Neddylation blockade protects endothelial cells against inflammation and ischemia-reperfusion injuries Authors: Vincent Mayoral, Pierre Braud, Mélanie Néel, Anne-Clémence Vion, Mariefleur Vaillant, Benoît Mesnard, Fabienne Haspot, Stéphanie Le Bas-Bernardet, Gaëlle Tilly, Nicolas Degauque, Julien Branchereau, Gilles Blancho, Sarah Bruneau Published: 2026 ABSTRACT BackgroundLong-term allograft survival is limited by the occurrence of chronic allograft dysfunction, mainly due to antibody mediated rejection and aspecific fibrosis. The latter is favored by ischemia-reperfusion injury (IRI), particularly in organs from extended criteria donors and from donors after cardiocirculatory arrest. The graft endothelium is highly sensitive to IRI that causes chronic damages and increases its immunogenicity. Thus, improvement of preservation methods and the treatment of organs prior to implantation (so called “graft preconditioning”) constitute a major clinical challenge to improve graft survival. Among the mechanisms implicated in the reactivity of endothelial cells, neddylation has emerged as a potential candidate of interest.MethodsTo study the potential of neddylation blockade as a preconditioning strategy, we used an in vitro model of microvascular endothelial cells (ECs) treated in conditions close to those of organ transplant preservation. ECs were subjected to hypothermic preservation (4 °C) in the presence of the neddylation inhibitor MLN4924, and following rewarming (37 °C), stimulated with TNAα and IFNγ to mimic sterile inflammation at reperfusion.ResultsWe found that preconditioning of ECs with MLN4924 had potent anti-inflammatory effects by blocking the induction of adhesion molecules (ICAM1, VCAM1, E-Selectin), of HLA class I and II, and of the pro-inflammatory cytokines and chemokines IL6, RANTES and Fractalkine. Functionally, this translated into inhibition of EC-mediated T lymphocyte activation and of T cell adhesion to ECs. MLN4924 preconditioning also blocked the induction of prothrombotic molecules (Tissue Factor and PAI-1), and platelet adhesion to ECs. Finally, MLN4924 enhanced the expression of the complement inhibitor CD55 and of the antioxidant molecule HO-1, and protected ECs from hypothermia and hypoxia induced cell death.ConclusionsOur results identify the neddylation inhibitor MLN4924 as a powerful protective molecule for endothelial cells and a promising candidate for graft preconditioning and protection of organs against IRI.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12964-026-02968-z. Background By improving patient quality of life and survival, solid organ transplantation stands as the elective treatment for patients suffering from end-stage organ failure. However, despite ongoing advances in the field, the availability of grafts remains a matter of concern [1]. To solve the current organ shortage, extended criteria donors and donors after circulatory death have to be used, despite a higher susceptibility of these grafts to ischemia-reperfusion injuries (IRI) [2–6]. Thus, improving preservation strategies of these vulnerable organs appears necessary. Lately, significant efforts have been made to improve graft preservation strategies, not only to maintain organ viability, but also to improve their quality [7]. Pharmacological supplementation of the preservation solution with protective molecules could present the advantage of targeting the graft only, especially its endothelium, avoiding many limitations of systemic drug delivery (toxicity, failure to reach the graft). This so-called “graft preconditioning” provides the opportunity to prevent or at least reduce IRI and dampen the immunogenicity of the graft endothelium. Combined with the current standards of hypothermic preservation methods, graft preconditioning by “supplemented” preservation fluid could represent an effective and easy strategy for graft quality improvement. In this work, we have studied MLN4924, a neddylation inhibitor, as a potential pharmacologic agent for graft preconditioning. Neddylation is a type of protein post-translational modification in which the ubiquitin-like protein NEDD8 (Neural precursor cell Expressed, Developmentally Downregulated protein 8) binds to target proteins and directs them to proteasome for degradation, through a process analogous to ubiquitination [8]. Neddylation has been particularly studied in cancer and MLN4924, a first-in-class small molecular inhibitor of this pathway has been previously investigated in several clinical trials [9–13]. There is evidence that neddylation blockade by MLN4924 abrogates NF-kB and mTOR signaling, key pathways that notably drive endothelial cell pro-inflammatory responses [14, 15]. MLN4924 has also been shown to induce HIF-1α accumulation [14], suggesting that it may improve the capacity of endothelial cells to control oxidative stress. Thus, considering ischemia-reperfusion injury pathophysiology, the use of MLN4924 could be considered in transplantation, for promoting cytoprotection against ischemia by impairing proinflammatory pathways. Here, we describe for the first time the use of MLN4924 in an in vitro cold ischemia-reperfusion model which will pave the way for its use for solid organ transplant preconditioning. Methods Antibodies and reagents The antibodies and qPCR probes used in this work are listed in Supplementary Tables S1, S2 and S3. Cell culture and study model Primary cultures of human dermal microvascular endothelial cells (HDMEC) and of human lung microvascular endothelial cells (HLMEC) from single donors were purchased from Promocell and primary cultures of human renal glomerular endothelial cells (HRGEC) and of human pancreatic microvascular endothelial cells (HPMEC) from single donors were purchased from iCelltis. All types of endothelial cells were cultured in complete endothelial growth medium EGM-2MV (Promocell). HDMEC from 3 donors, HRGEC from 2 donors and HLMEC and HPMEC from a single donor were used in this work, all between passages 