Characterization of a core binding site for ADAMTS-13 in the A2 domain of von Willebrand factor

Abstract

ADAMTS-13, a metalloprotease in plasma, specifically cleaves the Tyr-1605–Met-1606 bond in the A2 domain of von Willebrand factor (VWF) to regulate the polymer distribution of VWF in circulation, which is critical for primary hemostasis. A 73-aa peptide (VWF73) was previously identified as the minimal substrate cleavable by ADAMTS-13. In this study, VWF73 was enzymatically and chemically cleaved into shorter peptides, and the inhibition of cleavage of a VWF73-derived substrate by these purified peptides was measured in competition studies using a quantitative assay we recently reported. A 24-aa peptide encompassing Pro-1645–Lys-1668 (P′40–P′63) and situated 40 aa downstream from the cleavage site was the minimal peptide that could bind to and competitively inhibit ADAMTS-13 (K_i = 12 μM). This peptide and longer peptides encompassing this core sequence also inhibited the cleavage of multimeric VWF by ADAMTS-13. These results suggest the presence of a complementary extended binding site, or exosite, on ADAMTS-13. Mutation of Asp-1653 and Asp-1663 to Ala in this region significantly reduced the rate of cleavage of the substrate peptide, whereas the Glu1655Ala mutation caused an enhanced rate of cleavage. These results suggest that ionic interactions of the Pro-1645–Lys-1668 region with the exosite on ADAMTS-13 play a significant role in mediating substrate recognition.

The polymeric adhesive protein von Willebrand factor (VWF) plays a critical role in primary hemostasis (1). Its hemostatic function depends on the unique composition of multimers: large multimeric forms are crucial for shear-dependent platelet aggregation, and small multimers, such as dimers and tetramers, are unable to support platelet aggregation. The multimer distribution of VWF in plasma is regulated in part by ADAMTS-13, a metalloprotease (2–4) that cleaves the ultra-large form of VWF (ULVWF) (5) secreted by activated endothelial cells and platelets into the less adhesive form consisting of small multimers (6). A deficiency in ADAMTS-13 leads to inadequate processing and accumulation of the highly adhesive ULVWF in circulation (7, 8). ULVWF multimers unfold under high shear in arterioles and capillaries and bind to platelets to form pathological microvascular thrombi. Deposition of these pathological thrombi in the microvasculature is a characteristic feature of the disease thrombotic thrombocytopenic purpura (TTP). Familial TTP is caused by inherited mutations in the ADAMTS13 gene (9). A majority of patients with acute idiopathic TTP was found to contain circulating inhibitory autoantibodies to ADAMTS-13 (7, 8). Assessment of the ADAMTS-13 level in plasma and the level of inhibitory antibodies has since become an important tool in the diagnosis of TTP and in monitoring the disease process.

The ADAMTS-13 level in plasma was initially measured by directly assessing the loss of large VWF multimers (3) or the increase in abundance of cleaved fragments (4). Other indirect measurements, such as residual collagen binding (10) and residual ristocetin cofactor activity (11), also have been developed. These assays, involving lengthy procedures and the use of nonphysiological denaturants, are difficult to perform and often have large interassay variations. Kokame et al. (12) showed that a 73-aa peptide from the A2 domain of VWF (VWF73) was cleaved efficiently by ADAMTS-13. Cleavage of this peptide was more efficient than the entire A2 domain of VWF. It was hypothesized that VWF73 assumes a conformation comparable with that of the shear-induced accessible conformation of the A2 domain in VWF multimers. It was unusual that an extended peptide sequence, consisting of 10 aa before and 63 aa after the scissile bond, is necessary for recognition and cleavage by ADAMTS-13. To understand how ADAMTS-13 interacts with this extended peptide substrate, we further fragmented VWF73 into shorter peptides and identified regions that contribute to substrate recognition.

Results

A derivative of VWF73, consisting of a HRP conjugate of a biotinylated VWF78 sequence, previously designated as HRP-H-A2-B (13), was used as substrate for characterizing the activity of ADAMTS-13 in the presence of peptides derived from various regions of the A2 domain of VWF. In the absence of inhibitory peptides, HRP-H-A2-B was cleaved efficiently, with a K_m of 0.25 ± 0.03 μM, k_cat of 0.67 s⁻¹, and a catalytic efficiency (k_cat/K_m) of 3.05 × 10⁶ M⁻¹·s⁻¹ as determined by nonlinear least square curve fitting. The recombinant A2 domain of VWF (Val-1476–Gly-1672) was digested with cyanogen bromide to generate a mixture of peptides. This mixture exhibited strong inhibitory activity on the cleavage of HRP-H-A2-B by ADAMTS-13 (data not shown), indicating that treatment with cyanogen bromide did not destroy sequences that could bind to ADAMTS-13. Cyanogen bromide-derived peptides of A2 were separated by HPLC, and the separated peptides were individually tested for inhibitory activity. Only one tryptophan-containing peptide, encompassing Val-1607–Gly-1672, inhibited cleavage of HRP-H-A2-B (data not shown). Because this peptide is contained in VWF73, we carried out subsequent quantitative studies with peptides derived from H-A2-B.

