The Impact of RIPK1 Kinase Inhibition on Atherogenesis: A Genetic and a Pharmacological Approach

RIPK1 (receptor-interacting serine/threonine-protein kinase 1) enzymatic activity drives both apoptosis and necroptosis, a regulated form of necrosis. Because necroptosis is involved in necrotic core development in atherosclerotic plaques, we investigated the effects of a RIPK1S25D/S25D mutation, which prevents activation of RIPK1 kinase, on atherogenesis in ApoE−/− mice. After 16 weeks of western-type diet (WD), atherosclerotic plaques from ApoE−/− RIPK1S25D/S25D mice were significantly larger compared to ApoE−/− RIPK1+/+ mice (167 ± 34 vs. 78 ± 18 × 103 µm2, p = 0.01). Cell numbers (350 ± 34 vs. 154 ± 33 nuclei) and deposition of glycosaminoglycans (Alcian blue: 31 ± 6 vs. 14 ± 4%, p = 0.023) were increased in plaques from ApoE−/− RIPK1S25D/S25D mice while macrophage content (Mac3: 2.3 ± 0.4 vs. 9.8 ± 2.4%, p = 0.012) was decreased. Plaque apoptosis was not different between both groups. In contrast, pharmacological inhibition of RIPK1 kinase with GSK’547 (10 mg/kg BW/day) in ApoE−/− Fbn1C1039G+/− mice, a model of advanced atherosclerosis, did not alter plaque size after 20 weeks WD, but induced apoptosis (TUNEL: 136 ± 20 vs. 62 ± 9 cells/mm2, p = 0.004). In conclusion, inhibition of RIPK1 kinase activity accelerated plaque progression in ApoE−/− RIPK1S25D/S25D mice and induced apoptosis in GSK’547-treated ApoE−/− Fbn1C1039G+/− mice. Thus, without directly comparing the genetic and pharmacological studies, it can be concluded that targeting RIPK1 kinase activity does not limit atherogenesis.


Introduction
Rupture of vulnerable atherosclerotic plaques leads to acute clinical complications such as myocardial infarction or stroke and accounts for about 85% of cardiovascular deaths [1]. One of the main characteristics of a vulnerable plaque is the presence of a large necrotic core. In almost 90% of ruptured human plaques, the necrotic core comprises >10% of the plaque area while almost 65% of ruptured plaques contain a necrotic core that occupies >25% of the plaque [2]. Recent studies have demonstrated that regulated necrosis, in particular necroptosis, plays a significant role in the formation of a necrotic core during atherosclerotic plaque development [3][4][5][6][7][8][9].
Necroptosis, in contrast to apoptosis, leads to a pro-inflammatory state and can be induced by various stimuli, including oxidized LDL and TNFα [4,10]. Upon TNFα stimulation, the fate of the cell towards survival, apoptotic, or necroptotic pathways is dependent on the ubiquitination and phosphorylation profile of receptor-interacting

Mice
Standard ApoE −/− mice (Jackson Laboratory, Bar Harbor, ME, USA, 002052) were crossbred with C57BL/6 mice containing a phospho-mimetic S25D mutation in the RIPK1 coding sequence that strongly suppresses RIPK1 kinase activity [13]. The resulting ApoE −/− RIPK1 S25D/S25D mice and ApoE −/− RIPK1 +/+ controls (all female, 6-8 weeks old) were fed a western-type diet (WD; C1000 supplemented with 20% milkfat and 0.15% cholesterol, Altromin, Lage, Germany) for 16 weeks to induce plaque formation. Only female mice were used because plaque formation is more severe in females as compared to males in mouse models of atherosclerosis after 16 weeks WD [29,30]. Previous experiments in our research group confirmed that established plaques are present at this timepoint and not much is gained, regarding plaque phenotype, by feeding WD for longer periods in ApoE −/− mice. The animals were housed in a temperature-controlled room with a 12 h light/dark cycle and had free access to water and food. After 16 weeks WD, an overdose of sodium pentobarbital (250 mg/kg, i.p.) was administered and blood samples were collected via the retro-orbital plexus. Plasma levels of total cholesterol were measured using a commercially available kit (Randox Laboratories, Crumlin, UK). Non-responding mice to WD (plasma cholesterol <400 mg/dL) were excluded.
