Local ethics committee approval: 116/14 & 261/22 for peripheral blood from healthy probands, 94/14 for peripheral blood from polytrauma patients. Anonymized information on patient and proband data is published in Seibold et al., 2021 (14).

All mouse experiments were performed in adherence with the National Institute of Health Guidelines on the Use of Laboratory Animals and the European Union “Directive 2010/63/EU on the protection of animals used for scientific purposes” (animal experiment approval numbers: 1413, 1258). Immunohistochemistry staining for PRKD activity in mouse lungs was performed on samples from polytrauma as well as polytrauma with hemorrhagic shock mice acquired from a previous study described in Denk et al., 2018 (15). In brief, to mimic polytrauma in patients, anesthetized mice received a blunt chest trauma, head injury, femur fracture, and soft tissue injury. Additional hemorrhagic shock was induced by drawing blood to reach a mean arterial blood pressure of 30 mm Hg for 60 min upon which mice were resuscitated with a balanced electrolyte solution for 30 min. All mice were sacrificed 4 h post injury and organs as well as blood were harvested. Lungs were embedded in paraffin for further analysis (15). We have subsequently conducted immunofluorescence staining’s on embedded FFPE samples of lung tissue sections from 4 mice each with polytrauma or polytrauma with hemorrhagic shock as well as the respective sham controls as described in 2.5.

TxT experiments were conducted by inducing a defined bilateral lung contusion using an air pressure blast wave on anesthetized 12-week old male C57BL/6 animals (14). The level of pulmonary contusion was chosen based on histologic, cardiopulmonary as well as immunologic changes in earlier studies and was sufficient to induce a profound local and systemic inflammatory response, but without being lethal itself (16). Mice were sacrificed 4 h after the traumatic insult and blood plasma as well as tissue samples were frozen at -80°C for further analyses. The respective mouse experiments were conducted as part of a previous study published in Seibold et al., 2021 (14). Lysates of the lung tissue for Western blots were generated as described in 2.4.

Human umbilical vein cells (HUVECs, ATCC, CRL-1730) were maintained in endothelial cell growth medium MV2 (PromoCell) with MV2 supplement with 1% penicillin and streptomycin. HEK293T cells (ATCC) were maintained in DMEM with 10% fetal calf serum supplemented and 1% penicillin and streptomycin. All cells were kept at 37°C in 5% CO2. To mimic a traumatic microenvironment in vitro, we stimulated HUVECs with a polytrauma cocktail (PTC) of cytokines, chemokines and anaphylatoxins, consisting of Intereukin1-beta, Interleukin 6, CXC-motif-chemokine-ligand-8, complement component-3a, complement component 5a-des-Arg at concentrations measured in patient serum 24 h after polytrauma (17) (IL1β, 200 pg/mL; IL6, 500 pg/mL; CXCL8, 150 pg/mL; C3a, 500 ng/mL; C5a-des-Arg, 10 ng/mL). The role of PRKD in post-traumatic hyperinflammation is investigated in a number of experiments by employing the specific PRKD small molecule inhibitors CRT0066101 and Kb-NB-142-70 (9, 18). Both inhibitors work at nanomolar concentrations. For CRT0066101, IC50 values for PRKD1, 2, 3 are reported as 1, 2.5, or 2 nM, respectively (19). Kb-NB-142-70 inhibits PRKD1, 2, 3 with an IC50 of 28.3, 58.7 and 53.2 nM (20). In cell culture experiments both inhibitors were applied at a concentration 5 µM.

Organs were cut up into smaller pieces and immediately transferred into cryo-tubes that were subsequently stored on dry ice. For long-term storage the tubes were transferred to the freezer set to -80°C on the same day. Tissue lysates were prepared by addition of NP-40 Lysis buffer [150 mM sodium chloride, 1% NP-40, 50 mM Tris pH 8.0] (14) supplemented with 1 × Protease Inhibitor Cocktail and 1× Phosphatase Inhibitor Cocktail (both Sigma–Aldrich) to 20 mg of frozen tissue pieces in a 2 mL reaction tube. Homogenization of samples was achieved using a TissueLyser (Qiagen) at a frequency of 50 s^–1 for 5 min. The samples were centrifuged for 20 min at 12000 rpm at 4°C and the cleared lysate was stored at -20°C.

Immunofluorescence (IF) staining on lung tissue sections from mice was conducted as described previously (13, 21). The lung tissue sections (5µm thickness) were deparaffinized and rehydrated using xylene and a graded alcohol series. The masked epitope was revealed by heat-induced epitope retrieval using Dako antigen retrieval solution (Dako #S1699). Subsequently, the mix of primary antibodies CD34 (#ab8158 Abcam; dilution 1:100) and Phospho-PKD/PKCμ (Ser916) (#2051 Cell Signaling; dilution 1:100) diluted in TBS + 0,05% Tween was applied on tissue sections and incubated under humidified atmosphere overnight at 4°C. Following washing with TBS-Tween, tissue sections were incubated for one hour at RT with the combination of fluorophore labelled secondary antibodies (goat anti-Rabbit IgG (H+L)-Alexa Fluor™ 647, #A21244; goat anti-Rat IgG (H+L)-Alexa Fluor™ 568 ‚#A110077). Sections were mounted using ProLong™ Gold Antifade mount. Nuclei were stained with 4′,6-Diamidin-2-phenylindol (DAPI). Images were acquired by a Leica TCS SP8 confocal laser scanning microscope in sequential scan mode with equal settings using a HC PL APO CS2 40x/1.30 OIL objective and HyD detectors. pPRKD^S916 signal intensities (8-bit) in airy confocal sections were quantified by generating detection masks from the CD34 signal using NIH ImageJ with the following commands: Threshold, close, erode, analyze particles (10-infinity; show masks; add to manager; include holes; summary).

