Shrinking Weibel‐Palade bodies prevents high platelet recruitment in assays using thrombotic thrombocytopenic purpura plasma

Abstract Background Thrombotic thrombocytopenic purpura (TTP), caused by a genetic or autoimmune‐driven lack of ADAMTS‐13 activity, leads to high levels of the ultra‐large von Willebrand factor (VWF) multimers produced by endothelial cells, causing excess platelet recruitment into forming thrombi, often with mortal consequences. Treatments include plasma infusion or replacement to restore ADAMTS‐13 activity, or prevention of platelet recruitment to VWF. Objectives We tested a different approach, exploiting the unique cell biology of the endothelium. Upon activation, the VWF released by exocytosis of Weibel‐Palade bodies (WPBs), transiently anchored to the cell surface, unfurls as strings into flowing plasma, recruiting platelets. Using plasma from patients with TTP increases platelet recruitment to the surface of cultured endothelial cells under flow. WPBs are uniquely plastic, and shortening WPBs dramatically reduces VWF string lengths and the recruitment of platelets. We wished to test whether the TTP plasma‐driven increase in platelet recruitment would be countered by reducing formation of the longest WPBs that release longer strings. Methods Endothelial cells grown in flow chambers were treated with fluvastatin, one of 37 drugs shown to shorten WPBs, then activated under flow in the presence of platelets and plasma of either controls or patients with TTP. Result We found that the dramatic increase in platelet recruitment caused by TTP plasma is entirely countered by treatment with fluvastatin, shortening the WPBs. Conclusions This potential approach of ameliorating the endothelial contribution to thrombotic risk by intervening far upstream of hemostasis might prove a useful adjunct to more conventional and direct therapies.


| INTRODUC TI ON
Thrombotic thrombocytopenic purpura (TTP) is a severe disease resulting in multiorgan failure and, untreated, has a mortality of >90%.
TTP arises from a severe deficiency of ADAMTS-13 activity, 1-3 a circulating plasma metalloprotease, the only known substrate for which is von Willebrand factor (VWF).
VWF undergoes a complex biosynthesis in endothelial cells, culminating in the formation of tubules of multiple coiled ultra-large (UL-VWF) multimers that drive the formation of the rod-shaped endothelial secretory granules, Weibel-Palade bodies (WPBs). [4][5][6] Endothelial activation drives WPB exocytosis to release their content into the plasma, where flow unfurls the tubular coils of UL-VWF into long strings anchored to the endothelial surface by an unknown mechanism. 7 The transient strings are highly active in platelet recruitment and are incorporated into primary hemostatic structures or are cleaved by ADAMTS-13 into the smaller VWF multimers seen in plasma. 8 We recently showed that the lengths of VWF strings (but not VWF's multimeric state) change with the length (ie, size) of WPBs; that WPBs are formed in a range of sizes from 0.5 to 5 μm and that their size is affected by physiological cues, or by drugs that affect the modulators of WPB size (Golgi linkage, secretory trafficking rate, and VWF expression level). [9][10][11][12] In TTP, the lack (either genetic or immune mediated) of functional ADAMTS-13 precludes proteolysis of the just-secreted UL-VWF multimers. UL-VWF incorporated into relatively stable strings then forms platelet/plasma VWF hemostatic structures or escapes into circulating plasma. With UL-VWF persisting, excess platelet recruitment to strings or into thrombi then occurs. [13][14][15][16][17] We describe here a potential approach toward ameliorating congenital TTP (cTTP) that focuses on modulating endothelial cell function. We show, in these in vitro proof-of-concept experiments, that we can repurpose a representative of well-understood drugs, statins, that we previously identified as able to decrease WPB size, thus decreasing string lengths, to reduce the platelet recruitment under flow in vitro. Fluvastatin decreases the number and length of VWF strings by reprogramming the size of newly forming WPB in endothelial cells in vitro, 10,12 by an "organelle-directed medicine" approach, 18 producing results consistent with their known anti-inflammatory and anticoagulant 19,20 effects in vivo. Our proof-of-concept experiments comparing plasma from controls and patients with cTTP show that overcoming ADAMTS-13 deficiency and ablating excess platelet adhesion to VWF in cTTP might be possible using fluvastatin.

