Reactive-site mutants of N-TIMP-3 that selectively inhibit ADAMTS-4 and ADAMTS-5: biological and structural implications

Abstract

We have reported previously that reactive-site mutants of N-TIMP-3 [N-terminal inhibitory domain of TIMP-3 (tissue inhibitor of metalloproteinases 3)] modified at the N-terminus, selectively inhibited ADAM17 (a disintegrin and metalloproteinase 17) over the MMPs (matrix metalloproteinases). The primary aggrecanases ADAMTS (ADAM with thrombospondin motifs) -4 and -5 are ADAM17-related metalloproteinases which are similarly inhibited by TIMP-3, but are poorly inhibited by other TIMPs. Using a newly developed recombinant protein substrate based on the IGD (interglobular domain) of aggrecan, gst-IGD-flag, these reactive-site mutants were found to similarly inhibit ADAMTS-4 and ADAMTS-5. Further mutations of N-TIMP-3 indicated that up to two extra alanine residues can be attached to the N-terminus before the Ki (app) for ADAMTS-4 and ADAMTS-5 increased to over 100 nM. No other residues tested at the [-1] position produced inhibitors as potent as the alanine mutant. The mutants N-TIMP-3(T2G), [-1A]N-TIMP-3 and [-2A]N-TIMP-3 were effective inhibitors of aggrecan degradation, but not of collagen degradation in both IL-1α (interleukin-1α)-stimulated porcine articular cartilage explants and IL-1α with oncostatin M-stimulated human cartilage explants. Molecular modelling studies indicated that the [-1A]N-TIMP-3 mutant has additional stabilizing interactions with the catalytic domains of ADAM17, ADAMTS-4 and ADAMTS-5 that are absent from complexes with MMPs. These observations suggest that further mutation of the residues of N-TIMP-3 which make unique contacts with these metalloproteinases may allow discrimination between them.

Figure 1. Characterization of the gst-IGD-flag substrate

(A) Schematic representation of the gst-IGD-flag substrate and its products upon aggrecanase activity. The observed molecular masses are indicated in parentheses. (B) Time course of cleavage of gst-IGD-flag by ADAMTS-5-4. gst-IGD-flag (17 μM) was incubated with recombinant ADAMTS-5-4 (500 pM) at 37 °C. At different time points (0–32 h), reactions were stopped and analysed by SDS/PAGE (10% acrylamide). The 17 kDa product band (arrow) also stained positive for the aggrecanase generated neo-epitope ARGSV. (C) The product was analysed by densitometry to calculate the extent of cleavage, expressed as pixel volumes. (D and E) Cleavage of gst-IGD-flag by different amounts of enzyme ([E]). gst-IGD-flag (16 μM) substrate was incubated for 16 h at 37 °C with increasing amounts of ADAMTS-5-4 (0–2 nM). M, molecular mass markers (size in kDa).

Figure 2. The K_i (app) of N-TIMP-3 mutants for ADAMTS-4-2 and ADAMTS-5-4

N-TIMP-3 or the mutants (0–1 μM) were incubated with ADAMTS-4-2 [1 nM (A and C)] or ADAMTS-5-4 [500 pM, (B and D)] for 1 h at 37 °C, before the addition of gst-IGD-flag (17 μM). After 16 h at 37 °C, SDS/PAGE and densitometric analysis of the 17 kDa product band was used to determine the percentage of residual activity. Data were fitted to the tight binding equation to obtain the K_i (app).

Figure 3. Dose-dependent inhibition of GAG release upon IL-1α stimulation in porcine articular cartilage by N-TIMP-3 reactive-site mutants

Porcine articular cartilage explants were stimulated with IL-1α (10 ng/ml) for 3 days with the various TIMPs at the concentrations indicated (0.01, 0.1 and 1 μM). GAG released in the medium was measured by DMMB. n=3 for each treatment.

Figure 4. Effects of the N-TIMP-3 mutants on aggrecan and collagen degradation in porcine articular cartilage under long-term stimulation with IL-1α

Porcine articular cartilage explants were stimulated with IL-1α (10 ng/ml) with the various TIMPs at 0.1 μM for 38 days. The amount of GAG and hydroxyproline (Hypro) released into the medium was measured using the DMMB and DMBA assay respectively. The cumulative amounts of GAG (A) and hydroxyproline (B) loss against time were plotted. n=3 for each treatment.

Figure 5. Inhibition of GAG release but not hydroxyproline release of explanted human articular cartilage stimulated with IL-1α

Cartilage from the knee of an OA patient was stimulated with IL-1α (10 ng/ml)+OSM (50 ng/ml) and the various TIMPs (0.01, 0.1 and 1 μM concentrations) for 3 days. GAG (A) and hydroxyproline (Hypro) (B) released into the medium were taken as measures of cartilage degradation. n=3 for each treatment. Cartilage samples from the hips of five OA patients (the previous and four additional) were treated with IL-1α (10 ng/ml) and the various TIMPs at a concentration of 100 nM. Data from the different patients were normalized by taking the amount of (C) GAG released upon stimulation as 100 and (D) hydroxyproline (Hypro) released without stimulation as 100, shown as means±S.D., n=3 for each patient; *P<0.03.

Figure 6. In silico models of N-TIMP-3 and [−1A]N-TIMP-3 with ADAM17, MMP-1, ADAMTS-4 and ADAMTS-5

Computer-simulated lowest free energy ranked models of complexes of N-TIMP-3 (salmon ribbon) or [−1A]N-TIMP-3 (blue ribbon) with MMP-1 (A), ADAM17 (B), ADAMTS-4 (C) or ADAMTS-5 (D) are shown. The catalytic Zn²⁺ ion is shown in magenta, water-accessible enzyme surfaces are grey and regions which interact only with N-TIMP-3, [−1A]N-TIMP-3 or with both are in salmon, blue and green respectively. Phe³⁴ and the N-terminal extra alanine residues are shown with sticks. In the [−1A]N-TIMP-3 complexes with MMP-1 and ADAM17, the TIMP molecule is tilted and the interaction of Phe³⁴ of the inhibitor with the enzymes are lost. Both N-TIMP-3 and [−1A]N-TIMP-3 are bound to ADAMTS-4 and ADAMTS-5 in a similar tilted manner. Residues which make the deeper active-site clefts of ADAM17 (Met³⁴⁵), ADAMTS-4 (G³²²VST³²⁵) and ADAMTS-5 (G³⁷²HHS³⁷⁵) that interact with the TIMP and are not present in MMP-1 are highlighted in red.