: 4MKT)27, LTA4H:OPB-Pro (PDB ID: 4MS6)27, LTA4H:ARM1 (PDB ID: 4L2L)27 and LTA4H:4MDM complexes but are absent in the LTA4H:4MDM:OPB-Pro complex (Fig. S3). We analyzed that the LTA4H:4MDM:OPB-Pro complex created a lower dielectric environment, which induced the release of these non-catalytic water molecules. The loss of water molecules could afford more favorable binding entropies for nonpolar ligand binding, provided steric clashes are avoided.


Numerous human pathologies correlate with dysfunctional LTA4H activities that result in an accumulation of LTB4 and PGP5,8,14. Research studies suggest that PGP is a pro-inflammatory matrikine that is produced during host tissue injury18. While host proinflammatory responses seem to be strongly associated with elevated levels of PGP, a recent study questions the biological activity of PGP. Numao and co-workers suggest that the anti-inflammatory LTA4H AP activity may result from clearance of substrates other than PGP, implying that even at high levels, PGP might be biologically less relevant to inflammatory signaling19. Therefore, we employed a murine model of LPS-induced pulmonary neutrophilic inflammation to ascertain the biology of LTA4H AP activity independent of PGP. Previous studies have shown that IN LPS induces acute neutrophilic inflammation in lungs while simultaneously inducing the LTA4H enzyme levels29,30. Past studies showed that PGP levels declined rapidly within 24 h due to endogenous LTA4H AP activity after exposure to LPS5. Numao also reported that the levels of PGP observed in the murine lung exposed to LPS did not cause significant neutrophilic inflammation19. Therefore, we concluded that the murine pulmonary inflammation model induced by LPS was a suitable model to ascertain if at least a part of the LTA4H AP biology is unrelated to PGP. In this model, perturbation of the LTA4H AP activity by 4MDM will provide additional corroboration. We construed that if the LTA4H AP-mediated hydrolysis of PGP is a significant anti-inflammatory pathway, 4MDM should not alter LPS-induced neutrophilic inflammation because PGP, the putative target of the LTA4H AP activity, is cleared rapidly after LPS exposure. Our study demonstrated that LTA4H AP activity was selectively augmented by 4MDM when Ala-pNA or Pro-pNA were used as reporters, which corresponded to significantly reduced airway neutrophilia caused by IN LPS throughout the 5-day observation period. PGP levels in BALF were low or undetectable, suggesting that the biological effects of the augmented LTA4H AP activity is independent of PGP. Beneficial effects were also independent of LTB4 levels. These results led us to conclude that augmentation of the LTA4H AP activity brings broader biological effects beyond the clearance of PGP. These observations bring forth a speculation that the LTA4H AP activity is not exclusive to the clearance of PGP, and therefore raises the need for detailed enzymatic studies on the LTA4H AP activity with substrates other than PGP.

Findings from our studies reveal several previously underappreciated biochemical mechanisms. First, PGP interacts with LTA4H AP by a substrate-induced inhibition mechanism. Under this mechanism, increasing concentrations of PGP would negatively feedback the LTA4H AP activity. Besides accelerating PGP accumulation, the substrate-induced inhibition may cause persistent inflammation by inhibiting LTA4H AP activity, assuming robust LTA4H AP activity is required to maintain the anti-inflammatory functions of LTA4H. Second, the enzyme kinetics significantly diverged depending on substrates. Arg-pNA, Ala-pNA, and Pro-pNA interact with the LTA4H AP substrate site under three distinctive enzymatic mechanisms in the presence of 4MDM. While 4MDM primarily inhibited hydrolysis of Arg-pNA, 4MDM enhanced the hydrolysis of Ala-pNA and Pro-pNA. Of these three substrates, the enzyme kinetics of Pro-pNA was most intriguing as it followed a rarely observed hyperbolic catalytic activation (HCaA) mechanism, which is analogous to an “uncompetitive inhibition” mechanism where the modulator has higher affinity for the enzyme–substrate complex than to the enzyme. Surprisingly, PGP was further inhibited by the presence of 4MDM, which showed us that the structure of the N-terminus of the reporter group is insufficient to report the enzyme kinetics for corresponding peptides. Regardless, we speculate that the effect of PGP hydrolysis may not be relevant for our preclinical model.