3 and 6. To study the effects of a pretreatment with MLN4924 in conditions close to those of a graft preservation system, endothelial cell monolayers were incubated for 3 h at 4 °C with 10 µM to 100 µM MLN4924 (Calbiochem), or with DMSO as the vehicle control (indicated in the figures as NT for “no treatment”). Cells were then washed with PBS, switched to 37 °C in complete culture medium, and stimulated or not with 100 U/mL TNFα (recombinant human TNFα, R&D Systems) and 1000 U/mL IFNγ (recombinant human IFNγ, R&D Systems) in order to mimic the inflammatory reaction that occurs following reperfusion, for 4, 24 and 48 additional hours depending on the analyses. Gene expression studies To identify genes whose expression is regulated by the MLN4924 pretreatment, mRNA of HDMEC was extracted using the RNeasy isolation kit (Qiagen) and cDNA were generated using the M-MLV Reverse Transcriptase (Life Technologies) according to the manufacturer’s instructions. qPCR plates designed to analyze the expression of 92 genes of interest related to endothelial cell biology (RT² Profiler PCR Arrays “Human endothelial cell biology”, Qiagen) were used to examine endothelial cell responses. Relative gene expression levels were evaluated using ACTB, B2M, GAPDH, HPRT1 and RPL10 as references, according to the 2−ΔΔCt method as previously described [16]. Results obtained were validated by RT-qPCR as described in the supplementary section. T lymphocyte-mediated endothelial cell aggression assays Blood from healthy volunteers was collected in EDTA tubes at the Etablissement Français du Sang (EFS, Nantes, France) according to institutional guidelines with written informed consent (consent number CPDL-PLER-2023 037). PBMCs were separated from blood samples on a Ficoll gradient layer according to the manufacturer’s recommendations and untouched CD4 and CD8 T lymphocytes were isolated by magnetic separation on an AutoMACS cell separator (Miltenyi Biotec), using the CD4+ T Cell Isolation kit and the CD8+ T cell Isolation kit respectively (Miltenyi Biotec) according to the manufacturer’s protocol. These cells were activated for 3 days with human T-activator CD3/CD28 Dynabeads (Gibco) at a 1:1 ratio in complete RPMI medium. HDMEC were pretreated with 10 µM MLN4924 for 3 h at 4 °C in EGM-2MV, washed with PBS, and co-incubated at 37 °C for 24 h with activated CD4 or CD8 T cells at a 1:1 ratio in EGM-2MV medium. HDMEC were then washed extensively to remove T cells, collected by Trypsin/EDTA digestion, and their activation was assessed by analyzing the expression of ICAM-1, VCAM-1, HLA class I and HLA class II by flow cytometry as described above. Endothelial cell-mediated T lymphocyte activation assays To assess the effect of MLN4924 on the capacity of endothelial cells to activate allogenic T lymphocytes (non-professional antigen-presenting cell properties), HDMEC were treated for 3 h at 4 °C with 10 µM MLN4924, washed with PBS and activated for 24 h at 37 °C with TNFα (100 U/mL) and IFNγ (1000 U/mL). HDMEC were then washed to remove residual TNFα and IFNγ, and co-incubated with T lymphocytes (1:1 ratio) freshly isolated from the blood of healthy volunteers on an AutoMACS cell separator using a Human pan T cell isolation kit (Miltenyi Biotec), as per manufacturer’s instructions. After 18 h of co-incubation, T lymphocytes were collected and the expression of the activation marker CD69 was analyzed at their surface by flow cytometry. Expression of intracellular Granzyme B by activated CD8 T cells was also assessed by flow cytometry after fixation and permeabilization of the cells using the Cytofix/Cytoperm kit from BD Biosciences. Adhesion of T lymphocytes and monocytes to endothelial cells HDMEC pretreated with 10 µM or 100 µM of MLN4924 for 3 h at 4 °C, and further activated for 24 h at 37 °C with TNFα (100 U/mL) and IFNγ (1000 U/mL), were co-incubated for 1 h with PBMCs from healthy volunteers isolated on a Ficoll gradient layer from blood samples collected at the Etablissement Français du Sang (EFS, Nantes, France) (Ratio of 5 PBMC for 1 HDMEC). Nonadherent PBMCs were removed by 5 PBS washes, and HDMEC and adherent cells were harvested using Trypsin/EDTA and analyzed by flow cytometry using anti-CD3, anti-CD45, anti-CD31 and anti-CD14 fluorescent antibodies (Supplementary Table S1). Ten thousand cells per sample were analyzed and the adhesion index was calculated as the ratio of the percentage of CD45+ CD3+ T cells or CD14 + monocytes within each group, compared with the untreated group (no MLN4924 pretreatment and no TNFα/IFNγ activation). Platelet adhesion assays Human platelets were isolated from blood samples from healthy volunteers. Blood was drawn in Vacutainer tubes containing sodium citrate, acid citrate and glucose (ACD) and centrifuged at 180 g for 5 min at room temperature (RT) without brake to separate the platelet-rich plasma. Platelet-rich plasma was diluted in Tyrode buffer and further centrifuged at 150 g for 5 min at RT without brake to eliminate potential residual red blood cells, and finally at 500 g for 5 min at RT without brake to pellet platelets. Platelets were stained with 2.5 µmol/L calcein acetoxymethyl ester (calcein-AM, Life Technologies) for 15 min at room temperature in the dark. After washing with Tyrode buffer, platelets were co-incubated with HDMEC. For platelet adhesion assays in static conditions, stained platelets were incubated with treated HDMEC in complete EGM-2MV culture medium for 1 h at 37 °C. After 5 washes in PBS, adherent platelets were visualized by fluorescent microscopy on an Olympus IX71 inverted microscope (x40 magnification) with an Olympus DP72 camera using CellSens Viewer acquisition software (Olympus), and quantified using ImageJ software. For platelet adhesion assays under flow, HDMEC were cultured on 0.4 ibidi slides (IBIDI) and treated for 3 h at 4 °C with 10 µM MLN4924 without flow. Cells were then washed with complete medium, and unidirectional laminar shear stress was applied for 24 h