RIPK3 Protects Against Endothelial Activation and Vascular Permeability in a Mouse Model of Ischemia-Reperfusion Injury

Abstract

BACKGROUND:

RIPK3 (receptor-interacting protein kinase 3) has context-specific roles that are frequently associated with cellular damage and death. We previously found that hypoxia can trigger elevated levels of RIPK3 in endothelial cells (ECs), which contributes to lethal vascular rupture during mouse embryonic development. However, it is unknown whether elevated RIPK3 likewise compromises endothelial barrier function in adult vasculature under hypoxic conditions such as ischemia-reperfusion (I/R) injury.

METHODS:

Twelve-week-old male and female littermate control or inducible EC-specific Ripk3 knockout (Ripk3^iECKO) mice were exposed to surgical intestinal I/R injury. Clodronate liposomes were used to reduce circulating monocytes in vivo. Immortalized murine EC (MS1 [mile sven 1) and murine macrophage (BMA3.1A7 [bone marrow A clone 3.187]) lines were used for in vitro experiments.

RESULTS:

Ripk3^iECKO mice displayed an unexpected increase in small intestinal vascular permeability after I/R injury, rather than the decrease we predicted. Subsequent analyses using multiplex cytokine assays revealed significantly elevated levels of IL-6 (interleukin-6) in the serum and small intestinal tissue of I/R-injured Ripk3^iECKO mice. Upon TNFα (tumor necrosis factor-alpha) stimulation of immortalized murine Ripk3 knockout ECs grown in vitro, we found increased transcription and secretion of IL-6. These cells also expressed elevated levels of VCAM-1 (vascular cell adhesion molecule-1), which was likewise upregulated in the small intestines of Ripk3^iECKO mice. Using an IL-6 neutralizing antibody, we found that IL-6 triggered VCAM-1 elevation in Ripk3 knockout cells. This VCAM-1 expression correlated with enhanced macrophage binding to Ripk3 knockout cells and increased accumulation of leukocytes in Ripk3^iECKO small intestines following I/R injury. Importantly, reduction of circulating monocytes with clodronate liposomes led to rescue of IR injury–induced vascular permeability in Ripk3^iECKO mice.

CONCLUSIONS:

Endothelial RIPK3 suppresses EC activation and inflammation associated with IL-6 and VCAM-1 elevation to protect the vascular barrier in the context of intestinal I/R injury. Thus, endothelial RIPK3 plays surprisingly beneficial roles that reduce I/R injury–induced vascular dysfunction.

Highlights.

• Endothelial RIPK3 (receptor-interacting protein kinase 3) suppresses inflammation and protects vascular barrier integrity during intestinal ischemia-reperfusion injury in mice.

• RIPK3 deficiency elevates expression of the proinflammatory cytokine IL-6 (interleukin-6) and of VCAM-1 (vascular cell adhesion molecule-1) in endothelial cells.

• Monocyte reduction with clodronate liposomes reduces ischemia-reperfusion injury–induced vascular permeability in endothelial RIPK3-deficient mice.

• This study adds to a growing list of nonnecroptotic and beneficial roles for endothelial RIPK3.

See accompanying editorial by Glading

RIPK3 (receptor-interacting protein kinase 3) is a cytoplasmic serine-threonine protein kinase that was initially identified as a key component of the necroptosis cell death pathway^1–3 and has a growing list of nonnecroptotic functions that contribute to inflammatory diseases.⁴ Our laboratory has identified several nonnecroptotic roles for RIPK3 in endothelial cells (ECs). For example, we found that elevation of endothelial RIPK3 in midgestation mouse embryos promotes lethal vascular rupture without activating its critical downstream necroptosis component, MLKL (mixed lineage kinase domain-like).⁵ This finding is consistent with studies in neonatal and juvenile mice, in which RIPK3 destabilizes EC adherens junctions to impair developmental and pathological ocular angiogenesis.⁶ It has also been reported that RIPK3 utilizes both necroptotic and nonnecroptotic mechanisms to promote vascular permeability that facilitates metastatic tumor cell extravasation out of the bloodstream in adult mice.^7,8 These studies collectively indicate that excessive endothelial RIPK3 promotes detrimental vascular permeability through multiple mechanisms.

Endothelial RIPK3 protein expression is naturally high in midgestation mouse embryos at the stage in which they are most hypoxic, before the establishment of the fetal-placental circulation.^9,10 We reported that hypoxia can trigger Ripk3 transcription in embryonic ECs, particularly in conjunction with suppression of the inhibitory chromatin remodeling enzyme CHD4 (chromodomain helicase DNA-binding protein 4).⁵ Hypoxia has also been shown to contribute to elevated RIPK3 expression in cultured human hepatocyte and murine neuronal cell lines.^11,12 Moreover, several mouse models of ischemic diseases are ameliorated on a global Ripk3 knockout (Ripk3-KO) background,^13–15 suggesting that hypoxia-induced RIPK3 expression drives tissue pathology. However, whether hypoxia triggers RIPK3 upregulation in adult ECs and subsequent RIPK3-mediated vascular permeability has not yet been explored.

Intestinal ischemia-reperfusion (I/R) injury is a clinically important pathophysiological process resulting from a broad range of surgical and nonsurgical causes, including acute mesenteric ischemia,^16,17 shock,¹⁸ neonatal necrotizing enterocolitis,¹⁹ trauma,²⁰ and intestinal transplantation.²¹ During the ischemic phase of intestinal I/R injury, blood flow to the intestines is reduced due to obstruction of mesenteric arteries supplying blood to the small and large intestines.^17,22 The current therapeutic approach for addressing intestinal ischemia is to resolve the obstruction and restore blood flow to the intestinal tissue as soon as possible.²³ Although reperfusion of the tissue is crucial, this process can trigger further pathological damage to the intestine through the generation of reactive oxygen species that can mediate cell death and breakdown of the local mucosal and vascular barriers. Factors such as inflammatory cytokines produced in the reperfused intestinal region can then be transported in the bloodstream and cause damage to other regions of the small intestines and remote organs.^23–26 Because intestinal I/R injury triggers transient tissue hypoxia and vascular permeability, we identified this as an ideal model for investigating the hypoxic upregulation of RIPK3 in ECs and its subsequent effects on vascular integrity in adult mice.

In this study, we investigated the role of endothelial-specific RIPK3 in intestinal I/R injury using genetic, surgical, and molecular approaches. We found that hypoxia and I/R injury upregulate RIPK3 expression in ECs of adult mice. Interestingly, this upregulation is protective to the vasculature because deletion of endothelial Ripk3 promotes enhanced inflammation and vascular permeability following I/R injury. Therefore, endothelial RIPK3 plays a beneficial role in the context of intestinal I/R injury by helping to maintain vascular integrity.

Materials and Methods

Data Availability

All primary supporting data are available within the article and its Supplemental Material. Additional data and materials from this study are available from the corresponding author upon reasonable request. The Major Resources Table, which lists detailed information about the resources used in this study, can be found in the Supplemental Material.