H-A2-B was digested with cyanogen bromide, endopeptidase Lys-C, 2-(2-nitrophenylsulfenyl)-3-methyl-3-bromoindolenine (BNPS-skatole), and trypsin, and the resulting peptides were purified by HPLC (Fig. 1). The purity and identity of the peptides were confirmed by mass spectrometry. Cleavage of H-A2-B after the single Met-1606 by cyanogen bromide generated two peptides, M3 and M6. M6, which is 1 aa shorter than the product derived from cleavage of H-A2-B at the Tyr-1605–Met-1606 bond by ADAMTS-13, competitively inhibited the cleavage of HRP-H-A2-B (Fig. 2), with an inhibition constant (K_i) of 0.25 ± 0.04 μM as determined by nonlinear least-square curve fitting (Table 1). M3, the sequence from the P10–P′1 positions, did not show any inhibitory activity at concentrations up to 45 μM.

Fig. 1.

Schematic representation of peptides derived from H-A2-B. Peptides are represented by the bars. Amino acids represented by their single-letter codes at the beginning and end of each peptide were numbered according to the sequence in VWF. Peptides with inhibitory activities are shown in gray.

Fig. 2.

Inhibition of HRP-H-A2-B cleavage by M6. (A) Rate of substrate cleavage in the absence (●) and presence (■) of 0.6 μM M6. (B) Lineweaver–Burk plot of data shown in A. ●, absence of M6; ■, presence of M6.

Table 1.

Inhibition constants of peptides from Leu-1591 to Lys-1668 of VWF

Peptide	Position	Ki, μM
M3	P15–P′1	(>45)
M6	P′2–P′63	0.25 ± 0.04
K1	P′2–P′12	(>46)
K5	P′13–P′63	1.18 ± 0.52
K3	Biotin acceptor	—
W1	P′13–P′39	(>238)
W3	P′40–P′63	12 ± 2.5
W7	P′2–P′39	(>31)
R2	P′13–P′35	(>78)
R4	P′36–P′53	(>78)
R13	P′54–P′63	(>78)

The peptides shown correspond to those in Fig. 1, and the peptide positions are relative to the scissile P1–P′1 Tyr-1605–Met-1606 bond. Concentrations in parentheses represent the highest concentrations tested that showed no inhibition.

M6 was further cleaved by endopeptidase Lys-C into peptides K1, K3, and K5, and only K5 competitively inhibited ADAMTS-13 (K_i = 1.18 ± 0.52 μM) (Table 1). Because K3, the C-terminal biotinylated tag sequence, did not show any inhibition, this region was not responsible for the inhibitory activity observed with M6. Equimolar mixtures of K1 and K5 did not restore the low K_i of M6 (Fig. 3A), indicating that K1 must be covalently linked to K5 to show high affinity for ADAMTS-13. These results showed that the removal of K1 sequences in the P′2–P′12 positions is associated with a 5-fold decrease in affinity.

Fig. 3.

Inhibition of substrate cleavage by peptides. (A) Concentration-dependent inhibition of K5 (□) and equimolar mixtures of K1 and K5 (○) at a substrate concentration of 0.07 μM. (B) Concentration-dependent inhibition of W3 (□) and equimolar mixtures of W1 and W3 (○) at a substrate concentration of 0.07 μM.

K5 was further cleaved at Trp-1644 by BNPS-skatole, and the resulting peptides W1 and W3 were purified and studied. Only W3 competitively inhibited ADAMTS-13 (K_i = 12 ± 2.5 μM) (Table 1). Neither W1 nor W7, derived from BNPS-skatole cleavage of M6, inhibited ADAMTS-13. In mixing studies, the extent of inhibition by an equimolar mixture of W1 and W3 (Fig. 3B) did not differ significantly from that of W3 alone, indicating that W1 can only enhance the binding of W3 to ADAMTS-13 when it was covalently linked to W3. Removal of W1, corresponding to positions P′13–P′39, resulted in another 10-fold decrease in affinity for ADAMTS-13.

