|Prof. Dr. Joachim Grötzinger|
|Structure and Function of Proteins|
|Why control ADAM17 activity?|
To avoid malformation and disease, tissue development and homeostasis have to be precisely coordinated in time and space. This requires networks of tightly regulated signaling and interaction events, which are formed by exact interactions of proteins. To ensure proper coordination, proteins, functioning as messengers, receptors, agonists, antagonists, or adhesion molecules, have to be expressed and released at specific time-points. The transmembrane metalloprotease A Disintegrin and Metalloprotease 17 (ADAM17) ensures the timely release of cell surface molecules by a rapid, irreversible proteolytic switch, called shedding. This process generates two of the most important initiators of immune responses, namely, TNFα [1,2] and soluble (s)IL-6R , from their membrane-bound counterparts. Notably, prior to shedding, proTNF-α and IL-6R conduct regenerative processes [4-6], highlighting the drastic changes that ADAM17 activity elicits.
Figure 1: ADAM17 activity has to be tightly controlled to allow physiological processes and to prevent pathological situations. ADAM17 sheds more
than 70 different substrates ; therefore, the enzyme acts as a major sheddase in various physiological and
protective processes, such as immune defence, coagulation, and regeneration. By contrast, the uncontrolled release of such potent substrates is associated with
numerous pathological conditions, such as uncontrolled inflammation (autoimmune diseases and chronic inflammation) and cancer progression. Hence, ADAM17 is initially
kept in an inactive state, e.g., by interaction with inhibitory proteins such as β1-integrin [7-10]. Only upon
activation by various stimuli, such as LPS, ATP, and thrombin, does ADAM17 shed its substrates from the cell surface, thereby changing their activities.
Later, the enzyme is switched off. This switching off is catalysed by protein disulfide isomerases (PDIs) and is accompanied by a structural change within
the extracellular part of ADAM17, which prevents substrate recognition and shedding.
|Another prominent class of ADAM17 substrates are growth factors of the EGFR ligand family .
These substrates are expressed as membrane-bound preforms; therefore, they are not active prior to shedding. The significance of this process in development is
displayed by the perinatality of ADAM17-deficient mice, which show comparable phenotypes to EGFR ligand-deficient mice, including malformations of the skin, hair,
eye, heart, and lung .
In adult organisms, generation of soluble EGFR ligands is essential for proper regeneration processes .
These few examples (and ADAM17 processes more than 70 substrates) illustrate the necessity of this enzyme for appropriate physiological processes such as immune defence and regeneration. However, they also indicate that uncontrolled ADAM17 activity supports misguided inflammation, such as occurs in autoimmune diseases (e.g., rheumatoid arthritis), chronic inflammation (e.g., Crohn's disease), and cancer progression, due to mistimed and enhanced release of pro-inflammatory mediators and growth factors. Hence, ADAM17 has to be tightly controlled to avoid pathophysiological situations and, accordingly, it is of extraordinarily high therapeutic interest. A prerequisite to modulate ADAM17 activity is an in-depth knowledge of the molecular basis of its functioning, including its on and off switches, because ADAM17 has to be activated to become proteolytically active [13, 14] and inactivated to prevent damage. The cytoplasmic tail of ADAM17 is dispensable for its activation by numerous stimuli [15-17]; therefore, we focus on its extracellular part.
Figure 2: Schematic drawing of atypical and typical ADAMs.
(A) ADAM17 and its closest relative ADAM10 are atypical members of the ADAM family because their extracellular regions comprise a pro-domain, a catalytic domain, a disintegrin domain, and a membrane-proximal domain (MPD), as well as a small stalk region [18-20].
(B) By contrast, typical members of the ADAM family comprise a cysteine-rich domain and an EGF-like domain instead of the MPD . According to the type-I transmembrane protein topology of all ADAM proteases, the extracellular part ends in a single transmembrane region and a cytoplasmic tail .
|The extracellular part of ADAM17|
|In its mature form, ADAM17 lacks its pro-domain, which retains the zymogen in an inactive state and is removed during maturation in the Golgi apparatus [21,22]. As the names suggest, the catalytic domain cleaves the substrates and the disintegrin domain interacts with β1-integrin. This interaction (Figure 3) supports cell-cell contacts between fibroblasts and tumour cells  and keeps the enzyme inactive, most likely due to steric hindrance of substrate recognition [7-10]. Activation of β1-integrin leads to the release of ADAM17 accompanied by the catalytically active enzyme [8,9].|
Figure 3: (A) Inactive ADAM17 interacts with β1-integrin.
(B) This interaction supports cell-cell adhesion between lung carcinoma cells (NCI-H292) and murine embryonic fibroblasts (MEFs), since the presence of soluble disintegrin domain reduces adhesion between these cells .
(C) Upon stimulation, the ADAM17-β1-integrin complex dissociates and both molecules become active.
(D) If this precise interaction is indeed mediated by the disintegrin domain, addition of soluble recombinant disintegrin domain should increase the shedding activity of ADAM17 due to competitive binding of soluble disintegrin domain to β1-integrin.