Mouse Lines and Treatments

All mouse studies were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at the Oklahoma Medical Research Foundation (protocol 23-12). The mice used in this study were maintained on a C57Bl/6J background and were housed in a single room of the animal facility with a 12-hour light-dark cycle; food formulated with 20% protein and 4.5% fat (Purina LabDiet; 5053; PicoLab Rodent Diet 20) and water were provided ad libitum. Although sex differences and interactions were not determined, both female and male adult mice were used in all experiments to increase statistical power. Ripk3-floxed mice,⁵ Cdh5(PAC)-Cre^ERT2 mice (gift of Ralf Adams, Max Planck Institute for Molecular Biomedicine; Taconic; 13073),²⁷ RIPK3-GFP (green fluorescent protein) mice (Ripk3-gfp^fl/fl; The Jackson Laboratory; 030284),²⁸ and ROSA^mT/mG mice (Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Lou}/J; The Jackson Laboratory; 007576)²⁹ were previously described. Ripk3^fl/fl mice were crossed with Cdh5(PAC)-Cre^ERT2 mice to generate Ripk3^fl/fl;iCdh5(PAC)-Cre^ERT2 mice (ie, Ripk3^iECKO), and Ripk3^fl/fl littermates were used as controls (ie, Ripk3^WT). Tamoxifen (Sigma; T5648) dissolved in peanut oil (10-mg/mL stock concentration) was administered intraperitoneally at 2 mg/d per mouse for 3 consecutive days in 8-week-old mice; experiments were conducted 4 weeks later to allow for full gene deletion. A similar tamoxifen administration scheme was used to assess Cre activity in ROSA^mT/mG;iCdh5(PAC)-Cre^ERT2 mice. Genotyping for each allele was performed using primers listed in Table S1.

Intestinal I/R Injury

After mice were anesthetized using isoflurane (3%) and extended-release buprenorphine (1.0-mg/kg dose; ZooPharm; 1Z-74000-222510) was administered as an analgesic, a midline 3-cm laparotomy was performed, and the cecum and ileum were isolated. As described previously,³⁰ atraumatic microvascular clips (Roboz; RS-5420) were used to occlude first-order branches of the superior mesenteric artery feeding the ileum. Ischemia was maintained for 1 hour using 1% to 1.5% isoflurane anesthesia, and body temperature was maintained at 37 °C using a water-circulating heating pad. Tissues were kept moist by irrigation with sterile saline prewarmed to 37 °C throughout the 1-hour period. At the end of the ischemic period, the microvascular clips were removed, the small intestines were placed back in the abdominal cavity, and the abdominal wall was sutured. Mice were then placed in a clean cage for a 24-hour reperfusion period before euthanasia via carbon dioxide inhalation at a flow rate of 3 L/min followed by cervical dislocation. Unclamped adjacent vessels and their downstream small intestinal segments were considered as internal control regions. In a separate cohort of mice, sham I/R injuries were performed as described above without the occlusion of first-order branches of the superior mesenteric artery feeding the ileum.

Mesenteric Arterial En Face Preparations and Immunolabeling for Endothelial RIPK3 Expression

Primary and secondary mesenteric arterial branches of the superior mesenteric artery exposed to either 1-hour ischemia or 24-hour I/R injury along with adjacent control (unclamped) mesenteric arterial branches were isolated from Ripk3-gfp^fl/fl mice. Vessels were then carefully cut open longitudinally and pinned to a sylgard-coated 12-well plate for subsequent immunolabeling of GFP/RIPK3 in mesenteric arterial ECs.³¹ Mean fluorescence intensity was calculated by normalizing the total RIPK3 fluorescence intensity to the number of EC nuclei identified by Hoechst staining.

In Vivo Vascular Permeability

Vascular permeability was measured using an Evans blue dye (EBD) assay. Mice were first anesthetized using isoflurane (3%); 1% EBD (Sigma; E2129) was administered retro-orbitally (4-µL/g bodyweight) and allowed to circulate for 30 minutes. Afterward, mice were euthanized via carbon dioxide inhalation at a flow rate of 3 L/min followed by cervical dislocation; small intestines were harvested. Wet weights were first measured, and then, the small intestines were dried for 24 to 48 hours at 37 °C before dry weights were measured. Tissues were then incubated in 500-µL formamide for 48 hours at 56 °C to extract EBD from the tissues. Because perfusion efficiency can vary between organs and mice, we opted not to perfuse mice before organ harvesting. Instead, extravasated EBD was measured spectrophotometrically at 620 and 740 nm, and values were corrected for hemoglobin using the following formula: OD (optical density)₆₂₀−(1.426×OD₇₄₀+0.03), as previously described.³² The concentration of EBD extravasated from the small intestines was calculated using a standard curve of known EBD concentrations, and final values were normalized to organ dry weight. Permeability following intestinal I/R injury was reported as fold change from each genotype’s own baseline. FITC (fluorescein isothiocyanate; 70-kDa) dextran (Sigma; 46945) was also used to measure vascular permeability. FITC-Dextran was injected retro-orbitally at a final dose of 100-mg/kg bodyweight following a 24-hour I/R injury challenge. FITC-dextran was allowed to circulate for 30 minutes, and mouse tissues were perfusion-fixed with 4% paraformaldehyde. Organs were then collected and cryoembedded for sectioning and fluorescent microscopy.

In Vivo Circulating Monocyte Depletion

Monocyte depletion from circulating blood was first calculated in a wild-type (WT) cohort of mice injected intraperitoneally with either PBS liposomes or clodronate liposomes (Encapsula Nano Sciences; CLD-8901)³³ at 100-µL/10-g body weight. Liposomes were allowed to circulate for 48 hours before mice were euthanized via carbon dioxide inhalation at a flow rate of 3 L/min followed by cervical dislocation, and blood was collected by cardiac puncture. A HEMAVET (Drew Scientific; HV03586) was used to measure circulating monocyte numbers in the blood. Subsequently, monocyte depletion was performed in littermate control and Ripk3^iECKO mice, and I/R injury was performed as described above after clodronate liposomes had circulated for 48 hours.

EC and Blood Leukocyte Isolation From Mice

To isolate ECs, mouse lungs were enzymatically dissociated with collagenase B (Roche; 34324621, 1 mg/mL) and dispase II (Sigma; D4693, 1 mg/mL) for 50 minutes at 37 °C. DNase I (Ambion; AM2224) was added to the collagenase/dispase mix to remove genomic DNA. Next, sheep anti-rat Dynabeads (Invitrogen; 11035) coated with anti-CD31 antibody (BD Pharmingen; MEC13.3, 553370) were incubated with the lung single-cell suspensions for 45 minutes at 4 °C. ECs were then magnetically separated from the rest of the lung cells, and both EC and non-EC cell samples were placed in TRIzol Reagent for RNA isolation. To isolate peripheral blood leukocytes, mice were euthanized via carbon dioxide inhalation at a flow rate of 3 L/min followed by cervical dislocation, and blood was then collected by cardiac puncture with a syringe prefilled with 100 µL of 3.2% sodium citrate and placed in 10-mL ACK lysis buffer (Gibco; A1049201) for 15 minutes at 25 °C for erythrocyte lysis. Blood samples were then centrifuged, the supernatant was aspirated, and cells were suspended in RNase-free PBS. Samples were centrifuged again to remove the PBS before adding TRIzol Reagent for RNA isolation.