Cells from cell culture experiments were lysed directly in the well by addition of 700 µL QIAzol lysis reagent (Qiagen). The wells were incubated for 5 min. Next, the contents of the wells were transferred to a new tube and RNA was isolated with the miRNeasy^® Mini Kit (Qiagen) according to the manufacturer’s instructions.

RNA was extracted from cells with the miRNeasy^® Mini Kit (Qiagen) according to the manufacturer’s instructions. cDNA was prepared from 200 ng of total RNA, using the iScript™ cDNA Synthesis Kit (Biorad) for mRNA. qPCR reactions were performed with PowerUpTM SYBRTM Green Master Mix (Applied Biosystems) (14). The thermal cycling conditions were set as follows: 2 min at 95°C, followed by 45 cycles of 15 s at 95°C denaturation and 1 min at 60°C for the anneal/extension phase. The relative gene expression levels were calculated by the “delta-delta Ct-method”.

Blood samples were subjected to centrifugation at 2000 rpm for 10 min at RT to separate the plasma from the cellular fraction. Hereafter, the samples were centrifuged at 13000 rpm for 20 min and the resulting plasma supernatants were transferred to new LoBind reaction tubes (Eppendorf). The sEVs were purified from the plasma by size exclusion chromatography with Exo-Spin™ columns (Cell Guidance Systems) according to the manufacturer’s instructions as described previously (13, 14).

HUVECs were grown to a confluency of ~ 90%, cells were washed with PBS and incubated in fresh serum-free MV2 medium for a duration of 16 h at 37°C, 5% CO₂. The supernatants were harvested, processed by centrifugation and 0.2 µm filtration to remove debris, concentrated using ultrafiltration at 2000 rpm, 4°C by Vivaspin Turbo 15 (100 000 MWCO) units (Sartorius, #VS15T41) followed by precipitation and size exclusion chromatography with the Exo-Spin™ system (Cell Guidance Systems) according to the manufacturer’s instructions and as described previously in Seibold et al., 2021 (14).

Coimmunoprecipitation experiments (CoIPs) were performed as described previously (21) by precipitating VE-cadherin from HUVEC cell lysates and detection of coprecipitated binding partners Cortactin, Vinculin and β-catenin in the immunoprecipitated samples from shScramble, shPRKD1 stable HUVEC knockdown cell lines as well as cells treated with the PRKD inhibitor CRT0066101 (5 µM). Samples were analyzed by Western blot.

To generate total cell lysates, cells were treated with lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 5 mM MgCl₂, 1% Triton X-100, supplemented with 1 × Protease Inhibitor Cocktail and 1× Phosphatase Inhibitor Cocktail. The resulting supernatants were mixed with 5x Laemmli Sample Buffer and incubated at 95°C for 10 min. The samples were then subjected to SDS-PAGE followed by Western blotting and detection by ECL as described previously (14).

NTA was performed using a Nanosight NS300 device with the respective software (Malvern Panalytical). Isolated sEVs were diluted in particle-free sterile filtrated PBS and injected using a syringe pump. Particles were tracked under constant flow. Each sEV sample was measured three times for 60 s.

The purified sEVs were subjected to negative staining. To this end, 5 µL of each sample was added onto a glow discharged carbon-coated copper TEM grid and incubated for 1 min. Next, the grid was washed 3 x with 7 µL of Mili-Q water and stained 3 x with 7 µL of 3% uranyl acetate. During the last staining step, the uranyl acetate drop remained 30 s on the grid before the liquid was removed and the grid was left to dry for at least 15 min. The grids were imaged using a JEOL 1400 TEM.

Healthy donors were subjected to venipuncture and blood samples were collected in EDTA monovettes. The isolation of PMNs was conducted using Polymorphprep (Axis Shield) according to the manufacturer’s instructions. In brief, 5 mL of blood where overlayed on top of the same volume Polymorphprep in a 15 mL tube and centrifuged at 550 x g for 35 min at RT. The resulting layers containing the PMNs were pooled and mixed with RPMI media in a 1:10 ratio. The cells were pelleted by centrifugation at 400 x g for 10 min and afterwards the supernatant was discarded and the pellet was resuspended in RPMI media (9).

A number of 8x10⁴ HUVEC cells per insert were seeded onto Corning Costar Transwell filters with 5 μm pore size (Sigma-Aldrich). The cells were grown for 48 h at which point PTC, CRT0066101 (5 µM) or DMSO were added to the upper compartment of the filters and incubated for the indicated time periods. Then, the medium with the inhibitors was removed and 100 µl of fresh MV2 medium was added to the upper compartment of the filter. The filter inserts were transferred to a new plate with 800 µl of MV2 medium per well containing chemoattractant (fMLP, 100 nM) or DMSO.

Human PMNs were harvested as described in 2.14 and labeled with CellTracker Deep Red Dye (Invitrogen) for 1 h as stated by the manufacturer. Finally, a number of 5x10⁵ labelled PMNs was added to the upper compartment of the inserts to initiate the experiment. After 1 h of transmigration the filter inserts were removed and relative fluorescence intensities in the respective wells was measured using a Tecan M200Pro plate reader in scanning mode (excitation: λ = 630 nm, emission: λ = 660 nm).

