Regulation of ectodomain shedding of ADAM33 in vitro and in vivo.

To the Editor:

A disintegrin and metalloprotease (ADAM)33 is a susceptibility gene for asthma and bronchial hyperresponsiveness (BHR). ADAM33 is a transmembrane protein, but a soluble protein containing the metalloprotease domain (sADAM33) has previously been identified in bronchoalveolar lavage fluid (BALF) from subjects with asthma and its levels significantly and negatively correlated with FEV1, suggesting a role in the development of airflow obstruction. We have shown that sADAM33 in asthmatic BALF is enzymatically active and that the sADAM33 metalloprotease causes angiogenesis and myogenesis in vivo or ex vivo.2, 3 Furthermore, in vivo allergen challenge causes shedding of enzymatically active sADAM33 into BALF of wild-type mice, and in a transgenic mouse model, lung-specific sADAM33 expression causes airway remodeling, which enhances eosinophil recruitment with associated BHR following allergen sensitization and challenge. Although TGF-β is a trigger for sADAM33 release in vitro, there is no mechanistic information or evidence of in vivo relevance. Because it is not possible to test directly the importance of TGF-β in ectodomain shedding in human asthma, we characterized the mechanism(s) of the TGF-β–induced ectodomain shedding of murine ADAM33 and determined its importance for shedding of ADAM33 in vivo. Detailed methodology is provided in the Methods section in this article's Online Repository at www.jacionline.org.

Initially, we confirmed that murine ADAM33 was similar to human ADAM33 in its sensitivity to TGF-β–induced ectodomain shedding. As expected, TGF-β treatment caused a dose-dependent increase in sADAM33 in supernatants of Cos-7 cells expressing murine ADAM33 (see Fig E1, A and B, in this article's Online Repository at www.jacionline.org). The main band had a molecular weight of around 102 kDa, indicating that the entire ectodomain had been shed; however, further processing was also evident, suggesting loss of the inhibitory prodomain. Consistent with this, there was a significant increase in ADAM33 enzymatic activity in cell-free supernatants following TGF-β treatment (Fig E1, C). Pericellular proteolysis is frequently mediated by members of the ADAM or MT-MMP families that are sensitive to the broad-spectrum hydroxamic acid–based inhibitor, GM6001; however, we found that it did not affect the shedding (see Fig E2, A and B, in this article's Online Repository at www.jacionline.org) or activity (Fig E2, C) of ADAM33. We also confirmed that GM6001 had no effect on the activity of purified recombinant ADAM33 (Fig E2, D). Because ADAM33 has a unique substrate-binding site and its catalytic activity is insensitive to GM6001, we postulated that ADAM33 ectodomain shedding was autocatalytic. Consistent with this, mutation of E347A in the catalytic site suppressed shedding of sADAM33 both at baseline and in response to TGF-β (Fig 1, A) and this was accompanied by a significant reduction in enzymatic activity in cell-free supernatants (Fig 1, B). These data suggest that a substantial component of the shedding of sADAM33 is autocatalytic.

Fig E1

TGF-β enhances shedding of murine ADAM33. Cos-7 cells transiently transfected with full-length murine ADAM33 or green fluorescent protein (GFP) as control were treated with TGF-β as indicated. Soluble ADAM33 immunoreactivity and enzyme activity in cell-free supernatants were assessed by Western blotting (A) with densitometry (arbitrary units, AU) (B) or fluorescence resonance energy transfer peptide cleavage assay, respectively; enzyme activity is expressed as percent of activity found in supernatants from control untreated cells (C). Data are mean + SD and are representative of 4 independent experiments. MW, Molecular weight. **P < .01. ***P < .001.

Fig E2

The broad-spectrum MMP inhibitor GM6001 does not inhibit ADAM33 shedding. Cos-7 cells transiently transfected with full-length murine ADAM33 were preincubated with GM6001 (0-10 μM) for 1 hour before addition of 5 ng/mL TGF-β for 8 hours. Cell-free supernatants were assessed for soluble ADAM33 by Western blotting (A) with densitometry (arbitrary units, AU) (B) and by fluorescence resonance energy transfer peptide cleavage assay (C). Purified recombinant ADAM33 protein was incubated with GM6001 and tested for activity in a fluorescence resonance energy transfer peptide cleavage assay (D). Activity is presented as a percent of that measured in the absence of GM6001. MW, Molecular weight. Data are expressed as mean + SD and are representative of 4 independent experiments.

Fig 1

Regulation of ADAM33 shedding. Cos-7 cells expressing full-length catalytically active murine ADAM33 (A-F) or inactive ADAM33 E347A (Fig 1, A and B) were treated without or with TGF-β (5 ng/mL) for 8 hours and SMAD (SB431542) (Fig 1, C and D) or MAPK (PD98059) (Fig 1, E and F) inhibitors, as indicated. Soluble ADAM33 immunoreactivity and enzymatic activity in cell-free supernatants were assessed by Western blotting (inset) with densitometry (arbitrary units, AU) (Fig 1, A, C, and E) or in a fluorescence resonance energy transfer (FRET) peptide cleavage assay, respectively; in panel Fig 1, B, enzymatic activity in the supernatants is expressed as a percent of that measured in supernatants from untreated cells expressing full-length active ADAM33, whereas in panels D and F, activity is expressed as a percent of that measured in supernatants from TGF-β–treated cells. Data are expressed as mean + SD (n = 4 independent experiments). #P < .05, ##P < .01 vs control, *P < .05, **P < .01, ***P < .001.