In another series of experiments, female ApoE −/− mice with a heterozygous mutation (C1039G+/−) in the fibrillin 1 (Fbn1) gene were fed either WD or WD supplemented with RIPK1 inhibitor GSK'547 (10 mg/kg BW/day) starting at the age of 6-8 weeks and sacrificed after 20 weeks WD. They were fed WD for 20 weeks because it was previously reported that this induces an advanced plaque phenotype in ApoE −/− Fbn1 C1039G+/− mice [27,28]. The daily dosage of GSK'547 was based on previous food-based GSK'547 studies in mice [26,31]. For diets containing 10 mg/kg BW/day GSK'547, the plasma concentration of the drug varies between 10 ng/mL (at the trough of the eating cycle) and 100 ng/mL (at peak of the eating cycle), which provides 75% to 97% inhibition in vivo, respectively. For WD supplemented with 10 mg/kg BW/day GSK'547, plasma levels of over 100 ng/mL and 150 ng/mL were reported after 2 and 4 weeks, respectively, which is well above the IC50 of 13 ng/mL (32 nM) [31]. Female ApoE −/− Fbn1 C1039G+/− mice are essential because, similar to humans, the Fbn1 mutation frequently causes aortic dissection (and mortality) in male ApoE −/− Fbn1 C1039G+/− mice and not in female ApoE −/− Fbn1 C1039G+/− mice [28].
Blood leukocyte subsets were analyzed by flow cytometry as previously described [32]. All experiments were ethically reviewed by the Ethical Committee of the University of Antwerp (Code 2018-01) and carried out in accordance with European Directive 2010/63/EEC.

Echocardiography
Echocardiograms were performed on anesthetized ApoE −/− Fbn1 C1039G+/− mice (isoflurane, 4% for induction and 2% for maintenance) at the start (0 weeks WD) and at the end of the GSK'547 study (20 weeks WD) using a VEVO2100 (VisualSonics, Amsterdam, The Netherlands), equipped with a 25 MHz transducer. Body temperature was maintained at 36-38 • C and heart rate at 500 ± 50 beats/min. To study arterial stiffness, pulse wave velocity (PWV) was determined in the abdominal aorta using a 24 MHz transducer, as previously described [33]. Briefly, the aortic diameter (D) was measured on 700 frames-persecond B-mode images of the abdominal aorta in EKV mode. Subsequently, the aortic flow velocity (V) was determined by pulse wave Doppler tracing. PWV was then calculated via the ln(D)-V loop method using MathLab v2014 software (MathWorks, Natick, MA, USA).

Histological Analyses
The proximal ascending aorta and brachiocephalic artery were fixed in 4% formaldehyde (pH 7.4) for 24 h, dehydrated overnight in 60% isopropanol, and subsequently embedded in paraffin. The proximal ascending aorta was marked on the distal arch end and the brachiocephalic artery on the distal carotid end to ensure that they were always cut on the proximal side. Serial cross-sections (5 µm) of the proximal parts of the proximal ascending aorta and the brachiocephalic artery were prepared at random for histological analyses. Although plaque formation was also observed in other vascular beds (e.g., aorta arch, carotids, and coronary arteries), we focused on the proximal ascending aorta and brachiocephalic artery because here plaque formation was observed in >90% of the mice, in both ApoE −/− and ApoE −/− Fbn1 C1039G+/− mice. Moreover, plaques in the brachiocephalic artery enter an advanced, human-like stage more rapidly [34,35]. Atherosclerotic plaque size, necrotic core area (defined as acellular areas with a threshold of 3000 µm 2 ), internal elastic lamina (IEL) area, thickness of the tunica media, and degree of stenosis were