PFKFB3 gene deletion in endothelial cells inhibits intraplaque angiogenesis and lesion formation in a murine model of venous bypass grafting

Vein grafting is a frequently used surgical intervention for cardiac revascularization. However, vein grafts display regions with intraplaque (IP) angiogenesis, which promotes atherogenesis and formation of unstable plaques. Graft neovessels are mainly composed of endothelial cells (ECs) that largely depend on glycolysis for migration and proliferation. In the present study, we aimed to investigate whether loss of the glycolytic flux enzyme phosphofructokinase-2/fructose-2,6-bisphosphatase 3 (PFKFB3) in ECs inhibits IP angiogenesis and as such prevents unstable plaque formation. To this end, apolipoprotein E deficient (ApoE−/−) mice were backcrossed to a previously generated PFKFB3fl/fl Cdh5iCre mouse strain. Animals were injected with either corn oil (ApoE−/−PFKFB3fl/fl) or tamoxifen (ApoE−/−PFKFB3ECKO), and were fed a western-type diet for 4 weeks prior to vein grafting. Hereafter, mice received a western diet for an additional 28 days and were then sacrificed for graft assessment. Size and thickness of vein graft lesions decreased by 35 and 32%, respectively, in ApoE−/−PFKFB3ECKO mice compared to controls, while stenosis diminished by 23%. Moreover, vein graft lesions in ApoE−/−PFKFB3ECKO mice showed a significant reduction in macrophage infiltration (29%), number of neovessels (62%), and hemorrhages (86%). EC-specific PFKFB3 deletion did not show obvious adverse effects or changes in general metabolism. Interestingly, RT-PCR showed an increased M2 macrophage signature in vein grafts from ApoE−/−PFKFB3ECKO mice. Altogether, EC-specific PFKFB3 gene deletion leads to a significant reduction in lesion size, IP angiogenesis, and hemorrhagic complications in vein grafts. This study demonstrates that inhibition of endothelial glycolysis is a promising therapeutic strategy to slow down plaque progression.


Introduction
Atherosclerosis is a chronic inflammatory disease of the arterial wall and it is one of the most important causes of cardiovascular disease, including severe conditions such as coronary artery disease, myocardial infarction, heart failure, and stroke. Vein bypass grafting is a surgical procedure that uses large saphenous veins to bypass occluded atherosclerotic arteries, thereby allowing revascularization of an ischemic region of the heart or limbs [1]. Unfortunately, at least 40% of patients suffer from bypass failure within eight years after the procedure due to negative vascular remodeling and intimal hyperplasia [1][2][3][4]. Furthermore, vein grafts often present accelerated atherosclerosis with formation of unstable plaques and increased risk of rupture [5][6][7].
New small vessels can form inside vein grafts to fulfill an increased demand for oxygen and nourishment of the vessel wall. This event, which is further promoted by inflammatory conditions, leads to intraplaque (IP) angiogenesis and contributes to plaque instability in the vein graft [8,9]. Indeed, apolipoprotein E deficient (ApoE −/− ) mice undergoing a vein graft interposition of the carotid artery develop unstable plaques with extensive IP neovessels that are often dysfunctional or immature and contribute to lesion destabilization by enhancing leukocyte recruitment and accumulation of cholesterol and platelets [10,11].
Angiogenesis is an energy-intensive process that requires extensive metabolic functioning of endothelial cells (ECs) to support sprouting, migration, and proliferation [12]. Recent studies have shown that ECs in neovessels generate more than 85% of their ATP by glycolysis [13,14]. One of the rate-limiting checkpoints of glycolytic flux is the conversion of fructose-6-phosphate to fructose-1,6-bisphosphate by 6-phosphofructo-1-kinase. Phosphofructokinase-2/fructose-2,6-bisphosphatase (PFKFB) enzymes synthesize fructose-2,6-bisphosphate, an allosteric activator of 6-phosphofructo-1-kinase and the most potent stimulator of glycolysis. Of all PFKFB isoenzymes, PFKFB3 appears the major producer of intracellular fructose-2,6-bisphosphate in ECs. PFKFB3 is upregulated in ECs under inflammatory conditions and its pharmacological inhibition or gene silencing reduces pathological angiogenesis in response to injury and inflammation [15][16][17]. Previous findings have shown that inhibition of PFKFB3 leads to reduced EC migration and proliferation in vitro. Additionally, sprout number and length of EC spheroids significantly decrease after knocking out PFKFB3 [18].
We have recently reported that the partial glycolysis inhibitor 3PO [3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one] reduces IP angiogenesis and plaque formation [19]. However, the specific role of endothelial PFKFB3 in the context of IP neovascularization and lesion progression remains to be investigated. Therefore, in the present study we used a vein graft procedure in EC-specific conditional PFKFB3 knockout mice on an ApoE −/− background to test whether endothelial PFKFB3 is an important driver of IP angiogenesis and atherosclerotic lesion progression.