EC Culture

The MS1 (mile sven 1) adult murine pancreatic EC line (ATCC; CRL 2279) was cultured in DMEM (Gibco; 11960-044) supplemented with 10% fetal bovine serum (Fisher Scientific; 3SH30910.03) and 1% antibiotics (streptomycin/amphotericin; Sigma; A5955). Ripk3-KO MS1 cells, which were generated by CRISPR (clustered regularly interspaced short palindromic repeats)-Cas9 (CRISPR-associated protein 9) modification,³⁴ were cultured identically. TNFα (tumor necrosis factor-alpha; R&D Systems; 410-MT-010/CF) was administered at 20 ng/mL for designated times. For NRF2 (nuclear factor erythroid 2-related factor 2) studies, cells were incubated with 20-ng/mL TNFα or 50-μM tert-Butylhydroquinone (Sigma; 112941) for 1 hour before immunostaining for NRF2. To overexpress NRF2 (gene name: Nfe2l2), murine Nfe2l2 cDNA was cloned as an mRuby fusion construct into the XbaI/PmeI sites of the pBOB lentiviral plasmid (Addgene; 12337) by Gibson assembly. For lentivirus production, the NRF2-mRuby plasmid was mixed with psPAX2 (Addgene; 12260) and pMD2.G (Addgene; 12259) plasmids and transfected into HEK293-Lenti-X cells (Takara; 632180). Viral supernatants were titred to ≈1 MOI and then transduced into Ripk3-KO MS1 cells twice, 72 hours apart. Transduced cells were allowed to recover for 1 week before TNFα stimulation experiments commenced.

Monocyte Adhesion Assay

The BMA3.1A7 (bone marrow A clone 3.1A7) immortalized mouse macrophage line (Applied Biological Materials; T06730) was cultured in RPMI 1640 medium (Sigma; R8758) supplemented with 10% fetal bovine serum and 1% antibiotics. BMA3.1A7 cells were labeled with 4-µM calcein AM (AAT Bioquest; 22004), as previously described.^35,36 They were then incubated for 30 minutes at 37 °C on top of a monolayer of MS1 ECs that had been pretreated with 20-ng/mL TNFα for 24 hours. Nonadherent BMA3.1A7 macrophages were removed by multiple washing steps with Hanks Balanced Salt Solution (Gibco; 14175) before remaining cells were imaged using a Nikon TiE Eclipse epifluorescence microscope. Cells were then lysed in 1% Triton X-100 (Thermo Fisher Scientific; A16046.AE) for 10 minutes, and fluorescence was measured at an excitation wavelength of 490 nm and an emission wavelength of 525 nm and normalized to a control group in which MS1 ECs were not incubated with macrophages.

Ex Vivo Whole Blood Cytokine Release Assay

Cytokines in whole blood were measured using a cytokine release assay, as previously described.^37–39 Mice were first euthanized via carbon dioxide inhalation at a flow rate of 3 L/min followed by cervical dislocation, and blood was then collected through a cardiac puncture into tubes with the anticoagulant lepirudin (50 µg/mL; gifted by Dr Florea Lupu) to prevent blood clotting. Collected blood was then incubated with 1-µg/mL bacterial lipopolysaccharide (Sigma; L3012) or PBS for 2 and 4 hours. Plasma samples were collected at the end of each incubation period for downstream cytokine measurements using a multiplex immunoassay.

In Vivo and In Vitro Cytokine Measurements

Serum cytokines in mice were measured following a 4-hour I/R injury. Blood leukocyte cytokine release in plasma was measured after stimulating leukocytes with lipopolysaccharide, as described above. MS1 cells were stimulated with 20 ng/mL TNFα for 24 hours, and secreted cytokines were measured in the cell-free supernatant. Serum, plasma, or supernatant cytokines were measured using a custom ProcartaPlex Multiplex Immunoassay (Thermo Fisher). The plates were read on a Bio-Plex 200 array reader (Bio-Rad). Samples were run in duplicate, and cytokine levels were quantified by comparing to a 4PL algorithm standard curve specific to each analyte.⁴⁰

Immunofluorescence

Small intestines were collected from euthanized mice and briefly rinsed in PBS before fixing overnight with 4% paraformaldehyde at 4 °C. They were then incubated in 10%, 15%, and 20% sucrose (1 hour each) and placed in a 1:1 mixture of 20% sucrose and optical cutting temperature compound (Sakura; 4583) overnight. Small intestines were then embedded in optical cutting temperature compound the next morning in cryomolds, and 20-µm tissue sections were cut with a cryotome (Fisher Scientific; Epredia Microm HM525 NX Cryostat). Tissue sections were dried at 37 °C, washed in PBS for 5 minutes to remove optical cutting temperature compound, permeabilized for 15 minutes in 0.3% Triton X-100, and blocked in 10% donkey serum (Jackson ImmunoResearch Lab; 102644-006) in PBS for 1 hour. Primary antibodies were diluted in 3% BSA (Rockland; BSA-50) and 0.1% Triton X-100 in PBS and applied to sections overnight at 4 °C. Secondary antibodies were diluted in 3% BSA in PBS and applied to sections the next day for 1 hour at room temperature in the dark. Confocal Z-stack images were obtained with a Nikon C2 confocal microscope. Primary antibodies used for immunofluorescence were rabbit-anti-GFP (1:200, Novus Biological; NB600-303), rat-anti-VE-cadherin (vascular endothelial cadherin; 1:100, Abcam; ab91064), goat-anti-CD45 (1:50, R&D Systems; AF114), goat-anti-VCAM-1 (vascular cell adhesion molecule-1; 1:100, R&D Systems; AF643), rabbit-anti-NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells; p65; 1:700, Cell Signaling; 8242), rabbit-anti-NRF2 (1:300, gift from Dr Scott Plafker at OMRF),⁴¹ goat-anti-CD31 (1:100, R&D Systems; AF3628), and rat-anti-CD31 (1:100, BD Pharmingen; 553370). Secondary antibodies used were Cy5-donkey-anti-rabbit IgG (1:500, Jackson ImmunoResearch), Alexa488-donkey-anti-rat IgG (1:500, Thermo Fisher), and Cy3-donkey-anti-goat IgG (1:500, Jackson ImmunoResearch). Hoechst (20 µg/mL) was added to the secondary antibody incubation.^5,42 A similar staining protocol was used for cultured cells grown on coverslips (Electron Microscopy Sciences; 7229608) and fixed with 4% paraformaldehyde for 15 minutes.

Immunoblotting

Frozen mouse tissue samples or cells were first lysed in RIPA buffer (Sigma; R0278) that included Protease Inhibitor Cocktail (Thermo Fisher; 1860932) and Phosphatase Inhibitor Cocktail (Thermo Fisher; 78420) using a mini bead beater (DENTSPLY; C321001). Protein concentrations of lysates were then measured with a BCA Protein Assay Kit (Thermo Scientific; 23227) following the manufacturer’s instructions. Samples were diluted in 4X Bolt LDS Sample Buffer (Thermo Scientific; B0007) with 10X Bolt Sample Reducing Agent (Thermo Scientific; B0009) and heated at 70 °C for 10 minutes. Protein samples were separated on Bolt Bis-Tris Plus Protein 4% to 12% gels (Thermo Scientific; NW04120BOX) and then transferred to a nitrocellulose membrane using iBlot Transfer Stacks (Thermo Scientific; IB23001). The membranes were then blocked in 5% milk or 5% BSA (for phospho-proteins) dissolved in TBST, incubated in primary antibodies overnight at 4 °C, and secondary antibodies were applied the next day for 1 hour at room temperature. Bound antibodies were visualized using enhanced chemiluminescence detection reagents (Thermo Scientific; 34076 [West Dura], A38554 [West Atto]) and imaged using the iBright CL750 Imaging System (Thermo Scientific; A44116). Densitometric analysis was performed using National Institutes of Health ImageJ software. Primary antibodies used for immunoblotting were goat-anti-VCAM-1 (1:1000, R&D Systems; AF643), goat-anti-ICAM-1 (intracellular adhesion molecule-1; 1:2000, R&D Systems; AF796), rabbit-anti-NF-κB p65 (1:6000, Cell Signaling; 8242), rabbit-anti-phospho-NF-κB p65 (1:2000, Cell Signaling; 3033), and rabbit-anti-β-actin (1:4000, Cell Signaling; 4967). Secondary antibodies used were anti-rabbit-HRP (1:4000, Sigma; A6667) and anti-goat-HRP (1:4000, Sigma; A5420).

