Redefining the role of IL-18 in post-surgical recovery and sepsis: a key mediator of inflammation resolution

Abstract

Background

Timely resolution of innate immune responses activated by surgical intervention is crucial for patient recovery. While cytokines and innate immune cells are critical in inflammation resolution, the specific role of IL-18 in these processes remains controversial and underexplored.

Methods

We investigate determinants of successful recovery using peripheral blood samples from orthopedic surgery (ORT) patients (n = 33) at T0 (before surgery), T1 (24 h after surgery) and T2 (3 days after surgery). Monocytes from ORT patients underwent immunophenotyping together with bulk transcriptomic analysis. We found that IL-18 strongly defines the recovery immune signature. These results were further validated in vitro by comparing IL-18 and TNF-α effects on monocytes, and in 3D human intestine organoids together with single cell (sc)-RNAseq analysis.

Results

Transcriptomics of ORT monocytes revealed upregulation of ITG family integrins, namely ITGB3 and ITGB5, CXCL family chemokines, notably CXCL1-3, CXCL5, and SCL/TAL1 factor controlling differentiation and migration, but not pro-inflammatory genes. Similar changes were observed in IL-18 stimulated healthy donor monocytes in vitro, including an increase in CD11b, CD64, and CD86 levels, accompanied by increased phosphorylation of Akt but not NFκB. These changes were attenuated in the presence of TNF-α, thus showing a unique role of IL-18 when acting alone without its most frequent paired cytokine TNF-α. We further confirmed that IL-18 induces monocyte-macrophage transition and migration using human intestinal organoids. Finally, TNF-α/IL-18 ratio showed a high predictive value of clinical severity in septic patients.

Conclusions

We propose a novel role of IL-18 on monocyte migration and macrophage transition characterizing successful orthopedic surgery recovery, as well as the ratio of IL-18/TNF-α as a novel marker of inflammation resolution, with potential implications for patient monitoring and therapeutic strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-06652-7.

Introduction

Surgical interventions trigger the immune response and sterile inflammation, often progressing to systemic inflammatory response syndrome (SIRS). During this critical period, eventual infections can worsen inflammation, potentially leading to sepsis or septic shock [1]. Cytokines including IL-6, TNF-α and IL-1β are central regulators of immune activation and cell recruitment during SIRS [1–4]. However, unresolved inflammation and imbalanced cytokine production can lead to excessive monocyte/macrophage proliferation and migration causing tissue damage and poor outcomes [3, 5, 6]. Identifying cytokine signatures linked to inflammation resolution is crucial for improving postoperative recovery. In this context, major elective surgery like orthopedic surgery provides a practical and clinically relevant model to investigate the dynamics of inflammation resolution because of availability samples in time-controlled manner before and after surgery [7].

Successful resolution of inflammation moderates the immune response to allow a transition to tissue repair. Pro-resolving mediators such as lipoxins or resolvisns orchestrate this process by normalizing the acute responses and promoting the phagocytic clearance of tissue debris [8, 9]. Monocytes respond to various mediators and regulate both inflammation and resolution, by acquiring inflammatory, antigen-presenting, or clearance-related characteristics [5, 6, 10]. Despite this evidence, diagnostic markers of inflammation resolution remain lacking. Moreover, the role of monocytes in this transition is poorly defined, particularly in sterile inflammation induced by surgical interventions.

This study aims to identify the cellular and cytokine signatures associated with inflammation resolution during recovery from orthopedic surgery (ORT). Using peripheral blood samples from 33 surgery patients, we explore immune responses over time and validate findings in vitro and in an independent cohort of 31 septic patients, representing severe immune dysregulation. By focusing on monocyte function and cytokine interactions, we seek to uncover novel mechanisms underlying inflammation resolution and patient recovery.

Methods

Study design and patient data

Sex as a biological variable

Both men and women were recruited for this study. The findings of this study are therefore expected to be relevant to both sexes.

Study inclusion criteria and ethical approval

Thirty-three adult patients that underwent total hip arthroplasty or total knee arthroplasty or similar (referred to as ORT) at St Anne’s Hospital, Brno (Czech Republic) between September 2021 and May 2024 were eligible for the study. Patients undergoing ongoing chronic immuno-suppression therapy or oncological disorders were excluded from the study. Written informed consents were obtained from all recruited patients, and all procedures and protocols were approved by the institutional ethics committee of St Anne's University Hospital Brno (11G/2021).

