A 3D-Predicted Structure of the Amine Oxidase Domain of Lysyl Oxidase-Like 2

Abstract

Lysyl oxidase-like 2 (LOXL2) has been recognized as an attractive drug target for anti-fibrotic and anti-tumor therapies. However, the structure-based drug design of LOXL2 has been very challenging due to the lack of structural information of the catalytically-competent LOXL2. In this study; we generated a 3D-predicted structure of the C-terminal amine oxidase domain of LOXL2 containing the lysine tyrosylquinone (LTQ) cofactor from the 2.4Å crystal structure of the Zn²⁺-bound precursor (lacking LTQ; PDB:5ZE3); this was achieved by molecular modeling and molecular dynamics simulation based on our solution studies of a mature LOXL2 that is inhibited by 2-hydrazinopyridine. The overall structures of the 3D-modeled mature LOXL2 and the Zn²⁺-bound precursor are very similar (RMSD = 1.070Å), and disulfide bonds are conserved. The major difference of the mature and the precursor LOXL2 is the secondary structure of the pentapeptide (His652-Lys653-Ala654-Ser655-Phe656) containing Lys653 (the precursor residue of the LTQ cofactor). We anticipate that this peptide is flexible in solution to accommodate the conformation that enables the LTQ cofactor formation as opposed to the β-sheet observed in 5ZE3. We discuss the active site environment surrounding LTQ and Cu²⁺ of the 3D-predicted structure.

Keywords: lysine tyrosylquinone; lysyl oxidase–like 2; molecular dynamic simulation; molecular modeling.

Figure 1

The 2.4Å structure of a Zn²⁺–bound precursor LOXL2 (PDB: 5ZE3). (A) The amine oxidase domain of LOXL2 and precursor residues (Lys653 and Tyr689 in magenta) of the LTQ cofactor and Zn²⁺ (navy sphere) occupies the predicted Cu²⁺–binding site, His626–X–His628–X–His630 motif (in green). (B) Lys653 is located 16.6Å distance from Tyr689 (in solid grey line). Zn²⁺ is in tetrahedral coordination geometry.

Figure 2

The LTQ cofactor resides within 2.9Å of the active site Cu²⁺ in the catalytically–competent LOXL2 [25]. (A) 2HP covalently modifies the C5 carbonyl group of LTQ to form LTQ–2HP adduct. Ligation of LTQ–2HP adduct to the active site Cu²⁺ through O4 results in the formation of the monoanion form of LTQ–2HP (λ_max = 531 nm) at pH 8.0. (B) The active site Cu²⁺ in the mature LOXL2 can be titrated by PAR. W: water ligand to Cu²⁺.

Figure 3

A hydrophobic patch is found in the active site of the precursor LOXL2 structure (PDB: 5ZE3). (A) Phe635, Thr636, Tyr638, Ala654, and Phe656 comprise a hydrophobic patch and that seems to stabilize the short β–sheet (β8) in an unproductive conformation. Consequently, Lys653 is in the orientation to face away from Tyr689 to prevent the LTQ formation. (B) The same structure as (A) from a different angle. (C) Tyr689 and Phe656 are in π–π stacking interaction and within Van der Waals radii. (D) and (E) In addition, Lys653 is in hydrophobic interaction with His637.

Figure 4

Amino acid sequence alignment of (A) human LOX–family of proteins and (B) mammalian LOXL2s. The precursor residues (Lys653 and Tyr689) of LTQ are in pink, His residues are in green and conserved His residues are in bold. Residues that comprise the hydrophobic patch (Figure 3) are in blue. Numbering of residues are that of human LOXL2. Multiple sequence alignment was conducted by COBALT (https://www.ncbi.nlm.nih.gov/tools/cobalt/re_cobalt.cgi).

Figure 5

Overlay of the amine oxidase domains of LOX and LOXL2. (A) Overlay of AlphaFold 2–predicted human LOX (peptide backbone: in white; the LTQ precursors in yellow, His292-X-His294-X-His296: in yellow) and mouse LOXL2 (peptide backbone: in light blue; the LTQ precursors: in cyan, His628, His630, His632: in cyan). The major difference is the secondary structure of the heptapeptide (Val315–Ala321 in human LOX: β–sheet in yellow and Val650–Ala656 in mouse LOXL2: loop in cyan). In both structures, the precursor Lys residues (Lys320 in human LOX and Lys655 in mouse LOXL2) are within 4.0Å and 4.7Å from the C2 position of the precursor Tyr residues (Tyr355 in human LOX and Tyr691 in mouse LOXL2), respectively. (B) Overlay of the precursor Lys–containing heptapeptides, the precursor Tyr–containing peptides and His–X–His–X–His of the Zn²⁺–bound precursor LOXL2 (PDB: 5ze3, in light pink, residues are in magenta, Zn²⁺ is in slate), AlphaFold 2–prediced human LOX (in yellow) and AlphaFold2–predicted mouse LOXL2 (in cyan).