Real-Time Quantitative Reverse Transcription Polymerase Chain Reaction

Total RNA was extracted using TRIzol Reagent (Thermo Scientific; 15596018) and RNeasy Mini Kit (Qiagen; 74106). cDNA was prepared using an iScript cDNA Synthesis Kit (Bio-Rad; 1708891), and quantitative reverse transcription polymerase chain reaction (PCR) was performed using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad; 1725274), gene primers (Table S1), and a CFX96 Touch Real-Time PCR System (Bio-Rad) or CFX Opus 384 Real-Time PCR System (Bio-Rad). Target genes were normalized against at least 2 of the following housekeeping genes: Actb, Rn18s, Rpl13a, and Eife3.

Graphics

BioRender.com was used for generating the graphics in Figures 1C, 2A, and 2D and the graphic abstract.

Figure 1.

Endothelial RIPK3 (receptor-interacting protein kinase 3) deletion in established blood vessels causes elevated vascular permeability following intestinal ischemia-reperfusion (I/R) injury. A, Timeline for conditional deletion of endothelial Ripk3 (Ripk3^iECKO) in 8-week-old mice and subsequent baseline or intestinal I/R injury–induced vascular permeability measurements using Evans blue dye (EBD) in 12-week-old mice. B, Baseline vascular permeability measurements using EBD in Ripk3^WT and Ripk3^iECKO mice in the small intestine (n=5/group). C, Left, Schematic displaying small intestinal regions where vascular permeability was measured after I/R injury. Right, Small intestinal vascular permeability (fold change from each genotype’s own baseline) after 24-hour intestinal I/R injury in the injury region (I/R) and nearby control regions (adjacent control, distal control; n=6/group). D, Representative images of fluorescein isothiocyanate (FITC)-dextran (70 kDa) permeability (white arrows) in control regions of small intestinal tissue after 24-hour I/R injury. Scale bar, 100 µm. Data represent mean±SEM. P values were determined by the 2-tailed Student t test (B) and the 2-way ANOVA (C). SMA indicates superior mesenteric artery.

Figure 2.

Endothelial secretion of the inflammatory cytokine IL-6 (interleukin-6) is elevated in the absence of RIPK3 (receptor-interacting protein kinase 3). A, Schematic of mouse tissue collection and quantitative analysis of cytokines at 0 and 4 hours after ischemia-reperfusion (I/R) injury in Ripk3^WT and Ripk3^iECKO mice. B, Baseline IL-6 levels (left) and fold change (from baseline) following 4-hour I/R injury (right) in small intestinal tissue of Ripk3^WT and Ripk3^iECKO mice (n=3–4/group). C, Baseline IL-6 levels (left) and fold change (from baseline) following 4-hour I/R injury (right) in serum of Ripk3^WT and Ripk3^iECKO mice (n=3–4/group). D, Schematic of IL-6 measurement in stimulated whole blood and MS1 (mile sven 1) endothelial cells (ECs). Plasma IL-6 levels at 2 (E) and 4 hours (F) after ex vivo stimulation of whole blood collected from Ripk3^WT and Ripk3^iECKO mice. At each time point, blood was stimulated with PBS (left) or lipopolysaccharide (LPS; right); LPS data are graphed as fold change from PBS control treatments (n=3/group). G, IL-6 levels in cell supernatants following stimulation of wild-type (WT) and Ripk3 knockout (KO) MS1 ECs with 20-ng/mL TNFα (tumor necrosis factor-alpha) for 0 and 24 hours (n=3/group). H, Il6 transcripts measured by real-time quantitative reverse transcription polymerase chain reaction (qRT-PCR) in WT and Ripk3-KO MS1 ECs following stimulation with 20-ng/mL TNFα for 0, 4, and 8 hours (n=4/group). Data represent mean±SEM. P values were determined by the 2-tailed Student t test (B, C, E, and F) and the 2-way ANOVA (G and H).

Statistical Analyses

All statistical tests were performed using GraphPad Prism software, version 10.4.1. Results of all graphs are displayed as mean values±SEM unless indicated otherwise. Each data set includes at least a minimum sample size of n=3 biological replicates, which was sufficient for all inferential testing. Data analysis is followed as previously described.⁴³

Comparisons Between 2 Groups

A 2-tailed Student t test was performed for data with normal distribution and equal variances. A Welch t test was performed for data with normal distribution and unequal variances.

Comparisons Between >2 Groups

For data sets with one independent variable and equal variances (confirmed by the F test), a 1-way ANOVA followed by post hoc comparisons was used. For data sets with 2 independent variables and equal variances (confirmed by the Spearman test), a 2-way ANOVA with post hoc comparisons was performed. When assumptions of normality or equal variance were violated, data were transformed (ie, logarithmic transformation) before 2-way ANOVA and post hoc analysis to satisfy model assumptions.

A P value of ≤0.05 was considered significant. Outlier tests were performed in Prism, but no data were excluded in the final graphs. A summary of all statistical tests can be found in Table S5.

Results

Endothelial Ripk3 Deletion in Established Blood Vessels Causes Elevated Vascular Permeability Following Intestinal I/R Injury

Because we previously reported that embryonic endothelial RIPK3 expression can be elevated by hypoxia,⁵ we questioned whether endothelial RIPK3 also increases after ischemic challenge in adult mice. We addressed this by assessing RIPK3 expression in en face mesenteric artery preparations following intestinal ischemia and 24-hour reperfusion (Figure S1A). Because commercial RIPK3 antibodies give nonspecific immunostaining in our hands, we utilized a RIPK3-GFP reporter mouse²⁸ and immunostained for GFP. We found significantly elevated RIPK3/GFP expression in mesenteric ECs after 1 hour of ischemic clamping, and this elevated expression was maintained even after 24 subsequent hours of reperfusion (Figure S1A and S1B).