Cohort demographic and clinical characteristics

Intraoperative blood loss was generally less than 500 mL, ten patients developed hypotension which was managed with ephedrine bolus dose or continuous norepinephrine infusion. Three patients were administered atropine due to intraoperative bradycardia and one patient developed temporary dyspnea with oxygen desaturation below 90%. The demographic, clinical and surgery-related characteristics of the recruited patients are summarized in Table 1. The biochemical and blood differential parameters of the recruited patients are summarized in Table 2.

Table 1.

Demographic and clinical characteristics of the recruited patients

Recruited patients, n = 33
Sex
Women	23 (60.6%)
Men	10 (30.3%)
Demographics
Age (years)	73 (43–83)
BMI	30.24 (20.28–38.56)
Weight (kg)	83 (58–120)
Height (cm)	165 (148–189)
Smoker
Yes	7 (21.5%)
No	23 (56.5%)
Ex-smoker	3 (9%)
Operation type
Total hip arthroplasty	16 (48.4%)
Total knee arthroplasty	16 (48.4%)
Other (shoulder joint replacement)	1 (3%)
Anesthetic
Spinal anaesthesia	23 (69.6%)
General anaesthesia	10 (30.3%)
Major comorbidities
Autoimmune hypothyroidism	11 (33.3%)
Hyperlipidemia	9 (27.2%)
Hypertension	6 (18.1%)
Vertebrogenic Algic Syndrome	6 (18.1%)
Polyarthritis	3 (9%)
Operation characteristics
Duration of surgery (min)	80 (40–160)
Intraoperative fluids administration (L)	1.7 (0.2–4)
Bone cement implantation
Yes	29 (87.8%)
No	4 (12.1%)
Hypotension during surgery
Yes	11 (33.34%)
No	22 (66.67%)

Patient demographic and operation characteristics values represent the median, with quantile range (minimal and maximal value) defined in parentheses. All other values represent the number of patients, with the percentage of total patients defined in parentheses

Table 2.

Biochemical and blood differential parameters of recruited patients

Parameter	T0	T1	T2
Biochemical parameters
Creatinine (mg/dL)	63 (39–97)	77 (58–89)	N
Hemoglobin (HB) (g/L)	133 (82–164)	111 (78–137)****	105 (1.9–139)****
Urea (mg/dL)	4 (2.4–7.0)	5.1 (3.6–8.1)	N
Blood differential
Basophils (%)	0.65 (0.2–1.4)	0.25 (0.0–0.7)****	0.4 (0.1–0.7)*
Eosinophils (%)	2.5 (0.4–9.1)	0.2 (0–4.7)*	2.8 (0.1–11.8)$$
Erythrocytes (10*12/L)	4.38 (2.87–5.1)	3.58 (2.38–4.5)****	3.55 (2.81–4.39)****
Hematocrit (HT) (L/L)	0.399 (0.243–0.482)	0.334 (0.217–0.412)****	0.325 (0.262–0.416)****
Leukocytes (10*9/L)	6.1 (4–15.2)	9.4 (5.6–15.8)****	8.15 (4.9–13.2)***
Lymphocytes (%)	30.6 (14–47.4)	13.2 (4.8–26) ****	15.85 (7.4–27.4)***
Monocytes (%)	8.5 (2.9–15.2)	9.45 (2.7–16.9)	8.65 (4.7–12.4)
Neutrophils (%)	55.85 (41.2–81.4)	75.2 (34.4–89.8)****	70.05 (61.8–81.4)**
Nucleated red blood cells (NRBC) (%)	0 (0.0–0.2)	0 (0.0–0.1)	0 (0.0–0.2)
Platelet distribution width (PDW) (fl)	13.2 (8.8–15.7)	12.7 (9–15.5)	12.95 (9.3–16.8)$
Platelet crit (PCT) (mL/L)	2.7 (1.5–3.9)	2.3 (1.2–3.4)****	2.4 (1.1–3.9)***
Mean platelet volume (MPV) (fl)	10.9 (8.8–12.1)	10.8 (8.9–810.5)*	10.9 (1.6–12.6)$$
Thrombocytes (TROMB) (10*9/L)	246 (124–409)	210 (105–344)****	218 (98–353) ****

