Abstract
This disclosure is directed to methods and compositions to inhibit MASP protein activity using small molecule inhibitors. In one aspect, the disclosure is directed to methods for identifying inhibitors of MASP protein activity, including methods of screening capable of inhibiting MASP protein activity.
Claims
1. A method of screening for a small molecule capable of inhibiting MBL-associated serine protease (MASP) protein activity, the method comprising: (a) contacting a purified MASP protein with a selected small molecule in the presence of a polypeptide comprising at least one collagen-like binding domain of a mannose-binding lectin (MBL) or ficolin and Ca.sup.2+, wherein the small molecule is selected based on the indicated capacity of the small molecule to disrupt the binding between at least one of residues Glu216, His218, Tyr225, Asp226, Asp263 and Ser265 of the CUB2 domain of the MASP protein (with reference to amino acid residues 166 to 276 of human MASP-1 set forth in SEQ ID NO:6), and MBL or ficolin, and wherein the MASP protein is selected from the group consisting of MASP-1, MASP-2 and MASP-3; (b) determining the level of MASP protein activity of the MASP protein of step (a) in the presence of the small molecule, the peptide or ficolin and the Ca.sup.2+, wherein the level of the MASP protein activity is determined in a lectin pathway-specific assay by measuring lectin pathway-specific complement activation by the MASP protein; (c) determining the level of the MASP protein activity in the absence of the small molecule by the lectin pathway-specific assay; and (d) comparing the determined level of the MASP protein activity determined in step (b) with the level of the MASP protein activity determined in step (c), wherein a reduced MASP protein activity determined in the presence of the small molecule according to step (b) as compared to the MASP protein activity determined in the absence of the small molecule according to step (c) indicates the capacity of the small molecule to inhibit MASP protein activity.
2. The method of claim 1, wherein the small molecule that inhibits MASP protein activity disrupts the binding between at least one of the residues Glu216, Asp226, Asp263, and Ser265 of the MASP protein, and MBL or ficolin and Ca.sup.2+.
3. The method of claim 1, wherein the small molecule comprises an NH.sub.3 group.
4. The method of claim 1, wherein the small molecule is a small amine.
5. The method of claim 1, wherein the small molecule is a member selected from the group consisting of methylamine, ethylamine, lysine, dGMP—Deoxyguanosine monophosphate, dCMP—Deoxycytidine monophosphate, dAMP—Deoxyadenosine monophosphate, dTMP—Thymidine 5′-monophosphate, GMP—Guanosine monophosphate, AMP—Adenosine monophosphate, TMP—Thiamine monophosphate, UMP—Uridine monophosphate, dGTP—Deoxyguanosine triphosphate, dCTP—Deoxycytidine triphosphate, dATP—deoxyadenosine triphosphate, dTTP—Thymidine triphosphate, GTP—Guanosine triphosphate, CTP—Cytidine triphosphate, ATP—Adenosine Triphosphate, TTP—Thymidine triphosphate, and UTP—Uridine triphosphate.
6. The method of claim 1, wherein the one or more small molecules have molecular weight(s) of less than 1,000 daltons.
Description
DESCRIPTION OF THE DRAWINGS
(1) The foregoing aspects and many of the attendant advantages of this invention will become more readily appreciated as the same become better understood by reference to the following detailed description, when taken in conjunction with the accompanying drawings, wherein:
(2) FIGS. 1A-C demonstrate MASP-1 CUB2 binding to Ca.sup.2+ and MBL. (A) Gel filtration of purified MASP-1 CUB2. Equivalent amounts of CUB2 (0.2 mg) were loaded onto a Superdex 200 column in the presence and absence of 2 mM CaCl.sub.2. (B) Ca.sup.2+ binding to MASP-1 CUB2 measured by isothermal titration calorimetry. A representative experiment showing the energy released as CaCl.sub.2 is added to the CUB2 domain. Data were fitted to a model in which there is a single Ca.sup.2+ site. The K.sub.D was 0.49±0.01 mM in two independent experiments, with ΔH=−4.6±0.6 kJ/mol. (C) Binding of MASP-1 CUB2 to immobilized MBL using surface plasmon resonance. CUB2 was injected at 20, 10, 5 and 2.5 μM and data were fitted to a 1:1 binding model. Kinetic parameters were k.sub.on=131±13 M.sup.−1s.sup.−1 and k.sub.off=3.5±0.3×10.sup.−3 s.sup.−1 in two independent experiments.
(3) FIGS. 2A-C demonstrate the crystal structure of MASP-1 CUB2 in complex with a collagen-like peptide from MBL. (A) Structure of the unbound collagen-like peptide with the leading, middle and trailing strands in white, black, and grey. Numbering is based on the mature rat MBL polypeptide) (peptide 2 (SEQ ID NO: 2), pdb 3PON). The sequence of the peptide is shown below the structure, to illustrate the 1-residue stagger between each chain. Key residues involved in CUB2 binding are highlighted in red. (B) Structure of unbound MASP-1 CUB2 (crystal form 1, pdb 3POE). Secondary structural elements are indicated, and the Ca.sup.2+ is shown as a grey sphere. (C) Crystal structure of MASP-1 CUB2 bound to the 27-residue MBL collagen-like peptide 2 (pdb 3POB). The CUB2 domain is located mainly on the outside of the cone created by the four MBL subunits (inset).