Comparisons with previously published X-ray crystal structures of LTA4H:R-X-X tripeptides complexes and the LTA4H:4MDM complex can help explain our enzyme kinetics data from a structural perspective. X-ray crystal structures of LTA4H complexes demonstrate that RSR and RAR tripeptides bind the GXMEN motif in an extended β strand conformation31. The structures also reveal that Asp375, a key residue for argininyl-tripeptide hydrolysis, forms hydrogen bonding interactions with the guanidinium group of arginine to stabilize the N-terminal arginine residue and position the tripeptide for hydrolysis within the AP active site28,31. In addition, Arg563 and Lys565 are key residues for tripeptidase function as they cooperate with each other for strong alignment of the substrate at the binding site32. The crystal structure of [E296Q]LTA4H in complex with Arg-Ser-Arg (PDB ID: 3B7S) shows that the N-terminal arginine points toward Gln13631. The superposition of LTA4H:4MDM and [E296Q]LTA4H shows that the methoxy group of 4MDM likely hinders binding of Arg-X-X by physically clashing with N-terminal arginine atoms (Fig. S4), which agrees with our enzyme kinetics data that 4MDM inhibits N-terminal arginine substrate by a specific inhibition mechanism. The overlays of these X-ray crystal structures with the LTA4H:4MDM crystal structure demonstrates that 4MDM modulates peptidase activity through its influence on the hydrophobic interactions and rotational freedom of Q136. Also, the structural data suggest that nonpolar ligands are likely necessary for enhancing the peptidase activity due to more favorable binding entropies in the active site. Hence hydrophobic interactions and steric considerations are the primary factors influencing what modified kinetics may be possible. Hydrolysis of peptides with an N-terminus containing a smaller hydrophobic side chains are more likely activated by 4MDM, whereas larger or more polar side chains would likely be inhibited by 4MDM.

In conclusion, our studies demonstrate biochemical mechanisms of the LTA4H AP activity in details that were not previously attempted. Since the studies by Numao and, now, our studies are raising a possibility of more complex biological activities exerted by the LTA4H protein and its substrates19, a simple assessment of IC50 or AC50 to screen therapeutic molecules seems inadequate. Our data strongly suggest LTA4H AP activity, as modulated by 4MDM, is biologically relevant for anti-inflammatory responses. This response, however, appears independent of its action on PGP. These observations merit attention, because most previous methods seem incomplete and premature to determine even the simple therapeutic effects based on biochemical enzymatic activities. Development of any compounds targeting LTA4H will require complete enzymatic characterization as we have done in our recently published work25. Moreover, a single therapeutic agent such as 4MDM may experience substrate-dependent mechanisms, leading to highly divergent interactions (such as inhibition or activation) with LTA4H. This is clearly an underappreciated aspect of LTA4H biology. The potentially opposite therapeutic effects a single compound can generate in vivo complicates the search for therapies targeting LTA4H. More careful assessment and characterization of LTA4H enzymatic activities are much desired to further inform strategies for developing therapeutics for this important target. We recommend reexamining the mechanistic studies on inflammatory responses with regards to the complex activities of LTA4H and to base the discovery of new therapeutic agents on the distinct modification mechanisms of the enzyme.


Protein expression and purification

Human recombinant LTA4H with an N-terminal (His)6- and Xpress-tag was expressed in E. coli BL21(DE3) cells. Cells were cultured in LB medium containing 100 μg/mL ampicillin at 37 °C. Protein expression was induced with 1.0 mM isopropyl β-D-thiogalactopyranoside at OD600 = 0.6, and cells were grown for 20 h at 22 °C. Following harvesting, cells were resuspended in Histrap buffer A containing 10 mM imidazole, 200 mM NaCl, 10 mM Tris–HCl pH 8.0 and lysozyme. Cell lysates were clarified by centrifugation (14,000× g for 30 min) at 4 °C, then loaded onto a Histrap HP column (GE Healthcare). LTA4H was eluted with a linear gradient using Histrap buffer B containing 300 mM imidazole, 200 mM NaCl, and 20 mM Tris–HCl pH 8.0. LTA4H post Histrap was loaded onto a Mono Q column (GE Healthcare), washed with 10 mM Tris–HCl pH 8.0 buffer and eluted with a linear gradient of 0–0.5 M KCl containing 10 mM Tris–HCl pH 8.0 buffer. After elution, purified LTA4H was desalted using a desalting spin column (7 K Zeba spin column, Thermo Scientific) in 25 mM Tris pH 7.8 buffer, concentrated, then evaluated by SDS-PAGE to confirm purity.