Values represent the median, with quantile range (minimal and maximal value) defined in parentheses. All data were tested for normal distribution with the Shapiro–Wilk test and then subjected to Friedman analysis of variance with Dunn’s post-hoc test to correct for multiple comparisons. Patients with data missing from either of the time points were excluded for the calculations of statistical significance. *P < 0.05 for T0 vs T1 or T0 vs T2, **P < 0.01 for T0 vs T1 or T0 vs T2, ***P < 0.001 for T0 vs T1 or T0 vs T2, ****P < 0.0001 for T0 vs T1 or T0 vs T2, $P < 0.05 for T1 vs T2, $$P < 0.05 for T1 vs T2

Postoperative management

In the first 24 h post-surgery, patients received crystalloid infusions an average of 3.1 L (IQR 2800–3600) and had positive fluid balance. The maximal VAS (visual analogue scale), reporting the pain of patients, was 4–6 for most patients. Only exception were three patients that reported VAS 8–9. Only two patients had to receive blood transfusion. No patients experienced decrease in hemoglobin saturation with the need of oxygen supply during the first 24 h after the operation. Four patients developed hypotension or tachycardia during this time frame, of which two developed both conditions. Three patients with hypotension were administered noradrenaline and one patient with tachycardia was administered beta blockers and amiodarone. At 3 days following surgery (T2), five patients developed complications, none of which were infection related.

Blood sample preparation and cell culture

Blood sampling and processing

Samples of peripheral blood were obtained at three timepoints: before surgery (T0), and at 24 h (T1) and 3 days (T2) after ORT. Cytokine profiling was performed on samples obtained from 25 patients, and immunophenotyping of whole, stabilized blood was performed on samples obtained from 23 patients. RNAseq was performed from isolated monocytes from eight patients. Buffy coats from an additional 11 healthy blood donors were collected from the Department of Transfusion and Tissue Medicine, University Hospital, Brno (Czech Republic) and used for monocyte isolation. Eight of these samples were used for in vitro stimulation of monocytes with cytokines and the remaining three were used for monocyte co-cultivation experiments. Blood donors undergoing ongoing chronic immunosuppression therapy or with oncological disorders were excluded.

All blood samples were processed within 2 h of collection. Whole blood at a volume of 0.1 mL was stabilized by the Whole Blood Cell Stabilizer (Cytodelics AB, Sweden) at a ratio of 1:1 and incubated at room temperature (RT) for 15 min. Blood plasma was collected from blood by centrifuging at 2500 g for 15 min at 4 °C. Stabilized blood samples and blood plasma samples were stored at − 80 °C until use.

Monocyte isolation

Monocytes were isolated from 8 mL patient blood collected from ORT patients at the appropriate time points and enriched with the RosetteSep kit (STEMCELL Technologies, Vancouver, Canada) according to the manufacturer’s instructions. The blood was layered on Lymphoprep^® (Alere Technologies AS, Oslo, Norway) (density 1.077 g/mL) and centrifuged at 1200 g at 20 °C for 20 min. To isolate monocytes from healthy donors, buffy coats (~ 40 mL) were processed via the same method as above. Monocyte viability and yield was quantified with Luna™ automated cell counter (IMGEN Technologies, New York, Pennsylvania, USA). Monocyte preparations with > 85% viability and > 75% purity were used for the experiments. For molecular analysis, monocytes from patient samples were mixed with TRI Reagent® (Merck, Darmstadt, Germany) and stored at − 80 °C until further processing. For in vitro stimulation, monocytes from healthy donors were seeded at a cell density of 1 × 10⁶ of cells/mL in ultra-low attachment 24-well plates (Costar Life Sciences, Washington, DC, USA) in X-VIVO complete medium (BE02-060Q; Lonza, Cambridge, UK) and subjected to the appropriate stimulations.

Monocyte culture and stimulation with cytokines

Monocytes were stimulated with either TNF-α [11] (R&D Systems, Minneapolis, USA) at 10 ng/mL, IL-18 [12, 13] (R&D Systems, Minneapolis, USA) at 50 ng/mL or with both cytokines, all for 2 or 24 h as indicated. Unstimulated monocytes were used as the control group throughout the related in vitro experiments. The cells were subsequently collected for either RNA and protein isolation or stained for flow cytometry.