(4) FIGS. 3A-D provide the molecular basis of the Ca.sup.2+-dependent MASP-1 CUB2 binding to MBL. (A) The Ca.sup.2+-binding site in the unbound CUB2 domain, showing the six coordinating ligands arranged in a tetragonal bipyramid. The Ca.sup.2+ is shown as a grey sphere, water molecules as the small grey spheres, and ionic and electrostatic bonds by dotted lines. (B) The same view of the CUB domain in the CUB/collagen complex, showing the interaction with Lys46 of MBL. (C) Close-up of the CUB2/collagen interface. The electrostatic potential of the collagen is shown on a semi-transparent surface representation with some structural elements represented as ball and sticks. Key binding residues of CUB2 are shown as cylinders. Residue numbers of the MBL leading and trailing strands (with an apostrophe) are shown. (D) Sequence alignment of rat MASP-1 CUB2 (SEQ ID NO: 5) with the CUB2 domains of human MASP-1 (SEQ ID NO: 6), MASP-2 (SEQ ID NO: 7), C1r (SEQ ID NO: 8) and C1s (SEQ ID NO: 9). Symbols denote the degree of conservation: (*) identical, (:) a conservative substitution, and (.) a semi-conservative substitution. Secondary structure elements are shown above the alignment and numbering is based on the sequence of rat MASP-1. Residues involved in Ca.sup.2+ and MBL binding are highlighted. Two of the key collagen-binding residues not conserved in C1s, are highlighted in dark grey.
(5) FIGS. 4A-C provide the conserved mechanism of CUB/collagen interactions. (A) Sequence alignment of the CUB2 (SEQ ID NO: 5) and CUB1 (SEQ ID NO: 10) domains of rat MASP-1. Symbols denote the degree of conservation as described in FIG. 3D. Secondary structure elements are shown above the alignment. Residues involved in Ca.sup.2+ and MBL binding are indicated. (B) Superimposition of rat MASP-1 CUB2 and human MASP-1 CUB1 (from pdb: 3DEM27). The lysine-binding pocket is conserved in CUB1 reflecting a common mode of binding to the collagen-like domain of MBL and ficolins. Ca.sup.2+-coordinating residues occupy equivalent positions in CUB1 and CUB2, whereas differences in the binding loops L5 and L9 probably dictate the orientation of the collagen-like domain. (C) Comparison of collagen-bound MASP-1 CUB2 with four unbound structures (from crystal forms 1, 2A and B and 3; see FIGS. 7A and B and materials and methods). Ca.sup.2+ and Ca.sup.2+-coordinating residues are in equivalent positions in all crystal forms. Loop L5 shows a high degree of flexibility. Of the two extremes, only the “closed” conformation, where Tyr225 is pulled inside loop L5, is associated with binding (to the collagen peptide, methylamine, ethylamine and lysine in crystal form 3), whilst the “open” conformation, where Tyr225 is in a more solvent exposed position, is only observed in unbound structures.
(6) FIGS. 5A-E provide inhibition of both MASP binding to MBL and complement activation by small amines. (A) Surface plasmon resonance of MASP-1 binding to immobilized MBL in the presence and absence of amine inhibitors. (B) Inhibition data (including S.E.) expressed as a fraction of MASP-1 binding to MBL in the absence of inhibitors. IC.sub.50 values were 49±9, 193±39 and 228±43 mM for lysine, ethylamine and methylamine, respectively, from three independent experiments. (C) Inhibition of complement activation by lysine. Lectin pathway-specific complement activation was measured on mannan-coated plates by the amount of deposition of the membrane-attack complex. Data are expressed relative to the amount of deposition in the absence of lysine. Error bars represent the difference between duplicate measurements in a representative experiment. The IC.sub.50 from three independent experiments was 25±3 mM. (D and E) Structures of the lysine-binding pocket of MASP-1 CUB2 bound to; (D) methylamine and (E) ethylamine.
(7) FIGS. 6A-D provide the structural basis of complement activation in the lectin pathway. (A) Superimposition of the crystal structure of the MASP-1 CUB2/collagen complex with the structure of the CUB1-EGF-CUB2 homodimer (pdb: 3DEM) showing how MASP binds the collagen-like domains of MBL. Dotted circles highlight the positions of the predicted CUB1 MBL-binding sites for two additional MBL stems. (B) The stems crossover when extended to their full length (53 residues in rat MBL and 59 in human MBL) which is incompatible with the bouquet-like structure of MBL. (C) This discrepancy can be accommodated by distortion of the CUB-EGF-CUB2 domains, through flexion of the CUB2-EGF junction, thereby creating a “spring-loaded” complex. (D) Upon binding to a pathogen, release of the tension in the complex causes the stems to splay apart, pulling the SP domains into alignment; thereby permitting autoactivation.
(8) FIGS. 7A and 7B provide data collection and refinement statistics for CUB2-form 1, CUB2-form 2, CUB2-form 3, CUB2/methylamine, CUB2/ethylamine, peptide 1, peptide 2 and a complex.