A number of 2x10⁴ HUVEC cells per well were seeded in a 96-well plate and grown for 48 h. The supernatants of the wells were discarded and PTC, CRT0066101 (5 µM) or DMSO in MV2 medium was added to the appropriate wells and incubated for 4 h at which point the supernatant was removed. Human PMNs were harvested and labelled with CellTracker Deep Red as described 2.15. The labelled PMNs were added to the wells (1x10⁵ PMNs/well) and incubated for 1 h at 37°C. To remove non-adhesive PMNs, the plate was washed three times with RPMI and analyzed by measurement of relative fluorescence intensities using a Tecan M200Pro plate reader in scanning mode (excitation: λ = 630 nm, emission: λ = 660 nm).

A number of 8x10⁴ HUVEC cells per insert were seeded onto Corning Costar Transwell filter inserts with 5 μm pore size (Sigma Aldrich) and grown for 48 h. The compartment in the top of the insert was filled with 200 µL of MV2 medium containing Thrombin, PTC, CRT0066101 (5 µM) or DMSO and filters were transferred into a CellZscope TEER device (NanoAnalytics) where TEER was measured by impedance spectroscopy (14).

A number of 1.2x10⁵ HUVEC cells per insert were seeded onto Corning Costar Transwell filter inserts with 5 μm pore size (Sigma Aldrich) and grown for 48 h. At this point DMSO, Thrombin, PTC, CRT0066101 or Kb-NB-142-70 (both 5 µM) were added to the upper compartment of the filters and incubated as indicated. These solutions also contained FITC-albumin (Sigma–Aldrich) in a concentration of 40 µg/mL. After the incubation period, the filter inserts were removed and the FITC fluorescence was measured in the bottom well using a Tecan M200Pro plate reader (excitation: λ = 490 nm, emission: λ = 520 nm).

All ELISAs were performed according to the manufacturer’s instructions. For IL6 (Thermo Scientific), CXCL8 (R&D systems) and Leukotriene B₄ (Enzo, R&D systems) cell culture supernatants from HUVECs or shPRKD1-knockdown HUVEC cell lines were used. Standard curves were created using non-linear- 4P- regression.

HUVEC cells grown in 6-well plates and treated with DMSO, PTC, CRT0066101 (5 µM) or Kb-NB-142-70 (5 µM) as indicated. Cells were washed with PBS and detached by PBS supplemented with EDTA (Sigma–Aldrich) for 3 min. The cells were resuspended in FACS buffer (PBS, 5% FCS, 0.05% EDTA) that contained the fluorescently labelled primary-antibodies (Miltenyi Biotec). Staining was performed on ice for 1 h. Flow cytometry was done using an LSRII special order FACS System (BD Bioscience) or an Attune NxT analytical flow cytometer (Thermo Scientific). Data was evaluated and plotted with Flowjo (Agilent).

Lentiviruses were produced by transfecting HEK293T cells at 80% confluency using Lipofectamine 3000 (Invitrogen) with psPAX2 and pMD2.G packaging constructs as well as the respective shRNAs ( Supplemental Table S1 ) as described previously (22, 23). The supernatants containing the virus were harvested after 48 h and stored at -80°C. HUVEC cells were transduced with the respective lentivirus and subsequently selected with puromycin (6 µg/mL) for 6 weeks to generate stable shRNA cell lines.

For nucleofection of pcDNA3-PRKD1 (24) expression constructs, the HUVEC nucleofector kit (Lonza) was used according to the manufacturer’s instructions and as described previously (14).

Mass spectrometry was performed on sEVs from DMSO and CRT0066101 (5 µM) treated HUVECs that were isolated after 16 h of treatment (n = 5 samples per condition). The sample preparation and analysis were performed as described in Seibold et al., 2021 (14).

A number of 1x10⁵ HUVEC cells were seeded in a 24-well plate and transfected at 90% confluency with the indicated siRNAs using Lipofectamine 3000. After 48 h, the supernatants were discarded and the cells were treated with PTC, CRT0066101 (5 µM) or DMSO.

HUVEC cells were seeded into a 24-well plate at a concentration of 1x10⁵ cells/well. After 24 h, the supernatants were discarded and 5x10⁵ human PMNs were added, followed by PTC, CRT0066101 (5 µM) or DMSO treatment. This coculture was incubated for 3 h 45 min at which point fMLP was added (100 nM). After 15 min the supernatants were collected, centrifuged for 2 min at 2000 rpm at RT and stored at -20°C. Respiratory burst experiments with PMNs were performed using the Neutrophil/Monocyte Respiratory burst assay kit (#601130, Cayman Chemical) according to the manufacturer´s instructions. PMNs were incubated with DHR 123 reagent at a concentration of 1x10⁵ cells/mL in Assay Buffer (RPMI, 1 mM Calcium Chloride) for 30 min before they were added to HUVECs that had been seeded into a 24-well plate at a concentration of 1x10⁵ cells/well and grown for 24 hours. This coculture was incubated for 3 h 45 min at which time point the PMNs were harvested. Next, fMLP was added (100 nM) to the harvested PMNs and incubated for 15 min. Subsequently, PMNs were resuspended in FACS buffer, stored on ice and analyzed by flow cytometry using the FITC-A channel.