using the pumping system and control unit from IBIDI in the presence of 10 U/mL TNFα. Local shear stress was calculated using Poiseuille’s law and averaged to 3.5 dyn/cm². Labelled platelets were then slowly and continuously injected in the ibidi slide exposed to flow (100 µL/min) for 3 min. From one platelet sample, half was used for HDMEC preconditioned with MLN4924, and the other half for control non-preconditioned HDMEC. After adhesion, the slide was slowly and continuously washed from remaining non-adherent platelets for 1 min with culture medium. Platelet adhesion was directly imaged in an Eclipse Ti2 inverted microscope (Nikon) to count the number of adherent platelets per field of view. C3c deposition HDMEC pretreated with 10 µM MLN4924 for 3 h at 4 °C, and further activated for 24 h at 37 °C with TNFα (100 U/mL) and IFNγ (1000 U/mL), were washed with PBS and further incubated for 2 h at 37 °C with a pool of human sera from ten healthy volunteers diluted at 1:3 in M199 medium. Cells were then washed twice with PBS without Ca2+/Mg2+, and collected using 10 mM EDTA. C3c deposition at their surface was evaluated by flow cytometry using a rabbit anti-human C3c and a secondary AF488 anti-rabbit IgG antibody (Supplementary Table S1). Hypothermia and hypoxia-induced endothelial apoptosis HDMEC were incubated in UW preservation medium (Belzer MPS® UW) supplemented or not with 10 µM MLN4924, for 24 h at 4 °C in incubation bags containing a reagent mixture that chemically binds the oxygen quickly to create a 0% O2 atmosphere (Anaerocult™ A mini, Millipore), as model of static cold storage. After these 24 h of storage, HDMEC were washed with PBS and incubated at 37 °C in complete culture medium EGM-2MV as a model of reperfusion. After 4 h, cell apoptosis was analyzed by flow cytometry using Annexin V-APC and 7-AAD staining (BD Biosciences). Statistical analyses All data are presented as mean ± SEM. Statistical analyzes were performed using the Wilcoxon test, and P values < 0.05 were considered significant. Results Preconditioning of endothelial cells with MLN4924 inhibits their activation and proinflammatory responses To identify genes whose expression might be regulated in endothelial cells by a treatment with MLN4924 during the hypothermic preservation of an organ, we set up an in vitro model in which HDMEC were treated for 3 h at 4 °C with 10 µM MLN4924, and further washed and stimulated with TNFα/IFNγ for 4 h at 37 °C. The expression level of 92 genes related to endothelial cell biology were then analyzed using targeted qPCR arrays, which led to the identification of a panel of genes whose expression was strongly repressed by MLN4924 pretreatment in inflammatory conditions, notably genes encoding adhesion molecules (ICAM1, VCAM1, E-SELECTIN) and pro-inflammatory chemokines and cytokines (CCL2, CCL5, CX3CL1, IL6, IL1B) (Fig. 1A and Supplementary Table S4). These results were validated by qPCR in 5 independent experiments (Fig. 1B). Fig. 1Endothelial cell preconditioning with MLN4924 inhibits their activation. HDMEC were pretreated or not with MLN4924 10 µM for 3 h at 4 °C and further stimulated with TNFα and IFNγ for 4 to 24 h at 37 °C. A The expression levels of 92 genes related to endothelial cell biology were analyzed by specific qPCR arrays after 4 h of stimulation. The scatter plot shows the relative gene expression levels in cells preconditioned with MLN4924 vs. non-preconditioned cells (NT), after TNFα/IFNγ stimulation. Genes upregulated or downregulated more than five-fold in MLN4924-preconditioned cells are shown in red and blue respectively. The table displays the fold change in mRNA expression for each of these genes in preconditioned vs. untreated HDMEC. B Expression of genes coding for adhesion molecules (ICAM1, VCAM1, E-SELECTIN) and pro-inflammatory cytokines and chemokines (IL6, CCL5, CXC3CL1 and IL1B) was analyzed after 4 h of stimulation by RT-qPCR. Scatter plots show the fold change in mRNA expression in 5 independent experiments (with mean ± SEM). C Expression of adhesion molecules (ICAM1 and VCAM1) and of HLA class I and II molecules were analyzed by flow cytometry after 24 h of stimulation. Representative histograms are shown and scatter plots show the fold change in mean fluorescence intensity (MFI) from 7 independent experiments (with mean ± SEM). D Release of IL6, RANTES and Fractalkine in the culture medium were quantified by ELISA after 24 h of stimulation. Scatter plots show concentrations in pg/mL with mean ± SEM from 6 independent experiments. E Cell viability was assessed by flow cytometry using a fluorescent viability dye at 24 h of stimulation. The scatter plot shows the percentage of viable cells with mean ± SEM from 7 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 Endothelial cell preconditioning with MLN4924 inhibits their activation. HDMEC were pretreated or not with MLN4924 10 µM for 3 h at 4 °C and further stimulated with TNFα and IFNγ for 4 to 24 h at 37 °C. A The expression levels of 92 genes related to endothelial cell biology were analyzed by specific qPCR arrays after 4 h of stimulation. The scatter plot shows the relative gene expression levels in cells preconditioned with MLN4924 vs. non-preconditioned cells (NT), after TNFα/IFNγ stimulation. Genes upregulated or downregulated more than five-fold in MLN4924-preconditioned cells are shown in red and blue respectively. The table displays the fold change in mRNA expression for each of these genes in preconditioned vs. untreated HDMEC. B Expression of genes coding for adhesion molecules (ICAM1, VCAM1, E-SELECTIN) and pro-inflammatory cytokines and chemokines (IL6, CCL5, CXC3CL1 and IL1B) was analyzed after 4 h of stimulation by RT-qPCR. Scatter plots show the fold change in mRNA expression in 5 independent experiments (with mean ± SEM). C Expression of adhesion molecules (ICAM1 and VCAM1) and of HLA class I and II molecules were analyzed by flow cytometry after 24 h of stimulation. Representative histograms are shown and