Figure 6

Molecular modeling and energy minimization of 2HP–inhibited Δ1–2SRCR–LOXL2 in comparison to the structure after MD–simulation. (A) The initial 3D–model (in cyan) was generated by replacing Tyr689 with DPQ–2HP and crosslinking C2 with the ϵ–amino side chain of Lys653. The single bond between ϵ–nitrogen and δ–carbon of Lys653 and the aromatic ring of Tyr698 (LTQ–2HP) was not on the same plane (dihedral angle = 15°) and the –(CH₂)₄–NH– side chain of Lys653 was in a higher energy conformation. After manual rearrangement of the hexapeptide (His652–Lys653–Ala654–Ser655–Phe656–Cys657) and subsequent MD–simulation, the –(CH₂)₄–NH– side chain of Lys653 was in a more relaxed conformation (in magenta and light pink). (B) An overlaid view of Lys653 in the initial 3D–model (in cyan) and in the final structure after MD–simulation (in magenta) (C) Dihedral angles of Lys653 during the 10 ns MD–simulation (40 ps/frame): Χ1 (dihedral angle between N–Cα–Cβ–Cγ) in black; X2 (dihedral angle between Cα–Cβ–Cγ–Cδ) in red; X3 (dihedral angle between Cβ–Cγ–Cδ–Cϵ) in green; X4 (dihedral angle between Cγ–Cδ–Cϵ–Nζ) in blue. We observed pttp and mtmt rotamers among the ideal rotamers reported for Lys653 in the penultimate rotamer library: t stands for the side chain dihedral angles of 180°, p stands for the side chain dihedral angles of 65°, and m stands for the side chain dihedral angles of – 65° [31]. (D) An overlaid view of Lys653 after MD–simulation (in magenta) with the pttp rotamer (in green) (E) An overlaid view of Lys653 after MD–simulation (in magenta) with the mtmt rotamer (in slate).

Figure 7

MD–simulation of the 2HP–inhibited Δ1–2SRCR–LOXL2. (A) Plot of RMSD values of the Cα atoms of Δ1–2SRCR–LOXL2 backbone (in black) and the amine oxidase domain of LOXL2 (in red) during 10 ns of simulation. The MD–simulation for the former shows a gradual increase in the RMSD value with fluctuations, stabilizing at an average of 0.23 nm, while the latter quickly (≤1 ns) reaches a plateau of RMSD, stabilizing at an average of 0.14 nm. (B) Plot of RMSF values of the Cα atoms of Δ1–2SRCR–LOXL2 for 10 ns of simulation where fluctuations are ≤0.253 nm. (C) Top panel: Cα B–factor/residue of the MD–simulated Δ1–2SRCR–LOXL2 plotted against amino acid residue numbers. All peaks except the last peak at 755 (α3 helix) are loops. Bottom panel: Cα B–factor/residues of precursor LOXL2 plotted against amino acid residue numbers.

Figure 8

MD–simulation of the 2HP–inhibited Δ1–2SRCR–LOXL2. (A) A plot of eigenvalues versus the first 10 eigenvector indices obtained from the covariance matrix of Cα atoms over a stable trajectory of 10 ns of MD–simulation. (B) A plot of the Addison parameter (τ) values of the active site Cu²⁺ during 10 s MD–simulation indicates a square pyramidal coordination geometry. (C) Projection of the motion of Cα atoms of Δ1–2SRCR–LOXL2 along the first two principal eigenvectors during 0–1.08 ns (in red), 1.12–4.96 ns (in blue), and 5–10 ns (in green). (D) The Cα atoms corresponding to those in (C) are highlighted in the same colors.

Figure 9

Displacement of the Cα atoms in 2HP–inhibited Δ1–2SRCR–LOXL2 (A) and the amine oxidase domain of 2HP–inhibited LOXL2 (B) where regions of peptides/amino acids exceeding 3Å (A) or 2.5Å (B) are visualized by vectors (in teal). In (A), SRCR3 and SRCR4 domains are in cyan and in green, respectively, and the amine oxidase domain is in yellow. The structures shown are in the very first phase and vectors indicating the difference between the first and the last phase of MD–simulation.