Materials and methods
Animals EC-specific conditional PFKFB3 knockout mice (PFKFB3 fl/fl ) were generated by crossbreeding PFKFB3 fl/ fl mice with VE-cadherin (PAC)-Cre ERT2 mice (Cdh5 iCre ) [18]. Resulting mice were crossbred with ApoE −/− mice to generate an ApoE −/− PFKFB3 fl/fl Cdh5 iCre strain. All mice were on a C57BL/6 N background. ApoE −/− PFKFB3 fl/fl C-dh5 iCre mice (male, 6 weeks old) were injected with tamoxifen (0.1 g/kg body weight) for 5 consecutive days to induce PFKFB3 deletion in ECs, termed ApoE −/− PFKFB3 ECKO . ApoE −/− PFKFB3 fl/fl Cdh5 iCre control mice, further referred to as ApoE −/− PFKFB3 fl/fl mice, were injected with corn oil using the same protocol. All animal procedures were conducted according to the guidelines from Directive 2010/63/ EU of the European Parliament on the protection of animals used for scientific purposes. Experiments were approved by the ethics committee of the University of Antwerp (reference number 2017-96).

Vein graft surgery
ApoE −/− PFKFB3 ECKO and ApoE −/− PFKFB3 fl/fl mice were fed a western-type diet (Altromin, C1000 diet supplemented with 20% milkfat and 0.15% cholesterol, #100,171) for 4 weeks (Fig. 1). Next, vein graft surgery was performed as described [5,7,9]. Briefly, thoracal caval veins from donor ApoE −/− PFKFB3 ECKO or ApoE −/− PFKFB3 fl/fl mice were harvested. In the first group, ApoE −/− PFKFB3 fl/fl recipient mice received the caval veins from ApoE −/− PFKFB3 fl/ fl donor mice; in the second group, ApoE −/− PFKFB3 ECKO mice received the caval veins from ApoE −/− PFKFB3 ECKO mice. For each experiment, the right carotid artery of recipient mice was dissected and cut in the middle. On both the Fig. 1 Schematic overview of the experimental design. ApoE −/− PFKFB3 fl/fl Cdh5 iCre mice (6 weeks old) were injected with tamoxifen (0.1 g/kg body weight) for 5 consecutive days to induce PFKFB3 deletion in ECs. Control mice were injected with corn oil using the same protocol. After 2 weeks, mice were fed a western-type diet. Four weeks later, vein graft surgeries were performed. Mice were sacrificed 4 weeks after surgery proximal and distal artery end, a nylon cuff was sleeved and fixated with hemostatic clamps. The artery was everted around the cuffs and ligated with 8.0 sutures. Next, the caval veins were positioned over both cuffs, and ligated. Before surgery, mice were anesthetized with midazolam (5 mg/kg body weight, i.p., Roche), medetomidine (0.5 mg/kg body weight, i.p., Orion) and Fentanyl (0.05 mg/kg body weight, i.p., Janssen). After the procedure, mice were antagonized with atipamezole (2.5 mg/kg body weight, i.p., Orion) and fluminazenil (0.5 mg/kg body weight, i.p., Fresenius Kabi). Buprenorphine (0.1 mg/kg body weight, i.p., MSD Animal Health) was given after surgery to relieve pain. Animals were sacrificed under the aforementioned anesthesia 28 days after the graft procedure, followed by 2 min of in vivo perfusion-fixation.

