Endothelial ACKR3 drives atherosclerosis by promoting immune cell adhesion to vascular endothelium

Ackr3-floxed (Ackr3^fl/fl) mice provided by ChemoCentryx, Inc. (Mountain View, CA) were crossed with Apolipoprotein E deficient (Apoe^−/−) mice to generate Ackr3^fl/flApoe^−/− mice (> 10 generations). For tamoxifen-inducible, arterial endothelial cell-specific deletion of Ackr3, Ackr3^fl/flApoe^−/− mice were crossed with BmxCreER^T2 expressing mice (BmxCre) kindly provided by Dr. R. Adams (MPI Münster, Germany). Smooth muscle cell (SMC) specific deletion of Ackr3 was achieved by crossing Ackr3^fl/flApoe^−/− mice with SmmhcCreER^T2 expressing mice (SmmhcCre) kindly provided by Dr. S. Offermanns (MPI Bad Nauheim, Germany). The knockouts were induced with daily tamoxifen injections (Sigma; 1.5 mg per 20 g body weight, dissolved in corn oil) for 5 consecutive days, followed by a 4 or 12-week WD containing 21% fat and 0.15–0.2% cholesterol (Sniff diets). Tail samples were used for ACKR3 knockout genotyping in endothelial- and SMC-ACKR3-deficient mouse models with the following primers: forward primer: 5’ GAG TCA ATT GAG TGG GCA AGG 3’ and reverse primer: 5’ GCT ACA TTG CTT TCT TGA AGA AACC 3’ (Supp. Fig. S1A, B). For the bone marrow transplantation study (see details below), Cre^ERT2Ackr3^fl/flApoe^−/− mice (expressing the Cre ubiquitously) were used as donor mice, whereas Apoe^−/− mice were used as recipient mice. Cre^ERT2Ackr3^fl/flApoe^−/− mice were generated as follows: Cre^ERT2 (Gt(ROSA)26) mice were purchased from Taconic Laboratories and bred with Ackr3^fl/flApoe^−/− mice. In these mice, ACKR3 knockout genotyping was performed in spleen samples due to high blood content in this tissue (Supp. Fig. S1C). All mice were on a C57BL/6 J background. All animals were bred and housed in the local animal facility under specific pathogen free (SPF) conditions. Prior to the start of the WD, all mice were fed a normal chow diet. All animal experiments were approved by the local ethical committee (Regierung von Oberbayern, Sachgebiet 54, Germany; Az. 55.2.1.54-2532-177-2016 and ROB-55.2-2532.Vet_02-18-96).

All mice were irradiated twice in an X-ray machine (FAXITRON CP-160) with the program as follows: 5 Gy; kV160; mA 6.3; time 8.5; level 7. Mice were, after irradiation, treated with 1:1 neomycin (100 mg/ml in sterile dH2O) and polymyxin (0.1 g/ml in sterile dH2O) antibiotic solution for 4 weeks in their drinking water. After the irradiation, recipient Apoe^−/− mice were intravenously injected with bone marrow cells collected from Cre^ERT2−Ackr3^fl/flApoe^−/− and Cre^ERT2+Ackr3^fl/flApoe^−/− mice. Bone marrow cells from donor mice were prepared as follows: femurs and tibia of donor mice were collected without fractures and the bones were sterilized in 70% ethanol for 30 s and washed in PBS. The marrow of the bones was extracted under sterile conditions and a single cell suspension was prepared in sterile PBS by filtering the bone marrow with a 70 µm cell strainer. Syringes were prepared with 100uL of cell suspension (2.5 × 10⁶ cells per recipient mouse). After 4 weeks of recovery time, mice were treated with tamoxifen (1.5 mg per 20 g body weight, dissolved in corn oil) for 5 consecutive days. Mice were then subjected to 12 weeks of WD containing 21% fat and 0.15–0.2% cholesterol (Sniff diets).

