Cryo-EM reveals that iRhom2 restrains ADAM17 protease activity to control the release of growth factor and inflammatory signals

Abstract

A disintegrin and metalloprotease 17 (ADAM17) is a membrane-tethered protease that triggers multiple signaling pathways. It releases active forms of the primary inflammatory cytokine tumor necrosis factor (TNF) and cancer-implicated epidermal growth factor (EGF) family growth factors. iRhom2, a rhomboid-like, membrane-embedded pseudoprotease, is an essential cofactor of ADAM17. Here, we present cryoelectron microscopy (cryo-EM) structures of the human ADAM17/iRhom2 complex in both inactive and active states. These reveal three regulatory mechanisms. First, exploiting the rhomboid-like hallmark of TMD recognition, iRhom2 interacts with the ADAM17 TMD to promote ADAM17 trafficking and enzyme maturation. Second, a unique iRhom2 extracellular domain unexpectedly retains the cleaved ADAM17 inhibitory prodomain, safeguarding against premature activation and dysregulated proteolysis. Finally, loss of the prodomain from the complex mobilizes the ADAM17 protease domain, contributing to its ability to engage substrates. Our results reveal how a rhomboid-like pseudoprotease has been repurposed during evolution to regulate a potent membrane-tethered enzyme, ADAM17, ensuring the fidelity of inflammatory and growth factor signaling.

Keywords: ADAM17/iRhom2 sheddase complex; cryo-EM; growth factor signaling; immune signaling; rhomboid pseudoprotease.

Graphical abstract

Figure 1

Architecture of the human ADAM17/iRhom2 sheddase complex (A) Cryo-EM density map of the human ADAM17/iRhom2 sheddase complex, viewed from the side of the membrane (left) or from the extracellular space (right). (B) Overall structure of the human ADAM17/iRhom2 complex. The catalytic Zn²⁺ is shown as a green sphere. The approximate membrane boundaries are represented by gray bars. (C) Structure of iRhom2. Disulfide bonds in the IRHD (iRhom homology domain) are shown as spheres. (D) Transmembrane domains of iRhom2, viewed from the extracellular side. (E) Intramembrane interface between iRhom2 and ADAM17. (F) A diagram of the sheddase complex. ADAM17 is shown in yellow, with its prodomain in red. iRhom2 is colored blue. The first transmembrane domain of iRhom2 is labeled “1.” The key D475 residue (discussed below) is shown as a turquoise circle. The furin cleavage site is indicated. See also Figures S1 and S2.

Figure 2

Structure of ADAM17 and the mechanism of prodomain inhibition (A) Domain arrangement of ADAM17 in the sheddase complex. The catalytic Zn²⁺ is shown as a green sphere. The approximate membrane boundaries are represented by gray bars. MPD, membrane proximal domain. CANDIS, conserved ADAM17 dynamic interaction sequence. (B) Structure of ADAM17 prodomain and protease domain. The furin site, not resolved in the structure, is indicated by a dashed line. The inset shows a zoomed-in view of the catalytic center of the protease domain. (C) Interfaces between the prodomain and the rest of the complex. The extracellular domain of the ADAM17/iRhom2 complex is shown in surface representation. See also Figure S2.

Figure 3

Dissecting the interfaces between iRhom2 and ADAM17 (A) Four major interfaces between iRhom2 and ADAM17 are highlighted. Selected residues comprising the interfaces (within 5 Å distance) are shown in ball-and-stick representation. The ADAM17 prodomain is colored red, whereas the rest of ADAM17 is colored yellow. For highlighted side chains, we colored their oxygen atoms in red, nitrogen atoms in purple blue, and carbon atoms the same color as the main chain. (B) iRhom1/2 DKO HEK cells were transfected with empty vector (EV) or different iRhom2 single point mutants together with the ADAM17 substrate alkaline phosphatase (AP)-tagged amphiregulin (AREG). The growth medium was collected overnight and used for the AP-shedding assay. Substrate shedding (%), which represents ADAM17 shedding activity, was calculated by dividing the level of released alkaline phosphatase in the medium by the total alkaline phosphatase level. Error bars represent standard deviations (n = 3, three transfectants). A Dunnett’s test is performed by computing a Student’s t statistic for each transfection condition, and the statistic compares the transfection control (EV) and all iRhom2 mutants (D475R, L409W, S419W, E529R, E550R, H536A, A535W, and I386W) to the WT iRhom2 condition. ^∗∗∗∗p < 0.0001, ^∗∗p < 0.01, ^∗p < 0.05; ns, not significant. (C) Concanavalin A (ConA) enrichment was performed to the lysates from the AREG shedding assay to quantify levels in cells of the iRhom2 mutants as well as full-length, immature (ADAM17), and mature ADAM17 (mADAM17). (D–F) Cell lysates and anti-hemagglutinin (HA) immunoprecipitates were blotted for endogenous ADAM17, HA (iRhom2), and actin. In (E), EV and WT conditions are on the same immunoblot as mutant conditions, with superfluous lanes removed to make comparison easier. ^∗indicates non-specific signal. Data are representative of three independent experiments (B–F). See also Figures S3 and S4 an Table S1.