(9) FIG. 8 provides sequence alignment of the collagenous regions of MBL, ficolins and C1q. Amino acid sequence alignment of the collagenous regions of various rat and human MBLs (rat MBL-A, SEQ ID NO: 11, rat MBL-C, SEQ ID NO: 12, and human MBL, SEQ ID NO: 13) and ficolins (rat Ficolin-A, SEQ ID NO: 14; rat Ficolin-B, SEQ ID NO: 15; human H-Ficolin, SEQ ID NO: 16; human L-Ficolin, SEQ ID NO: 17; and human M-Ficolin, SEQ ID NO: 18) and the three chains of human C1q (human C1qA, SEQ ID NO: 19; human C1qB, SEQ ID NO: 20; and human C1qC, SEQ ID NO: 21). Symbols denote the conserved (*) residues in all molecules. Numbering is based on the sequence of rat MBL-A. Residues involved in MASP-1 CUB2 binding are highlighted.
(10) FIGS. 9A-9D provide a depiction of the binding pocket of MASP-1 CUB2 after co-crystallisation with small structure-guided inhibitors. Structures of the lysine-binding pocket of MASP-1 CUB2; (A) in the free protein, (B) bound to the MBL collagen peptide, (C) bound to methylamine and (D) bound to ethylamine.
(11) FIGS. 10A-B provide a CUB domain crystal form with an inter-molecular interaction mimicking the MBL mode of binding. (A) The crystal structure of MASP-1 CUB2, crystal form 3, shows three CUB2 molecules in the asymmetric unit forming a cloverleaf shape. (B) A close-up view of the three molecules shows a pseudo auto-inhibited form where K189 from each molecule projects into the binding pocket of its adjacent partner, mimicking the natural collagen/CUB interaction. One of the CUB molecules is represented as an electrostatic potential surface to highlight the charge and surface complementarity.
(12) FIGS. 11A-C provide the quality of the electron density map of MASP-1 CUB2 complexes. Stereo views of different crystal structures. (A) Backbone atoms of the complex of MASP-1 CUB2 with MBL collagen-like peptide 2. (B and C) Electron density (2Fo-Fc map contoured at 1.3σ) of (b) the MASP-1 CUB2/MBL peptide complex and (c) the MASP-1 CUB2/ethylamine complex.
DETAILED DESCRIPTION
Structural and Biochemical Characterization of Binding Partners
(13) Given the intrinsic flexibility and heterogeneity of mannan-binding lectins (MBLs) and MBL-associated serine proteases (MASPs), we chose the smallest binding unit for structural characterization of: a CUB domain and a collagen-like peptide derived from MBL. The CUB2 domain of MASP-1/-3 (subsequently called MASP-1 CUB2) was expressed in E. coli and refolded from inclusion bodies. It eluted significantly later from a gel filtration column in the presence of Ca.sup.2+ than in its absence, both confirming Ca.sup.2+ binding and suggesting that the Ca.sup.2+-bound form is more compact than the Ca.sup.2+-free form (FIG. 1A). Isothermal titration calorimetry revealed a single Ca.sup.2+-binding transition with a K.sub.D of 0.5 mM, which is lower than the concentration of Ca.sup.2+ in serum (1-2 mM), concomitant with the site being mostly occupied under physiological conditions (FIG. 1B). (Conceicao et al., Br. Med. J. 1:1103-1105 (1978)). MASP-1 CUB2 bound to MBL in a Ca.sup.2+-dependent manner (FIG. 1C) with a K.sub.D of 27 μM, as determined by surface plasmon resonance. As anticipated, the affinity was relatively weak compared to full-size MASP-1 binding to MBL (10-100 nM), where multiple CUB/collagen interactions contribute simultaneously. (Girija et al., J. Immunol. 179:455-462 (2007) and Chen and Wallis, J. Biol. Chem. 276:25894-25902 (2001)).
(14) We solved the crystal structure of the CUB2 domain in three different conditions (with one, two and three molecules in the asymmetric unit; see FIGS. 7A and 7B). It is a β-sandwich with two disulfide bonds and a Ca.sup.2+ at one end as shown in FIG. 2B. The Ca.sup.2+ has six coordinating ligands arranged in a tetragonal bipyramid, via the side chains of Glu216, Asp226, Asp263, the main chain of Ser265 and two water molecules as shown in FIG. 3A. These interactions fix the conformation of surrounding loops L5 and L9, accounting for the more compact structure of the domain on gel filtration.
(15) MASPs recognize a short motif in the collagen-like domains of MBLs and ficolins, comprising the sequence OGKXGP (SEQ ID NO: 3) (FIG. 8), where X is generally an aliphatic residue or a methionine residue, and the lysine residue is essential for binding. (Garija et al., supra (2007) and Wallis et al., J Biol Chem 279:14065-14073 (2004)). For crystallisation, we designed peptides of 24 (peptide 1; SEQ ID NO: 1) and 27 (peptide 2; SEQ ID NO: 2) residues, containing the binding motif at the centre, flanked by tandem GPO repeats that were predicted to form stable collagen helices at room temperature. The crystal structure of each peptide was solved independently, confirming the collagen-like properties. Peptide 2 (FIG. 2A) forms a straight, right-handed triple helix, 78 Å in length and ˜12 Å in diameter, with the chains coiled around each other with a characteristic one residue stagger, such that each residue has a unique chemical environment. Peptide 1 is almost identical, but slightly shorter (67 Å long).