Metabolic parameters
To determine whether ApoE −/− PFKFB3 ECKO mice exhibit an alteration in glucose metabolism and to characterize the metabolic phenotype, a glucose tolerance test (GTT) and insulin tolerance test (ITT) were done. To perform GTT, mice were fasted for 16 h, injected with a single dose of glucose (1 g/kg body weight, i.p.) and then glucose levels in peripheral blood (from tail) were determined after fixed time intervals (0-30-60-120 min) using a hand-held glucometer (OneTouch Ultra, range 20-600 mg/dL; Lifescan). For ITT, a single insulin dose was injected (Novorapid, 1 U/kg body weight, i.p.) in mice and blood glucose levels were monitored as in GTT. Liver enzymes, total cholesterol, and triglycerides were analyzed with an automated Vista 1500 System (Siemens Healthcare Diagnostics). Insulin and β-hydroxybutyrate in plasma samples were determined with a mouse insulin ELISA kit (80-INSMS-E01, ALPCO) and β-hydroxybutyrate assay kit (ab83390, Abcam), respectively.

Aortic sprouting
An aortic ring assay was performed as previously described [20,21]. In brief, murine thoracic aortas were dissected, cleaned under sterile conditions, transferred to 10 cm culture dishes, and cut into 0.5 mm thick rings with a sterile scalpel. After overnight starvation in serum-free Opti-MEM at 37 °C, ring segments were transferred into wells of a 96-well plate coated with 50 µL of a freshly prepared collagen type I solution (1 mg/mL). The aortic rings remained in Opti-MEM (supplemented with 2.5% fetal bovine serum and antibiotics) in the presence or absence of vascular endothelial growth factor (40 ng/mL, R&D Systems). Medium was replaced every 2 days. On day 6, rings were fixed with 4% paraformaldehyde and stained with von Willebrand factor antibody (PC054, Binding Site) that was added overnight prior to fluorescence microscopy imaging. The number of sprouts was counted for each ring and sprout numbers per ring were averaged for each group and graphed.

Mouse lung EC isolation
In order to check the efficiency of PFKFB3 deletion after tamoxifen injection, primary mouse lung ECs were isolated as previously described [22,23]. Briefly, 4 lungs were harvested, finely minced with scissors, and digested with 1.5 mg/ml collagenase Type I (Sigma-Aldrich #C0130) at 37 °C for 45 min (under gentle agitation). The digested cell suspension was filtered on a 70 µM sterile cell strainer, and spun at 400 g for 10 min. The pellet was resuspended in 2 ml of 0.1% bovine serum albumin and 50 µL magnetic dynabeads (ThermoFisher #11,035) precoated overnight with anti-mouse CD31 (BD Pharmingen #553,370) for EC-positive selection. After 20 min at room temperature under slow rotation, the bead-bound cells were recovered with a magnetic separator and washed five times with DMEM containing 10% fetal bovine serum. Cells were finally resuspended in 10 mL of complete DMEM medium (DMEM containing 20% fetal bovine serum, endothelial cell growth supplement, and antibiotics), seeded onto gelatin-precoated 10 cm plates, and grown for a few days to obtain enough protein material for western blotting.

Western blot analyses
Cells were lysed in an appropriate volume of Laemmli sample buffer (Bio-Rad) containing β-mercaptoethanol (Sigma-Aldrich) and boiled for 5 min. Protein samples were then loaded onto pre-casted Bolt 4-12% Tris-Bis gels (Invitrogen) and after electrophoresis transferred to Immobilon-FL PVDF membranes (Millipore) according to standard procedures. Membranes were blocked for 1 h with Odyssey blocking buffer (LI-COR Biosciences) diluted 1:5 with PBS. After blocking, membranes were probed overnight at 4 °C with primary antibodies diluted in Odyssey blocking buffer, followed by 1 h incubation with IRDye-labeled secondary antibodies at room temperature. Antibody detection was achieved using an Odyssey SA infrared imaging system (LI-COR Biosciences). The intensity of the protein bands was quantified using Image Studio software. The following primary antibodies were used: anti-β-actin (ab8226, Abcam) and anti-PFKFB3 (ab181861, Abcam). IRDye-labeled secondary antibodies (goat anti-mouse IgG, 926-68070, and goat anti-rabbit IgG, 926-32211) were purchased from LI-COR Biosciences.

