Cryo-EM structures of human A2ML1 elucidate the protease-inhibitory mechanism of the A2M family

Abstract

A2ML1 is a monomeric protease inhibitor belonging to the A2M superfamily of protease inhibitors and complement factors. Here, we investigate the protease-inhibitory mechanism of human A2ML1 and determine the structures of its native and protease-cleaved conformations. The functional inhibitory unit of A2ML1 is a monomer that depends on covalent binding of the protease (mediated by A2ML1's thioester) to achieve inhibition. In contrast to the A2M tetramer which traps proteases in two internal chambers formed by four subunits, in protease-cleaved monomeric A2ML1 disordered regions surround the trapped protease and may prevent substrate access. In native A2ML1, the bait region is threaded through a hydrophobic channel, suggesting that disruption of this arrangement by bait region cleavage triggers the extensive conformational changes that result in protease inhibition. Structural comparisons with complement C3/C4 suggest that the A2M superfamily of proteins share this mechanism for the triggering of conformational change occurring upon proteolytic activation.

Fig. 1. A2ML1 is a monomer and prevents access of large substrates to a conjugated protease.

A Thermolysin was incubated with the indicated molar ratios of A2ML1 to thermolysin with and without BAPN and the residual protease activity was then measured against β-casein (left) and gelatin without BAPN (right). For both substrates, a 2–3-fold excess of A2ML1 was required for full inhibition of thermolysin, whereas A2ML1 demonstrated little to no inhibition of thermolysin in the presence of BAPN. Data are represented as the mean values +/− standard deviation, n = 3 independently prepared samples. B SDS-PAGE analysis of A2ML1 cleaved using a 1:0.1 molar ratio of A2ML1 to thermolysin with a titration series of BAPN revealed increased cleavage of intact A2ML1 with increasing BAPN. This supports A2ML1’s dependence on covalent conjugation to the protease in order to achieve inhibition. This image is representative of duplicate experiments. Molecular weight markers are shown on the left-hand side in kDa. C, D Protease activity assays with chymotrypsin and HNE. In both cases, a higher molar ratio of A2ML1 to the protease is needed to inhibit the protease when casein is used as the substrate compared to the larger gelatin. Source data are provided in the Source data file, see Data availability. Data are represented as the mean values +/− standard deviation, n = 3 independently prepared samples. E, F Cleavage sites for the indicated proteases were determined by Edman sequencing (E) or LC-MS/MS (F), showing that most tested proteases cleave within the predicted bait region sequence with the exception of the Arg-specific proteases trypsin and plasmin.

Fig. 2. Structures of native A2ML1.

A Schematic representation of the domains of A2ML1 with disulfides and the internal thioester indicated. Notice that the LNK and BR are inserted in the MG6 domain, whereas the TE domain is inserted in the CUB domain. The MG8 domain is also known as the receptor-binding domain, RBD. B Local resolution of the native A2ML1 cryo-EM map in two orientations. C Fourier shell correlation indicates a resolution of 2.9 Å. D The plot of the particle orientation distribution demonstrates the presence of preferred orientations. E Example of EM map quality for native A2ML1 for residues in the MG2 domain. F As in panel E, but displaying the X-ray map obtained after molecular replacement and density modification at 4.4 Å resolution. G Native A2ML1 in a cartoon representation. The domains are colored as in panel A. To the right, the dashed red line indicates flexible BR residues not modeled. H Close-up on the inter-domain channel accommodating the C-terminal residues of the bait region.

Fig. 3. The structures of TEV-P-cleaved A2ML1.

A SDS-PAGE analysis of A2ML1 and empty trap TEV-P cleaved A2ML1-CE and B A2ML1-CC and -CA samples (to the right, the samples are shown after hydrophobic interaction chromatography (HIC) which showed co-elution of TEV-P with A2ML1 even without covalent protease trapping in the A2ML1-CA sample). The A2ML1-CE sample and post-HIC A2ML1-CC and -CA samples that are shown here were used for preparation of grids for cryo-EM; these samples were prepared once. Molecular weight markers on shown on the left-hand side of each gel in kDa. C Local resolution maps for the three different monomeric TEV-P cleaved A2ML1. Notice the significantly lower resolution for the TE and MG3 domains compared to the rest of the molecule. D Fourier shell correlation plots suggest resolution of 2.9 Å and 3.0 Å for A2ML1-CC and A2ML1-CE, respectively. E TEV-P conjugated A2ML1-CC in a cartoon representation with the domains colored as in Fig. 2A. The orientation is similar to that in panel C. The location of thioester glutamine is indicated as a red sphere marked Q973. For comparison, native A2ML1 is presented with the same orientation of the MG-ring in the right part of Fig. 2G. F EM map of the MG7 domain in A2ML1-CC confirming two disulfides bridges formed by four cysteines that are located close to each other. G EM map for two NAG residues linked to Asn120 in the linker connecting the MG1 and MG2 domains in A2ML1-CC.