Atherosclerotic plaque samples obtained during carotid endarterectomy were collected in the Maastricht pathology tissue collection (MPTC) in line with the Dutch code for proper secondary use of human tissue (http://​www.​fmwv.​nl) and the local Medical Ethical Committee (protocol number 16-4-181). This study conforms to the Declaration of Helsinki, all participants have given informed written consent prior to the inclusion. Sections were stained immunohistochemically with von Willebrand factor (Abcam) and CXCR7 (ACKR3) (ThermoFisher) antibodies.

Atherosclerotic lesion sizes were examined in plaque-prone areas such as the aortic root, arch and the aorta. Heart samples were fixed in 4% PFA and embedded in Tissue-Tek O.C.T. compound (Sakura) for cryo-sectioning (4 µm). Aortic arch samples were fixed likewise and embedded in paraffin for sectioning (4 µm). Aortic root and arch slides were stained with hematoxylin and eosin for lesion size assessments on ImageJ Software (three sections per mouse). Aortas were prepared en face and stained with Oil-Red-O for the quantification of lipid-laden lesion sizes on ImageJ Software. Plaque stability was examined in the Masson’s trichrome-stained aortic roots via Leica Software. Cellular composition of the lesions (macrophages, SMCs) as well as adhesion molecules (ICAM-1, 3E2, BD Biosciences) were studied via immunohistochemical stainings on the aortic root samples. Macrophages were identified with anti Mac-2 (M3/38, Cedarlane) antibody, SMCs were identified with anti-SMC (1A4, Dako) antibody and nuclei were stained with Hoechst (Invitrogen) solution (1:2000 in PBS) for 5 min or labeled with ProLong™ Diamond Antifade Mountant with DAPI (Thermofisher). NF-kB expression in atherosclerotic endothelial cells was assessed via immunohistochemical stainings on the aortic root samples with van Willebrand factor (vWF) (Abcam) and phospho-NF-kB p65 (SantaCruz) antibodies.

Leukocyte adhesion onto the ACKR3-deficient endothelium was analyzed by ex vivo perfusion of fluorescently labeled leukocytes on carotid arteries collected from control and BmxCre mice employing 2-photon excitation microscopy as described [58]. Isolated carotid arteries are mounted in vessel chambers and isolated and cell tracker green-labelled bone-marrow derived leukocytes are perfused due to the vessel. After washing, mounted carotid arteries were visualized using a LeicaSP5IIMP two-photon laser scanning microscope with a Ti: sapphire laser (Spectra Physics MaiTai Deepsee) tuned at 800 nm and a 20 × NA1.00 (Leica) water dipping objective. Spectral detection was performed using Hybrid Diode detectors (n = 4) tuned for maximum intensity of the (auto) fluorescence signal of the arterial wall and clear detection of cell tracker green: second harmonics generation of collagen (395–405 nm), autofluorescence; (470–490 nm), cell tracker green + autofluorescence; (500–550 nm), autofluroescence (570–600 nm). Adhered celltracker-positive leukocytes were counted manually by three observers in 3D sections of the arterial wall. 3D image processing was performed using LASX software including 3D analysis plugin (Leica).

Intravital microscopy was performed in the bifurcation of the left carotid artery by means of epifluorescence microscopy. The right jugular vein was canulated with a catheter for antibody and dye injection. After exposure of the left carotid artery, antibodies (1 μg) to CD11b (M1/70, Biolegend), Ly6C (HK1.4, Biolegend) and Ly6G (1A8, Biolegend) were sequentially administered to label myeloid cells, neutrophils and classical monocytes, respectively. Recordings were made 3 min after injection of each antibody. The diameters of the external carotid arteries were 268.8 ± 34.4 µm in the control group and 267.4 ± 37.1 µm in the endothelial ACKR3 deficient group. Intravital microscopy was performed using an Olympus BX51 microscope equipped with a Hamamatsu 9100-02 EMCCD camera and a ×10 saline-immersion objective. For image acquisition and analysis, Olympus Cell-R software was used.