Statistics
All data are expressed as mean ± SEM. Statistical analyses were performed using GraphPad Prism (version 9) and SPSS (version 25) software. Statistical tests are specified in the figure legends. Differences were considered significant at P < 0.05.

ECs of ApoE −/− PFKFB3 ECKO mice show impaired sprouting in an ex vivo mouse aortic ring assay
Previous findings have shown that inhibition of PFKFB3 leads to a reduction of EC migration and proliferation in vitro [18]. In line with these results, we found that vascular endothelial growth factor-induced sprouting in aortic rings of ApoE −/− PFKFB3 ECKO mice was 65% less as compared to ApoE −/− PFKFB3 fl/fl control mice (Supplemental Figure S1). This observation indicates a direct effect of endothelial PFKFB3 on angiogenesis.

PFKFB3 deficiency in ECs does not cause metabolic changes in adult mice
To evaluate whether EC-specific PFKFB3 gene deletion affects general metabolism, we analyzed plasma samples of ApoE −/− PFKFB3 fl/fl and ApoE −/− PFKFB3 ECKO mice after 4 weeks of western-type diet. No differences were observed in liver enzymes (γ-glutamyltransferase, alanine transaminase, alkaline phosphatase) and insulin (Table 1), indicating that PFKFB3 deletion in ECs has no obvious systemic side effects. Moreover, there were no differences in glucose and insulin tolerance tests after 12 weeks of western-type diet (Supplemental Figure S2), indicating that glucose absorption and insulin receptor sensitivity is normal in both strains. Also, body weight, cholesterol levels, and plasma triglycerides were not statistically different between both groups of mice (Table 1). In addition, levels of the ketone-body β-hydroxybutyrate were not changed in ApoE -/-PFKFB3 ECKO mice as compared to ApoE −/− PFKFB3 fl/fl mice (Table 1). These observations suggest that endothelial PFKFB3 deletion does not induce a metabolic switch from glucose to fatty acid-derived ketones and does not cause major side effects in ApoE −/− mice.

PFKFB3 deficiency in ECs inhibits neovascularization in vein graft lesions
Both ApoE −/− PFKFB3 fl/fl and ApoE −/− PFKFB3 ECKO mice displayed an intact endothelium 28 days after vein graft surgery (black arrows Fig. 3A, B). Circular-oriented VSMCs were seen  Fig. 3A, B). Foam cells, a small necrotic core, and cholesterol crystals were found particularly in vein grafts of ApoE −/− PFKFB3 fl/fl mice near the luminal side (asterisks, Fig. 3A, B). Furthermore, neovessels were found through the vein graft wall, predominantly in ApoE −/− PFKFB3 fl/fl mice. The newly formed vessels were often leaky as extravasated erythrocytes were found near and outside these microvessels (Fig. 3A, B).
As shown by a CD31 staining, the number of microvessels per lesion was reduced by 62% in ApoE −/− PFKFB3 ECKO versus ApoE −/− PFKFB3 fl/fl mice (Fig. 4A, B, E). IP microvessels were further characterized with α-SMA staining to detect the presence of a VSMC layer around the microvessels (Fig. 4C, D). VSMC coverage was observed in vein grafts from both ApoE −/− PFKFB3 ECKO and ApoE −/− PFKFB3 fl/fl mice (Fig. 4F). There was no statistical difference between the two groups (P = 0.05), although a trend of higher VSMC coverage in ApoE −/− PFKFB3 ECKO mice was observed. The organization of the microvessel network was further assessed by immunofluorescence confocal microscopy (Fig. 5). Staining of graft lesions with CD31 antibody revealed IP microvessels that were often covered by VSMCs as demonstrated by α-SMA staining (Fig. 5). These findings suggest that IP microvessels are able to reach a significant level of structural and multicellular complexity. Anti-TER-119 staining showed more erythrocyte infiltration into the lesions of ApoE −/− PFKFB3 fl/fl mice as compared to lesions of ApoE −/− PFKFB3 ECKO mice (Fig. 6A-C), suggesting increased IP vessel leakage in ApoE −/− PFKFB3 fl/fl mice.

