Interaction of Shiga Toxin with the A-domains and Multimers of von Willebrand Factor*

Background: VWF is a multimeric glycoprotein that causes platelet adherence and aggregation and is cleaved by ADAMTS-13. Results: Shiga toxin binds to VWF multimers, specifically the A1 and A2 domains, and decreases ADAMTS-13 cleavage rate at Tyr1605-Met1606 of the A2 domain. Conclusion: Shiga toxin binding to VWF impairs ADAMTS-13 cleavage. Significance: Delayed VWF cleavage may contribute to renal thrombotic microangiopathy in Shiga toxin-induced hemolytic uremic syndrome. Shiga toxin (Stx) produced by enterohemorrhagic Escherichia coli causes diarrhea-associated hemolytic-uremic syndrome (DHUS), a severe renal thrombotic microangiopathy. We investigated the interaction between Stx and von Willebrand Factor (VWF), a multimeric plasma glycoprotein that mediates platelet adhesion, activation, and aggregation. Stx bound to ultra-large VWF (ULVWF) secreted from and anchored to stimulated human umbilical vein endothelial cells, as well as to immobilized VWF-rich human umbilical vein endothelial cell supernatant. This Stx binding was localized to the A1 and A2 domain of VWF monomeric subunits and reduced the rate of ADAMTS-13-mediated cleavage of the Tyr1605-Met1606 peptide bond in the A2 domain. Stx-VWF interaction and the associated delay in ADAMTS-13-mediated cleavage of VWF may contribute to the pathophysiology of DHUS.


Shiga toxin (Stx) produced by enterohemorrhagic Escherichia coli causes diarrhea-associated hemolytic-uremic syndrome (DHUS), a severe renal thrombotic microangiopathy. We investigated the interaction between Stx and von Willebrand Factor (VWF), a multimeric plasma glycoprotein that mediates platelet adhesion, activation, and aggregation. Stx bound to ultra-large VWF (ULVWF) secreted from and anchored to stimulated human umbilical vein endothelial cells, as well as to immobilized VWF-rich human umbilical vein endothelial cell supernatant. This Stx binding was localized to the A1 and A2 domain of
The hemolytic-uremic syndrome (HUS) 2 is a leading cause of acute kidney failure in children worldwide (1,2). A majority of these cases are diarrhea-associated HUS (DHUS), and are characterized by episodes of bloody diarrhea with glomerular platelet-fibrin thrombi and associated renal thrombotic microangiopathy (1,3,4).
Increased platelet-mediated thrombus formation contributes to the thrombotic microangiopathy and acute renal failure in DHUS (3,(12)(13)(14)(15). Thrombocytopenia characteristic of DHUS and accumulation of platelets atop intact (i.e. non-desquamated) glomerular endothelial cells in the renal circulation have been shown in sophisticated and carefully timed morphological studies (13,15). These thrombi may be the result of increased secretion or persistence of von Willebrand Factor (VWF), a multimeric glycoprotein secreted by endothelial cells that mediates platelet adherence and aggregation (3,4,16). We have hypothesized that the hyper-adhesive ultra-large VWF (ULVWF) multimers secreted from the endothelial cells play a role in the pathophysiology of DHUS, a speculation supported by data obtained from in vitro microfluidic flow models and in vivo primate studies (4, 16 -18). ADAMTS-13 (a disintegrin and metalloprotease with thrombospondin domains 13), the plasma metalloprotease responsible for the cleavage of VWF into smaller less reactive forms, remains within a broad normal concentration range and does not provide an explanation for the prothrombotic pathologies during DHUS episodes (16, 19 -21). We have investigated the binding and functional interactions among Shiga-like toxin, VWF, and ADAMTS-13 to determine the molecular explanation for thrombus formation in DHUS.

EXPERIMENTAL PROCEDURES
Plasma Preparation-Human blood from unmedicated healthy donors was drawn into final concentrations of 0.38% sodium citrate or 5 mM EDTA. Blood was centrifuged at 150 ϫ g for 15 min at room temperature to isolate citrated and EDTA normal plasma (NP). All work on human VWF and human endothelial cells, including experiments in this study, have been specifically approved by the Rice University Institutional Review Board. Human tissues and blood samples were collected under a protocol approved by the Rice University Institutional Review Board. Donors provided their written informed consent to participate in the study.