HUVEC cells were seeded into a 24-well plate and incubated until 90% confluency had been reached. At this point, DMSO, PTC, CRT0066101 (5 µM) or Kb-NB-142-70 (5 µM) were added to the respective wells and incubated for 4 h, as indicated. These supernatants were then harvested and added to PMNs that were purified from whole blood. This mixture was incubated again for 4 h, then fMLP was added at a concentration of 100 nM. After 15 min the supernatants were collected by centrifugation for 2 min at 2000 rpm at RT and stored at -20°C. For the respiratory burst assays, the respective supernatants were added to PMNs pretreated with DHR 123 reagent at a concentration of 1x105 cells/mL in Assay Buffer (RPMI, 1 mM Calcium Chloride) for 30 min. PMNs were incubated for 3 h 45 min at which time point fMLP was added at a concentration of 100 nM. After 15 min the PMNs were collected by centrifugation, resuspended in FACS buffer, stored on ice and analyzed by flow cytometry using the FITC-A channel.

For confocal imaging, immunofluorescence staining was performed as described previously (25, 26). In brief, 1.5 × 10⁵ HUVEC cells were seeded on coverslips in 12-well plates and grown to confluency. Subsequently, HUVEC monolayers were treated with PTC, CRT0066101 (5 µM) or DMSO for 8 h. Samples were fixed using formaldehyde (3.75%) for 20 min at RT and permeabilized with Triton X-100 (0.1%) in PBS for 30 s. After blocking (5% FCS, 0.05% Tween-20 in PBS) for 20 min, samples were stained with primary antibodies against VE-cadherin, β-catenin overnight at 4°C. Samples were washed three times with PBS and incubated with secondary antibodies conjugated with the respective Alexa fluorophores as well as Phalloidin-Alexa-Fluor-647 (1:60) and DAPI (Thermo Scientific) for 1 h at RT. Coverslips were mounted in Fluoromount-G (Thermo Scientific). Intensity quantification was performed as described previously using NIH ImageJ (26). Experiments were analyzed by a confocal laser scanning microscope TCS-SP8-HCS (Leica) equipped with an HC PL APO CS2 40×/1.3 oil immersion objective. Images were acquired in sequential scan mode with equal settings within the linear range of detectors. Quantitative mean of ROI analysis was conducted by NIH ImageJ, averaging intensities from 3 similar-sized sub-ROIs per junction.

Unweighted enrichment analysis for MS data was performed using the EnrichR meta-analysis web tool (https://amp.pharm.mssm.edu/Enrichr/) to query the GO Biological process database with up- or down-regulated proteins (cutoff of (log 2≥ +1,5, ≤-1,5)).

Illustrations were created using Servier medical art templates (https://smart.servier.com/) with minor modifications according to terms of the creative commons attribution 3.0 license agreement (https://creativecommons.org/licenses/by/3.0/).

Quantification of Western blot bands was conducted using ImageJ “Gels Submenu” or “Gel plotting macros” (https://imagej.nih.gov/nih-image/manual/tech.html#analyze). Statistical analysis was performed using Prism software, version 9.4 (GraphPad, San Diego, CA). Graphs depict mean ± SEM for all conditions. Statistical significance: ns, not significant, *P = 0.05–0.01, **P = 0.01–0.001, ***P < 0.001, ****P < 0.0001.

The scope of the post-traumatic immune response is critically determined by the coordinated interaction of PMNs with the endothelium that is controlled at multiple levels in both cell types. We have recently shown that PRKD1 is implicated in modulating neutrophil cell motility upon trauma (9). Since PRKD is activated by cellular stresses, e.g., reactive oxygen species (27) and also modulates cell-cell adhesion (21), we were prompted to investigate, the role of PRKD in the context of the post-traumatic innate immune response of the endothelium. To this end, we stimulated human umbilical vein cells (HUVECs) in vitro with a polytrauma cocktail (PTC), consisting of Intereukin1-beta, Interleukin 6, CXC-motif-chemokine-ligand-8, complement component-3a, complement component 5a-des-Arg (IL1β, IL6, CXCL8, C3a and C5a-des-Arg) at concentrations measured in patient serum 24 h after polytrauma (17). Indeed, PRKD was activated and stayed active between 4 to 8 h after stimulation, as measured by pPRKD^S910-autophosphorylation of the kinase (26). After 16 h PRKD activity returned to almost basal levels ( Figure 1A ). PRKD activation at the 4 h time point correlates with a peak for cytokine transcripts in lungs after severe blunt chest trauma (14, 28). PRKD was also activated by the barrier-destabilizing agent thrombin (29). Compared to PTC, activation of PRKD by thrombin was fast, peaking 2 min after stimulation followed by a drop to almost basal levels within 30 min ( Figure 1B ). Excessive non–wound-related thrombin generation (30, 31) is a hallmark of acute traumatic coagulopathy (ATC) and associated with poor outcome (31). It is caused by a combination of tissue trauma with second hits, such as shock, hemodilution, hypothermia or acidosis (31). Thus, PRKD is activated by important trauma-associated stimuli in vitro with distinct kinetics. To examine activation in vivo, we used a well-established trauma model and measured PRKD activity and concomitant nuclear factor k-light-chain-enhancer of activated B cells (NFκB) activation by pP65^S536-phosphorylation to indicate pro-inflammatory signaling in lung lysates 4 h after blunt chest trauma (TxT) in mice (14). Figure 1C shows both, significantly increased pPRKD^S910- as well as NFκB pP65^S536- phosphorylation after TxT compared to sham controls. In vivo activation of PRKD in endothelial cells upon trauma was examined by immunohistochemistry in lung samples of mice that were previously subjected to severe polytrauma with and w/o hemorrhagic shock (PT +/- HS) to elicit a major trauma response (15). Polytrauma in combination with hemorrhagic shock is known to increase overall mortality in patients, whereby hemorrhagic shock significantly contributes to the systemic release of cytokines, such as IL6. In these experiments, increased costaining of pPRKD^S916-autophosphorylation with the endothelial marker CD34 was detected by confocal imaging concomitant with infiltration of PMNs in lung sections ( Figures 1D, E ). Also, polytrauma without hemorrhagic shock resulted in a similar trend ( Figure S1A ). Thus, our data demonstrate, PRKD is activated in endothelial cells upon trauma in vitro and in vivo ( Figures 1A–D ).

