Proprotein convertase subtilisin/kexin type 9 (PCSK9) Deficiency is Protective Against Venous Thrombosis in Mice

The effect of lipid lowering on the incidence of deep venous thrombosis (DVT) is controversial. The purpose of this study was to determine the effect of proprotein convertase subtilisin/kexin type 9 (PCSK9) deficiency on development of DVT in mice. Pcsk9 deficient (pcsk9 −/−) and wild-type (WT) littermates underwent partial inferior vena cava (IVC) ligation to induce venous thrombosis. 48 hours following IVC ligation, IVC thrombosis was evident in 60% of WT mice and 25% of pcsk9 −/− mice (p < 0.05). Analysis of IVC thrombosis revealed greater thrombus weight, length, myeloid cell recruitment, and more neutrophil extracellular trap formation (NETs) in WT compared to pcsk9 −/− mice. Intravital microscopy performed two hours following partial IVC ligation revealed that leukocyte firm attachment was increased in WT mice compared to mice undergoing a sham operation, however leukocyte attachment was reduced in pcsk9 −/− mice compared to WT mice. In conclusion, deficiency of PCSK9 is associated with protection from venous thrombosis. This protection is associated with reduced leukocyte recruitment and NET formation at the site of thrombosis.

Effect of PCSK9 deficiency on leukocyte attachment following IVC partial ligation. Leukocytes have been shown to be critical mediators of thrombosis in this DVT model. Since LDL may regulate inflammatory responses, mice with low LDL may show a reduced inflammatory response to injuries. To determine whether leukocyte attachment to the IVC was affected in pcsk9 −/− mice, intravital microscopy was performed to the site of stasis below the IVC ligation. Two hours following ligation, rhodamine-labeled attached leukocytes were readily apparent although they were reduced in pcsk9 −/− mice compared to WT mice (Fig. 1A,B), consistent with a protective, anti-inflammatory effect of PCSK9 deficiency. As a marker of leukocyte accumulation in the vein wall, CD45 expression was determined 48 hours following IVC ligation. CD45 expression was significantly lower in pcsk9 −/− mice compared to WT mice (Fig. 1C).
Effect of PCSK9 deficiency on venous thrombosis following IVC ligation. 48 hours following IVC ligation, visually apparent thrombus was evident in 12 of 20 (60%) WT mice but in only 5 of 20 (25%) pcsk9 −/− mice, p < 0.05. Accordingly, the weight and length of IVC thrombi were reduced in pcsk9 −/− mice compared to WT mice (Fig. 2). Formalin fixed thrombi were also analyzed for cellular composition. Immunohistochemical staining showed evidence of increased macrophage and especially neutrophil accumulation in the thrombi of WT mice compared to pcsk9 −/− mice (Fig. 3). As this DVT model has been shown to exhibit NETosis that contributes to clot formation, immunostaining for CRAMP and Cit-H3 was performed which was greater in WT mice (Fig. 4), suggesting reduced NETosis in pcsk9 −/− mice.
Potential circulating mediators of leukocyte chemoattraction and adhesion implicated in inflammatory vascular diseases were measured before and at the termination of the thrombosis experiment. MCP-1 levels were significantly elevated following IVC ligation but were not different between pcsk9 −/− mice and WT mice (Fig. 5A). In contrast, levels of sP-sel, CXCL1 and CCL3 were all reduced in pcsk9 −/− mice compared to WT mice following IVC ligation (Fig. 5B-D).