Because elevated endothelial RIPK3 is associated with blood vessel permeability and rupture,^5–7 we next investigated the role of endothelial RIPK3 in I/R injury–induced vascular permeability. We generated Ripk3^fl/fl;Cdh5(PAC)-Cre^ERT2 mice (hereafter called Ripk3^iECKO) to achieve tamoxifen-inducible Ripk3-KO in ECs, as we have previously described.^44,45 Tamoxifen was administered to 8-week-old littermate control (Ripk3^fl/fl; hereafter called Ripk3^WT) and Ripk3^iECKO mice, and 4 weeks later, we detected no Ripk3 mRNA expression in ECs isolated from Ripk3^iECKO lungs (Figure S2A and S2B). Notably, we found no difference in Ripk3 expression in peripheral blood leukocytes (Figure S2C), indicating that the Cdh5(PAC)-Cre^ERT2 line did not drive recombination in these cells under our deletion protocol. We also used a ROSA^mT/mG Cre reporter mouse²⁹ to assess the endothelial specificity of the Cdh5(PAC)-Cre^ERT2 transgene and found that Cre activity (GFP expression) was restricted to ECs and was not detected in tissue resident leukocytes (Figure S2D) or circulating blood cells (Figure S2E). We next measured baseline vascular permeability in 12-week-old littermate Ripk3^WT and Ripk3^iECKO mice using an EBD perfusion and leakage assay (Figure 1A). We found that vascular permeability was unchanged in the small intestines of Ripk3^WT and Ripk3^iECKO mice under baseline conditions (Figure 1B). However, when small intestinal EBD leakage was measured following 24 hours of I/R injury, we found that vascular permeability was elevated in the intestinal region downstream of clamped mesenteric vessels (ie, the I/R region) in both Ripk3^WT and Ripk3^iECKO mice (Figure 1C). Surprisingly, vascular permeability in unclamped adjacent control and distal control intestinal regions was also elevated in Ripk3^iECKO mice but not in littermate Ripk3^WT mice (Figure 1C). These differing observations between I/R and control regions suggest that the endothelial barrier in the I/R region may already be maximally compromised due to ischemic injury, inflammation, oxidative stress, and other I/R injury–related processes, which overrides additional effects from endothelial Ripk3 deletion. However, the intact EC barrier in intestinal control regions is clearly susceptible to enhanced vascular permeability mediated by endothelial Ripk3 deletion in an I/R injury context. In contrast, we found no changes in vascular permeability in sham-operated mice (Figure S3). In addition to quantifying EBD leakage, we also perfused mice with 70-kDa FITC-dextran after I/R injury and found that in Ripk3^iECKO mice, FITC-dextran leakage was enhanced in the submucosal regions of small intestinal tissue where postcapillary venules are located (Figure 1D, white arrows). Taken together, these results demonstrate that endothelial RIPK3 is elevated in I/R injury regions, and it plays an unexpected role in suppressing nearby vascular permeability.

Endothelial Secretion of the Inflammatory Cytokine Interleukin-6 Is Elevated in the Absence of RIPK3

Because some cytokines secreted by activated immune cells and ECs after reperfusion injury can promote vascular permeability,^23,46,47 we analyzed several such cytokines in the serum and small intestinal tissue of Ripk3^WT and Ripk3^iECKO mice at baseline and after I/R injury using a multiplex immunoassay (Figure 2A). We chose a time point of 4 hours after I/R injury (instead of 24 hours) for our analyses because many cytokine levels rise quickly after I/R injury.²³ Under baseline conditions, serum cytokine levels exhibited a general trend toward downregulation in Ripk3^iECKO mice (Figure 2B; Figure S4A through S4D). The cause and effect of this trend on baseline organ function were not obvious, so we focused instead on cytokine readings in mice with I/R injury. We found that IL-6 (interleukin-6) levels were significantly elevated in the small intestine (Figure 2B) and serum (Figure 2C) of Ripk3^iECKO mice after I/R injury, unlike the other cytokines that we analyzed (Figure S4A through S4D). We next measured circulating leukocyte numbers at baseline and after 4-hour I/R injury using a complete blood count test. No differences were detected in leukocyte numbers between Ripk3^WT and Ripk3^iECKO mice at baseline or after injury (Figure S5).

To determine the cell source of the increased IL-6 that we detected in Ripk3^iECKO mice after I/R injury, we collected whole blood from Ripk3^WT and Ripk3^iECKO mice via cardiac punctures, stimulated blood immune cells with bacterial lipopolysaccharide (or PBS) for 2 and 4 hours, and measured plasma IL-6 levels (Figure 2D). We did not observe differences in IL-6 levels secreted from Ripk3^WT versus Ripk3^iECKO blood at either time point (Figure 2E and 2F), indicating that circulating immune cells were not the source of the elevated IL-6 in Ripk3^iECKO mice after I/R injury. We next considered Ripk3-deficient ECs as a potential source of IL-6. We utilized the well-established immortalized mouse MS1 EC line,⁴⁸ in which we had previously deleted Ripk3 with CRISPR/Cas9 (ie, Ripk3-KO).³⁴ Because TNFα is one of the key inflammatory cytokines secreted during I/R injury,²³ we stimulated WT and Ripk3-KO MS1 cells with TNFα for up to 24 hours to simulate injury and measured secreted IL-6 in the supernatant (Figure 2D). Although no difference was detected at 4 hours (Figure S6), we found that Ripk3 deletion led to robustly enhanced IL-6 secretion at 24 hours after TNFα stimulation (Figure 2G). No comparable differences were seen in other cytokines we measured (Figure S4E through S4G). We also found that Il6 transcript levels were significantly elevated at 4 and 8 hours after TNFα stimulation in Ripk3-KO versus WT cells (Figure 2H). Together, these data indicate a role for endothelial RIPK3 in IL-6 production.

RIPK3 Does Not Regulate NF-κB Signaling in MS1 ECs

Because we found Il6 transcripts elevated in TNFα-stimulated Ripk3-KO MS1 ECs, we sought to identify a mechanistic explanation for this phenotype. The NF-κB signaling pathway is a key mediator of inflammation in ECs. TNFα stimulation of ECs activates canonical NF-κB signaling and results in the nuclear translocation of the transcription factors p65/p50 and p50/c-Rel, which mediate transcription of inflammatory cytokine and adhesion molecule genes.^49–51 Therefore, we assessed p65 phosphorylation and nuclear localization in TNFα-stimulated WT and Ripk3-KO MS1 ECs. We found no differences in p65 protein levels (Figure S7A and S7B), and although phospho-p65 levels increased at 10, 30, and 60 minutes after TNFα stimulation, they did so equivalently in both WT and Ripk3-KO cells (Figure S7C and S7D). Immunocytochemical staining revealed robust p65 nuclear localization after 10, 30, and 60 minutes of stimulation, but no obvious differences were detected between WT and Ripk3-KO cells (Figure S7E), indicating that canonical NF-κB signaling is not hyperactive in Ripk3-KO ECs.