Blood leukocyte tracking was performed according to the protocol described in Winter et al. [61]. Briefly, mice were injected with a rat anti-CD45 antibody (30-F11, BioLegend) and killed 2 h post-injection. Cryo-sectioned aortic root samples from the mice were then stained with an anti-rat secondary antibody to trace back the pre-labelled circulating immune cells which infiltrated into the aortic root lesions in the 2-h timeframe.

0.5% Evans blue (EVB) was prepared in 0.9% saline solution and the solution was filter-sterilized. Mice were injected with the EVB solution 40 mg/kg and sacrificed 30 min post-injection. Mice were then perfused with sufficient phosphate-buffered saline (PBS) and following organs were collected to analyze the retained amount of EVB: lung, aorta, aortic arch. Aortic arches were fixed with 4% paraformaldehyde (PFA) for 30 min and placed on objective slides to be imaged (EVB excitation peaks: 470 and 540 nm, emission peak at 680 nm). Tilescan z-stacks of the whole aortic arches were taken with a Leica SP8 3X confocal microscope tube for efficient EVB detection and processed using Imaris 8.4 (Oxford instruments). The endothelial Evans blue–positive signal was quantified as a volume after 3D segmentation based on appropriate fluorescence intensity thresholding. Aortas and lungs were weighed, air dried for half a day, fixed with formamide and cut into small pieces. The organs were then incubated at 56 °C shaking for 24 h to release the retained EVB. Optical density (OD) of the formamide solutions containing the EVB was then measured at 620 nm and normalized to tissue weights.

Cholesterol and triglyceride levels were quantified in EDTA-plasma samples using enzymatic assays (c.f.a.s. cobas, Roche Diagnostics) according to the manufacturer’s protocol.

Human Primary Coronary Artery Endothelial Cells (CC-2585), EGM-2 Bulletkit medium (CC-3162) and ReagentPack Subculture Reagents (CC-5034) were purchased from Lonza. The complete medium contained the following: 2% FBS, 0.2 mL hydrocortisone, 2 mL hFGF-B, 0.5 mL VEGF, 0.5 mL R3-IGF-1, 0.5 mL ascorbic acid, 0.5 mL hEGF, 0.5 mL hEGF, 0.5 mL GA-1000 as well as 5 mL penicillin–streptomycin antibiotic mix. Cells were kept at 37 °C and 5% CO₂ under sterile conditions in a humidified incubator. Subculturing and medium refreshing were performed according to the manufacturer’s protocol. For ACKR3 silencing studies, cells were transfected with either 30 nM negative control siRNA (Silencer™ Negative Control No. 1 siRNA, AM4611 ThermoFisher) or 30 nM ACKR3 siRNA (Silencer validated siRNA CXCR7, AM51331 Ambion by Life Technologies) with the following sequence: Sense GGAZGACACUAAUUGUUAGtt and Antisense CUAACAAUUAGUGUCAUCCtt. Transfection was performed via siPORT™ NeoFX™ Transfection Agent according to the manufacturer’s protocol (Invitrogen by Life Technologies). Cells were transfected for 48 h and the transfection efficiency was evaluated by droplet digital PCR. Cells were treated with 10 ng/mL TNFα 2–4 h before evaluating the impact of ACKR3 silencing. ACKR3-transfected HEK cells were kindly donated by Prof. Alexander Faussner and treated with tetracycline for 24 h to induce ACKR3 expression and with 10 ng/mL TNFα for 1 h before evaluating the impact of ACKR3 overexpression. NF-kB p65 phosphorylation was evaluated via ELISA (see above) on TNFα-stimulated ACKR3-induced HEK cells as well as on ERK inhibitor (SCH772984, Selleckchem, 2 µM) and Akt inhibitor (MK-2206-2HCl, Selleckchem, 20 µM) treated ACKR3 induced HEK cells. Static adhesion assay with THP-1 cells was performed as follows: HCAECs were transfected with 30 nM negative control siRNA or 30 nM ACKR3 siRNA for 48 h. In addition, negative control siRNA-transfected cells were also treated with ERK inhibitor (2 µM) and Akt inhibitor (10 µM) for 2 h prior to 10 ng/mL TNFα treatment for 2 h. THP-1 cells were stained with calcein for 30 min and then co-incubated with pre-treated HCAECs for 30 min after washing. HCAECs were then washed twice and the fluorescence measurement was performed at 485/535 nm wavelength with a TECAN infinite 200pro plate reader.