Discussion
VTE, including DVT and pulmonary embolism (PE), is a common cause of morbidity and mortality 13 . Current treatment strategies involve drugs that inhibit factors in the coagulation cascade 14 , which can lead to bleeding complications. Any drug useful in prevention of VTE, without the concomitant bleeding risk, could be clinically useful. Therefore, identification of risk factors for VTE distinct from the coagulation cascade is a high priority. The association between hyperlipidemia and VTE risk is controversial with some studies showing positive associations [15][16][17] and others neutral 18 . In support of a causal relationship between lipid levels and VTE, some trials of lipid lowering with statins have shown significant reductions in the incidence of VTE 7 . Multiple mechanism(s) could contribute to the link between lipids and reduced VTE risk including effects on platelet function [19][20][21] , changes in coagulation factors 16,22 , alterations in fibrinolysis 23,24 , and effects on inflammatory pathways 25,26 .
PCSK9 is a protease that targets the LDL receptor for degradation 27 . Humans with gain-of-function PCSK9 mutations have hyperlipidemia and heightened cardiovascular risk while those with loss-of-function mutations are protected from complications of atherosclerosis [27][28][29] . Clinical trials with antibodies have established their efficacy in reducing LDL 9 and clinical trials are now underway to determine their effects on cardiovascular complications. A recently published IVUS trial strongly supports a beneficial effect of PCSK9 inhibition on progression of atherosclerosis 30 .
Similar to statins, PCSK9 inhibition leads to upregulation of LDL receptors with resulting enhanced LDL clearance 27 . Mice with PCSK9 deficiency appear to have a similar phenotype to PCSK9 deficient humans with low levels of LDL 11 and protection from atherosclerosis 10 . Since lipid reduction may attenuate chronic inflammatory responses in arterial disease, reduced lipid levels may also have beneficial effects on inflammatory responses in venous disease. To determine whether the vascular benefits of PCSK9 inhibition might extend to venous disease, we tested the effects of PCSK9 deficiency in a mouse model of DVT. Partial ligation of the mouse IVC leads to occlusive thrombosis in the majority of mice on the C57BL6/J strain background at 48 hours following ligation 31 . This relatively chronic model induced by venous stasis may be particularly relevant to most DVT's that occur in humans. DVT in this model has been shown to involve leukocytes, platelets, coagulation factors and neutrophil extracellular traps 31 .
As expected, mice deficient in PCSK9 in this study showed reductions in total cholesterol, LDL, and HDL compared to control mice, even on standard chow diet. Circulating lipid levels play a critical role in inflammatory responses. For example, in mouse models of atherosclerosis, hyperlipidemia leads to marked upregulation of leukocyte-endothelial interactions that are increased even further in response to tissue injury or hemodynamic perturbations [32][33][34] . Whether very low cholesterol levels lead to reduced L-E interactions in a non-hyperlipidemic model is less clear. In the current study, partial ligation of the IVC triggered an increase in leukocyte adhesion inferior to the ligation as has been reported by others 31 . Leukocyte adhesion was reduced in mice with PCSK9 deficiency suggesting a beneficial role of very low LDL levels towards leukocyte recruitment in this model. However, we cannot rule out the possibility that other lipid-independent effects of PCSK9 could be affecting the phenotype in this model. Leukocyte accumulation in this model has been previously shown to play an important contributory role toward the subsequent development of stable venous thrombosis 31 . Consistently, mice with PCSK9 deficiency developed less IVC thrombosis compared to wild-type mice at 48 hours. Thrombi from PCSK9 deficient mice were also characterized by reduced myeloid markers. Leukocyte recruitment in this model is p-selectin dependent and associated with enhanced CXCL1 expression 30,31 . CXCL1 expression and release have previously been shown to be modified by lipids 35 . It could be that the altered circulating lipid profile in PCSK9 deficient mice might reduce expression and/or release of this neutrophil chemoattractant. In this study plasma levels of sP-selectin and CXCL1 were reduced in PCSK9 deficient mice. Neutrophils have been shown to promote thrombosis through generation of NETs 36 . Thrombi present in PCSK9 deficient mice showed reduced NET formation. Since NET formation has been shown by several groups to promote venous thrombosis, it is likely that by affecting myeloid cell recruitment following IVC stasis, PCSK9 deficiency leads to reduced subsequent NET formation. It is also possible that other undefined mechanisms play a role in this process. Although we did not observe an effect of PCSK9 deficiency on whole blood platelet aggregation, a recent paper reported an association between elevated PCSK9 levels and platelet reactivity in patients with acute coronary syndromes 37 . Unexpectedly, circulating neutrophil and lymphocyte counts were elevated in PCSK9 deficient mice, an observation that will require further study.
In summary, this study supports the hypothesis that the vascular benefits of lipid lowering therapy may extend to venous disease. Human data should be revealing with regards to arterial and venous endpoints in ongoing clinical trials of PCSK9 inhibition.
All animal use protocols complied with the Principle of Laboratory and Animal Care established by the National Society for Medical Research and were approved by the University of Michigan Committee on Use and Care of Animals. All animal experiments were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Model of inferior vena cava (IVC) thrombosis. A stenosis model of deep venous thrombosis (DVT) was
performed on male WT and pcsk9 −/− mice at 10 weeks of age. Mice were anesthetized with intraperitoneal (i.p.) injection of sodium pentobarbital (67 mg/kg). The surgical procedure was performed as described previously 38,39 . Briefly, the abdomen was opened through a ventral midline laparotomy. The small intestine was gently moved aside and draped with sterile saline-soaked gauze. The IVC was exposed gently and a metal spacer made from 30-gauge needle was placed on the outside of vessel. The IVC and spacer were ligated using a 6.0 nylon suture (Ethicon, Somerville, NJ) just below the left renal vein. The spacer was then removed to allow blood flow through the IVC. Side branches of the IVC were not ligated or manipulated during the procedure. For sham-operated mice, the IVC was exposed but no ligation was performed. 48 hours after surgery, mice were sacrificed with intraperitoneal (i.p.) injection of sodium pentobarbital (100 mg/kg). Presence, weight and length of thrombi from IVC were measured. Mice without thrombus were assigned a value of 0 for analysis of length and weight. Plasma   was injected intravenously via the tail vein 2 hours following partial IVC ligation. The abdomen was then reopened and labeled leukocytes were visualized with a Nikon FN1 fixed-stage microscopy system with X-cite for epi-fluorescence. The area of interest consisted of a 2-3 mm segment of IVC distal to the IVC ligation. Videos were recorded with a Photometrics Coolsnap Cascade 512B color digital camera system for 2 minutes and analyzed with the MetaMorph Premier software package (Molecular Devices). After intravital microscopy, mice were sacrificed with intraperitoneal (i.p.) injection of sodium pentobarbital (100 mg/kg). Adherent cells were defined by lack of movement for at least 2 minutes. The number of adherent cells was manually counted as overlaid cells (yellow) in pictures taken at 0 min (assigned green) and 2 min (assigned red) of recording.