analyzed on hematoxylin/eosin (H&E)-stained sections. The IEL was manually delineated, the IEL perimeter was measured, and subsequently, the IEL area was calculated using formulas for circle perimeter (2 *pi *r) and area (pi *r 2 ). The external elastic lamina (EEL) was manually delineated, the EEL perimeter was measured, and subsequently, the media thickness was calculated by subtracting the IEL radius from the EEL radius. Collagen and glycosaminoglycan content was determined on Sirius red and Alcian blue-stained sections, respectively. Apoptosis was analyzed using the ApopTag Plus Peroxidase In Situ Apoptosis Kit (Millipore, Burlington, VT, USA, S7101). For immunohistochemistry, the following antibodies were used: anti-Mac3 (BD Pharmingen, San Diego, CA, USA, 550292), anti-α-smooth muscle actin (α-SMA, Sigma-Aldrich, St. Louis, MO, USA, A2547), and anti-cleaved caspase 3 (Cell Signaling, Danvers, TX, USA, 9661). Images were acquired with an Olympus BX43 microscope, which was calibrated for each magnification. Plaque size was measured based on pixels per µm, which was determined during the calibration of the microscope. Per mouse, one section was analyzed. Plaques and tunica media of the vessel wall were manually delineated in ImageJ software (National Institutes of Health, Bethesda, MD, USA) to establish the region of interest (ROI). Further analyses within the ROIs were performed using color thresholding or manual counting (apoptotic cells).

Western Blotting
Tissues were homogenized in RIPA buffer containing protease and phosphatase inhibitors. Protein concentrations were determined using the BCA method. Samples were then 1:1 diluted in Laemmli sample buffer (Bio-Rad, Hercules, CA, USA) containing 5% βmercaptoethanol (Sigma-Aldrich) and heat-denatured for 5 min at 100 • C. Samples were loaded on Bolt 4-12% Bis-Tris gels (Invitrogen, Waltham, MA, USA) and after electrophoresis transferred to Immobilon-FL PVDF membranes (Millipore) according to standard procedures. Subsequently, membranes were blocked for one hour in Odyssey Li-COR blocking buffer. After blocking, membranes were probed with primary antibodies diluted in Odyssey Li-COR blocking buffer followed by 1 h incubation with IRDye-labeled secondary antibodies at room temperature. Membranes were visualized with an Odyssey SA infrared imaging system (Li-COR Biosciences, Lincoln, NE, USA).

Statistical Analyses
Statistical analyses were performed using GraphPad Prism 9 and SPSS software (version 27, SPSS Inc., Chicago, IL, USA). All data were expressed as the mean ± SEM, dots represent n samples from independent experiments or individual mice. Statistical tests are specified in the text and figure legends. Differences were considered significant when p < 0.05.

ApoE −/− Fbn1 C1039G+/− Mice Can Be Used as a Tool to Study Necroptosis in Atherosclerosis
The ApoE −/− Fbn1 C1039G+/− mouse is a recently developed model with a heterozygous mutation (C1039G+/−) in the fibrillin 1 gene (Fbn1), which results in fragmentation of the Because necrotic core formation was not altered in plaques of ApoE −/− RIPK1 S25D/S25D , the expression of apoptosis and necroptosis markers was evaluated in plaques of ApoE −/− RIPK1 S25D/S25D and ApoE −/− RIPK1 +/+ mice. Both immunohistochemical analysis of cleaved caspase 3 and TUNEL labeling did not show any differences in the frequency of apoptosis between both groups ( Figure 2F). Western blot analyses of P-MLKL and RIPK1 did not reveal differences either (data not shown).