Total RNA isolation from cell culture and mouse aortic arch samples was performed by commercially available RNA isolation kit from Zymoresearch (Direct-zol microprep kit) according to the manufacturer’s protocol. The quality (A₂₆₀/A₂₈₀) and the quantity (ng/µL) of the RNA was measured by NanoPhotometer N60/N50 (Implen). A ratio of ~ 2 for A₂₆₀/A₂₈₀ was accepted as good quality RNA.

RNA samples were diluted to the same concentration and the cDNA synthesis was performed via the commercially available iScript cDNA synthesis kit from Bio-Rad according to the manufacturer’s protocol.

PCR was performed on QX200 Droplet Digital PCR (ddPCR™) system from Bio-Rad. 20µL reaction mixes were prepared using 10μL of 2X ddPCR SuperMix for probes (No dUTP) (Bio-Rad), 1µL of 20X FAM-labeled primer/probe for the target gene, 1µL of 20X VIC-labeled primer/probe for the housekeeping gene, RNase-/DNase-free water and cDNA sample. Taqman primers were purchased from Thermofisher Scientific. Droplet generation was performed in the QX200 droplet generator (Bio-Rad) by adding 20µL reaction mix and 70µL droplet generation oil for probes (Bio-Rad) onto DG-8 cartridges covered with gaskets (Bio-Rad). 42µL of the droplet solution (containing up to 20,000 droplets) was transferred to the appropriate PCR plate (Bio-Rad) which was then sealed with a piercing foil using the PCR plate sealer (Bio-Rad). Cycling was performed in the ddPCR cycler with the following conditions: 10 min at 95 °C (enzyme activation), 30 s at 94 °C (denaturation) and 1 min at 60 °C (annealing/extension) for 40 cycles, 10 min at 98 °C (enzyme deactivation). The PCR plate was then proceeded to the droplet reader (Bio-Rad). Analysis was performed on QuantaSoft software (Bio-Rad).

Whole blood was collected from mice in EDTA-buffer tubes and red blood cell lysis was performed prior to stainings. Hematopoietic cell populations in blood samples were stained as follows: anti-CD45 (30-F11, eBioscience), anti-CD115 (AFS98, eBioscience), anti-Gr1 (RB6-8C5, eBioscience), anti-CD11b (M1/70, eBioscience), anti-B220 (RA3-6B2, eBioscience) and anti-CD3 (145-2C11, eBioscience). Cell populations were gated and analyzed as follows using FlowJo Software: leukocytes (CD45⁺), neutrophils (CD45⁺CD115⁻Gr1^high), monocytes (CD45⁺CD11b⁺CD115⁺), T cells (CD45⁺CD3⁺) and B cells (CD45⁺B220⁺). For cell culture experiments, the cells were washed and trypsinized at the end of the ACKR3 silencing treatments (72 h) to be collected as single cell suspensions. The samples were stained as follows: anti-ICAM (HA58, eBioscience), anti-VCAM (429, BD Biosciences) and anti-E-selectin (10E9.6, BD Biosciences). All samples were analyzed with FACS Canto II together with FACSDiva software (BD Biosciences).

Cell culture samples were lysed with M-PER™ Mammalian Protein Extraction Reagent (from ThermoFisher Scientific) according to the manufacturer’s protocol. Total protein levels in the lysates were measured via a commercially available BCA assay kit (Pierce™ BCA™ Protein-Assay from ThermoFisher Scientific) according to the manufacturer’s protocol.

Silenced and control cell culture samples were analyzed for the modifications in the MAPK pathway with the commercially available Human/Mouse MAPK Phosphorylation Array kit (AAH-MAPK-1-8) from RayBiotech, according to the manufacturer’s protocol.