Shiga toxin-1 (Stx-1) was purchased from List Biological Laboratories and used primarily at 0.1 nM except in FRET assays where a range of Stx-1 concentrations were tested. Cholera toxin (Ctx, Sigma-Aldrich) with an analogous A (22 and 5 kDa) B 5 (10.6 kDa) subunit structure to Stx-1 was used as a control toxin at 0.125 M (7). Toxin concentrations were consistent throughout experimentation. The purities of Stx-1 and Ctx were confirmed by 4 -15% SDS-PAGE, Coomassie staining and Western blot detection (goat anti-SLT-1B plus rabbit anti-goat IgG-HRP, rabbit anti-Ctx plus goat anti-rabbit IgG-HRP), and chemiluminescence.
Fluorescent Staining of Stimulated Endothelial Cells-Human umbilical vein endothelial cells (HUVECs) seeded on glass coverslips were washed with PBS, pH 7.4, and stimulated for 2 min with 100 M histamine in PBS. To prevent HUVEC-released ADAMTS-13 from cleaving secreted ULVWF strings, cells were stimulated in a relatively high incubation volume to cell surface ratio (i.e. 1 ml of buffer to 10 cm 2 of cells) for a short time period. Following stimulation, cells were incubated with 0.1 nM Stx-1 in 1% BSA/PBS with rabbit anti-human VWF and chicken anti-rabbit IgG-488 for 15 min. Cells were fixed for 10 min with 1% p-formaldehyde in PBS, washed 3ϫ with PBS, and then stained with either goat anti-SLT-1B plus donkey antigoat IgG-594 or mouse anti-SLT-1 plus goat anti-mouse IgG-594 for 15 min. HUVEC nuclei (blue) were detected with 1 nM DAPI. In cholera toxin experiments, the histamine-stimulated HUVECs were incubated with 0.125 M Ctx diluted in 1% BSA/ PBS plus goat anti-human VWF and donkey anti-goat IgG-488 for 15 min. Cells were fixed for 10 min with 1% p-formaldehyde in PBS, washed 3ϫ with PBS, and then stained with rabbit anticholera toxin plus chicken anti-rabbit IgG-594 for 15 min.
Quantification of Stx-1 Bound to HUVEC-secreted/anchored ULVWF Strings-HUVEC-anchored ULVWF strings detected with anti-VWF primary antibody plus fluorescent secondary IgG-488 were electronically traced in 488-nm (green) captured images at 600ϫ magnification, and the emitted fluorescent intensity was measured and integrated along the line. The xand y-coordinates of the traced ULVWF line were transferred to the corresponding 594-nm (red) captured images obtained using the detecting antibodies for Stx-1 (goat anti-SLT-1B or mouse anti-SLT-1) or in control experiments for cholera toxin (rabbit anti-cholera toxin) plus fluorescent secondary IgG-594 antibodies. The fluorescent intensity at 594 nm from these images was measured and integrated along the transferred line coordinates. To determine background 594-nm intensity, the line coordinates were trans-located ϳ20 pixels (ϳ2.3 m) parallel to its original position within each 594-nm image, and the fluorescent intensity was measured and subtracted.
Measurement of Stx-1 Binding to Recombinant VWF Adomains-Recombinant VWF A-domain proteins (A1, A2, A3) in concentrations ranging from 0.0625 to 1 M in 50 mM bicarbonate buffer, pH 9.6, were immobilized overnight at 4°C in a 96-well plate. The control wells were coated with 1% BSA in bicarbonate buffer, with subsequent dilutions to match the A-domain dilutions. The wells were washed with Tris-buffered saline, 0.05% Tween 20 (TBS-T) and blocked with 1-3% BSA/ PBS for one hour at 37°C. The blocked well were incubated with 0.1 nM Stx-1 in 1% BSA/PBS for 1 h at 37°C. Subsequently, wells were washed 3ϫ with TBS-T and incubated with goat anti-SLT-1B plus rabbit anti-goat IgG-HRP for 1 h at 37°C. Bound Stx-1 was detected with 3,3Ј,5,5Ј-tetramethylbenzidine substrate. Absorbance at 450 nm was determined with a Tecan Infinite M200 plate reader (Mannedorf, Switzerland). The experiment was conducted in triplicate.