Real-Time Polymerase Chain Reaction. RNA from the IVC was isolated using a QIAGEN RNeasy Mini
Kit (QIAGEN Inc., Valencia, CA). The primer sets were purchased from Applied Biosystems (Carlsbad, CA). RTPCR was performed using an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Carlsbad, CA). 100 ng of RNA and 1 μl of primer were used per reaction. 7000 System SDS Software and the 2 −ΔΔCT method 41 were used to analyze the results. Results were presented as fold change of transcripts for target normalized to internal control (GAPDH).
Immunohistochemistry. 48 hours following IVC ligation, the formed thrombi were isolated and fixed in zinc formation. Cellular components of thrombi were determined using corresponding antibodies on paraffin-embedded sections. Macrophages were identified with a rat anti-mouse F4/80 monoclonal antibody (1:100) (Abcam, Cambridge, MA). Neutrophils were identified with a rabbit anti-mouse myeloperoxidase (MPO) polyclonal antibody (1:200) (DAKO, Carpinteria, CA). The NET-related markers cathelicidin-related antimicrobial peptide (CRAMP) and citrullinated histone H3 (Cit-H3) were detected using a rabbit anti-mouse CRAMP polyclonal antibody (1:200) (Innovagen, Lund, Sweden) and a rabbit anti-mouse Cit-H3 polyclonal antibody (1:100) (Abcam, Cambridge, MA) as described previously 42 . Positive cells were detected with corresponding biotin-conjugated secondary antibodies. Stained cells were counted manually from five positively stained fields in each section using NIH ImageJ software and expressed as a percentage of positive cells per unit area.

Measurement of plasma factors.
To measure cholesterol and triglycerides levels, plasma samples were collected via retro-orbital bleeding using capillary tubes from WT and pcsk9 −/− mice at 10 weeks of age before procedures. The samples were analyzed in the Chemistry Core of the Michigan Diabetes Research and Training Center using Enzymatic-Colorimetric kits (Roche, Indianapolis, IN).
Cell counts and Whole blood aggregation. Complete circulating blood cell and platelet counts were obtained using a Hemavet 950 haematology system (Drew Scientific, Miami Lakes, Florida). For whole blood aggregation, 500μl of whole blood anticoagulated with 3.2% sodium citrate (1: 9) was diluted in same volume of 0.9% saline and incubated for 5 min at 37 °C. Collagen (5μg/ml) -stimulated whole blood aggregation was performed using a Whole Blood / Optical Lumi-Aggregation System (Chrono-Log Corp., Havertown, PA) according to manufacturer's instruction.
Statistical analysis. All data are presented as mean ± standard error. Statistical analysis was carried out using GraphPad Prism. To compare incidence of DVT in WT and pcsk9 −/− mice, the chi-square test was performed. For comparison between two groups, results were analyzed using unpaired t-test. For multiple comparisons, results were analyzed using one-way ANOVA followed by Tukey post-test analysis. Probability values of p < 0.05 were considered statistically significant.