NRF2 Overexpression Reduces IL-6 Production From Ripk3-KO MS1 ECs

The transcription factor NRF2 has antioxidant and anti-inflammatory roles in multiple cell types, including ECs.^52,53 In addition to translocating from the cytoplasm to the nucleus upon activation and promoting the transcription of antioxidant genes during oxidative stress,^52,54,55 nuclear NRF2 also inhibits Il6 transcription in macrophages⁵⁶ and ECs.⁵⁷ Therefore, we questioned whether NRF2 expression or activity was compromised in RIPK3-deficient ECs. Stimulated NRF2 nuclear translocation can occur between 30 minutes to 2 hours in ECs,⁵⁸ so we assessed NRF2 cytosolic and nuclear expression at baseline or after 1-hour stimulation with TNFα or tert-Butylhydroquinone, a known activator of NRF2. We found that although cytosolic NRF2 expression was comparable between the cell types, nuclear NRF2 expression was decreased in Ripk3-KO cells at baseline and after TNFα and tert-Butylhydroquinone stimulation (Figure 3A through 3D). When nuclear-to-cytosolic NRF2 ratios were calculated under all 3 culture conditions, this value was consistently and significantly decreased in Ripk3-KO compared with WT MS1 cells (Figure 3B through 3D). We next measured expression of the NRF2 target genes Hmox1 and Nqo1 at baseline and did not observe any changes in these mutant cells (Figure 3E and 3F). However, at 4 hours after TNFα stimulation, Hmox1 transcripts were significantly decreased, and Nqo1 transcripts trended downward in Ripk3-KO compared with WT MS1 cells (Figure 3E and 3F). Together, these data demonstrate that both NRF2 nuclear localization (assessed at 1 hour) and subsequent NRF2 transcriptional activity (assessed at 4 hours) are diminished in RIPK3-deficient ECs after TNFα stimulation.

Figure 3.

NRF2 (nuclear factor erythroid 2-related factor 2) overexpression reduces elevated IL-6 (interleukin-6) secretion from RIPK3 (receptor-interacting protein kinase 3)-knockout (KO) endothelial cells (ECs). A, Representative immunofluorescence images in gray scale depicting cytosolic vs nuclear NRF2 expression at baseline and 1 hour following 20-ng/mL TNFα (tumor necrosis factor-alpha) or tert-Butylhydroquinone (tBHQ) stimulation in wild-type (WT) and Ripk3-KO MS1 (mile sven 1) ECs. Scale bar, 20 µm. B through D, Mean NRF2 fluorescence intensity in cytosolic and nuclear compartments (left) and calculated NRF2 nuclear-to-cytosolic ratio (right) in WT and Ripk3-KO MS1 ECs at baseline (B) or after 1-hour stimulation with TNFα (C) or tBHQ (D; n=3/group). E and F, Transcript expression following stimulation with 20-ng/mL TNFα for 0 and 4 hours in WT and Ripk3-KO MS1 ECs of the following genes: Hmox1 (E) and Nqo1 (F; n=3–4/group). G, Il6 transcript expression in Ripk3-KO and Ripk3-KO+NRF2 overexpression (OE) MS1 ECs following stimulation with 20-ng/mL TNFα for 0 and 4 hours (n=3/group). H, IL-6 levels in cell supernatants following stimulation of Ripk3-KO and Ripk3-KO+NRF2 OE MS1 ECs with 20-ng/mL TNFα for 0, 4, and 24 hours (n=3/group). Data represent mean±SEM. P values were determined by the 2-tailed Student t test or the 2-way ANOVA (B–D) and the 2-way ANOVA (E–H).

To test whether elevating NRF2 could reduce IL-6 expression and secretion from RIPK3-deficient ECs, we generated a lentiviral construct encoding a NRF2-mRuby fusion protein and transduced it into Ripk3-KO MS1 cells (achieving ≈18%–20% transduction efficiency). To our surprise, we found that NRF2 overexpression increased Il6 transcripts in baseline and TNFα-stimulated Ripk3-KO ECs (Figure 3G), rather than decreasing them as we predicted from the literature.⁵⁶ Nevertheless, we proceeded with measuring IL-6 protein in supernatants from baseline and TNFα-treated MS1 cells, and we found that NRF2 overexpression reduced IL-6 production from TNFα-treated Ripk3-KO ECs (Figure 3H). Therefore, the elevated IL-6 found in the serum of Ripk3^iECKO mice and in TNFα-treated Ripk3-KO ECs may result from NRF2 dysregulation.

IL-6-Induced VCAM-1 Is Upregulated After Endothelial Deletion of Ripk3

IL-6 can induce adhesion molecule expression on ECs,⁵⁹ which subsequently leads to immune cell recruitment and extravasation that contribute to vascular permeability during an inflammatory response.^60–62 Therefore, we next assessed the expression of VCAM-1 and ICAM-1 in the small intestines of Ripk3^iECKO mice. Tissue immunoblots revealed that VCAM-1 protein levels were elevated in the small intestines of Ripk3^iECKO mice compared with Ripk3^WT mice at baseline (Figure 4A). Although an increasing trend was also observed in baseline ICAM-1 protein expression in Ripk3^iECKO small intestines, these levels were not significantly different from those seen in Ripk3^WT mice (Figure S8).

Figure 4.

VCAM-1 (vascular cell adhesion molecule-1) is upregulated in vivo and in vitro after endothelial deletion of RIPK3 (receptor-interacting protein kinase 3). A, VCAM-1 expression in the small intestine assessed by immunoblot in baseline Ripk3^WT and Ripk3^iECKO mice (n=7/group). B, VCAM-1 expression assessed by immunoblot in wild-type (WT) and Ripk3 knockout (KO) MS1 (mile sven 1) endothelial cells (ECs) stimulated with 20-ng/mL TNFα (tumor necrosis factor-alpha) for 24 hours (n=4/group). C, WT and Ripk3-KO MS1 cells were stimulated with either PBS or 20-ng/mL TNFα for 24 hours and immunostained for VCAM-1. White arrowheads in the insets indicate junctional localization of VCAM-1. Scale bar, 20 µm. Inset, 10 µm. D, VCAM-1 expression assessed by immunoblot in WT and Ripk3-KO MS1 ECs stimulated with 20-ng/mL TNFα for 24 hours in the presence of control IgG antibody or neutralizing IL-6 (interleukin-6) antibody (n=4/group). Data represent mean±SEM. P values were determined by the 2-tailed Student t test (A) and the 2-way ANOVA (B and D).

Next, we stimulated WT and Ripk3-KO MS1 cells with TNFα for 24 hours and found by immunoblotting that VCAM-1 protein expression was increased in Ripk3-KO cells (Figure 4B). In addition, we saw an elevation in VCAM-1 protein expression in Ripk3-KO MS1 cells when we performed immunocytochemistry on cells stimulated with TNFα for 24 hours, particularly at EC junctions marked by VE-cadherin (Figure 4C, white arrow heads). We also assessed Vcam1 transcription by quantitative reverse transcription PCR, but we saw no significant differences between WT and mutant cells at baseline and after 4 and 8 hours of TNFα stimulation (Figure S9). Therefore, VCAM-1 protein appears to accumulate in TNFα-stimulated Ripk3-KO ECs through posttranscriptional mechanisms.

Because VCAM-1 expression can be induced by IL-6,⁵⁹ we next assessed whether VCAM-1 elevation in Ripk3-KO MS1 cells was a result of increased IL-6 secretion from and signaling within those cells. IL-6 exerts its functions after binding to its receptor, IL-6Rα (IL-6 receptor α), and to the ubiquitously expressed 130-kilodalton signal-transducing β-receptor subunit, gp130 (glycoprotein 130). We first assessed transcripts for gp130 (Il6st) in WT and Ripk3-KO ECs at baseline and found no differences in expression (Figure S10A). We next assessed transcripts for the Il6ra, which were expressed in MS1 cells but showed no differences between WT and Ripk3-KO ECs (Figure S10B). In our MS1 cells, Il6st expression was 23^.5-fold higher than Il6ra expression, which is consistent with data from a recent RNA-Seq study of MS1 ECs.⁶³ Having established that IL-6 receptor complexes are present on MS1 cells, we next measured VCAM-1 protein expression in WT and Ripk3-KO ECs stimulated for 24 hours with TNFα in the presence of a control IgG antibody or a neutralizing IL-6 antibody. We found that VCAM-1 expression in Ripk3-KO ECs was significantly reduced by the neutralizing IL-6 antibody to levels comparable to those seen in WT ECs (Figure 4D). Therefore, IL-6 contributes to elevated VCAM-1 expression in Ripk3-KO ECs.