Equal amounts of protein samples obtained from tissue homogenates were mixed with Novex™ NuPAGE™ LDS sample buffer (4X) and boiled at 95 °C for 5 min. Lysates were then loaded onto 12% Mini-PROTEAN^® TGX™ Precast Protein Gels (Bio-Rad) along with Precision Plus Protein™ WesternC™ blotting standard (Bio-Rad). Western blot was performed with Mini Trans-Blot^® Cell and Criterion™ Blotter according to the manufacturer’s protocol (Bio-Rad). Briefly, protein separation was achieved by gel electrophoresis using running buffer (Tris/glycine/SDS from Bio-Rad) for about 1 h with 100–150 Volts (V). Proteins were then transferred from the gel to the membrane in transfer buffer (Tris/glycine buffer from Bio-Rad with 20% methanol added) for 1 h at 100 V. The membrane was then blocked with 5% bovine serum albumin (BSA) in Tris-buffered saline with Tween 20 buffer (TBST from Bio-Rad) for 1 h. Primary antibody incubation against p-Erk1/2 (Abcam) and p-p65-NFkB (SantaCruz) was carried out overnight at 4 °C. The membrane was then washed with TBST buffer and the HRP-labelled anti-rabbit secondary antibody (from Abcam) incubation was carried out for 1 h at room temperature. The membrane was washed again with TBST buffer and developed with SuperSignal™ West Pico PLUS Chemiluminescent Substrate (from Thermofisher). After the signal for the phospho-proteins was captured, the membrane was again washed in TBST, stripped for 30 min with Restore™ Western Blot Stripping Buffer (from Thermofisher) and blocked for 1 h with 5% BSA solution. The membrane was then incubated with antibodies against t-Erk1/2 (Abcam) and t-p65-NFkB (SantaCruz) for 1 h at room temperature and this was followed by again 1-h incubation of secondary antibody. The signal was captured as explained above. Analysis of phospho- and total protein signals was performed on ImageJ and the phospho-protein signal was normalized by dividing it to total protein signal.

All data are expressed as mean ± SEM. Statistical analyses were performed using GraphPad Prism version 7.0 or higher (GraphPad Software, Inc.). Outliers were identified with ROUT = 1 and normality of the data was tested via the D’Agostino–Pearson omnibus normality test. Statistical significance was tested via unpaired Student’s t test with Welch correction for normally distributed data and Mann–Whitney U test for non-normally distributed data. A result of < 0.05 for p value was considered statistically significant. *p < 0.05, **p < 0.01, ***p < 0.001.