Endothelial barrier stability is a critical pathophysiological parameter determining the passage of PMNs towards the site of injury (32, 33). Barrier integrity requires VE-cadherin adhesion contacts at the plasma membrane (34) and since PRKD1 has been implicated in the regulation of E-cadherin mediated cell-cell adhesion (21), we went on to characterize the functional state of the barrier using FITC-albumin flux experiments and transendothelial electrical resistance measurements (TEER) across differentiated HUVEC monolayers on transwell filters (14). A potential role of PRKD activity was examined using the selective PRKD small molecule inhibitors CRT0066101 and Kb-NB-142-70 (9, 18). Both inhibitors have been described to work at nanomolar concentrations. For CRT0066101, IC50 values for PRKD1, 2, 3 were determined at 1, 2.5, or 2 nM (19). The Kb-NB-142-70 inhibitor was shown to inhibit PRKD1 with an IC50 of 28.3 ± 2.3 nM, PRKD2 with 58.7 ± 4.2 and PRKD3 at 53.2 ± 3.5 nM (20). Thus, we have applied both inhibitors at 5 µM concentrations to evaluate if endothelial barrier stability is affected by inhibition of PRKD kinase activity. Interestingly, both inhibitors strongly increased basal endothelial barrier stability in FITC-albumin flux experiments across Transwell filters ( Figure 2A ). A similar observation was made by shRNA mediated knockdown of the PRKD1 isoform (shPRKD1), whereas shScramble was used as a control, suggesting that depletion of PRKD1 is indeed sufficient to stabilize barrier integrity ( Figure 2B ). Knockdown of PRKD1 by shRNA in stable cell lines was confirmed by Western blots ( Figure S1H ). Of note, at the CRT0066101 concentrations of 5 µM used in assays, overall cell viability and numbers of HUVEC cells were only mildly affected after 8 h ( Figure S1B ). To assess whether PRKD inhibition would revert destabilization of the endothelial barrier by thrombin or PTC, we performed additional TEER and FITC-albumin flux experiments. In line with Seibold et al., 2021 (14), destabilization of the endothelial barrier by thrombin was fast and peaked around 35 min after thrombin stimulation in TEER assays (T52 min). Inhibition of PRKD by CRT0066101 enhanced basal barrier stability and reverted barrier destabilization by thrombin ( Figure 2C ). FITC-albumin flux assays performed 2 h after thrombin stimulation verified the marked destabilization of the endothelial barrier by thrombin that was fully recovered upon cotreatment with CRT0066101 ( Figure 2D ). TEER measurement upon incubation of HUVEC cells with PTC revealed a slow, gradual decline in barrier stability that continued until the end of the measurement period at 8 h, where significant barrier destabilization was observed. Again, the barrier was stabilized when cells were incubated with PTC and CRT0066101 in combination as compared to PTC, on its own ( Figure 2E ). FITC-albumin flux also demonstrated a significant restabilization of the endothelial barrier upon CRT0066101 treatment of PTC samples after 4 h, corroborating the TEER data ( Figure 2F ). Thus, inhibition of PRKD during thrombin or PTC-mediated barrier destabilization was able to completely restore the endothelial barrier in vitro. The effect of PRKD inhibition on endothelial barrier stability was substantial and comparable to inhibition of Rho-associated protein kinase (ROCK), a major kinase regulator of barrier stability, when using the Y27632 inhibitor (10 µM) (35) ( Figure S1C ).