scatter plots show the fold change in mean fluorescence intensity (MFI) from 7 independent experiments (with mean ± SEM). D Release of IL6, RANTES and Fractalkine in the culture medium were quantified by ELISA after 24 h of stimulation. Scatter plots show concentrations in pg/mL with mean ± SEM from 6 independent experiments. E Cell viability was assessed by flow cytometry using a fluorescent viability dye at 24 h of stimulation. The scatter plot shows the percentage of viable cells with mean ± SEM from 7 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 The effect of the MLN4924 pretreatment on endothelial cell activation was then analyzed at the protein level. HDMEC were treated for 3 h at 4 °C with 10 µM MLN4924, and further activated for 24 h at 37 °C with TNFα/IFNγ. Flow cytometry analyzes showed that the MLN4924 pretreatment drastically inhibited the induction of ICAM1, VCAM1, HLA class I and HLA class II expression at the surface of HDMEC after 24 h of stimulation (Fig. 1C), which was still the case after 48 to 72 h (Supplementary Figure S1). MLN4924 pretreatment also abrogated the release of IL6, CCL5/RANTES and CX3CL1/Fractalkine by HDMEC in the culture medium (Fig. 1D). In line with these results, we confirmed that MLN4924 was able to target the proinflammatory NFκB pathway in our HDMEC model as observed in other cell types [17], as evidenced by a robust accumulation of phosphorylated IκBα (Ser32) in preconditioned stimulated cells (Supplementary Figure S2). While Ser32 phosphorylation typically precedes proteasomal degradation, MLN4924 is known to prevent this degradation by inhibiting the SCFβ E3 ubiquitin ligase, leading to stabilization of phosphorylated IκBα and thus sequestration of NFκB in the cytoplasm to prevent its transcriptional activity [18]. Importantly, HDMEC viability was not significantly impaired following MLN4924 pretreatment as compared to untreated cells, suggesting a low toxicity of this molecule on the endothelium in these treatment conditions (Fig. 1E). All these results except for IL6 regulation were confirmed in primary human renal glomerular endothelial cells (HRGEC), as well as in primary human lung (HLMEC) and pancreas (HPMEC) microvascular endothelial cells (Supplementary Figures S3, S4 and S5). Altogether, these results demonstrate that neddylation inhibition during hypothermic preservation may dampen EC inflammation during the reperfusion phase. Preconditioning of endothelial cells with MLN4924 inhibits lymphocyte-endothelial interactions To determine whether MLN4924 could prevent endothelial cell aggression by activated T lymphocytes, ECs were treated with 10 µM of MLN4924 for 3 h at 4 °C, and subsequently co-incubated at 37 °C with purified CD4 T cells or CD8 T cells from healthy volunteers that had been previously activated for 3 days by anti-CD3 and anti-CD28 stimulation. After 24 h, the activation of ECs was evaluated by flow cytometry. As illustrated in Fig. 2A, pretreatment of ECs with MLN4924 inhibited the induction of ICAM1, VCAM1, HLA class I and HLA class II by both activated CD4 and CD8 T lymphocytes, suggesting that this preconditioning was able to protect them against T cell aggression. Of note, T cell-mediated activation seemed to preferentially induce HLA II expression, while TNFα/IFNγ stimulation was more efficient to increase ICAM1 and VCAM1 expression at the surface of ECs. Such differences likely reflect the distinct signaling intensities inherent to soluble cytokine exposure versus the localized, multi-signal nature of T cell-endothelial cell interactions [19, 20]. Fig. 2Endothelial cell preconditioning with MLN4924 inhibits their proinflammatory responses. A HDMEC pretreated for 3 h at 4 °C with 10 µM MLN4924 were co-incubated at 37 °C with activated T CD4 and T CD8 lymphocytes from healthy volunteers. After 24 h of coculture, expression of ICAM1, VCAM1, HLA I and HLA II was analyzed at the surface of HDMEC by flow cytometry. Scatter plots show the fold change in MFI from 8 independent experiments (with mean ± SEM). B HDMEC pretreated for 3 h at 4 °C with 10 µM MLN4924 and further stimulated with TNFα and IFNγ for 24 h were co-incubated for 18 h with freshly isolated allogenic T CD4 and T CD8 cells from healthy volunteers. T cell activation was then evaluated by analyzing CD69 expression at their surface by flow cytometry, as well as intracellular Granzyme B expression within CD69+ T CD8 cells. Representative flow cytometry plots are shown, and scatter plots indicate the percentage of CD69+ T CD4 or T CD8 cells from 6 independent experiments, and the percentage of Granzyme B+ cells within the CD69+ T CD8 population from 4 independent experiments (with mean ± SEM). C HDMEC were pretreated for 3 h at 4 °C with 10 µM or 100 µM MLN4924, stimulated with TNFα and IFNγ for 24 h at 37 °C, and further co-incubated for 1 h at 37 °C with allogenic PBMCs from healthy volunteers. After extensive washes, adherent T lymphocytes and monocytes were quantified by flow cytometry. The scatter plots indicate the adhesion index of CD45+ CD3+ T cells (T CD3 adhesion) and of CD14 + monocytes (Monocyte adhesion) to HDMEC from 8 (T cells) and 5 (monocytes) independent experiments (with mean ± SEM). *P < 0.05, **P < 0.01 Endothelial cell preconditioning with MLN4924 inhibits their proinflammatory responses. A HDMEC pretreated for 3 h at 4 °C with 10 µM MLN4924 were co-incubated at 37 °C with activated T CD4 and T CD8 lymphocytes from healthy volunteers. After 24 h of coculture, expression of ICAM1, VCAM1, HLA I and HLA II was analyzed at the surface of HDMEC by flow cytometry. Scatter plots show the fold change in MFI from 8 independent experiments (with mean ± SEM). B HDMEC pretreated for 3 h at 4 °C with 10 µM MLN4924 and further stimulated with TNFα and IFNγ for 24 h were co-incubated for 18 h with freshly isolated allogenic T CD4 and T CD8 cells from healthy volunteers. T cell activation was then evaluated by analyzing CD69 expression at their surface by flow cytometry, as