Fig 2: Overview of BTP regulation by PCPE-1 and PCPE-2.a PCPE-1 has no effect on the BTP-dependent cleavage of non-collagenous substrates while PCPE-2 acts as a direct and potent inhibitor of these cleavages. Our results indicate that the CUB domains of PCPE-2 interact with BMP-1 auxiliary domains (CUB1 and CUB2) to allosterically inhibit the activity of the protease domain. b In the case of fibrillar procollagens, PCPE-1 enhances the C-terminal maturation of procollagens I-III through the binding of CUB1CUB2 to C-propeptides. PCPE-2 binds procollagens and BMP-1 with similar affinities but cannot interact with both partners simultaneously, thereby precluding the efficient processing of procollagens. The net effect (enhancement, no effect, inhibition) therefore depends on the relative concentrations of the three partners (procollagen, protease, PCPE-2). PCPE-1 and -2 are represented in their 2-domain (CUB1CUB2) configuration while BTPs are shown in the BMP-1 configuration (monomers with CUB1-CUB2-EGF-CUB3 auxiliary domains). Note that the incompletely-characterized case of fibrillar collagens V and XI is not represented.

Fig 3: The effect of PCPE-2 on procollagen C-terminal maturation is concentration-dependent and involves a direct interaction with procollagens.a Mini-procollagens I, II and III (430 nM) were incubated with 21 nM BMP-1 in the absence or presence of 430 nM PCPE-1 or/and PCPE-2 in a total volume of 30 µl for 1 h. Reaction products were analyzed by SDS-PAGE (4–20% polyacrylamide gel; non-reducing conditions) and Coomassie Blue staining. Enhancement factors estimated by densitometry are indicated below the gel. Gels are representative of n = 3 independent experiments. #Trimeric N-terminal cleavage product of mini-procollagen III (when visible). b Cleavage of CPIII-Long (360 nM) by BMP-1 (20 nM) for 1 h in the absence or presence of PCPE-1, PCPE-2, the CUB domains of PCPE-2, the NTR domain of PCPE-2 or combinations of PCPE-1 and PCPE-2 (or its domains). Molar ratios of PCPEs and domains to CPIII-Long are indicated above the gel and enhancement factors are reported below the gel. The gel is representative of n = 3 independent experiments. c Quantification of CPIII-Long (360 nM) processing by BMP-1 (16.5 nM) in the presence of increasing amounts of PCPE-1 and PCPE-2 (0–2.2 µM). Analysis was by SDS-PAGE (reducing conditions) on a 15% acrylamide gel with Sypro Ruby staining and quantification of individual substrate bands (means ± SD; n = 3 independent experiments). d Superposition of a model of PCPE-2 CUB domains (green) on the X-ray structure (PDB code 6FZV) of the complex between the CUB domains of PCPE-1 (orange) and C-propeptide III (blue). Calcium ions are represented as cyan spheres. e PCPE-2 binding to immobilized mini-procollagen III (435 RU). Increasing concentrations of PCPE-2 were injected (0.78–50 nM, prepared as serial two-fold dilutions). The best fit of the binding response at equilibrium (steady-state conditions) is also shown (inset). Kinetic constants were outside instrument specifications and could not be determined. f Successive injections of PCPE-2 and PCPE-1 on immobilized mini-procollagen III (435 RU) compared to successive injections of PCPE-2 alone. Source data for all graphs are provided as a Source Data file.

Fig 4: Interactions of PCPE-2 with procollagen and BMP-1 are mutually exclusive but the binding sites are only partially overlapping.a Competition experiments consisting of an injection of 200 nM PCPE-2 for 60 s followed by an injection of PCPE-2 (200 nM) alone or in combination with 50, 100 or 200 nM BMP-1 for another 60 s. The immobilized protein was C-propeptide III (351 RU). b Same as in (a) with PCPE-1 instead of PCPE-2. c Binding of WT and mutant CUB1CUB2* domains of PCPE-2 (50 nM, obtained from MBP-CUB1CUB2 fusion constructs) on immobilized mini-procollagen III (437 RU) as determined by SPR. Bar graph shows means ± range of n = 2 independent experiments performed in duplicates on two different surfaces with 437 and 651 RU of immobilized mini-procollagen III. d Binding of WT and mutant MBP-CUB1CUB2 domains of PCPE-2 (50 nM) on immobilized BMP-1 (1302 RU) as determined by SPR. Bar graph shows means ± range of n = 2 independent experiments performed in duplicates on two different surfaces with 1302 and 1492 RU of immobilized BMP-1. e Quantification of BMP-1 activity on the fluorogenic peptide in the presence of WT and mutant MBP-CUB1CUB2 domains of PCPE-2 (% of BMP-1 activity relative to control obtained in the absence of MBP-CUB1CUB2 domains); means ± SD of n = 4 independent experiments performed in duplicates except for WT (n = 6) and F87A (n = 3). Mutants were compared to WT using one-way ANOVA and Dunnett’s post-test. f Effect of WT and mutant MBP-CUB1CUB2 domains of PCPE-2 (388 nM) on the cleavage of LDLR ectodomain (388 nM) by BMP-1 (17 nM). Activities relative to the BMP-1 alone condition are indicated below the gel. The gel is representative of n = 2 independent experiments. Uncropped gel and source data for all graphs are available in Supplementary Fig. 16 or as a Source Data file.

Fig 5: The CUB domains of PCPE-2 are responsible for its inhibitory activity and the non-catalytic domains of BMP-1 are required for inhibition.a Quantification of BMP-1 activity on the fluorogenic peptide Mca-YVADAPK(Dnp)-OH in the presence of increasing concentrations of PCPE-2, PCPE-2 3C and various deletion constructs of PCPE-2 or PCPE-1. BMP-1 concentration in the experiment was 7 nM. Means ± SD (number of independent experiments run in duplicate indicated above each bar). b The IC50 of PCPE-2 was measured in the same conditions with PCPE-2 concentrations between 0.07 and 690 nM and the corresponding apparent inhibition constant (Kiapp) was calculated using the Morrison equation. Means ± SD of n = 3 independent experiments run in duplicate. c Effect of PCPE-2 (at the indicated molar ratios to protease) on the cleavage of the fluorogenic peptide by deletion mutants of BMP-1 (7 nM except for BMP-1[cat]: 8 nM). Means ± SD (number of independent experiments run in duplicate indicated above each bar). d Comparison of the binding of 50 nM of PCPE-1, PCPE-2 and its domains (CUB1CUB2, CUB2NTR, NTR and CUB1) on immobilized BMP-1 (974 RU). e Comparison of the binding of 50 nM BMP-1 and its deletion mutants on immobilized PCPE-2 (646 RU or 418 RU) after normalization with the maximum response obtained with BMP-1. f Fits of sensorgrams obtained with the kinetic (black dotted lines; heterogeneous ligand) or steady-state (inset) model when increasing concentrations of PCPE-2 (1.56-100 nM, prepared as serial two-fold dilutions) were injected over immobilized BMP-1 (974 RU). g Same as in (f) for the CUB1CUB2 domains of PCPE-2. Source data for all graphs are provided as a Source Data file.