| Patient selection
Nine patient episodes with cTTP under the care of a single tertiary center were identified from the UK TTP Registry (Medical Research Ethics Committee Numbers 08/H0810/54 and 08/H0716/72) and characterized here. 21 cTTP was defined as patients with ADAMTS-13 activity <10 IU/dL, no evidence of anti-ADAMTS-13 IgG antibodies and the identification of an ADAMTS-13 mutation causing cTTP. The female patients were diagnosed in adulthood (median age, 26), the male in infancy, aged 6. Two of the six women (patient 1 and patient 2) were pregnant at the time of sampling and were also resampled after pregnancy, respectively, 6 and 7 months after delivery. One patient (patient 3) was a new presentation, diagnosed immediately postpartum and sampled 1 month after delivery. All patients were routinely monitored with laboratory parameters in normal range not suggestive of acute or subacute disease and received regular plasma infusion therapy, typically 10 mL/kg Octaplas (Octapharmal Lachen, Switzerland). One patient was treated with intermediate purity factor VIII concentrate (BPL-8Y), ≈15 IU/kg (see Table 1). For all cases, the following parameters were included: demographic information, full blood count, ADAMTS-13 activity, ADAMTS-13 antigen, VWF activity, VWF antigen, flow-based assay parameters.

| Immunocytochemistry
After the flow assay, fixed HUVECs and platelets in μ-slides were

| Confocal microscopy
Imaging was performed using a spinning-disc Ultraview Vox confocal microscope with a 20× objective and 1.5× tube lens. A tiled image panel was taken comprising 36 images in a 3 × 12 grid with 30 μm z depth.

| Image analysis
VWF and platelet adherence to the HUVEC monolayer was quantified by image analysis in Fiji (ImageJ). Each field of view (35 images per sample) was assessed independently. The z-stacked images were transformed into a maximum intensity Z-projection, and the fluorescence signal of VWF and CD41 then thresholded and converted into binary values. The area of each image occupied by VWF particles or platelets was measured using the "Analyse particle" function within ImageJ. In order to segment VWF strings in images taken under laminar flow, we developed a Python script. 25 It implements a set of convolution matrices to remove nonstring VWF agglomerates and highlight regions of discrete contiguous signal aligned with flow direction. We used a threshold to produce a binary image and measured the properties of these unique regions.

| Statistical analysis
Statistical analyses were performed using Prism software version 7 (GraphPad Software, La Jolla, CA, USA). The tests used to assess statistical significance are indicated in the figure legends. Abbreviations: Fragmin, low-molecular-weight heparin; N/A, not applicable; p.p., post pregnancy.

| RE SULTS
immediately before plasma infusion therapy, was <5 IU/dL (normal range, 64-132 IU/dL), whereas VWF antigen and activity levels were both above the normal range for all patients, except for patient 4 and patient 2 (postpregnancy; see Table 1).
To test the effect of statins on platelet and plasma VWF recruitment in the presence of plasma from pooled controls or cTTP pa- Our analyses of string length and number found that the total cumulative string length increased almost twofold in the presence of patient as compared to control plasma ( Figure 1A, B). Increases ranged from 1.47-fold in patient 1 after pregnancy to 3.89-fold in patient 2, consistent with strings being more stable in the absence of ADAMTS-13. 13 We also confirmed that a 24-hour pretreatment with fluvastatin decreased the total cumulative length to below no-statin levels in assays using both control and patient plasma ( Figure 1A, B). The median string length ( Figure 1C) was also significantly longer in strings formed with TTP plasma, but did not fall as far as the total string length when treated with fluvastatin, suggesting that we are differentially suppressing the very longest, and most functionally active 10 fraction of strings. We also show a simplified metric; strings longer than 25 μm ( Figure 1D there is a rough relationship between which patients' plasma was used in the assay, the cumulative length, the fraction of strings longer than 25 μm and the number of strings seen in the presence or absence of fluvastatin treatment. This encourages us to believe that our assay is capable of discriminating between the small changes (shown in Table 1) in the plasma content between patients, and thus likely to be of utility.
We also analyzed comparative platelet recruitment. We found an average 3.87-fold increase in the area covered by platelets recruited to the untreated endothelial monolayer incubated with patient plasma as compared to control plasma (Figure 2A), but platelet recruitment was significantly inhibited by fluvastatin not only in the presence of control but even more dramatically in the presence of patient plasma in flow assays (Figure 2A, B). combining not only the VWF present in strings, but also exocytosed, surface-retained VWF that has not formed strings, plus plasma VWF recruited to the cell surface. In vivo these could all potentially contribute to sustained platelet adhesion and thrombus formation ( Figure 3A, B). 27 We conclude that these mutually supportive data reinforce the hypothesis that fluvastatin decreases the endothelial capacity for plasma VWF and platelet recruitment, significantly diminishing its potential prohemostatic capacity and effectively countering at least these parameters caused by the loss of a normal level of ADAMTS-13 function as observed in the plasma of patients with TTP.