(16) Structure of the MASP-1 CUB2/MBL Complex
(17) Crystals of MASP-1 CUB2 in complex with peptide 2 were grown in the presence of Ca.sup.2+ and diffracted to 1.8 Å resolution. The structure was solved by molecular replacement, using the structures of the free components as search models. The final refined model had R and R.sub.free values of 19.8 and 25.2% respectively; see methods and FIGS. 7A and 7B. CUB2 binds to two of the three OGKLGP (SEQ ID NO: 3) motifs in the central part of the collagen peptide (FIG. 3C). The leading and trailing chains of the collagen account for 56% and 44% of the interface (buried area), respectively; the middle chain is not involved in binding (Table 1). Major contacts are made by Lys46, Leu47 and Pro48 of the leading chain and Hyp44, Gly45, Lys46 and Leu47 of the trailing chain. The interface buries just 326 Å.sup.2 of solvent-accessible collagen surface and 305 Å.sup.2 of the CUB domain, which is remarkably small for a protein/protein complex and smaller than the handful of protein/collagen complexes for which structures are available in the structure databank. See, e.g., Emsley et al., Cell 101:47-56 (2000) and Hohenester et al., Proc. Natl. Acad. Sci. USA 105:18273-18277 (2008). The side chain of Lys46 of the leading strand plays a key role by extending from the peptide to contact three of the Ca.sup.2+-coordinating residues in the CUB domain, the carboxylate groups of Glu216 and Asp263 and the hydroxyl group of Ser265, and thus forms a bridge from the collagen ligand via the CUB to the Ca.sup.2+ (FIGS. 3B and 3C). Although the positions of these residues are essentially unchanged upon collagen binding, the Ca.sup.2+ is vital for orientating the binding loops as emphasized by comparison with the structure of the CUB1-EGF-CUB2 fragment of MASP-2 in the Ca.sup.2+-free form, in which much of loop L5 in CUB2 is disordered. (Feinberg et al., EMBO J 22, 2348-2359 (2003)). Thus, the binding arrangement provides a clear rationale for the Ca.sup.2+-dependence of MASP-1 binding to MBL and ficolins.
(18) TABLE-US-00001 TABLE 1 MBL residue Leading Trailing Hyp44 0 54.2 Gly45 0 8.6 Lys46 106.5 32.2 Leu4 9.7 48.2 Gly48 0.1 0 Pro49 65.0 0 Hyp50 1.9 0
(19) Relatively few additional contacts stabilize the complex. Hyp44 of the trailing chain, forms two important water-mediated hydrogen bonds to the carbonyl group of Asp217 and the carboxylate of Glu216 (FIG. 3C). These interactions would not be possible if Lys46 of the middle or lagging strand were to sit in the binding pocket, because a leucine residue would occupy the position of the hydroxyproline, probably explaining the unique conformation observed in the crystal structure. Tyr225 of the CUB domain forms extensive hydrophobic interactions at the binding interface and its hydroxyl group forms a hydrogen bond with the backbone carbonyl of Leu47 of the leading chain of the collagen. His218 also forms part of the hydrogen-bonding network, forming backbone contacts to the carbonyl of Gly45 of the trailing chain and the amine of Glu220 of the CUB. The latter interaction probably stabilizes the conformation of loop L5, which remains in close proximity to the collagen peptide, thereby reducing solvent accessibility to the binding interface. The side chain of Glu220 in loop L5 is sandwiched between the side chain of Lys46 and the amine of Leu47 of the trailing chain and is partially buried at the interface (FIG. 3C). However, the distances between each group is >3.5 Å, so Glu220 probably contributes little towards binding.
(20) To investigate the role of conformational changes on binding, we compared the structures of the bound and unbound components. Comparison of five distinct structures of CUB2 (FIG. 4C), reveals significant differences in the position of the binding loop L5, implying that this region is likely to be relatively flexible in the unbound MASP. In particular, the positions of Tyr225, His218, and Glu220, diverge by several Å, with movement of up to 6.8 Å for the Tyr Oη atom. Interestingly, loop L5 is in essentially the same position (referred to as the closed conformation) in all structures of CUB2 bound to a ligand (collagen, methylamine, ethylamine or lysine in CUB2 crystal form 3), probably because of the role of Tyr225 in forming hydrophobic contacts with the ligand. Apart from loop L5, the bound and unbound CUB domains are very similar. There are also no major changes in the collagen peptide upon binding; in particular, the triple helix does not bend, as has been observed in a complex of the I domain of α2β1 integrin38.
(21) Structure Guided Inhibition of MASP Interactions
(22) The molecular details revealed by our structure are in excellent agreement with previous biochemical results, both explaining the importance of Ca.sup.2+ and demonstrating how the OGKLGP motif (SEQ ID NO: 3) and the lysine residue in particular are vital. (Girija et al., J. Immunol. 179:455-462 (2007) and Wallis et al., J. Biol. Chem. 279:14065-14073 (2004)). Given the small size of the interface and the dominant role of Lys46, we sought further confirmation of the structure by testing the sensitivity of full-length MBL/MASP-1 complexes towards the free amino acid and small primary amines. Remarkably, lysine inhibited binding albeit relatively weakly (IC.sub.50 of ˜50 mM; FIG. 5), and ethylamine and methylamine were also inhibitors. Moreover, lysine inhibited complement activation itself in a lectin pathway specific assay (IC.sub.50 of ˜25 mM), probably by dissociating MBL/MASP complexes in serum. To probe the mechanism of inhibition, we co-crystallised the CUB2 domain with each of the smaller amines. The inhibitors are clearly evident in the lysine-binding pocket of the resulting crystal structures, forming equivalent contacts to Glu216, Asp263 and Ser265, and packing against Tyr225 (FIG. 5D, 5E, AND 9), thus occupying the binding site for MBL. The propensity for lysine to interact in this way was further demonstrated in a crystal structure of the CUB domain alone, in which a lysine side chain from each of three domains of the asymmetric unit cell projects into the binding pocket of its adjacent partner, mimicking the natural collagen/CUB interaction (FIGS. 10A and 10B).