Fig. 4. In cleaved A2ML1, the thioester and CUB domains associate through a conserved hydrophobic interface.

A Conformation of the monomer from the A2M-MA tetramer and C3b for comparison with A2ML1 in Fig. 3E with the thioester glutamines indicated by the red spheres. In both A2ML1 and A2M, the TE and MG2 domains interact, whereas in C3b, the TE domain is instead adjacent to the MG1 domain. Notice the green MG8 domain in C3b that is stably associated with the remaining parts of C3b. A unique C-terminal domain in C3b, C345C, is not shown. B Comparison of the conformation of the MG7-CUB-TE domains in A2ML1, A2ML1-CC, and C3b. All structures were superimposed through their MG7 domains to emphasize the movement of the CUB and TE domains upon proteolytic activation of A2ML1 and C3. C Expanded views of the interactions formed by the TE domain with the CUB and MG2 domains, top, and bottom, respectively. Putative polar interactions are indicated by dotted lines. D Alignment of regions of the CUB and TE domains in mammalian A2ML1 supports that the interface is very highly conserved. The full alignment is presented in Supplementary Fig. 11.

Fig. 5. The MG8 domain is flexibly attached to the rest of cleaved A2ML1.

A The two most extreme states observed in a 3D variability analysis of the A2ML1-CC with map and models superimposed. Except for the TE and CUB domains, only limited differences between the two states are observed. In the magnified view, the extreme positions of the TE domain are compared. B–D SDS-PAGE analyses of selected fractions from the SEC runs with native, methylamine-activated, and TEV-P cleaved A2ML1t-fMG8. Native A2ML1 was used as a control during SDS-PAGE analysis. The identities of the specific bands on the SDS-PAGE gels were confirmed by LC-MS/MS. In native A2ML1t-fMG8, the MG8 domain co-elutes with the rest of A2ML1 during SEC despite furin cleavage between CUB and MG8 (panel B), while the MG8 domain elutes independently of the rest of A2ML1 during SEC after activation by methylamine or TEV-P (panels C, D). This experiment was performed once.

Fig. 6. The conjugated protease has rotational freedom in an environment containing flexible extended regions.

A Comparison of the EM maps for A2ML1-CC and A2ML1-CE with a docked TEV-P. The absence of EM map density close to the thioester suggests rotational freedom and multiple possible conjugation sites on the protease. Strikingly, weak density is present in the volume of the docked TEV-P in A2ML1-CE, possibly due to a few preferred conformations of the LNK region in the empty trap state. B Hypothetical full model of native A2ML1. LNK residues not modeled in any of the structures were taken from the alphafold2 prediction. The bait region was modeled by hand in Coot with exposure of the experimental cleavage sites presented in Fig. 1E, F. C Hypothetical full model of cleaved A2ML1. Regions in LNK and the N-terminal part of the bait region (BR-N) not modeled in A2ML1-CC were modified manually in Coot starting from their conformation in panel B.

Fig. 7. A conserved mechanism for the conformational change in A2MF proteins.

A The BR-C retracts through the channel upon bait region cleavage and associates with a novel binding site formed by the MG6 and MG7 domains. The BR-C is presented in a back view compared to the front view in Fig. 2H. B Upon retraction of the BR-C, the MG3 and MG7 domains can move toward the LNK region to close the BR-C accommodating channel. Arrows indicate direction and magnitude of movement between the native (left) and cleaved state (right). C Magnified view of the BR-C in A2ML1-CC featuring the tight recognition of Arg730. D Equivalent view of the Nt-α′ region in complement C4b. E A hypothetical series of events from reversible protease-A2ML1 association and formation of the covalent bond (conjugation) between cleaved A2ML1 and the protease (steps 1–6). In step 3, the protease has cleaved and associates non-covalently with A2ML1. The BR-C has not retracted, and no overall conformational change has taken place. In step 4, the BR-C has retracted and the MG2-3-LNK delimited channel is open. In step 5, the channel has closed upon movement of the MG3 and MG7 domains. In step 6, the conformational change is complete, and the TE domain has formed a covalent bond to the protease. The protease may escape at any step between bait region cleavage and conjugation; steps 7–8 show one scenario where the protease escapes at a point late in the conformational change.