K5 was cleaved by trypsin into R1, R2, and R4, and none of these tryptic peptides exhibited appreciable inhibition of ADAMTS-13 (Table 1). An equimolar mixture of R1 and R4 also failed to reconstitute competitive inhibition comparable with that by W3. These studies showed that the smallest peptide identified in this study that could independently bind to ADAMTS-13 was W3, which is situated at positions P′40–P′63 from the cleavage site. The presence of sequences from P′2–P′39 positions increased the affinity for ADAMTS-13. However, sequences from this region by themselves, such as the isolated peptides W7, K1, and W1, were unable to bind to ADAMTS-13 with high affinity unless they were linked to W3.

The three inhibitory peptides, M6, K5, and W3, also inhibited the cleavage of multimeric VWF by ADAMTS-13 (Fig. 4). Increasing concentrations of these peptides progressively inhibited the conversion of large VWF multimers to dimers and trimers, consistent with the notion that the binding site on ADAMTS-13 for these inhibitory peptides is involved in the recognition of the multimeric VWF substrate. The concentration of W3 that showed significant inhibition was higher than those of M6 and K5, consistent with the higher K_i for this peptide determined in the inhibition studies.

Fig. 4.

Cleavage of multimeric VWF by ADAMTS-13 in the presence of inhibitory peptides. (A) Inhibition by M6. Lane 1, VWF without digestion; lane 2, VWF digested with ADAMTS-13 in plasma; lane 3, digestion with 2.1 μM M6; lane 4, 4.2 μM M6; lane 5, 8.4 μM M6; lane 6, 12.6 μM M6. (B) Inhibition by K5. Lanes 1 and 2, same as in A; lane 3, 2.1 μM K5; lane 4, 4.2 μM K5; lane 5, 11 μM K5; lane 6, 21 μM K5. (C) Inhibition by W3. Lanes 1 and 2, same as in A; lane 3, 11 μM W3; lane 4, 22 μM W3; lane 5, 43 μM W3; lane 6, 65 μM W3.

Cleavage of multimeric VWF and VWF73 by ADAMTS-13 was sensitive to high ionic strength, particularly chloride anions (14). We reasoned that ionic interactions may play a role in the interaction of ADAMTS-13 with its substrate, including the core sequence W3. The contribution of the six charged amino acids in the W3 region, Asp-1653, Glu-1655, Arg-1659, Glu-1660, Asp-1663, and Lys-1668, was assessed individually in H-A2-B by alanine mutation analysis. In these studies, the rate of cleavage by partially purified ADAMTS-13 was quantitated by SDS/PAGE, followed by staining with Sypro orange, and fluorescence intensity quantitation by fluorescence imaging. Although Arg1659Ala, Glu1660Ala, and Lys1668Ala mutations in H-A2-B did not significantly change the cleavage rate by ADAMTS-13, Asp1653Ala and Asp1663Ala mutations significantly reduced the rate of cleavage by 60% and 42%, respectively (Fig. 5). Interestingly, the Glu1655Ala mutation enhanced the rate of cleavage by 24%. These results are in agreement with the idea that ionic interactions involving residues 1653, 1655, and 1663 play a role in substrate recognition in the P′40–P′63 region.

Fig. 5.

Time-dependent cleavage of H-A2-B mutants. Shown are data for wild type (wt, ●), D1653A (■), D1663A (▴), E1655A (♦), R1659A (◊), E1660A (□), and K1668A (○).

Discussion

Studies on the initial rates of cleavage of HRP-H-A2-B showed that the reaction followed typical Michaelis–Menten kinetics with a K_m of 0.25 μM. This K_m value is approximately 6-fold lower than that of VWF76 (K_m = 1.6 μM) (15), and 13-fold lower than that of the fluorescent substrate FRETS-VWF73 (K_m = 3.2 μM) (16). The difference with the VWF76 substrate may be attributed to the dissimilarity in assay conditions. In this regard, although measurements in this study were carried out in reactions containing 5 mM Hepes, studies on VWF76 were carried out in the presence of 150 mM NaCl, which was known to inhibit the activity of ADAMTS-13 in vitro (14). The difference with the FRETS-VWF73 substrate may be due to the substitution of amino acids in the P7 and P′5 positions for the introduction of a fluorescence-emitting group and a corresponding quenching group to promote fluorescence energy transfer in FRETS-VWF73. Another possibility is that covalent attachment of HRP to H-A2-B at a unique site N-terminal to the VWF73 sequence may have oriented VWF73 in a favorable conformation for cleavage. The low K_m exhibited by the HRP-conjugated substrate HRP-H-A2-B enabled us to carry out cleavage rate measurements with a smaller amount of ADAMTS-13, at room temperature, and in brief incubations.