(23) Yet further support for our structure is provided by the excellent agreement with the available mutagenesis data for human MASP-1. (Teillet et al., J. Biol. Chem. 283:25715-25724 (2008)). For example, the mutation Tyr225Ala completely disrupts binding to H-ficolin and reduces binding to MBL and L-ficolin significantly, whereas His218Ala reduces binding to all three proteins. Interestingly, Ser265Ala inhibits binding towards the ficolins but not to MBL, showing that loss of the hydrogen bond between Ser265 and Lys46 can be tolerated, at least in part, presumably because the remaining two salt-bridge contacts of the lysine side chain (to Glu216 and Asp263) are still possible.
(24) A Conserved Mode of Ca.sup.2+-Dependent CUB/Collagen Binding
(25) The key binding residues (His218, Y225, Glu216 and Asp263 and Ser265) of MASP-1 CUB2 are strictly conserved in MASP-2, implying comparable binding to MBL (FIG. 3D). Moreover, these residues are also conserved in CUB2 of C1r, supporting a common binding mechanism in initiating complexes of the lectin and classical pathways. Although residues forming the Ca.sup.2+-binding site are present in CUB2 of C1s, which does not bind to C1q, His and Tyr residues (at positions 218 and 225 of MASP-1) are replaced by Ala and Leu, respectively. These findings are entirely compatible with current models of the C1 complex, in which two molecules each of C1r and C1s form a heterotetramer that presents six binding sites to C1q, via the CUB1 and CUB2 domains of C1r and CUB1 of C1s. (Bally et al., J Biol Chem 284:19340-19348 (2009) and Phillips et al., J Immunol 182:7708-7717 (2009)). Conversely, CUB2 of C1s is located at the C1s/C1s interface, close to the centre of the complex and does not participate directly in C1q binding.
(26) The key features of the CUB2 binding site are conserved in CUB1 (FIG. 4A), and important binding residues are also conserved in CUB1 of MASP-2, C1r and C1s (data not shown). See, also Teillet et al., J. Biol. Chem. 283:25715-25724 (2008). Moreover, structure superimposition of rat MASP-1 CUB2 with human MASP-1 CUB1 reveals substantial overlap, in which the Ca.sup.2+ superimpose, and the side chains of Glu49, Asp102, and Ser104 substitute for Glu216, Asp263 and Ser265, respectively (FIG. 4B). By analogy with CUB2, the carboxylate groups of these acidic residues and the hydroxyl of the serine are ideally configured to bind to Lys46 of the collagen. In addition, Tyr56 substitutes for Tyr225, but the equivalent of His218 in loop L5 is missing. Instead, Phe103 of loop L9 and Tyr52 of loop L5 probably participate in the interaction. Lys46, thus acts as a central pivot in all CUB/collagen interactions, and the neighboring contacts dictate the orientation of the collagen to enable coordinated binding to MBL.
(27) The central features of the CUB/collagen structure described here are remarkably similar to binding observed between vitamin B.sub.12-bound gastric intrinsic factor and cubilin, confirming a common binding mechanism for Ca.sup.2+-dependent CUB interactions. (Andersen et al., Nature 464:445-448 (2010)). Notably, in each complex, binding is multivalent and involves the crucial interaction of a basic side chain to acidic residues coordinating a Ca.sup.2+ within the CUB-binding partner, with residues from equivalent neighboring loops also contributing towards binding.
(28) The data presented here reveals for the first time how MBL/MASP and ficolin/MASP complexes assemble. Combining our CUB/collagen complex with the existing structure of the human CUB1-EGF-CUB2 dimer26, the MASP domains take the form of the cross-bar of an uppercase letter “A”, with each CUB2 domain holding a collagen-like stem to form the sides (FIG. 6A). (Wallis et al., Immunobiology 215:1-11 (2010)). Conserved contacts between CUB1 domains and additional collagen-like stems would permit up to four separate interactions simultaneously (FIG. 6A, broken circles). By bridging separate stems, the MASP would be receptive to any changes imparted following engagement of the surface of a pathogen by the CRDs of MBL or the fibrinogen-like domains of ficolins. Although compatible with the architectures of MBL and ficolins, crucially, the angle between the collagen-like stems (˜80°; FIG. 6A) is greater than that which would permit unconstrained binding. This is reflected by the stems converging with only four GXY repeats distal to the MASP-binding site, rather than the seven or more repeats present in all MBLs and ficolins (FIG. 6B). Given the rigidity of the collagen-like stems, which are tethered at the N-termini, the most likely way that binding could be achieved is through distortion of the CUB1-EGF-CUB2 dimer. Of the two domain junctions within this portion of the MASP, only EGF-CUB2 is likely to permit significant movement, because the interface is comparatively small. (Feinberg et al., EMBO J. 22, 2348-2359 (2003)). By contrast, CUB1 and the EGF-like domain are essentially fixed through extensive hydrophobic contacts at the dimer interface. (Teillet et al., J. Biol. Chem. 283:25715-25724 (2008) and Feinberg et al., EMBO J. 22, 2348-2359 (2003)). Overall therefore, complexes would be under tension, or “spring loaded”, as they circulate in serum (FIG. 6C). We propose that when MBL or ficolins engage with the surface of a pathogen, release of this tension helps to drive the changes that trigger activation. The most likely scenario is that extension of the EGF-CUB2 junction, concurrent with splaying apart of the collagen-like stems, pulls the SP domains together to allow reciprocal activation (FIG. 6D). MBLs and some ficolins possess an interruption or kink in the collagenous domain, which could also become deformed on binding to MASPs. However, this region can be removed from MBL without affecting its ability to activate complement and some ficolins do not possess such a kink.