HUVECs-Primary HUVECs were isolated from umbilical veins as described previously (4). Cells were seeded in flasks or glass coverslips coated with attachment factor (Invitrogen). HUVECs were grown in MCDB 131 (Invitrogen) and supplemented with 3% penicillin-streptomycin, 0.2 mM L-glutamine, and low serum growth supplement (Invitrogen). From previous studies, we know that the secretion and anchorage of hyper-adhesive ULVWF multimeric strings from HUVECs and glomerular microvascular endothelial cells are similar (4).
VWF Depletion of HUVEC Supernatant-Cell supernatant was collected after 1-3 days from HUVECs maintained in serum-free MCDB 131 medium supplemented with insulintransferrin-selenium (Invitrogen) plus glutamine and penicillin-streptomycin. Cell supernatant was depleted of VWF by overnight mixture at 4°C with rabbit anti-human VWF IgG coupled to CNBr-activated Sepharose beads (Sigma-Aldrich). The anti-VWF coupled Sepharose beads were prepared following the manufacturer's instructions.
Measurement of Stx-1 Binding to Immobilized VWF Secreted from HUVECs-Samples of HUVEC supernatant (containing VWF) and VWF-depleted HUVEC supernatant were diluted in 50 mM bicarbonate and immobilized overnight at 4°C on a 96-well plate. The wells were then blocked with 1-3% BSA/PBS overnight at 4°C. The TBS-T washed wells were incubated with 0.1 nM Stx-1 for 1 h at 37°C. The wells were washed, and the amount of bound Stx-1 was determined by immunoassay using goat anti-SLT-1B plus rabbit anti-goat IgG-HRP and 3,3Ј,5,5Ј-tetramethylbenzidine substrate detection at 450 nm. The experiment was conducted in triplicate.

Stx-1 Binds to Secreted and Anchored ULVWF Strings from
Stimulated HUVECs-Stx-1 binding to endothelial cell-anchored ULVWF was demonstrated in HUVECs stimulated with 100 M histamine and in the presence of either 0.1 nM Stx-1 or 0.125 M Ctx under static conditions. Ctx, with an analogous AB 5 subunit structure to Stx-1, was used as a negative control. Histamine was chosen as the stimulating agent to maintain consistency when comparing amounts of Stx-1 and Ctx binding upon HUVEC-secreted/anchored ULVWF strings. The cells were fixed and stained for either VWF and Stx-1 or for VWF and Ctx. Fluorescent microscopy demonstrated that Stx-1 bound along HUVEC secreted and anchored ULVWF strings, whereas Ctx did not bind (Fig. 1). Quantification of these images demonstrated that anchored ULVWF strings had 8-fold more Stx-1 attached than Ctx (Fig. 2A). These results were verified with both goat polyclonal and mouse monoclonal antibodies against Stx. The purity of toxins and specificity of antibodies are shown in Fig. 2B. In the absence of secreted ULVWF strings, Stx-1 (and Ctx) bound to the surfaces of unstimulated HUVECs (supplemental Fig. S1).
Stx-1 also Binds to Immobilized, VWF-rich HUVEC Supernatant-VWF-rich HUVEC supernatant samples with a range of VWF antigen from 100 -750 ng/ml were immobilized and compared with samples completely and partially (50%) depleted of VWF antigen. Partial depletion of VWF reduced Stx-1 binding by 30%, and full depletion reduced binding by 90% (Fig. 3).

Stx-1 Binds to Recombinant VWF A-domains A1 and A2-
Binding of Stx-1 to recombinant VWF A-domains (A1, A2, and A3) was assessed by ELISA. Stx-1 bound predominately to immobilized VWF A1 and A2 domains (Fig. 4A). Stx-1 binding to A1 was 15% greater than to A2, and 3-fold greater than to the A3 domain. The binding of Stx-1 to VWF A1 and A2 was concentration-dependent with saturation at 0.25 M. Purity of A-domain proteins is shown in Fig. 4B.