well as intracellular Granzyme B expression within CD69+ T CD8 cells. Representative flow cytometry plots are shown, and scatter plots indicate the percentage of CD69+ T CD4 or T CD8 cells from 6 independent experiments, and the percentage of Granzyme B+ cells within the CD69+ T CD8 population from 4 independent experiments (with mean ± SEM). C HDMEC were pretreated for 3 h at 4 °C with 10 µM or 100 µM MLN4924, stimulated with TNFα and IFNγ for 24 h at 37 °C, and further co-incubated for 1 h at 37 °C with allogenic PBMCs from healthy volunteers. After extensive washes, adherent T lymphocytes and monocytes were quantified by flow cytometry. The scatter plots indicate the adhesion index of CD45+ CD3+ T cells (T CD3 adhesion) and of CD14 + monocytes (Monocyte adhesion) to HDMEC from 8 (T cells) and 5 (monocytes) independent experiments (with mean ± SEM). *P < 0.05, **P < 0.01 We also investigated whether MLN4924 preconditioning modulates the antigen-presenting function of ECs. ECs were treated with MLN4924, activated with TNFα/IFNγ, and after 24 h of stimulation, they were co-cultured with freshly isolated allogenic T lymphocytes for an additional 24 h. The activation of CD4+ and CD8+ T cells was then evaluated by analyzing the expression of CD69 at their surface by flow cytometry. As shown in Fig. 2B, ECs activated with TNFα/IFNγ were able to activate allogenic T lymphocytes, as observed by the induction of CD69 on a subset of alloreactive cells, although it was not sufficient to induce T cell proliferation (data not shown), probably because of the lack of adequate costimulatory molecules on ECs. About 25% of CD8 T cells activated by both MLN4924 preconditioned HDMEC and non-preconditioned HDMEC expressed Granzyme B, suggesting the cytotoxicity of these cells on endothelial cells (Fig. 2B). However, pretreatment of ECs with MLN4924 significantly reduced activation of both CD4 and CD8 T cells (Fig. 2B), suggesting impaired antigen-presenting capacity of ECs following neddylation inhibition, and a probable subsequent decrease in T cell-mediated EC damage. Given the fact that ECs pretreated with MLN4924 did not upregulate adhesion molecules upon TNFα and IFNγ stimulation, we next evaluated whether this resulted in the inhibition of T lymphocyte adhesion to ECs. To do so, HDMEC pretreated for 3 h at 4 °C with MLN4924 at 10 µM or 100 µM and subsequently stimulated for 24 h with TNFα/IFNγ were co-incubated with freshly isolated human PBMC from healthy volunteers. After 1 h of co-incubation, nonadherent PBMC were removed by extensive PBS washing, and adherent leukocytes were collected and analyzed by flow cytometry. As illustrated in Fig. 2C, pretreatment of ECs with MLN4924 significantly inhibited T lymphocyte adhesion to these cells upon activation, and even completely abrogated it at high doses (100 µM), while it had no effect on monocyte adhesion in this setting. Collectively, these results highlight the ability of MLN4924 preconditioning to impair T cell-endothelial interactions and protect ECs from immune activation. Preconditioning of endothelial cells with MLN4924 inhibits their prothrombotic responses We next studied whether MLN4924 pretreatment could prevent EC prothrombotic responses, suggested by our first qPCR screening that pointed at a downregulation of Tissue Factor (F3) and Plasminogen Activator Inhibitor 1, PAI-1 (SERPINE1) gene expression in MLN4924 treated cells (Fig. 1A). This finding was confirmed by qPCR, which showed that MLN4924 pretreatment completely blocked the induction of Tissue Factor and PAI-1 mRNA expression upon TNFα/IFNγ stimulation of ECs (Fig. 3A). We also analyzed the expression of membrane bound Tissue Factor at EC surface by flow cytometry and found that its induction by TNFα/IFNγ stimulation was completely abrogated by MLN4924 pretreatment (Fig. 3B). To determine whether this regulation had functional antiaggregant consequences, we analyzed platelet adhesion to ECs in vitro, and found that MLN4924 pretreatment resulted in a significant decrease in platelet adhesion compared with untreated cells in static conditions, which was confirmed in endothelial cells cultured under flow (Fig. 3C). Collectively, these data demonstrate that MLN4924 preconditioning effectively suppresses the prothrombotic phenotype of ECs, supporting its potential for the prevention of thrombosis after transplantation. Fig. 3Endothelial cell preconditioning with MLN4924 inhibits their pro-thrombotic responses. HDMEC were pretreated with 10 µM MLN4924 for 3 h at 4 °C and further stimulated with TNFα and IFNγ at 37 °C. A The mRNA expression of F3 and SERPINE-1 were analyzed by RT-qPCR after 4 h of stimulation. Scatter plots show the fold change in mRNA expression from 6 independent experiments (with mean ± SEM). B Expression of membrane bound tissue factor was analyzed after 24 h of stimulation by flow cytometry. A representative histogram is shown, and the scatter plots shows the fold change in MFI from 6 independent experiments (with mean ± SEM). C After 24 h of stimulation, HDMEC were washed and co-incubated with freshly isolated and calcein-AM stained platelets from healthy volunteers. After 1-hour, non-adherent platelets were washed away and remaining adherent platelets were visualized by immunofluorescence microscopy. Representative photomicrographs show adhering platelets (green) on HDMEC (magnification x40). The scatter plot represents the fold change in the number of adherent platelets from 6 independent experiments (with mean ± SEM). Platelet adhesion to HDMEC was also assessed under flow and is expressed on the scatter plot as the number of adherent platelets per field of view from 3 independent experiments (with mean ± SEM). *P < 0.05, ***P < 0.001 Endothelial cell preconditioning with MLN4924 inhibits their pro-thrombotic responses. HDMEC were pretreated with 10 µM MLN4924 for 3 h at 4 °C and further stimulated with TNFα and IFNγ at 37 °C. A The mRNA expression of F3 and SERPINE-1 were analyzed by