Crystallization, data collection and structure determination

For LTA4H:4MDM complex, LTA4H (27 mg/ml, MW = 69.3 kDa) was co-crystallized with 4MDM. Crystals were obtained from drops containing 60–75 mM magnesium formate dihydrate and 19–23% PEG3350 at 22 °C. For LTA4H:4MDM:OPB-Pro complex, LTA4H (15 mg/ml) was co-crystallized with N-(4-oxo-4-pyrrolidinyl-butanoyl)-proline (OPB-Pro, Aurora Fine Chemical, LLC) and 4MDM. Crystals were obtained from drops containing 60–90 mM magnesium formate dihydrate and 20–25% PEG3350 at 22 °C. Both crystal diffraction data were collected at 100 K in-house at the WRAIR X-ray Diffraction Facility using a Bruker Microstar rotating anode X-ray generator with a Pt 135 CCD detector. For data collection, crystals were cryoprotected in mother liquor with the addition of 25% ethylene glycol and frozen in a N2 gas stream. The data was reduced with the Proteum software from Bruker, and the structure was determined by molecular replacement using the program Phaser within the PHENIX suite. The LTA4H structure (PDB ID: 4MS6) was used as a search model for phasing following the removal of water and ligands. 4MDM and Zn2+ molecules were built within Fo-Fc density using coot with refinement in PHENIX.

AC50/IC50 of 4MDM with Ala-pNA, Arg-pNA and Pro-pNA

For AC50/IC50 determination, 150 μL reactions contained 10 μg/ml LTA4H, 4MDM and 1xPBS pH 7.2. The concentration of 4MDM was varied from 0 to 8000 μM. After a 10-min incubation period at room temperature, 50 μL of 4.0 mM L-alanine-p-nitroanilide (Ala-pNA, Chem-Impex International Inc.), L-arginine-p-nitroanilide (Arg-pNA, Alfa Aesar), or L-proline-p-nitroanilide (Pro-pNA, Chem-Impex International Inc.) was added to the reaction for each experiment. The LTA4H AP enzyme reaction was continuously monitored at A405 for 15 min with 10 s intervals at 30 °C immediately following the addition of 50 μL of substrate solution. Replicates (n = 3) were run for each 4MDM combination.

Kinetic assay with Ala-pNA, Arg-pNA, Pro-pNA and PGP in the presence of 4MDM

Assays were performed in 200 µL volumes containing LTA4H, 4MDM, and substrate in 1xPBS buffer pH 7.2 using a 96-well plate (Corning Costar). Ala-pNA, Arg-pNA, Pro-pNA solutions were prepared at various concentrations (0.5–8.0 mM) in the above buffer. 4MDM in 5% DMSO was added to each well at various concentrations (0–2.0 mM) and then 10 µg/ml LTA4H was added to the well with incubation at 30 °C for 10 min. Enzyme activity was measured by continuously monitoring the increase in absorbance at 405 nm for 30 min with 10 s intervals at 30 °C immediately following the addition of the substrate. Six replicates were run for each 4MDM and substrate combination and all readings were measured using a Bio-Tek Powerwave. PGP assays were performed at 30 °C in 1xPBS buffer pH 7.2 using a 96-well plate (Corning Costar). Proline-glycine-proline (PGP, Biomatik) was prepared at various concentrations (100–1000 μM) in the above buffer. 4-MDM dissolved in 5% DMSO were added to each well with various concentrations (0–0.32 μM) and then LTA4H (31.25 ng/ml) was added to 200 μL well for 30 min incubation at 30 °C. The enzymatic reactions were stopped at different time points, then the enzyme activity was measured using the fluorescamine derivatization method (Tecan Spark 10 M Spectrophotometer). The standard curve for Gly-Pro was determined, in advance, and used for enzyme activity data analysis.