Human induced pluripotent stem cells (iPSCs) cultivation and intestinal organoid (IO) differentiation

iPSCs (WiCell, DF19-9-7T) [14] were cultured in tissue culture dishes coated in Cultrex Stem Cell Qualified Reduced Growth Factor Basement Membrane Extract (R&D Systems), and maintained in mTeSR Plus medium (STEMCELL Technologies) supplemented with penicillin and streptomycin (Pen/Strep) (500 U/ml) at 37 °C, with 5% CO₂ and 95% humidity. The medium was changed every two days and the cells were passaged using TryPLE (Gibco, Brooklyn, NY, USA) upon reaching ~ 80% confluence. After dissociation and dilution, the cells were placed in maintenance media containing a RHO/ROCK pathway inhibitor (10 μM, STEMCELL Technologies) for 24 h, after which the media was replaced for media not containing any of the inhibitor.

Human Intestinal organoids (IOs) were characterized and differentiated from iPSCs as previously described [15–17]. In brief, on day 0, iPSCs were induced to undergo definitive endoderm differentiation using RMPI1640 medium (Gibco) supplemented with Activin A (100 ng/mL; R&D Systems). After 24 h (day 1) the medium was replaced with fresh RMPI1640 supplemented with Activin A (100 ng/mL) and 0.2% HyClone-defined FBS (GE Healthcare Bio-Sciences, Barrington, Illinois). On days 3 and 4 the medium was replaced with fresh RMPI1640 supplemented with Activin A (100 ng/mL) and 2% HyClone-defined FBS. The transition to mid- and hind-gut was induced by daily maintenance of the definitive endoderm in RPMI1640 supplemented with 15 mM HEPES, 2% HyClone-defined FBS, 500 ng/ml FGF4 (R&D Systems), and 500 ng/mL WNT3a (R&D Systems) until 3D spheroids formed. Then, the spheroids were collected, embedded in a drop of Cultrex Membrane Extract Type 2 (R&D Systems), and maintained in IOs complete medium [Advanced DMEM F12 (Gibco) supplemented with B27 supplement (Thermo Fisher Scientific), GlutaMAX supplement (Thermo Fisher Scientific), P/S (500 U/mL), 15 mM HEPES, 500 ng/mL R-Spondin 1 (R&D Systems), 100 ng/mL Noggin (R&D Systems), 100 ng/mL EGF (R&D Systems, Minneapolis, USA)]. IOs cultured for approximately 50 days were used in this study.

Stimulation of IOs

IOs were stimulated in complete IOs medium without growth factors. Stimulations were performed using TNF-α at 10 ng/mL [11] (R&D Systems) and/or IL-18 at 50 ng/mL [12, 13] (R&D Systems). Unstimulated IOs were used as the control group throughout the organoids -related experimental procedures. IOs were collected after 4 h of stimulation for RNA isolation and qPCR or for 24 h for immunofluorescent staining. The untreated control IOs were cultivated with complete IOs medium for an equivalent amount of time.

Co-cultivation of monocytes and IOs

For co-cultivation experiments, IOs were extensively washed with cold PBS to completely remove the Cultrex embedding. Isolated monocytes were mixed with IOs (10⁵ monocytes per IO) and seeded together in a drop of Cultrex. Stimulations were performed with IL-18 and/or TNF-α as described above. Unstimulated monocytes and IOs were used as controls for the co-cultivation experiments. After 24 h, the co-cultures were extensively washed with cold PBS and dissociated as described below. Monocytes were also co-cultivated with Cultrex, but without the presence of IOs to serve as a control. Cultivation of monocytes with Cultrex, the extracellular matrix of IOs, without the IOs themselves, didn’t alter the monocyte subsets (Fig. S3D) or the MFI of CD68 and CD206 (Fig. S3E, F).

IO dissociation for flow cytometry analysis

IOs were washed with cold Hank's balanced salt solution (HBSS; Gibco, Brooklyn, NY, USA) to remove the Cultrex and then cut with scissors into small pieces. The cells were dissociated in TrypLE (Gibco) with constant shaking for 10 min at 37 °C. The cell suspension was filtered through a 70-μm strainer and washed with cold HBSS + 2% FBS. The suspensions were centrifuged 300g for 10 min at 4 °C, resuspended in PBS buffer, stained with the appropriate antibodies and subjected to flow cytometry as described below.