Figure 4

Mutations at D475 in iRhom2 drive unstimulated ADAM17 activation and degradation (A) HA-based immunoprecipitates and lysates were blotted for ADAM17, HA (iRhom2), and actin. (B) iRhom1/2 DKO HEK cells were transfected with different iRhom2 variants together with AP-tagged AREG or TNF as ADAM17 substrates. Medium was collected overnight and 2 μM GW280264X (GW) or GI254023X (GI) were used when indicated. Error bars represent standard deviations (n = 3, three transfectants). A Dunnett’s test is performed by computing a Student’s t statistic for each transfection condition compared with the WT iRhom2 condition. ^∗∗∗∗p < 0.0001. (C) Concanavalin A (ConA) enrichment was performed to the lysates from the AREG shedding assay. (D) Western blots of iRhom1/2 DKO HEK293 cells transfected with iRhom2 mutants together with wild-type (WT) mScarlet-tagged ADAM17. Cells were treated with bortezomib (BTZ, 1 μM), or bafilomycin A1 (BafA1, 1 μM) for 12 h before harvesting. DMSO was used as a solvent control. (E) Quantifications of the western blots from three independent experiments of (D) using ImageJ. Error bars represent standard deviations (n = 3, three independent experiments). A Dunnett’s test is performed by computing a Student’s t statistic for each transfection condition compared with D475R+DMSO condition. ^∗∗∗p < 0.0001, ^∗∗p < 0.01; ns, not significant. Data are representative of three independent experiments (A–D). See also Figure S3.

Figure 5

Activation of ADAM17 and its regulation by iRhom2 (A) Cryo-EM density (gray) and the structure of the tripartite complex (isolated ADAM17 prodomain, mature ADAM17, and iRhom2). (B) iRhom1/2 DKO HEK cells were transfected with either empty vector (EV), WT, or mutant HA-tagged iRhom2, and WT or mutant ADAM17 (indicated in red, with prodomain tagged with V5 and cytoplasmic domain tagged with mScarlet). Anti-HA immunoprecipitation (HA-IP) was performed to capture the ADAM17 prodomain in complex with iRhom2. Samples were blotted for mScarlet (ADAM17), prodomain (V5), and HA (iRhom2). A reducing sample buffer was used to separate the prodomain from the complex for analysis with western blots. (C) iRhom1/2 DKO HEK cells were transfected with WT or D475R iRhom2 together with WT or R58A ADAM17, with a V5 tag inserted before the furin cleavage site and an mScarlet tag at the C terminus. HA-based immunoprecipitates and lysates were blotted with V5 (full-length ADAM17 and its prodomain), HA (iRhom2), and actin. (D) iRhom1/2 DKO HEK cells were transfected with WT iRhom2 together with WT or R58A ADAM17 constructs. Cells were treated with 200 nM PMA for the indicated time (15 mins or 30 min) before harvesting. DMSO was used as solvent control. (E) Two-dimensional class averages of the mature ADAM17 complex and mature ADAM17/iRhom2 fusion. (F) Structural analyses of the mature ADAM17/iRhom2 fusion. To show the flexible protease domain, the density map is also lowpass filtered and contoured at a lower threshold (white). On the right, the structure of the full-length ADAM17/iRhom2 complex is docked into the density to demonstrate the movement of the protease domain. For clarity, the prodomain structure is not shown. (G) Movements of the mature ADAM17 extracellular domain. Structures of the full-length ADAM17/iRhom2 and two mature ADAM17 fusions are superimposed. Data are representative of three independent experiments (B–D). See also Figure S5.