Our data suggest that PRKD prominently modulates endothelial barrier stability, even under basal conditions ( Figures 2A, B ). To uncover the underlying molecular mechanisms, we tested the expression of relevant cell adhesion and tight junction proteins upon inhibition of PRKD by CRT0066101 for 4 and 8 h, respectively. There were no changes in protein expression, even though PRKD pS910-autophosphorylation was abrogated ( Figure S1D ). Next, we examined VE-cadherin and β-catenin levels at adherens junctions using quantitative immunofluorescence and confocal microcopy in confluent HUVEC monolayers ( Figure 3A, B ). Upon exposure of cells to PTC for 8 h, adherens junctions appeared destabilized, as indicated by significantly lower intensities of VE-cadherin and β-catenin at the junctions. These effects were completely reversed by CRT0066101. Of note, in line with loss of adhesion complexes also junctions were physically disrupted in the PTC condition (arrow heads). Adherens junctions are stabilized by linking engaged VE-cadherin-β-catenin complexes to the underlying actin cytoskeleton (21). F-actin linkage is achieved by actin-binding proteins, such as Vinculin or Cortactin, both of which were described to further bind β-catenin (21). As we have previously shown regulation of this process for E-cadherin adhesion complexes downstream of PRKD1 phosphorylation (21), we performed coimmunoprecipitation (co-IP) experiments to investigate molecular interactions in VE-cadherin adhesion complexes upon stable knockdown of PRKD1, or inhibition of PRKD activity by CRT0066101 ( Figures 4A, B ). In line with our previous findings (21), Vinculin, Cortactin, and β-catenin coprecipitated with VE-cadherin after knockdown of PRKD1, suggesting improved F-actin binding in the absence of PRKD. Similar effects were evident following PRKD inhibition ( Figure 4A ). For the E-cadherin adhesion complex, we have previously demonstrated that these molecular changes are mediated by abrogating phosphorylation of the PRKD substrate Cortactin at S298 (21). In addition to PRKD, also ROCK and RhoGTPases are known regulators of adhesion complex stability (14). We thus investigated expression of the endothelial cell relevant barrier-regulatory transcripts ROCK1 and RhoB upon PTC and CRT0066101 treatment of HUVEC cells and observed a significant upregulation in PTC-treated samples (14). PRKD inhibition significantly reduced transcript levels and cotreatment of PTC & CRT0066101 significantly reversed the respective upregulation ( Figure 4C ). To corroborate these findings, we have also shown downregulation of ROCK1 upon CRT0066101 incubation on protein level ( Figure 4D ). Thus, our data indicate that active PRKD is involved in barrier destabilization by multiple mechanisms, including remodeling of VE-cadherin adhesion complexes and induction of the Rho-ROCK-signaling axis downstream of PTC.

To migrate towards an injured site, PMNs need to adhere to the endothelium, both to undergo rolling-ball migration (36), but also to initiate diapedesis (37, 38). On the endothelial side, adhesion is facilitated by the surface expression of cellular adhesion molecules, such as ICAM-1, VCAM-1 or E-selectin (10). Indeed, we found that treatment of HUVEC monolayers with PTC significantly increased PMN adhesion after 1 h. This effect was abrogated in the presence of CRT0066101, implying that PRKD activity is also involved in controlling neutrophil adhesion to the endothelium ( Figures 5A, B ). PTC markedly increased transcript levels of cellular adhesion molecules, like ICAM-1, VCAM-1 and SELE (E-selectin) after 8 h ( Figure 5C ). In line with adhesion studies, CRT0066101 treatment impaired expression of all cellular adhesion molecule transcripts and reverted PTC-mediated upregulation of cellular adhesion molecules on the RNA level. Surface expression of cellular adhesion molecules on endothelial cells was detected by flow cytometry after 2, 4 and 8 h of PTC treatment ( Figure 5D ). Again, PTC markedly increased their respective surface presentation and the PRKD inhibitor CRT0066101 normalized expression levels. At the example of ICAM-1, this regulation was further verified using the Kb-NB-142-70 inhibitor ( Figure 5E ) as well as for VCAM-1 with knockdown of PRKD1 ( Figures 5F , S1H ). Since all three cellular adhesion molecule transcripts are known NFκB target genes (39–41) and PRKDs are described to modulate NFκB activity under certain conditions (42), we determined pP65^S536-phosphorylation as a marker for NFκB activation. In line with cellular adhesion molecule expression profiles, increased pP65^S536-phosphorylation was detected following PTC stimulation, which was prevented by cotreatment with CRT0066101 ( Figure 5G ). Cellular adhesion molecule expression in endothelial cells therefore is regulated downstream of PRKD1 and NFκB after stimulation with pro-inflammatory cytokines and anaphylatoxins.

NFκB also controls expression of important post-traumatic pro-inflammatory mediators e.g., IL6 and CXCL8 (43, 44). Indeed, the NFκB inhibitor BAY 11-7082 (45) impaired the PTC-induced increase in IL6, CXCL8 and of all three cellular adhesion molecule transcripts ( Figure 6A ). In addition, important pro-inflammatory, neutrophil-relevant cytokines/chemokines, such as IL6, CXCL8, CXCL2, CXCL5, C-C-chemokine ligand 2 (CCL2), IL1β, but also C3 were upregulated by PTC, and this upregulation was reverted by CRT0066101 ( Figure 6B ). In most cases, the PRKD inhibitor even reduced basal transcript levels. This is of relevance, as CXCL2/5/8 (46–48) and CCL2 (49) act as neutrophil chemoattractants and IL1β as well as IL6 have pro-inflammatory functions (50, 51). We have also quantified release of IL6 and CXCL8 by HUVEC cells using ELISAs, which demonstrated the same regulation that could be further corroborated at the example of CXCL8 using the second PRKD inhibitor Kb-NB-142-70 ( Figures 6C, D ). The concurrent regulation of the transcripts IL6, CXCL8, IL1β and ICAM-1 could also be demonstrated by knockdown and re-expression of PRKD1. Depletion of PRKD1 in HUVECs reverted the upregulation of the respective transcripts upon PTC stimulation and re-expression of PRKD1 in the knockdown cells reestablished a significant up-regulation upon PTC treatment ( Figures 6E , S1H ). Of note, we were able to titrate PRKD1 expression from complete knockdown back to endogenous levels, providing clear evidence for regulation of the transcripts downstream of PRKD1. In addition, we also quantified CXCL8 secretion by ELISAs as part of the same experiments with congruent results ( Figures 6F , S1H ).