Immunoassays and protein analyses

Immunophenotypization using flow cytometry

Flow cytometry profiling of stabilized whole blood and isolated monocytes samples

To determine the levels of immune surface markers in stabilized blood, the samples were thawed at 37 °C, diluted in Fixation buffer (Cytodelics AB) at a ratio of 1:5 and incubated at RT for 10 min. To lyse the red blood cells, the samples were diluted 1:4 in Lysis buffer (Cytodelics AB) and incubated at RT for up to 15 min. The cells were then washed with Wash buffer (Cytodelics AB) before blocking nonspecific binding for 15 min with Fc receptor (FcR) blocking solution (Miltenyi, Auburn, CA, USA) in MACS buffer (PBS with 2% FBS and 2 mM EDTA). The cells were stained for 30 min in MACS buffer using the indicated antibodies at the dilutions provided in Supplementary Table 1.

For the immunophenotyping of in vitro stimulated, and monocytes co-cultivated with 3D human intestinal organoids (described below), single cell suspensions were stained with the appropriate antibodies similar to the procedure described above as well as with the live-dead fixable dye (ThermoFischer Scientific,Waltham, MA, USA). Acquisition was performed on a BD FACSymphony™ A1 (BD Biosciences) and a Sony SA3800 (Sony Biotechnology, San Jose, CA, USA) and the data were analyzed using FlowJo^® v10.10 software (FlowJo, LLC Ltd, Ashland, OR, USA).

Immunofluorescence staining of IOs

After the appropriate stimulations as described above, IOs were fixed in 4% paraformaldehyde (PFA) for 20 min at RT and washed three times with phosphate-buffered saline (PBS). For histological slides, IOs were first dehydrated in 15% sucrose overnight at 4 °C. The samples were then frozen in Tissue Freezing Medium (Leica Biosystems, Nussloch, Germany) in isopropanol cooled to −80 °C and were stored at − 80 °C. The frozen samples were cut and rehydrated in PBS for 10 min. IO sections were permeabilized with PBS + 0.5% Triton-X100 for 15 min at RT and then washed three times with IF buffer (PBS + 0.2% Triton-X100 + 0.05% Tween-20). The sections were then blocked with IF buffer + 2.5% bovine serum albumin (BSA) for 1 h at RT and were labeled with antibodies against e-cadherin, and ZO-1 overnight at 4 °C. Subsequently the sections were submitted to secondary labeling with AlexaFluor conjugates (AlexaFluor 488 and AlexaFluor 546 respectively). All antibodies and dilutions are available in Supplementary Table 2. Unstained slides that underwent the same process were used as negative controls to evaluate the quality of the staining. The samples were washed three times with IF buffer before staining with DAPI for 5 min at RT. The slides were washed for a final time before images were captured under a Zeiss LSM 780 confocal microscope at 10 × magnification. The images were processed using ImageJ software.

Plasma cytokine profiling using bead array

The plasma levels of the following human cytokines and chemokines were determined by the LEGENDplex Human Inflammation panel 1 (BioLegend, San Diego, CA, USA): interferon (IFN)-α2, IFN-γ, interleukin (IL)−1β, IL-6, IL-8, IL-10, IL-12p70, IL-17, IL-18, IL-23, IL-33, monocyte chemoattractant protein (MCP)−1 and tumor necrosis factor (TNF)-α. The measurements and plasma dilutions were performed according to the manufacturer's guidelines. The samples were then acquired on a FACS Canto II (BD Biosciences, New Jersey, NJ, USA) and, the resulting data were analyzed using the LEGENDplex software. To control potential batch effects between measurements, cytometer setup and tracking (CST) beads (BioLegend, San Diego, CA, USA) were used before each experiment, to calibrate cytometer parameters.

ELISA

ELISA was performed to detect IL-18BPα in supernatants from the in vitro stimulated monocytes, using an IL-18BPα Duoset kit (R&D Systems) according to the manufacturer’s protocol. The results were obtained using a Thermo Multiskan plate reader (ThermoFischer Scientific).