ApoE −/− Fbn1 C1039G+/− Mice Can Be Used as a Tool to Study Necroptosis in Atherosclerosis
The ApoE −/− Fbn1 C1039G+/− mouse is a recently developed model with a heterozygous mutation (C1039G+/−) in the fibrillin 1 gene (Fbn1), which results in fragmentation of the elastin fibers in the media of the vessel wall and increased arterial stiffness. Moreover, ApoE −/− Fbn1 C1039G+/− mice develop advanced atherosclerotic lesions with human-like features such as large necrotic cores, intraplaque (IP) neovessels, IP hemorrhages, IP inflammation, and spontaneous plaque rupture leading to myocardial infarction and stroke [27,28]. Because necroptosis in experimental atherosclerosis has so far only been studied in classical mouse models such as the ApoE −/− or LDLr −/− mouse [3][4][5][7][8][9]24], we analyzed the expression of necroptosis proteins in plaques of ApoE −/− and ApoE −/− Fbn1 C1039G+/− mice. To this end, both ApoE −/− and ApoE −/− Fbn1 C1039G+/− mice were fed a WD for 6-24 weeks. Compared to ApoE −/− mice, ApoE Fbn1 C1039G+/− mice developed 50% larger plaques with a significantly larger necrotic core after 24 weeks WD (two-way ANOVA followed by Sidak's post hoc test between genotypes per timepoint, p = 0.03) as compared to ApoE −/− mice ( Figure 3A,B). Necrotic core formation in ApoE −/− Fbn1 C1039G+/− mice started after 6 weeks WD, whereas in ApoE −/− mice the necrotic core was only detectable after 12 weeks WD ( Figure 3B). At baseline (0 weeks WD), very low levels of RIPK1, RIPK3, and MLKL were observed in the atherosclerosis-prone aortic arch of both ApoE −/− and ApoE −/− Fbn1 C1039G+/− mice but the expression levels increased during plaque progression ( Figure 3C). In plaques of ApoE −/− Fbn1 C1039G+/− mice the expression of RIPK3 and MLKL started to increase significantly after 6 weeks WD, in contrast to plaques of ApoE −/− mice where no significant increases were observed compared to 0 weeks WD. Accordingly, plaques of ApoE −/− Fbn1 C1039G+/− mice expressed significantly higher levels of MLKL and RIPK3 as compared to ApoE −/− mice. The expression of RIPK1 was significantly increased after 24 weeks WD in plaques of ApoE −/− Fbn1 C1039G+/− mice and after 12 weeks WD in plaques of ApoE −/− mice. No significant difference in RIPK1 expression was observed between the two genotypes. Because macrophages express RIPK1, RIPK3, and MLKL, the elevated expression levels of these markers may be caused by increased macrophage infiltration during plaque formation. However, a significant increase in the P-MLKL/MLKL ratio was also observed in plaques of both ApoE −/− and ApoE −/− Fbn1 C1039G+/− mice ( Figure 3C), demonstrating that MLKL is actively phosphorylated in the plaques and suggestive of necroptosis initiation. Despite significantly higher levels of MLKL in ApoE −/− Fbn1 C1039G+/− mice as compared to ApoE −/− mice, no significant difference in the P-MLKL/MLKL ratio was observed.

Discussion
Despite recent findings showing that RIPK1 is a central driver of inflammation in atherosclerosis [4,7,8], the role of RIPK1 in the progression and destabilization of atherosclerotic plaques is not straightforward and is complicated by its dual nature, namely a scaffolding function (regulating pro-survival signaling and inflammatory gene expression) versus kinase activity (promoting cell death). In the present study, RIPK1 S25D/S25D mice containing an S > D mutation in the kinase domain of RIPK1 were crossbred with ApoE −/− mice to investigate the role of RIPK1 kinase activity in experimental atherosclerosis. The S25 residue is phosphorylated by IKKs and the S > D mutation mimics this phosphorylation event. As a consequence, it strongly represses RIPK1 kinase activity both in vitro and in vivo [13]. We were able to confirm the inhibition of necroptosis in BMDMs isolated from RIPK1 S25D/S25D mice after treatment with LPS/zVAD-fmk. Interestingly, besides the inhibition of RIPK1 kinase activity, NF-κB and p38 activation were significantly decreased in RIPK1 S25D/S25D BMDMs as compared to BMDMs from RIPK1 +/+ controls. This may be attributed to the fact that LPS, a TLR4 ligand, was used to induce necroptosis in vitro. After TLR4 binding, LPS induces Erk1/2 expression in a RIPK1 kinase-dependent but necroptosis-independent way. Erk1/2 in turn induces NF-κB-and cFos-dependent inflammatory gene expression [36]. It should be noted that induction of necroptosis in atherosclerotic plaques can be stimulated by other factors including atherogenic ligands such as oxLDL, through direct upregulation and activation of RIPK3 and MLKL, and by cytokine (TNFα) secretion associated with sterile inflammation [4].