Vascular cells, more specifically endothelial (EC) and smooth muscle (SMC) cells are described to express ACKR3 [5, 25]. Furthermore, we could also reveal ACKR3 expression in ECs of human atherosclerotic lesions (Fig. 1A). To study the role of vascular ACKR3 in atherosclerosis we crossbred Ackr3^fl/fl mice with BmxCreER^T2 (arterial EC specific) and SmmhcCreER^T2 (SMC specific) expressing mice. To establish hematopoietic ACKR3 deficient animals (htACKR3) we transplanted bone marrow from animals with a systemic ACKR3 deficiency (UniCre^ERT2 Ackr3^fl/fl) into Apoe^−/− mice. All mice used were Apoe^−/− background. Mice lacking endothelial ACKR3 (EC-ACKR3) were subjected to 4-week and 12-week WD in order to evaluate early onset and advanced atherosclerotic lesion formations (aortic root and arch), respectively (Fig. 1B, C). Analyses revealed that EC-ACKR3 deficiency significantly decreased lesion sizes in the aortic roots of mice after 4 weeks of WD (Fig. 1D, E) and significantly reduced lesion sizes in both aortic roots and arches after 12 weeks of WD (Fig. 1F–I). In SMC-specific and hematopoietic ACKR3 deficient mice (after 12 weeks WD), however, we observed no differences in atherosclerotic lesion sizes in the aortic roots, arches and the aorta (Supp. Fig. S2A-N). Further analyses of the advanced lesions disclosed that EC-ACKR3 deficient mice had significantly less macrophages within their lesions (Fig. 1J, K) while the macrophage content of SMC-specific and hematopoietic ACKR3 deficient mice remained unchanged (Supp. Fig. S3A-C). SMC content was unaffected in all three models (Supp. Fig. S3D-F), whereas lesional collagen content was significantly increased in SMC-ACKR3-deficient (Supp. Fig. S3G) and EC-ACKR3-deficient (Fig. 1L, M) lesions but not in hematopoietic ACKR3-deficient lesions (Supp. Fig. S3 H, I). Necrotic core size in EC-ACKR3-deficient lesions was reduced (Fig. 1N). Of note, lesional SMC content correlated significantly with the collagen density (Supp. Fig. S4A), suggesting that SMCs may be responsible for the enhanced plaque stability through collagen production. Plasma lipid levels did not differ between control and EC-ACKR3 mice (Table 1) as well as in control, SMC-ACKR3 and hematopoietic ACKR3 deficient mice (data not shown). Interestingly, mice lacking EC-ACKR3 developed leukocytosis (Table 1). While CXCL12 levels did not differ in the plasma (Supp. Fig. S4B) between control and EC-ACKR3-deficient mice, circulating CXCL12 levels correlated significantly with atherosclerotic lesions sizes in the control mice, whereas this correlation was lost in the absence of endothelial ACKR3, suggesting that ACKR3–CXCL12 interactions drive atherosclerosis (Fig. 1O). Hence, EC-specific loss of ACKR3 clearly protects against atherosclerosis under hyperlipidemic conditions.

Since our experimental model involves modification of the vascular endothelium through a receptor knockout, we hypothesized that endothelial ACKR3 may be involved in leukocyte extravasation, which may explain the changes in the lesional macrophage content between control and EC-ACKR3-deficient mice. To test if leukocyte infiltration into the lesions is affected by EC-ACKR3 deficiency, we followed the protocol described in Winter et al. [61]. Briefly, 4-week WD-fed control and EC-ACKR3-deficient mice were injected intravenously with an anti-CD45 antibody and the presence of this antibody was tracked in the atherosclerotic lesions as an indicator of leukocyte infiltration (Fig. 2A). We observed significantly less infiltrated leukocytes in the lesions of EC-ACKR3-deficient mice (Fig. 2B, C), suggesting that endothelial ACKR3 is involved in the regulation of leukocyte extravasation. Following these findings, we hypothesized that EC-ACKR3 may mediate leukocyte infiltration by affecting endothelial permeability. The permeability of the endothelial barrier is a critical factor impacting the trafficking of blood-borne elements into the sub-endothelial space. To investigate whether EC-ACKR3 deficiency leads to ‘leakiness’ of the vascular endothelium, we injected EC-ACKR3-deficient and control animals with EVB and quantified the amount of EVB that penetrated through the endothelium (Fig. 2D). No significant differences were observed in EVB quantities in the aortic arches (Fig. 2E, F), aortas (Fig. 2G) or in the lungs (Fig. 2H) of the control and EC-ACKR3 deficient mice, demonstrating that EC-ACKR3 does not affect vascular integrity.

Next, the potential impact of EC-ACKR3 on leukocyte–endothelium adhesion was tested to elucidate EC-ACKR3-mediated decrease in arterial immune cell invasion. To this end, we performed an ex vivo arterial perfusion assay (Fig. 3A). Viable carotid arteries from control and EC-ACKR3-deficient mice were collected and perfused ex vivo with fluorescently labeled leukocytes. Leukocyte adhesion to arteries with endothelial ACKR3 deficiency was markedly decreased suggesting that ACKR3 is involved in the regulation of the endothelial adhesion processes (Fig. 3B, C). To further confirm our findings, we tested the impact of endothelial ACKR3 on vascular adhesion in vivo via intravital microscopy (Fig. 3D). Indeed, deficiency of EC-ACKR3 strongly limited the adhesion of leukocytes, which was tested in different subsets: myeloid cells (Fig. 3E), classical monocytes (Fig. 3F) and neutrophils (Fig. 3G) were investigated individually and all subsets were observed to adhere significantly less to the arteries of the EC-ACKR3 deficient mice.