Exacerbated Endothelial Activation in the Absence of RIPK3 Leads to Increased Leukocyte Adherence That Contributes to Enhanced Vascular Permeability

We questioned whether the elevated IL-6 and VCAM-1 that we detected in the absence of RIPK3 correlated with increased leukocyte recruitment to the small intestine. We first addressed this question in vivo by immunostaining for CD45⁺ leukocytes in the small intestines of Ripk3^WT and Ripk3^iECKO mice at 24 hours after I/R injury. We found significantly increased numbers of CD45⁺ cells in the Ripk3^iECKO mice (Figure 5A and 5B). We next visualized and quantified macrophage adhesion to an EC monolayer in vitro. We coincubated fluorescently labeled macrophages (BMA3.1A7) with WT or Ripk3-KO MS1 cells that had been previously stimulated with TNFα. After coincubation for 30 minutes, we found increased macrophage adhesion to Ripk3-KO cells compared with WT ECs (Figure 5C through 5E).

Figure 5.

Exacerbated endothelial activation in the absence of RIPK3 (receptor-interacting protein kinase 3) leads to increased leukocyte adherence that contributes to vascular permeability. A, CD45⁺ leukocyte accumulation was assessed by immunostaining of small intestinal tissues from Ripk3^WT and Ripk3^iECKO mice following 24-hour ischemia-reperfusion (I/R) injury. Scale bar, 100 µm. B, CD45⁺ cells, as in A, were quantified from multiple confocal Z-stack images taken from each mouse tissue by normalizing the CD45⁺ expression area to the total cell nuclear area in the small intestines. Individual data points are averages of 2 to 4 imaged and quantified sections per mouse (n=5/group). C, Representative images of immortalized BMA3.1A7 macrophages labeled with calcein AM fluorescent dye (green) and incubated on top of a monolayer of wild-type (WT) and Ripk3-knockout (KO) MS1 (mile sven 1) endothelial cells (ECs) that were previously treated with 20-ng/mL TNFα (tumor necrosis factor-alpha) for 24 hours. Cells were washed multiple times before imaging. Scale bar, 20 µm. D, The number of macrophages adherent to ECs was quantified in each acquired image (n=4 images/group). E, Following imaging (as in C), cells were lysed with Triton X, and fluorescence was measured using a plate reader. Values were normalized to readings from wells in which BMA3.1A7 cells were incubated with untreated WT MS1 ECs (dotted line; n=3/group). F, Timeline for detecting vascular permeability with Evans blue dye (EBD) after clodronate-liposome treatment and intestinal I/R injury; tamoxifen was administered at 8 weeks of age. G, Small intestinal vascular permeability (fold change from each genotype’s own baseline) after 24-hour intestinal I/R injury; measurements are shown in the injury region (I/R) and control regions (control, distal control) following clodronate-liposome treatment (n=4–5/group). H, CD45⁺ leukocyte accumulation was assessed by immunostaining in small intestinal tissues from Ripk3^WT and Ripk3^iECKO mice following 24-hour I/R injury and clodronate-liposome treatment. I, CD45⁺ cells, as in G, were quantified from multiple confocal Z-stack images taken from each mouse; CD45⁺ expression area was normalized to total cell nuclear area in the small intestines. Individual data points are averages of 3 to 4 imaged and quantified sections per mouse (n=3/group). Data represent mean±SEM. P values were determined by the 2-tailed Student t test (B, D, E, and I) and the 2-way ANOVA (G).

Because circulating leukocytes (eg, monocytes) can compromise the vascular barrier during injury and inflammation via adhesion-mediated signaling and extravasation,^62,64–67 we next sought to evaluate the contribution of these cells to vascular hyperpermeability in Ripk3^iECKO small intestines following I/R injury. Administration of clodronate liposomes to WT mice for 48 hours did not efficiently deplete small intestinal F4/80⁺ macrophages (Figure S11A), but it did deplete 47% of circulating monocytes (Figure S11B). Because both Ripk3^WT and Ripk3^iECKO mice treated with this clodronate-liposome scheme for 48 hours displayed difficulty tolerating isoflurane anesthesia during the ischemic period of our I/R injury surgical procedure, we continued with the same concentration of clodronate liposomes rather than increasing it to achieve higher monocyte/macrophage depletion. We performed this clodronate-liposome administration scheme in Ripk3^WT and Ripk3^iECKO mice followed 48 hours later by I/R injury surgeries and subsequent EBD leakage assays to measure vascular permeability (Figure 5F). We found that the elevated vascular permeability previously detected in the small intestinal regions of Ripk3^iECKO mice (Figure 1C) was rescued after clodronate-liposome treatment (Figure 5G). We also found fewer CD45⁺ leukocytes in the small intestines of both Ripk3^WT and Ripk3^iECKO mice (Figure 5H), and the significant accumulation that had been seen in IR-injured Ripk3^iECKO intestines without clodronate treatment (Figure 5B) was reduced to WT levels (Figure 5I). Therefore, these findings suggest that adherence of circulating leukocytes to blood vessels in IR-injured Ripk3^iECKO mice contributes to vascular permeability in the small intestine.

Discussion

Previous studies addressing the role of RIPK3 in I/R injury models have reported mixed and organ-specific effects. For example, global Ripk3-KO mice display smaller myocardial lesions in a heart I/R injury model, while brain lesions generated in a major cerebral artery occlusion model are comparably sized to those in WT mice.¹⁴ Meanwhile, pharmacological inhibition of RIPK1, the upstream regulator of RIPK3 in the necroptosis pathway, attenuates intestinal epithelial necroptosis in an intestinal I/R injury model.^68,69 These inconsistent findings highlight the limitations of using global knockout mice and systemic pharmacological treatments to decipher cell-type–specific and organ-specific roles for RIPK3 in I/R injury models.

Our current study addresses the role of endothelial RIPK3 in intestinal I/R injury–induced vascular permeability. Based on previous work by our laboratory and others,^{5,13,15,68,69} we predicted that RIPK3 would be elevated in ECs after ischemic injury and would subsequently promote vascular permeability. While we did find RIPK3 elevated in mesenteric vessels exposed to ischemia and I/R, we were surprised to find vascular permeability increased after I/R injury in Ripk3^iECKO mice compared with littermate controls. This increase in vascular permeability correlated with (1) elevated IL-6 in serum and small intestinal tissue, (2) enhanced expression of the endothelial adhesion molecule VCAM-1, and (3) immune cell accumulation in small intestinal tissue. Altogether, our data demonstrate that endothelial RIPK3 plays a beneficial role in maintaining vascular integrity after I/R injury.