(29) Accordingly, in one aspect, the disclosure provides a method of screening for one or more small molecules capable of inhibiting MASP protein activity. The method comprises contacting a MASP protein, or a portion thereof, with at least one candidate small molecule inhibitor of MASP protein activity in the presence of a polypeptide comprising at least one collagen-like binding domain of a mannose-binding lectin (MBL) or ficolin, and Ca.sup.2+. The method also comprises determining, at least one of (i) the presence or affinity of binding between the MASP protein, or portion thereof, and the polypeptide comprising the at least one collagen-like binding domain, or (ii) the level of MASP protein activity in the presence and absence of the at least one candidate small molecule inhibitor. A reduced binding between the MASP protein and the polypeptide comprising the collagen-like binding domain of MBL, and/or reduced MASP protein activity determined in the presence of a candidate small molecular in step (b) compared to the level of binding and/or MASP protein activity determined in the absence of the candidate small molecule inhibitor indicates the capacity of the small molecules to inhibit MASP protein activity.
(30) In one embodiment, the level of MASP protein activity is determined by performing an assay to determine the ability of the MASP protein to activate the lectin or classical complement pathway. In one embodiment, the method comprises the use of a full length MASP protein, and the level of MASP protein activity is determined by performing an assay to determine the ability of the MASP protein to activate complement. Exemplary full length amino acid sequences for MASP proteins are provided herein, and set forth as SEQ ID NOS: 22-28. Other known naturally occurring isoforms and the present screening method can be carried out with such additional isoforms.
(31) In one embodiment, the binding activity of a complement C1r/C1s, Ugegf and BMP1 (CUB1 or CUB2) domain of the MASP protein to a collagen-like domain of the MBL or ficolin is determined. In one embodiment, the collagen-like domain comprises at least one or more of the amino acid sequence OGKXGP, wherein X is an aliphatic amino acid residue or methionine residue. In one embodiment, the binding activity of the CUB1 or CUB2 domain of the MASP protein to the collagen-like domain of the MBL or ficolin is determined using surface plasmon resonance.
(32) In one embodiment, the small molecule with an indicated capacity to inhibit MASP protein activity disrupts the binding between at least one of residues Glu216, His218, Lys225, Asp263, and Ser265 of the MASP protein, with reference to SEQ ID NO: 6 (human MASP-1 CUB domain), and MBL or ficolin. In one embodiment, the small molecule with an indicated capacity to inhibit MASP protein activity disrupts the binding between at least one of residues Glu216, Asp226, Asp263, and Ser265 of the MASP protein, with reference to SEQ ID NO: 6, and Ca.sup.2+.
(33) A person of ordinary skill in the art will appreciate that the present method can incorporate known high throughput method to test a plurality of candidate MASP inhibitors.
(34) MBL/MASP and ficolin/MASP interactions represent attractive targets for therapeutics aimed at inhibiting complement activation in ischemia-reperfusion injury and other complement-dependent disorders. The C1 complex is also an important drug target because of its pathogenic role in various neurodegenerative disorders including Alzheimer's disease. (Tenner and Fonseca, Adv. Exp. Med. Biol. 586:153-176 (2006)). The mechanistic basis of complex formation presented here provides a starting point for the design of such therapeutics, which would prevent complement activation at its earliest stage, before the amplification steps characteristic of downstream events. It would also allow selective targeting of a single pathway, rather than complete shutdown of complement, thereby causing minimal disruption to immune function. It is particularly encouraging that the CUB/collagen interface can be targeted by small molecules, highlighting the potential of this approach for the development of novel therapeutics via library and/or fragment-based drug design strategies.