PFKFB3 deficiency in ECs reduces lesion size of vein grafts
Twenty-eight days after vein graft surgery, the lesion area in the graft was decreased by 36% in ApoE −/− PFKFB3 ECKO mice as compared to ApoE −/− PFKFB3 fl/fl mice (Fig. 7A,  B), suggesting a significant role of endothelial PFKFB3 in lesion formation and/or progression. Moreover, vein graft stenosis, lesion thickness, and necrotic area were significantly reduced in vein grafts of ApoE −/− PFKFB3 ECKO mice (Fig. 7C-E).

PFKFB3 deficiency in ECs decreases macrophage infiltration
Macrophage accumulation was mainly observed underneath the luminal EC or around the endothelium of microvessels in vein graft lesions of ApoE −/− PFKFB3 ECKO mice. In vein graft lesions of ApoE −/− PFKFB3 fl/fl mice, macrophages appeared much more diffuse in the vascular wall (Fig. 8A). Quantification of the macrophage infiltration showed a significant decrease in ApoE −/− PFKFB3 ECKO mice (Fig. 8B). Analysis of vascular cell adhesion molecule-1 (VCAM-1) expression at the luminal side of the vein graft did not reveal significant differences in ApoE −/− PFKFB3 ECKO mice as compared to ApoE −/− PFKFB3 fl/fl mice (Supplemental Figure S3 A-C).

RT-PCR reveals an elevated M2 macrophage signature in vein grafts of ApoE −/− PFKFB3 ECKO mice
To obtain more insight into the mechanisms underlying the observed phenotype in vein grafts of ApoE −/− PFKFB3 ECKO mice, real-time RT-PCR reactions were performed (Fig. 9). Neither the expression of VCAM-1 and intercellular adhesion molecule-1 (ICAM-1) nor the expression of hypoxia markers C-X-C chemokine receptor 4 (CXCR4) and vascular endothelial growth factor A (VEGFA) significantly changed in vein grafts of ApoE −/− PFKFB3 ECKO mice as compared to vein grafts of ApoE −/− PFKFB3 fl/fl mice. mRNA levels of M1 macrophage markers CD38 and G-protein coupled receptor 18 (GPR18) did not change either. However, mRNA expression of the M2 markers Early growth response protein 2 (Egr2) and Arginase-1 (Arg1) were elevated in vein grafts of ApoE −/− PFKFB3 ECKO mice.