In order to understand how EC-ACKR3 affected endothelial adhesion in mice, we examined adhesion molecule expression in atherosclerotic lesions by immunohistochemistry and observed significantly less intracellular adhesion molecule-1-positive (ICAM-1 +) cells in the aortic root lesions of EC-ACKR3-deficient mice (Fig. 4A, B). As ACKR3 is also expressed in human atherosclerotic endothelium (Fig. 1A), we sought to investigate whether ACKR3 can regulate adhesion molecule expression in human endothelial cells as well. In order to study this potential human link, we silenced ACKR3 in human coronary artery endothelial cells (HCAECs) (Fig. 4C, D; Supp. Fig. S5A, B) and investigated its impact on adhesion molecule expression upon TNF-α stimulation. Flow cytometry analysis revealed that silenced HCAECs express significantly less ICAM (Fig. 4C, E) and vascular cell adhesion molecule VCAM (Fig. 4F) compared to control samples. In line with this finding, expression of these adhesion molecules correlated significantly with the expression of ACKR3 in the HCAECs (Supp. Fig. S5C, D). Although ACKR3 has long been regarded as a non-signaling receptor [6, 40, 57, 60], further research in recent years revealed that ACKR3 can signal through β-arrestin and mitogen activated protein kinase (MAPK) pathways, as well as NF-kB [1, 27, 31, 33, 38, 41, 42, 67]. In order to investigate whether the ACKR3 deficiency dependent decrease in adhesion molecule expression could be due to the downregulation of these pathways, we performed a MAPK phosphorylation antibody array on ACKR3 silenced and control HCAECs. ACKR3 silencing revealed significant downregulation of the key mediators in the MAPK pathway including decreased ERK1/2 phosphorylation (Fig. 4G). We further confirmed reduced ERK1/2 phosphorylation in silenced cells via western blot (Supp. Fig. S5E). Moreover, we and Li et al. have previously shown that ACKR3 can regulate PPAR-γ expression in adipose tissue [18, 32]. Here we show that this is also true for HCAECs as ACKR3 silenced cells showed increased expression of the anti-inflammatory PPAR-γ (Fig. 4H), which has been established to inhibit the expression of ICAM and VCAM when activated [43, 49]. Furthermore, PPAR-γ is known to be a negative regulator of NF-κB [45, 50, 54], which induces the transcription of ICAM and VCAM [29, 52]. In line with this, decreased phosphorylation of the NF-κB p-65 subunit in ACKR3 silenced HCAECs was confirmed via quantitative ELISA analysis (Fig. 4I) as well as western blot (Supp. Fig. S5F). To further confirm these findings in our mouse model, we quantified phospho-NF-kB p-65 expression in atherosclerotic endothelial cells via immunohistochemically stained aortic root lesions of 4-week WD fed mice (Fig. 4J). Our results indicated that EC-ACKR3 deficient mice had significantly less expression of phospho-NF-kB p-65 in their atherosclerotic endothelial cells (vWF +) (Fig. 4K), suggesting that ACKR3 modulates cell adhesion to endothelial cells most likely through NF-kB activation. The impact of ACKR3 on NF-κB activation was also confirmed with ACKR3-transfected HEK cells (Supp. Fig. S5G); cells with induced ACKR3 expression showed significantly higher NF-κB p-65 subunit phosphorylation compared to control cells, whereas the NF-κB phosphorylation was dampened by ERK and Akt inhibitors (Fig. 4L). These results suggest that ACKR3 activates NF-κB through ERK and Akt. The functional role of ERK and Akt was further validated using an adhesion assay in which inhibition of both ERK and Akt dampened THP-1 cell adhesion onto HCAECs in a similar manner as ACKR3 silencing (Fig. 4M). The effects of EC-ACKR3 on ECs in the context of atherosclerosis are summarized in Fig. 5.