Initial studies on peptides derived from cyanogen bromide digestion of the entire A2 domain of VWF showed that sequences outside the VWF73 region had no effect on the activity of ADAMTS-13. This is consistent with previous truncation studies (12) showing that removal of these sequences promotes exposure and cleavage of the VWF73 region. Peptides outsides the VWF73 region also did not enhance cleavage of VWF73-derived substrate, ruling out the possibility that these sequences would bind to sites in ADAMTS-13 that allosterically regulate proteolytic activity.

Peptide W7 showed no inhibition at concentrations up to 31 μM. We were unable to assess the properties of this peptide at higher concentrations due to its limited recovery from the BNPS-skatole digest of M6. It is possible that this peptide would exhibit inhibition at higher concentrations. Nevertheless, the K_i would be significantly higher than those of the three peptides M6, K5, and W3, encompassing the core binding region.

Of all of the peptides studied, M6, which is 1 aa shorter than the product of proteolysis, exhibited the strongest inhibition. The K_i of 0.25 μM for M6 indicated that it competed favorably with the substrate (K_m = 0.25 μM) under the conditions used in this study. This strong inhibition by M6 also suggests that product inhibition might be a key feature in the regulation of ADAMTS-13 activity in vivo. Consistent with strong product inhibition, cleavage of multimeric VWF by ADAMTS-13, promoted by the presence of denaturants, rarely approach completion despite prolonged incubation. Together with shear-induced exposure of the cleavage site, product inhibition may regulate the extent of proteolysis and the multimer distribution of VWF in circulation.

The progressive increase in K_i that correlates with deletion of sequences in the P′2–P′39 region confirms that, although sequences in this region did not bind to ADAMTS-13 by themselves, they improve affinity of the core W3 sequence in the P′40–P′63 region when they were linked. These results are consistent with the proposal that substrate interaction initiates with binding of the core sequence in the P′40–P′63 region, which aligns sequences in the contiguous P′2–P′39 region for additional interaction, and positions the scissile bond to the active center of the enzyme. Also consistent with the notion that the entire P′2–P′63 region interacts with ADAMTS-13, mutations in P′33 and P′35 led to an approximate 2-fold decrease in the catalytic efficiency, and mutations in the P′9, P′10, and P′12 positions led to a 3- to 6-fold reduction in the catalytic efficiency (15).

The requirement of an extended substrate sequence and evidence of interaction along its length with ADAMTS-13 suggests that there is a complementary extended binding site, or exosite, on ADAMTS-13 adjacent to or contiguous with the catalytic center. The location of this exosite on ADAMTS-13 is not known. It has been shown that autoimmune antibodies from acquired TTP patients interact with a common epitope(s) located in the Cys-rich and spacer domain of ADAMTS-13. Deletion of the Cys-rich and spacer domains in recombinant ADAMTS-13 also resulted in loss of proteolytic activity toward multimeric VWF and peptide substrates, such as VWF73. These observations suggest that the exosite in ADAMTS-13 that plays a role in substrate recognition may involve sequences in the Cys-rich and spacer domain of ADAMTS-13. Specific inhibition of ADAMTS-13 by targeting the catalytic center and this putative exosite may be one way of reducing the excessive degradation of VWF in patients with type 2A von Willebrand disease, in whom mutations predispose the VWF molecules to excessive degradation by ADAMTS-13 (reviewed in ref. 17).

Materials and Methods

Peptide Fragmentation.

Peptides (1–5 mg) were cleaved with 2% cyanogen bromide in 10% formic acid and 6 M guanidine hydrochloride for 18 h as described (18). Digestion of peptides with endoprotease Lys-C and trypsin (Roche Diagnostics, Indianapolis, IN) was performed in the presence of 0.1 M ammonium bicarbonate with enzyme to peptide weight ratios of 200:1 and 100:1, respectively (19). Cleavage of peptides with BNPS-skatole was performed as described (20).

Peptide Isolation and Quantitation.