RT-qPCR after 4 h of stimulation. Scatter plots show the fold change in mRNA expression from 6 independent experiments (with mean ± SEM). B Expression of membrane bound tissue factor was analyzed after 24 h of stimulation by flow cytometry. A representative histogram is shown, and the scatter plots shows the fold change in MFI from 6 independent experiments (with mean ± SEM). C After 24 h of stimulation, HDMEC were washed and co-incubated with freshly isolated and calcein-AM stained platelets from healthy volunteers. After 1-hour, non-adherent platelets were washed away and remaining adherent platelets were visualized by immunofluorescence microscopy. Representative photomicrographs show adhering platelets (green) on HDMEC (magnification x40). The scatter plot represents the fold change in the number of adherent platelets from 6 independent experiments (with mean ± SEM). Platelet adhesion to HDMEC was also assessed under flow and is expressed on the scatter plot as the number of adherent platelets per field of view from 3 independent experiments (with mean ± SEM). *P < 0.05, ***P < 0.001 Preconditioning of endothelial cells with MLN4924 induces cytoprotective responses Our first qPCR screening of genes regulated in ECs upon MLN4924 pretreatment identified HMOX1, coding for Heme Oxygenase 1 (HO-1), a well-known cytoprotective and anti-oxidant molecule, as markedly upregulated in treated cells (Fig. 1A). We confirmed this upregulation at the mRNA level by additional qPCR analyzes (Fig. 4A) and at the protein level by Western blot (Fig. 4B). In addition, we analyzed the expression of complement regulatory molecules in ECs preconditioned with MLN4924 and found that CD55 was significantly upregulated at the mRNA level (Fig. 4C) and at the protein level as shown by flow cytometry (Fig. 4D). However, this increase was not sufficient to prevent C3c deposition at the surface of preconditioned ECs (Fig. 4E). Fig. 4Endothelial cell preconditioning with MLN4924 induces cytoprotective responses. HDMEC were pretreated for 3 h at 4 °C with 10 µM MLN4924 and further stimulated for 4–24 h with TNFα and IFNγ. A The expression of Heme Oxygenase was analyzed at the mRNA level by RT-qPCR after 4 h of stimulation. The scatter plot represents the fold change in mRNA expression (with mean ± SEM) from 6 independent experiments. B Heme oxygenase expression was analyzed at the protein level by Western blot after 24 h of stimulation. A representative blot from 3 independent experiments is shown, and the scatter plot represents the quantification of protein bands by densitometry, expressed as the ratio of HO-1/β-actin (with mean ± SEM). C Expression of the complement inhibitory molecule CD55 was analyzed at the mRNA level by RT-qPCR after 4 h of stimulation. The scatter plot shows the fold change in mRNA expression (with mean ± SEM) from 6 independent experiments. D Expression of CD55 was also analyzed at the protein level by flow cytometry after 24 h of activation. A representative histogram is shown, and the scatter plot represents the fold change in MFI from 7 independent experiments (with mean ± SEM). E Deposition of complement C3c at the surface of HDMEC co-incubated with human serum was assessed by flow cytometry. A representative histogram is shown and the scatter plot represents the fold change in MFI from 5 independent experiments (with mean ± SEM). F HDMEC were stored for 24 h at 4 °C in complete hypoxia (0% O2) in UW preservation medium containing or not 10 µM MLN4924. After a PBS wash, they were switched back to 37 °C in normoxia and in complete culture medium. Cell apoptosis was evaluated after 1 h by flow cytometry using annexin V and 7-AAD staining. Representative flow cytometry plots are shown, and the scatter plot represents the percentage of Annexin V+ apoptotic cells in 6 independent experiments (with mean ± SEM). *P < 0.05, **P < 0.01 Endothelial cell preconditioning with MLN4924 induces cytoprotective responses. HDMEC were pretreated for 3 h at 4 °C with 10 µM MLN4924 and further stimulated for 4–24 h with TNFα and IFNγ. A The expression of Heme Oxygenase was analyzed at the mRNA level by RT-qPCR after 4 h of stimulation. The scatter plot represents the fold change in mRNA expression (with mean ± SEM) from 6 independent experiments. B Heme oxygenase expression was analyzed at the protein level by Western blot after 24 h of stimulation. A representative blot from 3 independent experiments is shown, and the scatter plot represents the quantification of protein bands by densitometry, expressed as the ratio of HO-1/β-actin (with mean ± SEM). C Expression of the complement inhibitory molecule CD55 was analyzed at the mRNA level by RT-qPCR after 4 h of stimulation. The scatter plot shows the fold change in mRNA expression (with mean ± SEM) from 6 independent experiments. D Expression of CD55 was also analyzed at the protein level by flow cytometry after 24 h of activation. A representative histogram is shown, and the scatter plot represents the fold change in MFI from 7 independent experiments (with mean ± SEM). E Deposition of complement C3c at the surface of HDMEC co-incubated with human serum was assessed by flow cytometry. A representative histogram is shown and the scatter plot represents the fold change in MFI from 5 independent experiments (with mean ± SEM). F HDMEC were stored for 24 h at 4 °C in complete hypoxia (0% O2) in UW preservation medium containing or not 10 µM MLN4924. After a PBS wash, they were switched back to 37 °C in normoxia and in complete culture medium. Cell apoptosis was evaluated after 1 h by flow cytometry using annexin V and 7-AAD staining. Representative flow cytometry plots are shown, and the scatter plot represents the percentage of Annexin V+ apoptotic cells in 6 independent experiments (with mean ± SEM). *P < 0.05, **P < 0.01 Finally, we studied whether the preconditioning of ECs with MLN4924 could prevent damages to these cells induced by cold storage and hypoxia. To do so, ECs were incubated