Importantly, atherosclerosis was exacerbated in ApoE −/− mice carrying the RIPK1 S25D/S25D mutation as compared to ApoE −/− RIPK1 +/+ controls. No differences in the necrotic core, cleavage of caspase 3, TUNEL, and necroptosis markers were observed, meaning that cell death was not significantly changed. This contrasts with two previous studies by Karunakaran et al. [4,8] in which ApoE −/− mice were treated with RIPK1 kinase inhibitor Nec1s or antisense oligonucleotides (ASOs) against RIPK1. The authors reported that Nec1s reduced the plaque area in the aortic root and overall lesion burden in the aorta of ApoE −/− mice [4]. Moreover, a significant decrease in the absolute P-MLKL-positive area was observed in aortic lesions of Nec-1s-treated mice, which could not be detected with immunohistochemistry in the present study. However, since Nec1s reduced the lesion area, it is unclear whether relative P-MLKL expression, and thus necroptosis, is truly inhibited in the plaques. Karunakaran et al. also reported that administration of RIPK1-specific ASOs resulted in a decreased, but not absent, expression of RIPK1 in ApoE −/− mice [8]. In this way, a basal level of RIPK1 was maintained to preserve minimal pro-survival NF-κB signaling and to prevent spontaneous cell death. This is important because we previously reported that a complete RIPK1 knock-out in macrophages of ApoE −/− mice results in impaired NF-κB signaling, increased apoptotic cell death, and plaque progression [7]. Interestingly, the authors reported that RIPK1-specific ASOs did not prevent necroptotic cell death at all in isolated macrophages but reduced inflammatory gene expression. In ApoE −/− mice, this led to smaller atherosclerotic plaques and lower levels of inflammatory cytokines [8]. Together, these studies stress the complex involvement of RIPK1 either as a pro-survival scaffold or as an active kinase in atherosclerosis, which is even further complicated depending on the stage of plaque development. Karunakaran et al. hypothesized that RIPK1 is mainly involved in earlier stages of plaque development, hence their studies cover a period of only 8-10 weeks of western diet (WD), as opposed to the more advanced plaques that were obtained after 16 weeks WD in the present study.
ApoE −/− RIPK1 S25D/S25D mice were characterized by increased levels of glycosaminoglycans both in the plaque and the tunica media, suggesting there is a general change in vascular extracellular matrix (ECM) that is not limited to plaques. Since VSMCs are the main producer of ECM components, it is tempting to propose that a change in VSMC phenotype is involved. Most likely, the RIPK1 S25D/S25D mutation promotes a switch in VSMCs from a contractile to synthetic phenotype so that more ECM is produced. Elevated glycosaminoglycan levels have high lipoprotein binding capacity, thereby contributing to plaque expansion. Blood vessel properties were also affected by the RIPK1 S25D/S25D mutation as brachiocephalic arteries of ApoE −/− RIPK1 S25D/S25D mice were dilated. This explains why significantly larger plaques did not lead to a significant increase in stenosis in ApoE −/− RIPK1 S25D/S25D mice. Vessel dilation combined with turbulent blood flow (due to plaques) point towards positive vascular remodeling, as first described by Glagov [37,38]. As mentioned above, RIPK1 kinase activity is linked to TLR4 signaling in response to DAMPs. TLR4 signaling is involved in atherogenesis and is known to contribute to expansive arterial remodeling [30,33,34]. These observations may link RIPK1 to the observed vessel changes, but the exact mechanism remains to be elucidated.