Murine model of acute lung inflammation induced by LPS

All methods were carried out in accordance to ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines and relevant regulations. Ethical approval was granted from the University of Virginia Institutional Review Board (IRB) #3527. Pathogen-free 8–10 week old female C57BL/6 mice were purchased from the Jackson Laboratory. The Institutional Animal Care and Use Committee at the University of Virginia approved all experiments. All compounds are > 95% pure by HPLC analysis. Mice were anesthetized by IP ketamine and xylazine mix. All mice received intranasal (IN) lipopolysaccharide (LPS) (from Escherichia coli O55:B5, Sigma-Aldrich) diluted in 50 μL of sterile saline on Day 0 (20 μg/mouse), while one half of the mice received IN 4MDM in CDX-PBS (vehicle) and the other half received IN vehicle alone. Mice were then treated daily with IN 4MDM or vehicle. Mice were treated on Days 0 (baseline before exposure to LPS) then once daily intranasally on Days 2–4 (Fig. 1A). Whole lung bronchoalveolar lavage (BAL) was performed with our previously published method6,33. Cell pellets of the BAL fluid were stained with antibodies specific to CD45 labeled with PerCP (BD Bioscience), CD11b labeled with APCcy7 (BD Bioscience), and Ly6G labeled with PE (BD Bioscience). Flow cytometry was performed using previously published gating method6,33. Representative pictures of the H&E with peri-bronchial leukocytes infiltration were taken. Levels of LTB4 in BAL fluid were measured with a commercial ELISA kit (R&D systems). LTA4H AP activity in BAL fluid was measured by using a previously published method6.

BALF sample cleanup for PGP quantification using GC/MS

BALF samples were first filtered to remove salts and unwanted biological components present within the sample matrix. This removal was accomplished through the use of Waters C-18 Plus Short Cartridge Sep-Paks, with 360 mg of sorbent (Waters Coporation). Briefly, cartridges were conditioned by preforming three washes with acetonitrile (FisherScientific) followed by three washes with a 0.1% (v/v) trifluoroacetic acid (TFA) (Chem-Impex Internation) in MilliQ water solution, using a 1 mL Air-Tite All-Plastic Norm-Ject syringe (Air-Tite Products Co Inc) to dispense each solvent. BALF samples were then acidified with 1 μL of TFA, after which 300 μLs were drawn up and pushed through the cartridge at a rate of 1–3 drops per second. An additional two washes with the 0.1% (v/v) TFA in MilliQ water solution was then performed to remove any components in the sample matrix not compatible with the LC–MS analysis. Finally, bound PGP was eluted from the cartridge using a 20% (v/v) acetonitrile in MilliQ water solution. Eluate was dried in a SpeedVac, and resuspend in 300 μLs of a 0.1% (v/v) formic acid (FA) (Sigma Aldrich) in MilliQ water solution.

GC/MS instrumental analysis

PGP quantification was carried out on a TSQ Quantum Ultra Triple Quadrupole mass spectrometer (ThermoFisher Scientific) connected to an Accela HPLC pump (ThermoFisher Scientific). Samples were injected onto a 1 mm × 150 mm Hypersil Gold 3 μm particle C-18 reversed phase column (ThermoFisher Scientific) using an Accela autosampler (ThermoFisher Scientific). Mobile phase A consisted of 0.1% FA in MilliQ water. Mobile phase B consisted of 0.1% FA in methanol. An isocratic gradient of 90% A and 10% B at a flow rate of 50 μLs min−1 was employed for the first 6 min of the run. Then, following the primary elution of PGP at 3.5-min mark, the gradient shifted to 100% B and held there for an additional minute before returning to initial conditions for the duration of the run, allowing for column regeneration. The mass spectrometer was run in the selected reaction monitoring (SRM) mode to detect PGP present in BALF samples. Transitions pairs used for SRM analysis have been provided in Table S3. Stock solutions of PGP were prepared from lyophilized PGP (Bachem, Bubendorf, Switzerland) in phosphate buffered saline (PBS) pH 7.4 (ThermoFisher Scientific) at six concentrations, ranging from 156 to 5000 pg mL−1. Standards were prepared fresh and extracted the same day as BALF samples to ensure consistent processing.


Prism v9.2.0 (Graphpad) was used for all statistical analyses. Murine study results were analyzed by two-way ANOVA with Bonferroni corrected subgroup analyses. P valued less than 0.05 was considered significant. Enzyme kinetic samples were prepared in six replicates (n = 6), and percent consumptions for each well was calculated and selected for the initial velocity using the pNA standard curve line of best fit. The coefficient of variation is within 5%.