Protein isolation and western blotting

For western blotting, whole protein lysates were prepared from Trizol-lysed samples containing 2–3 million monocytes, according to the manufacturer’s protocol. After isolation, total proteins were resuspended in RIPA 1 × buffer (ThermoFischer Scientific) containing 1% SDS and a Protease and Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific). Samples were heated to 50 °C for 30 min to completely dissolve the proteins and the total protein quantity was determined by Pierce bicinchoninic acid assay (ThermoFischer Scientific, Waltham, MA, USA). Protein extracts were boiled in one-third volume of Laemmli buffer (Biorad Laboratories, Berkeley, CA, USA) and left to migrate on a 10% poly-acrylamide gel (Biorad Laboratories) at 120 V. Proteins were then transferred to a 0.2 mm pore PVDF membrane (Thermo Fisher Scientific) with a voltage of 10 V for 10 min. The membranes were blocked with 5% (w/v) BSA in TBS-T buffer [20 mM Tris–HCl pH 7.5, 137 mM NaCl, and 0.1% (v/v) Tween 20] for 30 min at RT. Immunoblotting was performed to detect phospho-Akt, Akt1 and phospho-NFκB1/p50; anti-β-actin was used as an internal control. Primary and secondary antibodies were diluted in 5% BSA in TBS-T buffer. All respective antibodies are listed in Supplementary Table 1 (Table S1) [15].

RNA-related processes and sequencing

RNA isolation

A total of 2–3 million TRI Reagent^®-lysed monocytes were treated with 20% chloroform and centrifuged at 12,000g for 15 min at 4 °C before extracting total RNA using a Qiagen RNeasy mini kit (QIAGEN, Düsseldorf, Germany) according to the manufacturer’s instructions. The isolated RNA was quantified using a Nanodrop Spectrophotometer Q5000 (Thermo Fisher Scientific, Waltham, MA, USA). The purity and integrity of the RNA was validated using a Bioanalyzer2100 RNA nano 6000 chips (Agilent Technologies, Santa Clara, CA, USA). Only RNA that displayed adequate quantity (> 50 ng/μL) and RIN > 8 was used for library construction.

Reverse transcription and quantitative PCR

Reverse transcription of total RNA was performed using a High-Capacity cDNA Reverse Transcription Kit (ThermoFischer Scientific) according to the manufacturer’s protocol. The cDNA samples were diluted at a 1:3 ratio with ultrapure RNAse-free water, and then mixed with TaqMan™ Gene Expression Master Mix (Applied Biosystems, Foster City CA, USA) and the appropriate TaqMan™ probes (ThermoFischer Scientific) for quantitative PCR (qPCR). The following genes were analyzed: TAL1, (Hs01097987_m1), JAM3 (Hs00230289_m1), ITGB3 (Hs01001469_m1), TNFA (Hs00174128_m1), IL1B (Hs01555410_m1), IL18 (Hs01038788_m1), CXCL8 (Hs00174103_m1), CCL2 (Hs00234140_m1).

Sample preparation and bulk or single-cell RNAseq (scRNAseq)

For the bulk RNAseq, total RNA was extracted from isolated monocytes of ORT patients at T0, T1 and T2, with the Qiagen RNeasy mini kit (QIAGEN, Düsseldorf, Germany) as described above.

For the single-cell (sc) RNAseq, single cell suspensions were prepared from IOs co-cultivated with monocytes and stimulated appropriately as described above, with TNF-α and/or IL-18 for 24 h. Unstimulated co-cultures were used as controls throughout the experiment. Monocyte preparations with purity > 85% was used for this experiment. Single cell suspensions were prepared by first washing the co-cultures with cold HBSS (Gibco), before dissociation with TrypLE (Gibco) and filtering through a 70-μm strainer, as described above. Dead cells and doublets were removed using a Dead cell removal kit (Miltenyi Biotec, Cologne, Germany), according to the manufacturer’s instructions. Briefly, cell suspensions were mixed with MicroBeads for 15 min and then separated over a MACS column in the magnetic field of a MACS separator (Miltenyi). The viable cell fraction was collected in the flow-through. The cells were then counted using a hemocytometer. The starting number of cells for all conditions was ≥ 400.000 cells. The cells were fixed in with Chromium Next GEM Single Cell Fixed RNA Sample Preparation Kit and hybridized with Chromium Next GEM Single Cell Fixed RNA Human Transcriptome Probe Kit according to the manufacturer’s guidelines (10X Genomics, Pleasanton, CA, USA). After the hybridization, cell suspensions were washed three times with post-hybridization wash buffer. After final washing steps, cells were resuspended and passed through 30 μm filter. The cell concentration was counted and 10 K cells from each sample were used for the single cell library construction.

Library construction and sequencing

Bulk RNA libraries were prepared using a QuantSeq 3ʹ mRNA-Seq Library Prep Kit FWD (Lexogen, Vienna, Austria) with 250 ng RNA template. The libraries were sequenced using a QuantSeq FWD kit (Lexogen) and at a sequencing depth of ~ 13 M.