(E) Soluble disintegrin domain increases constitutive background shedding of amphiregulin from the cell surface of a breast cancer cell line (MDA-MB-231). These results confirm that the inhibitory interaction between β1-integrin and ADAM17 is mediated by this specific domain, but that dissociation of the ADAM17-β1-integrin complex is not the only requirement for ADAM17 activation .
|The MPD is supposed to play a key role in the regulation of this enzyme [18, 19, 23-26, 17, 27-30]. In line with this, the MPD controls and coordinates multimerisation  and substrate recognition  (Figure 4). Notably, ADAM17 recognises its substrates in distinct ways, depending on their membrane topology. An ADAM17 deletion mutant, which solely comprises the MPD and lacks the catalytic and disintegrin domains and the transmembrane and cytoplasmic regions, is sufficient to bind the IL-6R and IL-1RII substrates, both of which have a type-I protein topology, in contrast to the type-II transmembrane protein TNFα (Figure 4B–D). In accordance, replacement of the MPD of ADAM17 with the EGF-like domain of ADAM22 generates a shedding-incompetent chimera (Figure 4I).|
(A) Full-length HA-tagged ADAM17 can recognise, bind, and precipitate its substrate, myc-tagged IL-1RII .
(B) A GPI-anchored construct containing the MPD and stalk region of ADAM17, termed MPD17-GPI, tagged with PC can precipitate the type-I proteins IL-1RII (C) and IL-6R (D), but not the type-II protein TNFα. These data indicate that there are marked differences in the recognition mode of at least type-I and type-II class substrates .
(E) MPD17-GPI interacts with ADAM17 substrates; therefore, it functions as a competitive inhibitor of the full-length molecule. To examine this, HEK293 cells, which endogenously express ADAM17, were transfected with control or MPD17-GPI-encoding vectors. To obtain proper shedding, cells were stimulated with PMA, DMSO as a control, or PMA plus a metalloprotease inhibitor (either GM6001 or marimastat). As indicated by the red bar, MPD17-GPI significantly decreases the shedding of all tested substrates. The shedding activity of cells transfected with control DNA was set at 100%.
(F) An interaction between MPD17-GPI and TNFα could not be detected by coimmunoprecipitation experiments; therefore, this inhibitory effect might be due to dimerisation of ADAM17, thereby generating inactive ADAM17-MPD17-GPI complexes. Hence, we tested whether ADAM17 forms multimers by cotransfecting and coimmunoprecipitating PC- and HA-tagged full-length molecules, and it was found that ADAM17 indeed forms multimers or at least dimers.
(G) Next, we tested in the same experimental setting whether the MPD with the stalk region (MPD17-GPI) is able to do so, and in fact the MPD17-GPI alone is able to dimerise.
(H) To prove that the MPD is needed for shedding, a chimera was generated in which the MPD of ADAM17 was replaced by the EGF-like domain of ADAM22 (ADAM17EGF22).
(I) The enzymatic activity of this chimera was tested in comparison to that of wild-type ADAM17. The constructs were transfected together with IL-1RII, as the ADAM17 substrate, into ADAM17ex/ex MEFs. In these MEFs, ADAM17 expression is reduced by 95%. After 1 day, shedding activity was tested as in HEK293 cells (E). The resulting data indicate that ADAM17 in which the MPD is replaced completely lacks shedding competency because it shows comparable results to eGFP-transfected control cells.
|The specific thiol switch within ADAM17|
|Due to the impact of the MPD in ADAM17 regulation and activity, we were interested in what it actually looks like. Hence, we produced recombinant MPD using bacterial expression systems for structural determination by multidimensional heteronuclear NMR spectroscopy. Soluble expressed MPD is a disulfide isomer that is only partially structured and is a flexible elongated domain (Figure 5C). The MPD comprises a classical thioredoxin motif (C600KVC), which is essential for ADAM17 shedding activity , and exogenous PDI supresses ADAM17 activity [18,30]. Therefore, we hypothesise that PDI targets this specific motif. Indeed, PDI catalyses disulfide isomerisation within the flexible open MPD . Two disulfide bonds thereby undergo specific isomerisation. This process changes the affected disulfide pattern from a sequential arrangement (C600-C630 and C635-C641), which is present in the elongated MPD structure, to an overlapping pattern (C600-635 and C630-641), which exists in a compact, rigid-structured MPD. Whereas the latter isoform is associated with the inactive enzyme, the flexible form is associated with active ADAM17 . Hence, the MPD acts as the off-switch of ADAM17, which is operated by exogenous PDI .|
(A) The MPD of ADAM17 exists in two conformers, which differ in terms of their disulfide bond pattern and can be separated by reverse phase HPLC. Intriguingly, the open MPD can be converted into the closed MPD by reduced PDI .
(B) The HSQC spectrum of the open conformer shows fewer signals than the closed conformer. These signals represent amide groups of the protein in a defined environment; therefore, these data indicate that the open MPD is less structured than the closed MPD. Intriguingly, signals present in the HSQC of the flexible open MPD appear at the same position as those of the closed isomer, indicating that these parts have identical structures.
(C/D) The N-terminal regions of both isomers show identical structures .
(C) By contrast, the C-terminal part is flexible in the open MPD, but (D) well-structured in the closed MPD.
(E) The differences in the MPD structures are due to different disulfide bond connections. The open flexible MPD contains a sequential arrangement of the two disulphide bond connections (C600-C630 and C635-C640). PDIs catalyse the isomerisation of these specific disulphide bonds into an overlaying arrangement (C600-C635 and C630-C640) .
(F) The closed MPD is linked to inactivated ADAM17. Overexpression of PDIA6 reduces ADAM17 activity after PMA stimulation compared to that in mock-transfected control cells.
(G) PDI abrogates the substrate recognition of ADAM17. Treatment of an inactive ADAM17 variant (HE-ADAM17) with PDI abolishes binding of ADAM17, as shown by Western blotting.
|Model of ADAM17 inactivation by PDI|
Our current knowledge allows us to propose the following model of ADAM17 activity regarding type-I substrates, which can be possibly not fulfil by type-II
substrates due to the contrariwise orientation of the protein backbone (Figure 6).
Figure 6: Current model of ADAM17 substrate recognition and inactivation.
|Back to Top|