for 24 h at 4 °C in hypoxia (0% O2) in UW preservation medium, and subsequently washed and switched back to 37 °C in normoxia and in complete endothelial medium, as a model of static cold storage followed by reperfusion. EC apoptosis was then assessed using Annexin V and 7-AAD staining by flow cytometry. We found that the addition of MLN4924 at 10 µM in the preservation medium during the cold storage phase was able to markedly decrease EC apoptosis (Fig. 4F). This suggests that the preconditioning of grafts with MLN4924 during cold storage will have implications for the prevention of endothelial damage associated with ischemia-reperfusion injuries. Discussion Improving graft preservation remains a major clinical challenge in the current context of a persistent shortage of organs available for transplantation that has led to increase the use of organs from extended criteria donors and from donors after cardiac arrest – both highly sensitive to ischemia-reperfusion injuries. These past two decades, organ transplant preservation techniques have been improved, with the introduction of machine perfusion of organs (as opposed to static cold storage in a preservation solution). Hypothermic machine perfusion of transplants from deceased donors has beneficial effects on delayed graft function or early dysfunction and on long term graft survival [21–26]. Ex vivo normothermic machine perfusion using oxygenated perfusion solutions has also been tested and has the advantage of restoring a physiological environment, and thus offers the possibility of better preserving organs (less ischemia), “reconditioning” them (ATP replenishment) and evaluating them ex vivo prior to transplantation [27]. However, keeping organs under normothermic perfusion for long periods poses problems for the elimination of toxic products derived from cellular activity. Besides, this procedure is costly, requires a lot of human resources, and is still difficult to implement on a routine basis. An attractive strategy to further improve graft preservation consists in the treatment of organs with protective agents during the preservation phase, so called “graft preconditioning”. The addition of protective pharmacological molecules to the preservation solution is easy to implement, and makes it possible to target only the graft, notably its endothelium, thus avoiding numerous off-target effects. Here, we identify MLN4924, a small molecular inhibitor of the neddylation pathway, as a promising candidate for endothelial protection with strong potential for clinical implementation in graft preconditioning. In this work, we used an in vitro model of treatment of microvascular endothelial cells in order to reproduce the conditions of a hypothermic organ preservation (preconditioning at 4 °C), followed by warm reperfusion (37 °C) in the presence of the inflammatory factors TNFα and IFNγ (model of sterile inflammation following transplantation). This model enabled us to evaluate the impact of the neddylation inhibitor MLN4924 as a preconditioning agent during the cold preservation phase on later endothelial responses. While the current average time of hypothermic preservation for organ transplants lasts from 2 to 14 h depending on the organ, we first used a relatively short preconditioning time of only 3 h at 4 °C with 10 µM MLN4924 in our molecular analyses, which was defined in preliminary experiments as the optimal treatment dose and duration to obtain a maximal anti-inflammatory effect of MLN4924 on endothelial cells, with low associated cell death (data not shown). Indeed, longer incubation times of ECs in hypothermia resulted in increased cell death in the absence of MLN4924 treatment in vitro, compromising mRNA and protein analyses due to the poor quality of samples from untreated cells. However, it is very likely that longer MLN4924 preconditioning of organs will have equally powerful anti-inflammatory effects on graft endothelial cells. One of the most striking effects of MLN4924 in our model is the strong inhibition of endothelial cell activation and inflammation. Indeed, pretreatment of endothelial cells with 10 µM MLN4924 for 3 h at 4 °C prior to stimulation with TNFα and IFNγ was able to completely abrogate the induction of the adhesion molecules ICAM1, VCAM1 and E-SELECTIN, to inhibit the upregulation of HLA class I and the induction of HLA class II at the surface of ECs, and to strongly decrease the expression and release of IL6, RANTES and Fractalkine. MLN4924 preconditioning also protected ECs from T CD4 and T CD8 lymphocyte-mediated aggression. As a result, preconditioning with MLN4924 could prevent the adhesion of T lymphocytes to ECs, and markedly inhibited EC-mediated allogenic T lymphocyte activation. These anti-inflammatory effects are certainly mostly due to the blockade of the NFkB pathway by MLN4924, which is a major signaling pathway driving endothelial cell activation [28, 29]. Importantly, these effects were obtained after a simple short pretreatment of cells at 4 °C and lasted for as long as 48 h, even though MLN4924 was washed away after pretreatment. This suggests that graft preconditioning with MLN4924 will be able to block sterile inflammation following reperfusion at transplantation. This is particularly interesting since this short pretreatment modality shows little toxicity on ECs, in contrast to prolonged treatments with MLN4924 at 37 °C that significantly impaired EC viability (data not shown). Interestingly, while MLN4924 preconditioning of ECs was able to inhibit adhesion molecule expression at their surface and subsequent T lymphocyte adhesion, it had no effect on monocyte adhesion in our in vitro assays. This discrepancy may be explained by our observation that MLN4924 failed to inhibit the secretion of CCL2 and IL8 (data not shown). Because monocyte adhesion is highly dependent on chemokine-mediated activation of integrins [30], the persistence of these signals might compensate for the reduction in ICAM1, VCAM1 and E-Selectin