Thus, our data demonstrate that activation of PRKD1 in endothelial cells downstream of PTC is implicated in the regulation of cellular adhesion molecules as well as neutrophil-relevant cytokines/chemokines, and it critically destabilizes the endothelial barrier. Therefore, we determined whether trans-endothelial migration of neutrophils would be affected by measuring the passage of fMLP stimulated PMNs across differentiated confluent HUVEC monolayers. PRKD1 depletion significantly impaired PMN migration with and w/o a chemotactic fMLP gradient after 90 min ( Figure 6G ). In addition, we have pretreated HUVEC cells for 4 h with PTC and/or CRT0066101, before performing transendothelial migration assays against fMLP gradients. Here, CRT0066101 reverted increased PMN migration upon PTC treatment, supporting our previous molecular findings ( Figures 6H, I , S1H ).

The extent of the post-traumatic immune response is critically dependent on the self-organized establishment of secondary chemotactic gradients by neutrophils. After extravasation and chemotaxis towards injuries, neutrophil behavior switches to highly coordinated unidirectional motion and clustering at wound cores (11, 52). This swarming process relies on paracrine release of the lipid attractant LTB₄ (11). Yet, it currently remains unclear how neutrophil activation and chemoattractant synthesis are coordinated in the context of intercellular crosstalk with the endothelium. Since LTB₄ biosynthesis (53) in neutrophils can be stimulated by chemokines, such as CXCL8 (54), we wondered whether endothelial cells and the PRKD1 signaling axis would also contribute to the regulation of secondary LTB₄ gradients. Moreover, we have recently shown that post-traumatic endothelial sEVs are vital propagators of the inflammatory response after traumatic injury (14), and PRKD1 can quantitatively and qualitatively alter sEV secretion (13). We therefore performed in vitro coculture studies with PMNs and HUVEC cells to investigate intercellular communication. LTB₄ production by PMNs alone was not dependent on PTC treatment or PRKD activity, respectively ( Figure 7A ). However, upon coculture of PMNs with HUVECs, LTB₄ generation was significantly increased by PTC stimulation. This increase was prevented by CRT0066101 coincubation, as measured by ELISAs in culture supernatants ( Figure 7B ). Of note, these effects were even further potentiated by more than 50-fold when HUVEC cells were treated with PTC/CRT0066101 separately, and supernatants were transferred to fMLP activated PMNs to induce LTB₄ synthesis ( Figures 7C, D ). Thus, some components in the PTC treated supernatants of HUVEC cells robustly induced LTB₄ production by fMLP activated PMNs, which was impaired upon PRKD inhibition. To validate this regulation for PMNs, we first excluded that the HUVEC cell culture supernatants contained LTB₄ ( Figure S1I ). In addition, we confirmed increased LTB₄ production by supernatants upon PTC stimulation of HUVEC cells and inhibition by coincubation with the second PRKD inhibitor Kb-NB-142-70, as well as after knockdown of PRKD1, which was rescued by PRKD1 re-expression ( Figures 7E, F , S1H ). Since the HUVEC cells release large amounts of additional CXCL8 upon PTC stimulation ( Figure 6C ), we wondered whether this would contribute to the increase in LTB₄ generation by PMNs following incubation with PTC stimulated HUVEC supernatants. To this end, we have initiated a depletion of CXCL8 from HUVEC cells prior to induction with PTC by siRNAs. As shown in Figure 7G (left side), we were able to strongly impair CXCL8 release into HUVEC supernatants upon PTC stimulation. However, LTB₄ production was only partially inhibited by CXCL8 knockdown by 22% ( Figure 7G , right side). Thus, additional components in the HUVEC supernatants are likely to be involved in triggering the major increase in LTB₄ biosynthesis of PMNs.

To evaluate whether soluble factors or intercellular communication by EVs are responsible for triggering LTB₄ synthesis in PMNs, we separated HUEVC supernatants into the respective fractions using 100,000g ultracentrifugation. Indeed, ultracentrifuged supernatants with soluble factors from PTC treated HUVECs significantly increased LTB₄ release by PMNs, and the respective supernatants from PTC & CRT0066101 cotreated HUVECs partially reversed LTB₄ synthesis. Interestingly, the EV pellet fraction alone barely enhanced LTB₄ release, and a PRKD-dependent regulation of LTB₄ was not detectable. However, when the respective supernatant and pellet EV fractions were recombined, LTB₄ biosynthesis was prominently increased beyond the levels achieved by soluble factors, alone. The same increase in LTB₄ synthesis was not observed in the PTC & CRT0066101 cotreatment group ( Figure 8A ). Thus, both secreted soluble factors and EVs from endothelial cells are likely required to work in concert for efficient LTB₄ biosynthesis by PMNs. To corroborate involvement of EVs, and in particular sEVs, described as vital regulators of post-traumatic inflammation (14), we treated HUVEC cells with PTC and the sEV biogenesis inhibitor GW4869, which blocks ceramide synthesis through inhibition of neutral sphingomyelinase (nSMase) (14, 55). The respective supernatants were harvested and incubated with PMNs for LTB₄ production experiments ( Figure 8B ). Vice versa, we have also employed an inhibitor of macropinocytosis (Amiloride) to block this major route for sEV uptake (56), in order to impair the transfer of HUVEC derived PTC-sEVs to PMNs ( Figure 8C ). Again, both approaches significantly impaired LTB₄ production upon treatment with PTC supernatants and therefore support modulation of LTB₄ biosynthesis in PMNs via endothelial cell derived sEVs ( Figures 8B, C ). These data were further corroborated by knockdown studies using siRNAs to deplete the essential endosomal-sorting-complex-required-for-transport (ESCRT)-III component VPS24 to impair ESCRT-dependent sEV biogenesis and the nSMAses, SMPD2/3 to inhibit ESCRT-independent sEV release (57) from HUVEC cells ( Figures 8D , S1F–G ). In line with the inhibitor experiments, this strategy also strongly impaired LTB₄ production in PMNs upon incubation with HUVEC derived PTC-supernatants. This suggests a vital role for post-traumatic HUVEC-sEVs in LTB₄ synthesis by PMNs, which further requires secretory signals, such as endothelial CXCL8 for execution.