Stx-1 Inhibits ADAMTS-13-mediated Cleavage of Tyr 1605 -Met 1606 of the A2 Domain-The effect of Stx-1-VWF interaction upon ADAMTS-13-mediated cleavage of the Tyr 1605 -Met 1606 peptide bond in the A2 domain was assessed using a synthetic FRETS-VWF73 substrate containing a 73-amino acid sequence of the A2 domain that includes the Tyr 1605 -Met 1606 peptide bond cleaved by ADAMTS-13. The addition of 0.1 nM Stx-1 to NP reduced the ADAMTS-13-mediated cleavage rate of the FRETS-VWF73 substrate by 35% (Fig. 5). The cleavage of FRETS-VWF73 by plasma containing ADAMTS-13 inactivated by EDTA, which binds the Zn 2ϩ and Ca 2ϩ required for enzyme activity, is shown for comparison. The inset in Fig. 5B demonstrates that the inhibitory effect of Stx-1 is maintained even at lower concentrations.

DISCUSSION
We previously described a delay in ADAMTS-13-mediated cleavage of HUVEC-anchored ULVWF multimeric strings with adherent platelets under flowing conditions in the presence of nanomolar concentrations of Stx and suggested that this effect may contribute to the thrombotic microangiopathy in DHUS (4). The experiments described in this study demonstrate that Stx binds to the A1 and A2 domains of VWF and that binding results in a decreased rate of ADAMTS-13-mediated cleavage of the 1605-1606 peptide bond in the A2 domain of VWF. The results provide evidence in support of the importance of Stx-VWF interaction in the pathophysiology of DHUS.
The FRET VWF cleavage assay was conducted using plasma with ADAMTS-13 concentrations within a normal range, as found in DHUS patients (19 -21, 28). The inhibitory effect of 0.1 nM Stx on ADAMTS-13-mediated cleavage of the Tyr 1605 -Met 1606 peptide bond occurred within minutes of Stx addition and was likely the result of Stx binding to the VWF A2-derived 73-amino acid peptide due to the acute nature of this effect. This suggests that a delay in cleavage of VWF can be achieved without decreased ADAMTS-13 levels (26).
The importance of Stx-VWF interaction is further supported by results from our 2005 study (4) where secreted ULVWFs stimulated by Stx-1 and Stx-2 were visualized by adherent platelets. The presence of platelets upon Stx-occupied ULVWF indicate that Stx binding upon VWF does not obstruct platelet binding, meaning Stx-occupied ULVWF are still prothrombotic (4).
Stx-1 bound to surfaces of unstimulated HUVECs, presumably to Gb3 receptors in the absence of secreted ULVWF strings (1605-1606 peptide bond supplemental Fig. S1). Although Stx-1 and Stx-2 bind to Gb3 on both HUVECs and glomerular micro-  Fig. 1 was determined from the intensities of fluorescent antibodies against Stx-1 or Ctx detected along the ULVWF strings. The ULVWF strings fluorescently detected with anti-VWF antibody plus secondary in 488 nm (green) images were electronically traced, and the x-and y-coordinates of these points were transferred to the corresponding images at 594 nm (red) detected antibodies directed against Stx-1 or Ctx. The intensities were measured and integrated along these points. The background intensities, measured from within these 594 nm images at parallel locations displaced from the original position, were subtracted. Values are means of 12-16 images ϩ S.D. (n ϭ 6 for each condition). p value was calculated with Student's t test. B, purity of Stx-1 and Ctx was confirmed through 4 -15% SDS-PAGE by Coomassie staining and Western blot detection either with goat anti-Stx plus rabbit anti-goat IgG-HRP or with rabbit anti-Ctx plus goat anti-rabbit IgG-HRP followed by chemiluminescence. Lanes contain 200 ng of Stx, 200 ng of Ctx, and 15 g of plasma proteins (NP). vascular endothelial cells and Stx-1 to the HUVEC-secreted/anchored ULVWF multimeric strings (as shown in this work), our data shows that Gb3 does not successfully compete with ULVWF for the binding to Stx-1 (4). Although the relative binding affinities of Stx for Gb3 and HUVEC-secreted/anchored ULVWF strings are not known, our study demonstrates that the binding affinity of Stx for the HUVEC-anchored ULVWF strings must be high. This is because after a 15-min incubation with 10 Ϫ10 M Stx-1, a sufficient quantity of the toxin attached to the HUVEC-anchored ULVWF strings to allow easy detection by immunofluorescence/ microscopy, even when Gb3 binding sites on the HUVECs were also available.