TGF-β activates multiple signals including mothers against decapentaplegic homolog (SMAD) and non-SMAD pathways. Inhibition of SMAD signaling using SB431452 significantly suppressed TGF-β1–induced shedding of immunoreactive and enzymatically active ADAM33 (Fig 1, C and D). In contrast, the MAP2K/mitogen-activated protein kinase kinase inhibitor PD98059 dose-dependently increased shedding of immunoreactive and enzymatically active forms of sADAM33 (Fig 1, E and F). These data suggest a complex effect of TGF-β signaling, with SMAD activation being stimulatory for ADAM33 shedding, whereas mitogen-activated protein kinase (MAPK) activation has a negative regulatory effect. The latter effect may be mediated via TIMP-3, a known inhibitor of ADAM33 enzymatic activity, which is reported to be induced by TGF-β via MAPK signaling. Thus, inhibiting MAPK activation with PD98059 may inhibit TIMP3 expression and release ADAM33 from the effects of this natural inhibitor leading to increased autocatalytic shedding and augmented enzyme activity. Although ectodomain shedding can be regulated by natural protein inhibitors, other mechanisms including membrane trafficking, protein maturation, and substrate presentation have also been shown to be important for the regulation of other sheddases. Whether release of ADAM33 is preceded by its membrane trafficking or maturation and/or by assembly into higher order complexes requiring additional protein interactions to exosites located either on the ADAM33 or adapter proteins remains to be determined.

Epithelial injury leads to release of TGF-β, and we have postulated that this is the source of the TGF-β that drives the increase in sADAM33 observed in response to allergen challenge in vivo. To test this hypothesis, we used BALF from mice in which lung epithelial Tgfb1 was conditionally deleted in bronchial epithelial club cells before intranasal administration of either 25 μg house dust mite extract or recombinant murine IL-33. After house dust mite challenge, lower levels of sADAM33 could be detected in the BALF of Tgfb1 mice compared with littermate controls (Fig 2, A) and it also contained less sADAM33 enzymatic activity (Fig 2, B). In the same way, when mice were challenged with IL-33, BALF from Tgfb1 mice had a lower level of sADAM33 immunoreactive protein (Fig 2, C) and enzymatic activity (Fig 2, D). Of note, exogenous TGF-β alone was ineffective at stimulating shedding of ADAM33 in vivo. This might be explained either by the fact that TGF-β could not pass through the intact epithelium of unchallenged mice and/or by the low dose and short half-life of the growth factor, which did not allow sufficient time for it to pass through the epithelium to reach the subepithelial mesenchymal cells where ADAM33 is expressed.

Fig 2

Epithelial TGF-β enhances shedding of ADAM33 in vivo. Epithelial (Epi)Tgfb or littermate Tgfb control mice were challenged with intranasal house dust mite (HDM) extract (A and B) or recombinant IL-33 (C and D) and, where indicated, recombinant TGF-β (Fig 2, C and D). Soluble ADAM33 immunoreactivity and enzyme activity in BALF were assessed by Western blotting (Fig 2, A and C) or FRET peptide cleavage assay, respectively; enzyme activity is expressed as relative fluorescent units per minute (RFU/min) (Fig 2, B and D). FRET, Fluorescence Resonance Energy Transfer. Data are mean + SD (n = 4-6 mice per group). Data are representative of 2 independent experiments. ##P < .01, ###P < .001 vs control, *P < .05, ***P < .001.

Epithelial-derived IL-33 induces rapid release of TGF-β into the airways to enhance migration of innate lymphoid cells (ILCs) and development of a robust ILC2 response that initiates allergic immunity. Because we found that epithelial-derived TGF-β is also required for ectodomain shedding of ADAM33, it may be speculated that sADAM33 can contribute to the recruitment or activation of ILCs or other innate immune cells, either by affecting matrix turnover or by promoting growth factor or chemokine shedding. Of note, polymorphisms in TGFB, IL33, and ADAM33 genes have each been associated with asthma susceptibility, yet each has a small overall effect on disease development. The involvement of 3 susceptibility gene products in epithelial responses to allergens highlights how they may cooperate to amplify the downstream asthmatic responses.

Identification of the involvement of TGF-β in ectodomain shedding of ADAM33 in an in vivo model strengthens the case for exploring how human polymorphic variation in the ADAM33 gene is linked to asthma pathogenesis. Four single nucleotide polymorphisms (S1, S2, T1, and T2) encode amino acid substitutions in the transmembrane and cytoplasmic domain of ADAM33 and have been associated with asthma. Although the intracellular domain of ADAM33 is relatively short, it is very rich in prolines, having a putative SH3 binding site where the T2 SNP is located, a casein kinase I/II phosphorylation site, and an MAPK consensus sequence that is likely to be important for regulation of ADAM33 function, especially as we have identified a negative regulatory role for MAPK in our current studies. Further work is required to determine whether this effect is direct and involves ADAM33 phosphorylation or indirect via inhibitors such as TIMP3. Alternatively, one mutation Ala395Val is located within the catalytic domain, which may directly affect catalytic activity.

In summary, we have provided direct evidence that epithelial-derived TGF-β is an important regulator of ectodomain shedding of enzymatically active ADAM33 from the mesenchyme. This process appears largely to be autocatalytic and involves SMAD signaling, but is negatively regulated by MAPK signaling. These findings highlight the importance of epithelial-mesenchymal cross-talk in asthma pathogenesis and underscore the potential for co-operation between different asthma susceptibility genes to drive disease pathogenesis.