| DISCUSS ION
In this study, we explored the potential use of reprogramming endothelial cells into producing shorter WPBs that contain VWF with lowered platelet/plasma VWF-recruiting capacity. We have chosen congenital TTP as a simple genetic disorder causing massively increased thrombotic risk and morbidity, but this approach could potentially be extended to the study of immuno-mediated TTP, as it acts by circumventing the lack of ADAMTS-13 functionality. Current therapies for TTP include plasma infusion and plasma exchange to replenish the level of ADAMTS-13 reducing UL-VWF multimers.
For congenital TTP, early diagnosis is possible, allowing for plasmareplacement prophylaxis to begin. However, if dosing and uptake is not optimal, a high risk of acute episodes with ischemic stroke and other consequences exist, as well as nonovert symptoms like lethargy, headache, mental disorders, and abdominal pain. 21 The disadvantages of such therapies include frequent hospital visits (every 1-2 weeks), the possibility of fluid overload and of developing an allergic reaction.
People with mild asymptomatic thrombocytopenia/subacute TTP episodes thus tend to defer or avoid the therapy, with long-term effects including silent thrombotic organ failure, cognitive impairment, and depression. 28 The arrival of recombinant ADAMTS-13 (now in phase III) 29 will allow regular and possibly home treatment, but availability may be limited and cost implications significant, indicating a need for further/adjunct options. Additional tools for overcoming the consequences of losing ADAMTS-13 activity could therefore be of benefit.
Statins are drugs designed to lower cholesterol by inhibiting 3-Hydroxy-3-methylglutaryl-coenzyme A reductase, but that also have pleiotropic beneficial effects, 19,30 leading to a reduced risk of cardiovascular disease and thrombosis and a decrease in all-cause mortality, 31 yet without causing bleeding. 32 In addition to these effects, statins also unlink the stacks of cisternae forming the Golgi ribbon, where WPBs are formed, thus enforcing the formation of shorter WPB organelles without affecting multimerization of VWF. 9 The shorter WPB reduce the ability of agonist-driven released VWF to recruit plasma VWF and platelets, 10 thereby supporting the reported antithrombotic effect of statins, which include, but are not limited to, increase in nitric oxide production, and decrease in tissue factor and plasminogen activator inhibitor 1. [33][34][35] We tested the ability of fluvastatin-treated endothelial cells to We previously identified statins as WPB-shortening drugs, 10 and showed that they diminish formation of the long platelet-recruiting VWF strings that form after exocytosis and reduce recruitment of plasma VWF to the endothelial surface. We here take this further by now showing that fluvastatin's WPB-mediated antithrombotic effect can also entirely overcome the enormously detrimental amplifying effect on these two aspects of hemostatic initiation occurring when using plasma samples of patients with cTTP. These data suggest a potential antithrombotic with an entirely different mode of action.
While progress of this work requires a clinical study, we provide not only a rationale supporting an additional therapy for cTTP but also a supplementary rationale for the mode of action of this drug in more common cerebrovascular disorders.

ACK N OWLED G M ENTS
The authors thank Bernhard Laemlle and Laura Benjamin for reading the manuscript. The work was partly funded by Shire (funding to MAS) and partly by a grant (MC_UU_00012/2) from the Medical Research Council to DFC.