Peptides were purified by reverse-phase HPLC on a C-18 column (μBondapak; Waters, Milford, MA) connected to a Waters HPLC system. Peptides, eluted with a 0–80% acetonitrile gradient (21), were detected by absorbance at 214 or 280 nm. The peptides (1 μg) were analyzed by liquid chromatography-electrospray ionization mass spectrometry at the Mass Spectrometry Center of the University of Washington. The concentrations of peptides were determined by amino acid analysis using fluorescamine (22) on the hydrolysates obtained after incubation of the peptides in 6 M HCl under argon at 110°C for 20 h. The concentrations of large peptides, H-A2-B, M6, and K5, were also determined by the BCA agent (Pierce, Rockford, IL). Trp-containing peptides were also quantitated spectrophotometrically by using an ε₂₈₀ of 5.56.

Inhibition Studies.

Kinetic analyses on reaction rates in the presence of peptides were performed under two conditions. Initially, various amounts of peptide were added to a fixed amount of substrate for the determination of IC₅₀. Reaction rates were determined as described (13). Subsequently, with peptides that showed inhibition, additional studies were performed by varying the substrate concentration in the presence of a fixed inhibitory peptide concentration. The reaction rates were fitted to the appropriate equations by nonlinear least squares with KaleidaGraph. Kinetic data were also analyzed in Lineweaver–Burke plots to verify competitive inhibition.

Mutagenesis.

Site-directed mutagenesis of H-A2-B was performed by PCR with a QuikChange II mutagenesis kit (Stratagene, La Jolla, CA). The oligonucleotide primers used for mutagenesis were 5′-TATCCTCATCCAGGCCTTTGAGA-CGCTCCC-3′ for the Asp1653Ala mutation, 5′-ATCCAGGACTTTGCGACGCTCCCCCGA-3′ for the Glu1655Ala mutation, 5′-TTTGAGACGCT-CCCCGCAGAGGCTCCTGA-CCT-3′ for the Arg1659Ala mutation, 5′-ACGCTCCCCCG-AGCGGCTCCTGACCT-3′ for the Glu1660Ala mutation, 5′-GAGAGGCTCCTGCCCTGGTGCTGCAG-3′ for the Asp1663Ala mutation, and 5′-GACCTGGTGCTGCAGGCGCTTCTGAACGACAT-3′ for the Arg1668Ala mutation. All mutations were confirmed by DNA sequencing. Mutant H-A2-B peptides were expressed and purified as described (13).

Digestion of Multimeric VWF.

Normal human plasma was desalted on a PD-10 column (GE Healthcare, Piscataway, NJ) in 10 mM Hepes, pH 7.4. VWF was digested with the ADAMTS-13 in the desalted plasma by incubation for 15 h at 37°C in 5 mM Hepes, pH 7.4/7 mM CaCl₂/1 M urea with or without peptides. The reaction was terminated by methanol (final concentration of 10%) and placed on ice for 30 min. Precipitates were collected by centrifugation at 8,000 × g for 10 min and dissolved in gel sample buffer (50 mM Tris·HCl, pH 6.8/8 M urea/2 mM EDTA). The samples were heated at 80°C for 10 min and applied to 1% agarose gels prepared according to the method of Warren et al. (23). The separated VWF multimers were transferred to a PVDF membrane and visualized by the alkaline phosphatase substrate 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (Kirkegaard & Perry Laboratories, Gaithersburg, MD) after incubation with a polyclonal rabbit anti-human VWF antibody (Accurate Chemical and Scientific, Westbury, NY) and alkaline phosphatase-conjugated anti-rabbit IgG (Sigma, St. Louis, MO).

Digestion of Mutant Substrate Peptides.

Normal and mutant H-A2-B substrate peptides were reduced by DTT and reacted with iodoacetamide as described (24). The blocked peptides were purified by HPLC as described above. Peptides (1.5 μM) were digested with partially purified ADAMTS-13 (equivalent to 0.1 μg/ml) (25) at 22°C, and aliquots were withdrawn at 15-min intervals. The digestion was stopped by the addition of SDS/PAGE sample buffer containing EDTA and DTT (final concentrations of 10 and 20 mM, respectively). The digested products were separated by SDS/PAGE using 16% polyacrylamide gels containing tricine (Invitrogen, Carlsbad, CA). The separated peptides were stained with the fluorescent dye Sypro orange diluted in 10% acetic acid as recommended by the manufacturer (Invitrogen), and the intensity of blue fluorescence (excitation at 450 nm) was determined in a Storm 840 imaging system (GE Healthcare) and analyzed with the ImageQuant 5.2 program.

Abbreviations

VWF

von Willebrand factor

TTP

thrombotic thrombocytopenic purpura

BNPS-skatole

2-(2-nitrophenylsulfenyl)-3-methyl-3-bromoindolenine.