EXAMPLES
Experimental Methods
(35) Expression, Production and Refolding of the MASP-1 CUB2 Domain
(36) The cDNA encoding MASP-1 CUB2 (residues E165-A277 of the mature polypeptide, isoform 1) was cloned into the NcoI-EcoRI sites of plasmid pET28A (Novagen) and transformed into E. coli BL21 DE3. Cells were grown to an OD.sub.595 of 0.8 in 2×YT medium and expression was induced by addition of IPTG to a final concentration of 1 mM. After four hours at 37° C. with shaking, cells were harvested by centrifugation at 4000×g for fifteen minutes, and resuspended in Bugbuster® master mix reagent (40 ml reagent/L of culture medium; Novagen), containing a protease inhibitor cocktail tablet (Roche). The cell suspension was incubated with gentle shaking for 15 minutes at room temperature and the cells were disrupted by sonication. The insoluble fraction, containing MASP 1 CUB2 inclusion bodies was pelleted by centrifugation at 20,000×g for twenty minutes, at 4° C. and washed twice with 1:10 diluted Bugbuster® mix and subsequently with 2M urea in 50 mM Tris pH 8.0 containing 0.5 M NaCl and 1 mM EDTA. Purified inclusion bodies were solubilized in 8M urea in 50 mM Tris pH 8.0, 5 mM DTT to a final concentration of 2 mg/ml. The CUB domain was refolded by drop dilution into 50 mM Tris/HCl, pH 8.5, containing 240 mM NaCl, 10 mM KCl, 2 mM MgCl.sub.2, 2 mM CaCl.sub.2, 0.4 M sucrose and 1 mM DTT, to a final concentration of 0.1 mg/ml. After three days, the sample was dialyzed against 25 mM Tris/HCL at pH 8.0, containing 25 mM NaCl and bound to a Q-Sepharose column (20 mL) equilibrated with the same buffer. Protein was eluted using a gradient of NaCl (25 mM to 1 M over 40 ml) and fractions containing the CUB domain were concentrated by filtration. Finally, protein was loaded onto a 16/60 Superdex 75 gel filtration column (GE healthcare) equilibrated in 10 mM Tris buffer (pH 7.5) containing 10 mM NaCl and 2 mM CaCl2. Fractions were pooled and stored at −80° C.
(37) Proteins and Peptides
(38) Rat MBL-A (referred to as MBL throughout) was produced as described previously. (Wallis and Drickamer, supra (1999)). MASP-lent (referred to as MASP-1) is a modified form of rat MASP-1 in which the zymogen cleavage site has been replaced by an enterokinase recognition site (DDDDK; SEQ ID NO: 4). It is secreted as a zymogen but cannot autoactivate, and binds to MBL with comparable affinity as native MASP-1. (Chen and Wallis (2004)). Two collagen peptides were synthesized and HPLC purified by Pepceuticals Ltd. Peptide 1 is Ac(GPO).sub.3-GKL-(GPO).sub.4NH.sub.2 (SEQ ID NO: 1) and peptide 2 is Ac(GPO).sub.4-GKL-(GPO).sub.4NH.sub.2 (SEQ ID NO: 2) (predicted Tm=26.9 and 36.0° C. respectively). (Persikov et al., J Biol Chem 280:19343-19349 (2005)).
(39) Crystallization
(40) Crystals were grown using the sitting-drop method by mixing equal volumes of protein and reservoir solution (1.2+1.2 μL). All proteins were prepared in 10 mM Tris pH 7.5, containing 10 mM NaCl and 2 mM CaCl.sub.2. For the complex, CUB2 (5 mg/ml; 0.38 μM) was mixed with a 1.5-fold molar excess of trimeric peptide. The reservoir solution contained 28% PEG 8000 and 50 mM Tris/HCl, pH 8.0 and 4 mM CaCl.sub.2. CUB2 was crystallized in three different conditions to give different crystal forms (see FIGS. 7A and 7B). Reservoir solutions were: 1) 24% PEG 8K in 50 mM Tris/HCl, pH 9.0 with 20 mM CaCl.sub.2; 2) 1.5 M lithium sulphate in 100 mM Tris, pH 8.5 and 3) 24% PEG 8K in 50 mM Tris/HCl, pH 9.0 with 200 mM MgCl.sub.2, respectively. Both collagen peptides were crystallized using a reservoir solution of 1.6 M ammonium sulphate in 100 mM MES, pH 6.0 but yielded different crystal forms. Crystals of complex, collagen peptide and CUB2, condition 3, grew at 20° C. and crystals of the CUB domain, conditions 1 and 2, grew at 6° C. All crystals were briefly transferred to reservoir solution containing 20% glycerol before cryoprotection in liquid nitrogen, and were maintained at 100 K during data collection. Complexes of CUB2 bound with ethanolamine, methanolamine and lysine were crystallized in buffer condition 2, supplemented with 1 M amine (Sigma). Although crystals grew in the presence of all three amines, and novel density was observed in the binding pocket, lysine could not be resolved in the resulting structure, probably reflecting more than one binding conformation.
(41) X-Ray Crystallography
(42) For the MASP-1 CUB2/MBL collagen-like peptide structure (PDB 2POB), diffraction data were collected at Diamond beamline 104 (λ=0.9184) and recorded on ADSC Q315 CCD detectors. Data were processed with XDS). (Kabsch, J. Appl. Cryst. 26:795-800 (1993)). Phases were determined using phaser, the model was optimized using cycles of manual refinement with Coot and maximum likelihood refinement in Refmac5—all software are part of the CCP4 suite. (Collaborative Computational Project, Number 4, Acta Cryst. D50:760-763 (1994)). The final Ramachandran plot shows 95.8% of residues in favored regions, and 4.2% in allowed regions, as defined by PROCHECK (CCP4 suite).
(43) All eight structures determined in this study have been submitted to the Protein Data Bank with the accession numbers listed in FIGS. 7A and 7B available at the rcsb Protein Data Bank Website.