The interaction between Stx and VWF (instead of Stx and ADAMTS-13) was clearly demonstrated by several means; Stx was capable of binding to HUVEC-secreted/anchored ULVWF multimeric strings, large soluble VWF multimers that were immobilized, and recombinant VWF A1 and A2 domains. The interaction between Stx and VWF A2 demonstrated in the ELISA likely explains the reduction in the rate of cleavage at Tyr 1605 -Met 1606 by ADAMTS-13 and may account for the persistence of hyper-adhesive HUVEC-secreted/anchored ULVWF strings for a time period sufficiently long to allow excessive platelet adherence and aggregation.
Stx-1 was further examined for binding to immobilized fibrinogen; however, the reactivity level was similar to that of its nonspecific interaction with immobilized BSA. This study focused on VWF because binding of Stx upon VWF creates an effect that is uniquely prothrombotic by preventing a natural cleavage, as opposed to binding upon other proteins, which may not produce an effect as heavily implicated in thrombosis.
The role of glycosylation in Stx-VWF binding (analogous to Stx binding on Gb3) was considered. Our experiments demonstrated that Stx binds to both non-glycosylated recombinant A-domain and glycosylated VWF multimers, indicating that the interaction can occur in the absence of carbohydrate chains.
The interaction between Stx and VWF may also influence binding to platelet GPIb␣. We have demonstrated previously that the structure of the A1 domain in isolated and full-length VWF is changed when bound to collagen (29). This conformational change is associated with an increase in binding capacity of VWF for GPIb␣ (30). We hypothesize that similar to collagen, the interaction between Stx and ULVWF alters the struc-  Equal volume samples of citrated NP at 37°C was subsequently added to both conditions. EDTA plasma (EDTA), which inactivates ADAMTS-13, was added to the substrate alone for comparison. Cleavage rate was assessed with a FRET assay. A, results from three experiments were normalized, averaged, and displayed. Both NP and NPϩStx-1 achieved R 2 of 0.97, demonstrating good linear fit. B, the cleavage rates were determined from the slopes generated by linear regression of data in A. Each error bar represents the mean ϩ S.D. (n ϭ 3). p value (*, p Ͻ 0.05 compared with NP) was calculated with Student's t test. Inset shows the inhibitory effects of low range 1 ϫ 10 Ϫ11 M to 6.25 ϫ10 Ϫ13 M Stx-1 addition normalized to the cleavage rate of NP without Stx-1 addition Ϯ S.D. (n ϭ 4). ture of the A1 domain and increases binding affinity for GPIb␣. Furthermore, Stx, LPS, and cytokines, which are increased in DHUS, also increase secretion/anchorage of ULVWF multimeric strings from glomerular microvascular endothelial cells (4,9,16,18,31,32). These mechanisms, together with the inhibitory action of Stx on ADAMTS-13 cleavage, presumably form part of the mechanism that contributes to the pathophysiology of thrombotic microangiopathy in DHUS.
In this study, we describe a novel and clinically significant molecular interaction between Stx and VWF. Stx binds to A1 and A2 or possibly to peptide regions near the A1-A2 junction. This binding obstructs the access of ADAMTS-13 to the Tyr 1605 -Met 1606 A2 cleavage site. The resulting delay in cleavage of endothelial cell-anchored ULVWF multimers, by increasing the time available for platelet adhesion, activation, and aggregation, provides a possible explanation for thrombotic microangiopathy in DHUS.