Discussion
IP angiogenesis is frequently observed inside human vein graft lesions and is recognized as a contributing factor of plaque vulnerability [3,6,10,11,24]. In the present study, we crossed EC-specific PFKFB3 knockout mice with ApoE −/− mice to investigate the role of EC glycolysis modulation in vein graft IP angiogenesis. To our knowledge, this is the first study using a conditional EC-specific PFKFB3 knockout mouse in the context of advanced atherosclerosis.
First of all, we did not observe any adverse effects or changes in general metabolism after PFKFB3 deletion in ECs. Circulating liver enzymes, blood glucose, insulin, and total cholesterol were not affected. Also glucose and insulin tolerance tests were similar in ApoE −/− PFKFB3 fl/fl versus ApoE −/− PFKFB3 ECKO mice. Ketone-body β-hydroxybutyrate was not changed in both groups. These findings suggest that PFKFB3 deletion in ECs does not lead to severe side effects or to a major metabolic switch in ApoE −/− mice. In line with these findings, recent evidence indicates that the immune cell distribution in peripheral blood and lymphoid organs is unaffected after systemic treatment of mice with PFKFB3 inhibitor PFK158 [25].
Next, and in line with previous studies, we found that PFKFB3 deletion impaired vessel sprouting from aortic rings. Along these lines, ApoE −/− PFKFB3 ECKO mice showed a significantly reduced number of microvessels in vein graft lesions, albeit without a clear impact on hypoxia markers such as CXCR4 and VEGFA. Moreover, IP microvessels in ApoE −/− PFKFB3 ECKO showed a trend toward higher VSMC coverage and less leakage of erythrocytes inside the graft lesion. These findings are consistent with recent in vitro and in vivo observations showing that PFKFB3 inhibition reduces VE-cadherin endocytosis and promotes normalization of the endothelial barrier by tightening EC junctions [16]. Apart from PFKFB3 inhibition, we recently reported that also atorvastatin promotes IP vessel maturation in vein grafts by preventing VE-cadherin internalization and increasing pericyte coverage [9].
We also observed a reduction in plaque size in vein grafts of ApoE −/− PFKFB3 ECKO mice, which suggests that PFKFB3 may play a direct role in plaque progression. Such compelling possibility corresponds with data from in vitro studies showing that PFKFB3 is linked to pro-inflammatory signaling of ECs in response to blood flow shear stress. Indeed, turbulent blood flow in atheroprone regions leads to inhibition of Krüppel-like Factor 2 activity, which correlates with PFKFB3 upregulation, increased EC glycolysis, and inflammatory activation [26]. The importance of PFKFB3 in plaque progression has also been suggested by a recent study showing a positive correlation between PFKFB3 expression and an unstable plaque phenotype in both carotid and coronary plaques in humans [25]. Furthermore, administration of the PFKFB3 inhibitor PFK158 in mice led to a reduction in advanced plaques with a vulnerable phenotype and an increase in plaque stability [25]. The reduction of IP angiogenesis in ApoE −/− PFKFB3 ECKO mice, as described in this study, is also in line with previous findings in our group showing decreased IP angiogenesis following administration of the glycolysis inhibitor 3PO in a mouse model of advanced atherosclerosis [19]. In this perspective, it is worth mentioning that 3PO significantly reduced initiation of plaque formation in a preventive study design.
Most interestingly, we detected a reduction in the percentage of macrophage infiltration in vein graft lesions of ApoE −/− PFKFB3 ECKO mice. This finding is in agreement with the presence of crosstalk between EC metabolism and macrophages in pathological conditions, as previously reported [27,28]. For example, in tumor settings a metabolic competition for glucose between EC and macrophages reduces EC hyperactivation and prevents abnormal vessel leakage [29]. However, both immunohistochemical stains and RT-PCR analysis of endothelial adhesion molecules did not reveal significant differences in ApoE −/− PFKFB3 ECKO mice as compared to ApoE −/− PFKFB3 fl/fl mice. PFKFB3 inhibition also abolishes the inflammatory response caused by lipoprotein(a) with concomitant attenuation of transendothelial monocyte migration in atherosclerotic plaques [30]. It is therefore possible that the observed reduction in macrophage infiltration in vivo is in part due to an improved restoration of EC junctions after PFKFB3 deletion, as mentioned above. Interestingly, vein grafts of ApoE −/− PFKFB3 ECKO mice revealed an elevated M2 macrophage signature. In particular the canonical M2 macrophage marker Arg-1 was strongly upregulated. Changes in CD38 and Gpr18, which are exclusive M1 markers [31] could not be demonstrated. M2 macrophages resolve inflammation and are usually associated with lesion regression, which corresponds to the improved phenotype of ApoE −/− PFKFB3 ECKO vein grafts. However, it is presently unclear what drives M2 polarization in our experimental settings. We previously reported that glycolysis inhibitor 3PO promotes a macrophage M2 phenotype by stimulating the expression of Arg1 and the exclusive M2 marker Egr2 [19]. Given that VE-cadherin-Cre (used to delete PFKFB3 in ECs) may also show recombinase activity in hematopoietic cells [32] we cannot rule out the possibility that also macrophages are (partially) PFKFB3 deficient. Importantly, myeloid knockdown of PFKFB3 does not affect the size and composition of plaques in atherosclerotic mice [33], suggesting that endothelial PFKFB3 deficiency and not myeloid PFKFB3 deficiency remains a key condition to suppress lesion formation and to obtain an improved lesion phenotype.
Altogether, our findings indicate that endothelial PFKFB3 plays a critical role in IP angiogenesis and lesion progression, and that PFKFB3 inhibition is a promising approach to prevent plaque development and to reduce the complications of vein bypass grafting.