expression. Another key effect of MLN4924 in our studies is its ability to protect ECs from hypothermia and hypoxia induced apoptosis. Indeed, the addition of MLN4924 in the preservation medium during incubation of ECs for 24 h at 4 °C under hypoxic conditions (0% O2) before switching them back to 37 °C in normoxic conditions was able to strongly decrease subsequent EC apoptosis. This finding supports the potential of MLN4924 preconditioning to confer protection against ischemia-reperfusion injuries. This was rather unexpected since MLN4924 is well known for its proapoptotic properties in cancer cells [31], and has been shown to block both NFkB [17] and mTOR signaling [15], two pathways classically known to drive EC survival [32, 33], by inhibiting the neddylation and further degradation of their intrinsic regulators DEPTOR [15] and IkBα respectively [14]. However, studies have also shown that hypoxia-induced endothelial apoptosis was due to NFkB-mediated Bcl-2 suppression [34], which could explain in part the anti-apoptotic effect of MLN4924 in this setting. In addition, we have found that preconditioning of ECs with MLN4924 resulted in a marked and sustained upregulation of the protective antioxidant molecule HO-1, which is well known to protect ECs against apoptosis [35], and thus probably greatly participates in the mechanisms of EC protection against cell death. Consistent with these results, an anti-apoptotic effect of MLN4924 has also been described in other models such as cerebral endothelial cells [36] or hepatocytes [37]. Another interesting effect of MLN4924 preconditioning was the induction of the complement inhibitor CD55 (DAF) at the surface of ECs. This increase in CD55 expression is likely a consequence of the increase in HO-1 expression upon MLN4924 treatment, as shown in other models [38]. These factors are also known to be implicated in the accommodation phenomenon (EC resistance to antibody aggression), for which MLN4924 could also be of interest [7]. However, despite significant, CD55 increased expression was not sufficient to inhibit complement C3c deposition in our in vitro assays, probably because these cells already expressed very high basal levels of cell surface complement inhibitors, including CD55, in this in vitro setting. Nevertheless, CD55 upregulation may confer a more significant protective advantage against complement-mediated injury in a physiological “whole organ” context. At early times after transplantation, the endothelium is exposed to a broader array of complement triggers that may exceed the protective threshold of basal inhibitor levels, making even modest increases in CD55 functionally relevant. Finally, we found that MLN4924 preconditioning was able to block the induction of the prothrombotic molecules Tissue Factor (F3) and PAI-1 (SERPINE1), probably again through inhibition of the mTOR and NF-kB pathways [39–41], and subsequently inhibited adhesion of platelets to ECs, which suggests that this treatment could decrease the thrombosis risk following transplantation. This could have interesting implications for the preconditioning of pancreatic transplants notably. Indeed, the pancreas is extremely sensitive to ischemia-reperfusion injuries, and early pancreatic allograft failure is mainly due to thromboses that occur in 10% of recipients and are favored by prolonged ischemic times [42]. Altogether, our study identifies neddylation blockade as a very promising strategy for graft preconditioning, involving a synergy of anti-inflammatory and cytoprotective mechanisms. Recent studies have demonstrated the protective effects of neddylation blockade on ischemia-related oxidative stress in other models. Notably, MLN4924 was shown to protect cardiomyocytes against myocardial IRI by restoring autophagic flux [43], and to reduce ischemic brain injury by limiting neutrophil trafficking through ICAM1 and P-Selectin downregulation [36]. Our results expand upon these findings by showing that the anti-oxidant responses triggered in MLN4924-preconditioned ECs are also characterized by a strong induction of HO-1, subsequent resistance to ischemia-reperfusion associated apoptosis, and a massive inhibition of endothelial inflammatory and thrombotic responses. Consequently, MLN4924 may protect the microvasculature not only by preventing the initial wave of inflammatory infiltration but also by enhancing the survival of the endothelium itself under the metabolic stresses of cold storage and reperfusion. MLN4924 has already reached phase 3 clinical trials for patients with myeloid malignancies, suggesting low toxicity of this drug [44]. Importantly, its use in an ex vivo perfusion setting will ensure that only the graft is targeted, with limited “side effects” and lower toxicity. A limitation of our study is that this work was performed in an in vitro model of “ischemia-reperfusion” on microvascular endothelial cells, and the effect of MLN4924 pretreatment on other graft cell types was not directly assessed. In the kidney for instance, another important target of ischemia-reperfusion injuries is the tubular epithelial cell, and it would be interesting to determine whether MLN4924 treatment provides similar protection in this compartment. One study has evaluated the effect of neddylation blockade in lipopolysaccharide (LPS)-induced kidney damage and has showed that MLN4924 treatment of a proximal tubular cell line could inhibit the release of inflammatory cytokines by these cells upon LPS stimulation [45]. MLN4924 treatment at low doses was also protective against LPS-induced acute kidney injury in mice [45]. This suggests that MLN4924 preconditioning might also have protective effects on tubular cells against ischemia-reperfusion injuries in the transplant setting. Future preclinical studies on the preconditioning of whole organ transplants will follow to confirm the potential of MLN4924 as a preconditioning agent. Supplementary Information Supplementary Material 1. Supplementary Material 1.