Having demonstrated the requirement for post-traumatic sEVs and PRKD signaling in HUVEC cells for LTB₄ biosynthesis in PMNs, we next investigated how CRT0066101 would alter PTC induced sEV secretion. We have previously shown that PTC enhances the release of pro-inflammatory sEVs from HUVEC cells that carry cytokine and cellular adhesion molecule transcripts to propagate inflammation to other endothelial cells (14). Crosstalk of these sEVs with PMNs, however was not studied in detail. To identify HUVEC sEV cargos implicated in LTB₄ production that are also changed by PRKD inhibition, we conducted a MISEV analysis (58) in combination with proteome profiling. In line with our previous work in cancer cells (13), inhibition of PRKD significantly increased sEV release, as measured by nanoparticle tracking analysis (NTA) as well as sEV protein markers ( Figures 9A, B ). Presence of nanovesicles in samples was corroborated by transmission electron microscopy (TEM) ( Figure 9C ). Interestingly, EnrichR meta-analysis (59) for GO terms of mass spectrometry (MS) data on the respective sEVs obtained from CRT0066101 treated HUVEC cells, demonstrated a significant enrichment of signatures associated with neutrophil degranulation, activation, and immunity ( Figure 9D ). Moreover, we identified a downregulation of the vital leukotriene biosynthesis enzyme LTA₄ hydrolase (LTA₄H) ( Figure 9F ) in CRT0066101 derived HUVEC-sEVs. LTA₄H is essential to catalyze the last step of LTB₄ synthesis (53). We therefore tested whether depletion of LTA₄H from HUVEC cells prior to sEV production would impact on LTB₄ levels released by PMNs. Indeed, we show a significant reduction in LTB₄ biosynthesis from PMNs upon LTA₄H knockdown in HUVECs ( Figure 9G ). Next, we determined LTA₄H levels in HUEVC-sEVs upon PTC and/or CRT0066101 treatment. LTA₄H was strongly increased in sEVs after PTC treatment, and this was reversed by PRKD inhibition ( Figure 9H ). We verified the transfer of LTA₄H to PMNs using PTC-treated HUVEC supernatants as well as the respective purified sEVs ( Figures 9I, J ). Furthermore, the relevance of these findings for patients was corroborated by treating PMNs from healthy donors with plasma-sEVs isolated from polytrauma patients (injury severity score ≥ 16) or healthy probands and then determining the respective LTA₄H levels in cell lysates. The plasma samples were obtained from four patients at the time of shock room admittance (T0) as well as 4, 12 and 24 h after hospitalization. Relative concentration and size distribution of sEVs after purification by size exclusion chromatography was measured by nanoparticle tracking analysis, as part of our previous study in Seibold et al., 2021 (14). Concerning transfer of LTA₄H, we indeed demonstrated significantly higher levels of LTA₄H expression in PMNs after incubation with polytrauma patient-sEVs when compared to healthy proband-sEVs for the 12 h timepoint, whereas LTA₄H after 4 h was already increased and after 24h level decreased again from the 12 h peak. ( Figure 9K ). In conclusion, our data suggest that PRKD alters endothelial sEV cargo composition to induce LTB₄ biosynthesis in PMNs via transfer of LTA₄H, which was also confirmed for polytrauma patient-sEVs.

Besides a vital function as secondary chemoattractant, LTB₄ released from PMNs also contributes to neutrophil activation (60). Therefore, we evaluated the neutrophil respiratory burst as a terminal readout for release of reactive oxygen species following activation (61). In line with a role of PTC in the coactivation of fMLP stimulated PMNs, the respiratory burst was slightly increased by PTC stimulation, but also by CRT0066101, when PMNs were measured on their own ( Figure 10A ). However, in coculture with HUVECs, the respiratory burst was strongly increased by PTC treatment and this effect was substantially reversed upon PRKD inhibition with CRT0066101 ( Figure 10B ). The respiratory burst of PMNs can be triggered by LTB₄ (62). Therefore, we blocked LTB₄ biosynthesis by the inhibitor Safingol (63) in PMNs, which prevented the increase in respiratory burst by PTC derived HUVEC supernatants ( Figure 10C ). To corroborate these findings, we have used CP-10569 & LY255283 (1 µM) to inhibit the high and low affinity LTB₄ receptors BLT1/2 (64), respectively, thus blocking upstream activation of PMNs. Also, these inhibitors significantly impaired the PTC supernatant mediated respiratory burst alone, but more efficiently in combination, suggesting that both pathways are utilized ( Figure 10D ).

In summary these data indicate that activation of the PRKD1 axis downstream of post-traumatic inflammatory mediators reprograms intercellular communication of endothelial cells to control the magnitude of the post-traumatic immune response from PMNs via LTB₄ and the induction of the PMN respiratory burst.