Clearly, transgenic and knock-out models of RIPK1 should be applied with caution, given the multitude of regulatory events that may be affected [39]. Therefore, the monoselective, new generation RIPK1 kinase inhibitor GSK'547 was included in the present study to treat atherosclerotic ApoE −/− Fbn1 C1039G+/− mice, which is a model of advanced, human-like atherosclerosis. By introducing the ApoE −/− Fbn1 C1039G+/− model, we did not aim to directly compare the pharmacological approach with the genetic study in ApoE −/− RIPK1 S25D/S25D mice. Instead, we prefer ApoE −/− Fbn1 C1039G+/− mice for pharmacological studies because signs of necroptosis mainly occur in the more advanced stages of atherosclerosis. Indeed, the expression of phosphorylated MLKL is increased in advanced human fibroatheroma as opposed to early lesions [4]. Moreover, RIPK3 deletion in LDLr −/− mice resulted in smaller plaques and less necrosis in the aortic root as compared to LDLr −/− RIPK3 +/+ controls after 16 weeks WD but not in earlier plaques after 8 weeks WD [3]. In the present study, we confirmed that expression levels of necroptosis proteins MLKL, RIPK1, and RIPK3 as well as the phosphorylation of MLKL increase during the growth of the plaque and the development of the necrotic core in ApoE −/− Fbn1 C1039G+/− mice. Indeed, Western blot analysis showed very weak bands of P-MLKL, MLKL, and RIPK1 at baseline, and quantification of these samples should be interpreted cautiously. However, clear bands were observed after 12 and, especially, after 24 weeks WD, underlining the occurrence of necroptosis. Treatment with GSK'547 was limited to 20 weeks WD because the survival rate of ApoE −/− Fbn1 C1039G+/− mice can drop below 50% after 20-25 weeks on WD due to myocardial infarction and stroke [28,40], and because 20 weeks WD was reported to suffice to induce an advanced plaque phenotype in ApoE −/− Fbn1 C1039G+/− mice [27,28].
Although GSK'547 efficiently inhibited necroptosis in vitro, this effect could not be confirmed in plaques of ApoE −/− Fbn1 C1039G+/− mice. On the contrary, plaques of GSK'547treated mice showed higher expression levels of P-MLKL than untreated controls. Plaques contain different necroptosis ligands such as oxLDL, TNFα, and other DAMPs, and some of them are able to bypass RIPK1 kinase activity for necroptosis induction [41,42]. Changes in the plaque area and composition were not observed after GSK'547 treatment, yet GSK'547treated ApoE −/− Fbn1 C1039G+/− mice showed increased TUNEL and cleaved caspase 3 positivity in plaques, suggesting a switch to RIPK1 kinase-independent apoptosis as compared to untreated controls. Initially, apoptosis is preferred over necrotic cell death in atherosclerotic plaques as long as apoptotic cells are efficiently cleared, a process called efferocytosis. However, efferocytosis is impaired in advanced atherosclerosis, which results in accumulation and secondary necrosis of apoptotic bodies, contributing to plaque progression and growth of the necrotic core [7,[43][44][45]. According to a recent study, GSK'547 treatment has a stage-dependent impact on atherogenesis. GSK'547 alleviates systemic inflammation in the early stages of atherosclerosis and reduces the plaque area after 2 weeks of treatment but exacerbates plaque formation after 4 weeks of treatment. Long-term GSK'547 treatment promotes macrophage accumulation and foam cell formation by upregulating the expression of several lipid metabolism-related genes. Moreover, GSK'547 inhibits ApoA1 synthesis in the liver and reduces plasma HDL levels, which contribute to plaque development [31].
In conclusion, in vitro inhibition of RIPK1 kinase activity is efficient in macrophages with a RIPK1 S25D/S25D mutation. However, plaque necrosis is not changed in ApoE −/− RIPK1 S25D/S25D mice and plaque progression is even worse as compared to control mice. The increased plaque size is mainly due to increased deposition of ECM components, although a relationship between RIPK1 kinase inhibition, ECM modifications, and vascular remodeling remains to be elucidated. Pharmacological inhibition of RIPK1 kinase with GSK'547 in ApoE −/− Fbn1 C1039G+/− mice did not alter the plaque area and composition but induced a transition to apoptosis. Therefore, without directly comparing the genetic and pharmacological studies, it can be concluded that targeting RIPK1 kinase activity is not an ideal approach to prevent plaque progression and plaque destabilization.