Figure 10

Comparison of amine oxidase domains and metal–binding sites of the 3D–modeled 2HP–inhibited LOXL2. (A) and the crystal structure of Zn²⁺–bound precursor (PDB:5ZE3) (B). Overall structures are very similar (RMSD = 1.347). In 3D–modeled structure (A), the cyan–colored secondary structure features in the precursor (B) are missing (having become loops). In the 2HP–ihibited LOXL2 (A), the Cu²⁺–coordination environment is a distorted square pyramidal with O4–Cu²⁺ in a Jahn–Teller axis and N1 and N3 of LTQ–2HP and His628 and His630 comprising the square bottom. In the precursor structure (B), Zn²⁺ is in tetrahedral coordination geometry having O4, His626, His628 and His630 at each tip of the tetrahedron. The cysteine–paring pattern of the five disulfide bonds are totally conserved in both structures.

Figure 11

Molecular modeling and energy minimization of a 3D–model of the resting form of Δ1–2SRCR–LOXL2 in comparison to the structure obtained from MD–simulation. (A) An overlaid view of LTQ and Cu²⁺–binding site in four distinct configurations (frame 72: in green; frame 150: in cyan; frame 200: in yellow; frame 251: in pink in (B)). (B) Dihedral angles of Lys653 throughout the MD–simulation Χ1 (dihedral angle between N–Cα–Cβ–Cγ) (black) X2 (dihedral angle between Cα–Cβ–Cγ–Cδ) (red) X3 (dihedral angle between Cβ–Cγ–Cδ–Cϵ) (green) X4 (dihedral angle between Cγ–Cδ–Cϵ–Nζ) (blue). We observed four stable rotamers (ptmt, tttt, ttmm, tttp) among the ideal rotamers reported for Lys653: t stands for the side chain dihedral angles of 180°, p stands for the side chain dihedral angles of 65°, and m stands for the side chain dihedral angles of −65° [31]. (C) An overlaid view of Lys653 in frame 72 after MD–simulation (in green) with the ptmt rotamer (in slate). (D) An overlaid view of Lys653 in frame 125 after MD–simulation (in cyan) with the tttt rotamer (in slate). (E) An overlaid view of Lys653 in frame 170 after MD–simulation (in yellow) with the ttmm rotamer (in slate). (F) An overlaid view of Lys653 in frame 225 after MD–simulation (in pink) with the tttp rotamer (in slate).

Figure 12

A 3D–modeled structure of the resting form of LOXL2: (A) The active site structure. (B) The Cu²⁺–coordination environment is a slightly distorted square pyramidal with Wax–Cu²⁺ in a Jahn–Teller axis and Weq, His626, His628 and His630 comprise the square bottom. O4 carbonyl of LTQ is 2.7Å from Weq. (C) Left: Overlay of the resting form of LOXL2 (LTQ and three His in magenta, peptide in pink, Cu²⁺ in golden sphere), 2HP–inhibited LOXL2 (LTQ–2HP and three His in cyan, peptide in pale cyan, Cu²⁺ in slate sphere) and the Zn²⁺–bound precursor (Tyr687 and three His in green, peptide in pale green, Zn²⁺ in grey sphere). Right: Overlay of Tyr687, LTQ–2HP and LTQ. (D) Acidic residues (Asp/Glu) and His residues in the active site. In orange stick: conserved acidic residues. In orange line: acidic residues in LOXL2 that are not conserved in the human LOX–family of proteins. In yellow stick: conserved acidic residue as a part of Ca²⁺–binding site. His623 and His653 (in cyan stick) are conserved. His637 (in blue line) is conserved in human LOX–family of proteins (Figure 4A) but not in mammalian LOXL2 (Figure 4B).

Figure 13

Electrostatic potential on the surface of the 3D–modeled catalytic domain of the resting form of LOXL2 (A), 2HP–inhibited LOXL2 (B) and the crystal structure of the Zn²⁺–bound precursor LOXL2 (C). LTQ, LTQ–2HP and Tyr689 (all in cyan) are exposed to the solvent. The acidic groove surrounding the LTQ cofactor and Cu²⁺ most likely accommodates substrate (Lys–containing peptides) binding.