(44) Structure of the Collagen Peptide
(45) Crystals of the peptide 2 (SEQ ID NO: 2) diffracted to 1.5 Å resolution and the structure was solved by molecular replacement using the Gly-Pro-Hpr collagen peptide structure as a search model (pdb id: 1CAG48). (Bella et al., Science 266:75-81 (1994)). The C-terminal end of the trailing chain is not well resolved and shows higher temperature factor due to the absence of neighboring chains. The side chains of Lys46 and Leu47 residues in the middle of the collagen peptide point into the solvent. Lys46 of the leading chain shows nice density up to the CE, as a consequence of stacking against Hyp44 of the trailing chain. In contrast, we observe poor density for the side chain of Lys46 from the other two chains, which do not appear to make intrachain contacts.
(46) Structure of the Rat MASP-1 CUB2
(47) Uncomplexed CUB2 was crystallized in three different conditions. The initial structure (crystal form 1) was solved by molecular replacement using the human MASP-1 CUB2 module (pdb id: 3DEM25) as a search model, and refined to 1.5 Å resolution (FIG. 7A). (Teillet et al., J. Biol. Chem. 283:25715-25724 (2008)). The identity of the Ca.sup.2+ was verified by its anomalous signal using CuKα radiations. Interestingly, we find a second Ca.sup.2+ ion in our anomalous signal coordinated by Ser190, Glu192 and 4 water molecules. This Ca.sup.2+ is likely due to the presence of relatively high concentration of Ca.sup.2+ in the crystallization buffer and is not relevant here. Furthermore, this region of the protein is usually at the interface with the EGF module in the full-length MASP, which would not permit Ca.sup.2+ binding. In this structure, the “open” conformation of loop L5 is stabilized via Tyr225 and Glu220 making contact with another molecule in the crystal lattice.
(48) We succeeded in obtaining another crystal form, crystal form 2, with two CUB molecules in the asymmetric unit (FIG. 7B). Interestingly, although the molecules are otherwise almost identical, loop L5 adopts the “closed” conformation in one molecule, and the “open” form in the other (crystal form 2A and 2B respectively, as described in FIG. 5C). This condition was also used to co-crystallize CUB2 with methylamine, ethylamine and lysine. In each case, only the CUB2 molecule in the “closed” conformation in the asymmetric unit is bound to the amine, the CUB in the “open” form was ligand free.
(49) In the third crystal form, crystal form 3, there are three CUB molecules in the asymmetric unit (FIG. 9). All three are almost identical, with loop L5 in the closed conformation in each case. Lys189 projects from one CUB into the binding pocket of its adjacent neighbor, mimicking the CUB/collagen interaction.
(50) Gel Filtration
(51) Gel filtration was carried out on a 16/60 Superdex 200 column (GE healthcare) in 50 mM Tris/HCl pH 7.4, containing 150 mM NaCl at a flow rate of 1 ml/min.
(52) Isothermal Titration Calorimetry
(53) The CUB2 domain was made calcium free by EDTA treatment and dialyzed extensively into 50 mM Tris/HCl, pH 7.4 containing 150 mM NaCl. Freshly prepared 2 mM CaCl.sub.2 in the same batch of assay buffer was titrated into CUB2 domain (114 μM) in the cell of a VP-isothermal titration calorimeter (Microcal) equilibrated at 25° C. The data were fitted to a 1:1 binding model using the Origin software supplied by the manufacturer, after subtraction of the heats of dilution of the calcium. Because binding was weak, the experiment was carried out under low ‘c-value’ conditions (where the concentration of the protein in the cell is lower than that of the K.sub.D). Hence, during fitting the stoichiometry was fixed at 1 (the stoichiometry expected for the CUB2 domain based on examination of the CUB1-EGF-CUB2 crystal structure) and the K.sub.A and ΔH variables were allowed to float.
(54) Surface Plasmon Resonance
(55) MBL (0.025 mg/ml) was immobilized onto the surface of a GLM sensor chip (˜10,000 response units) at pH 4.5, using amine-coupling chemistry. Binding to CUB2 was measured on a ProteOn® XPR36 (BioRad) in 10 mM Tris/HCl pH 7.4, containing 140 mM NaCl, 5 mM Ca.sup.2+ and 0.005% Tween-20 at 25° C., at a flow rate of 25 μL/min. For inhibition assays, MBL was immobilized on to a CM-5 sensor chip (˜14,000 response units) and binding was measured on a Biacore® 2000 (GE Healthcare) in 20 mM Tris/HCl pH 7.4, containing 140 mM NaCl, 5 mM Ca.sup.2+, and 0.005% SP-20. MASP-1 (100 nM) was premixed with inhibitor and flowed over the chip surface at a rate of 5 μL/min at 25° C. Changes in response units caused by differences in sample composition were subtracted from all data. To account for changes in ionic strength of the buffer due to the inhibitor, binding was measured in the presence of equivalent concentrations of NaCl. Binding was >60% maximal at all concentrations tested (up to 550 mM).
(56) Complement Assay
(57) Lectin pathway specific complement activation was measured using the Wieslab® MBL pathway kit (Euro-Diagnostica), which measures deposition of the membrane attack complexes on a mannan-coated plate. The lectin pathway is functional, even under conditions of high ionic strength (1M NaCl). (Petersen et al., J Immunol Methods 257:107-116 (2001)).
(58) While illustrative embodiments have been illustrated and described, it will be appreciated that various changes can be made therein without departing from the spirit and scope of the invention.