Notably, this is not the first time we have found beneficial roles for endothelial RIPK3. We previously reported that RIPK3 promotes developmental angiogenesis in the embryonic hindbrain and newborn retina by regulating growth factor receptor trafficking.⁴⁵ We also reported that apolipoprotein E–deficient Ripk3^iECKO mice fed a high-fat diet develop more atherosclerotic lesions in their descending aortas than littermate controls.⁴⁴ Importantly, this increased atherosclerosis correlated with elevated expression of the adhesion molecule E-selectin in cultured RIPK3-knockdown MS1 ECs. Therefore, our previous atherosclerosis study correlates with the current study in terms of revealing protective roles for endothelial RIPK3 in inflammatory challenge models and with intriguing similarities in the regulation of adhesion molecules that mediate immune cell accumulation.

Inflammatory cytokines are known contributors to vascular permeability after I/R injury,^23,46,47 and our current study reveals that IL-6 is significantly elevated after I/R injury in the serum and small intestinal tissue of Ripk3^iECKO mice. We also found that IL-6 protein and transcripts were elevated in Ripk3-KO MS1 cells after stimulation with TNFα. Although ECs are a known source of IL-6 under inflammatory conditions,⁷⁰ our data are the first to indicate that RIPK3 serves as a brake on endothelial IL-6 production. IL-6 uses 2 mechanisms to mediate its biological effects. Classical IL-6 signaling occurs when the cytokine binds to a cell surface complex consisting of IL-6Rα and gp130. Cells that do not express membrane-bound IL-6Rα, including ECs, undergo IL-6 signaling when the cytokine binds a soluble form of IL-6Rα, which subsequently interacts with membrane-bound gp130 to mediate transsignaling.⁷¹ Classical IL-6 signaling typically regulates homeostatic processes, while IL-6 transsignaling occurs in cardiovascular diseases and mediates inflammation.⁷¹ This proinflammatory transsignaling also promotes vascular permeability through activation of intracellular JAK (Janus kinase)/STAT3 (signal transducer and activator of transcription 3) proteins.^70,72–75 Therefore, we think that the elevated IL-6 that we detected in the serum of Ripk3^iECKO mice contributed to increased I/R–induced vascular permeability through transsignaling on ECs. We also suspect that ECs are a major source of the elevated IL-6 that we found in Ripk3^iECKO mice after I/R injury, which would be consistent with our in vitro data. Although immune cells are typically the principal sources of this cytokine in inflammatory pathologies,⁷⁶ we found no evidence of Ripk3 deletion in immune cells from Ripk3^iECKO mice. Likewise, whole blood removed from the mutant mice and stimulated ex vivo did not show elevated IL-6 secretion compared with blood from control mice. It is important to note that IL-6 is garnering significant attention for its ability to promote cytokine storms, particularly in the context of severe infection (eg, influenza virus), acute respiratory distress syndrome associated with COVID-19, and CAR T-cell therapy.^70,77 Therefore, it will be interesting to learn in the future whether endothelial RIPK3 regulates IL-6 production in these and additional inflammatory settings.

In our attempt to define how endothelial RIPK3 suppresses IL-6 secretion from ECs in an I/R context, we found intriguing connections to the antioxidant transcription factor NRF2. NRF2 was an interesting candidate for us because I/R challenge causes oxidative stress that stimulates anti-inflammatory NRF2 activity.⁷⁸ Moreover, NRF2 has been shown to suppress Il6 transcription in ECs.⁵⁷ We predicted that diminished NRF2 activity might account for elevated IL-6 transcription and secretion from Ripk3-KO ECs. Our results were complicated. We saw less NRF2 nuclear localization in TNFα-stimulated Ripk3-KO ECs, which was consistent with the elevated IL-6 secretion from those cells. However, while NRF2 overexpression gratifyingly reduced IL-6 secretion from TNFα-stimulated Ripk3-KO ECs, it paradoxically increased Il6 transcription. Therefore, more work is needed in the future to address the tantalizing connections between RIPK3, NRF2, and IL-6 in ECs. Specifically, it will be interesting to determine whether RIPK3 influences NRF2 activity directly in an I/R challenge setting, potentially through serine and threonine phosphorylation events, which are key regulators of NRF2 antioxidant activity.⁷⁹ Moreover, nontranscriptional mechanisms through which NRF2 suppresses IL-6 production in stimulated ECs will require further investigation.

Because VCAM-1 serves as a docking site for circulating leukocytes that can subsequently extravasate out of the bloodstream,^77,80,81 the regulation of adhesion molecule expression is highly relevant to tissue inflammation. Moreover, leukocyte binding to adhesion molecules induces intracellular signaling that compromises endothelial barrier function and promotes vascular permeability.^62,64–67 We found that macrophages adhered more robustly to cultured Ripk3-KO MS1 cells than to WT cells, and our in vivo immunostaining indicated that CD45⁺ leukocytes were more abundant following I/R injury in Ripk3^iECKO small intestines than in those from control littermates. Considering these findings, we were particularly interested to find that clodronate liposomes significantly rescued I/R-induced vascular permeability in Ripk3^iECKO mice. We found fewer CD45⁺ leukocytes following I/R injury and clodronate-liposome treatment in Ripk3^WT and Ripk3^iECKO mice. Notably, resident leukocyte and macrophage depletion was not impressive in these tissues with our clodronate liposome administration scheme. Therefore, we think that circulating leukocytes, specifically monocytes, which were depleted by ≈47% with clodronate liposomes, were more impactful contributors to permeability than were tissue resident leukocytes in the small intestine after I/R challenge.

Collectively, this study highlights a beneficial role for endothelial RIPK3 in the complex pathophysiological response to intestinal I/R injury. It also demonstrates that RIPK3 regulates endothelial IL-6 secretion and vascular adhesion molecule expression, which may have implications for multiple inflammatory conditions. Future studies are warranted to investigate how endothelial RIPK3 deletion impacts other organs following I/R injury, particularly given the well-established heterogeneity of ECs across different vascular beds.⁸² It is likely that organ-specific responses to endothelial Ripk3 deletion may uncover additional, distinct mechanisms beyond those identified in this study. Findings from our current study can be added to the growing list of nonnecroptotic functions for RIPK3 in ECs and should raise awareness that the use of pharmacological necroptosis inhibitors for treating ischemic diseases could have unintended effects on the vasculature.

Limitations of the Study

While our blocking antibody studies indicated that autocrine endothelial IL-6 signaling promoted VCAM-1 protein expression in vitro, we have not yet demonstrated this connection in vivo. Future studies will require genetic depletion of endothelial Il6 in I/R-injured Ripk3^iECKO mice to reveal how this cell-specific cytokine impacts VCAM-1 expression, leukocyte accumulation, and vascular permeability in the small intestine.

Article Information

Acknowledgments

The authors acknowledge Drs Florea Lupu and Ravi Keshari for their guidance and technical advice with cytokine assays and Drs Scott and Kendra Plafker for generously sharing the NRF2 (nuclear factor erythroid 2-related factor 2) antibody and providing their expertise in NRF2 biology. Figures 1C, 2A, and 2D and the graphic abstract were created using BioRender.com.

Sources of Funding

This work was funded by grants from the National Institutes of Health to C.T. Griffin (R35HL144605) and to Lijun Xia (P20GM139763) and the American Heart Association to C.F. Johnson (20POST35120396).

Disclosures

None.

Supplemental Material

Tables S1–S5

Figures S1–S11

Supplementary Material

atv-46-148-s001.pdf ^{(5MB, pdf)}

atv-46-148-s002.pdf ^{(4.9MB, pdf)}