For the construction of scRNAseq libraries, single-cell gel beads in emulsion (GEMs) were created using the Chromium Next GEM Single Cell Fixed RNA Gel Bead Kit and Chromium instrument (10X Genomics, Pleasanton, CA, USA). Single cell libraries were constructed with Chromium Next GEM Single Cell Fixed RNA Hybridization & Library Kit (10X Genomics) according to manufacturer. Libraries were equimolarly pooled and were sequenced on an Illumina NovaSeq X device.

Bioinformatic analysis

Cytokine profiling bioinformatic analysis

The cytokine profiling dataset was subjected to principal component analysis with the R package FactoMineR 2.11 [18] and then to Sparse Partial Least Squares Discriminant Analysis (sPLSDA) with the R package mixOmics 6.26.0 [19].

Bulk RNAseq bioinformatic analysis

The bulk RNAseq raw reads were pre-processed on a multithreaded UNIX-based server running an Arch Linux x86_64 operating system. The raw reads were mapped to the reference genome with HISAT2 2.2.1 [20] and Samtools 1.13 [21], using the GCA_000001405.15 gene annotation from the Dec. 2013 assembly of the human genome (hg38, GRCh38 Genome Reference Consortium Human Reference 38). The total number of counts per gene was calculated using HTseq 1.99.2 [22]. Differential expression analyses were performed in the R v4.3.3 environment, using the DESeq2 1.42.1 [23]. pipeline. The differentially expressed genes (DEGs) were filtered for low counts and considered to be significantly differentially expressed if they demonstrated a log-fold change |LFC|≥ 1 and P value ≤ 0.05. Gene ontology (GO) analysis was performed on the significant DEGs with clusterProfiler 4.10.1. [24]. DEGs for both bulk and single-cell RNAseq were annotated to the MSigDB gene ontology gene sets. The ggplot2 3.5.1., complexHeatmap 2.18.0. and ggVENNdiagram 1.5.2. packages were used for visualization [25, 26], and barplots were generated in Graphpad Prism 8.4.2.

ScRNAseq bioinformatic analysis

Cell barcodes and transcripts were quantified using CellRanger 8.0.1 multipipeline with GRCh38 genome reference and Chromium Human Transcriptome probe set v1.0.1. The clustering, cell annotation and DEG analysis were all performed according to the Seurat 5.1.0. pipeline [27] Annotation of cell clusters was done by use of the gene sets available in the R package SingleR 2.6.0. [28], based on gene expression pattern similarity with human gene populations. All analyses were done in the R v4.4.0. environment.

Statistical analysis and sample size

Prism^® (GraphPad Software LLC Ltd, La Jolla, CA, USA) software and R v4.4.0. were used for statistical analyses. Data were tested for normal distribution with the Shapiro–Wilk test and statistical tests were applied as appropriate. For all analyses regarding the cohorts and monocyte stimulation experiments, paired Friedman analysis of variance was applied to compare between the time points of the same patient. The Dunn’s post-hoc test was used to correct for the pairwise comparisons. For the analyses concerning IOs, unpaired Kruskal-Walli’s analysis of variance was used with Dunn’s post hoc test. All data are represented as individual values in dots, along with the median value. After normality testing, a one-way non-parametric paired ANOVA was performed for both the comparisons between patient time points and for the in vitro monocyte experiments. Dunn’s test was used as a post-hoc test.

For the co-cultivation experiments, after normality testing, a one-way non-parametric unpaired ANOVA was performed with Dunn’s as a post-hoc test, in order to recapitulate the increased variability of IOs of different differentiation or co-cultivation batches. For the bulk and single-cell RNAseq data, all statistics were computed during the bioinformatic analysis with the built-in DESeq2 and Seurat pipelines respectively. Any deviation from the abovementioned statistical tests is described in the figure legend or the appropriate section of the results. The level of statistical significance was determined as follows: *(P < 0.05), **(P < 0.01), ***(P < 0.001) and ****(P < 0.0001).

Sample size was selected according to each procedure. In order to capture the biological and technical variability derived from cytokine profiling and flow cytometry measurements, 23–25 patient samples were used. For the sequencing and in vitro experiments, the sample size was 8–11, in order to account for the high variability observed in human studies and for the complexity of the model in the case of co-cultivation of IOs with monocytes.

Results

Increased IL-18 levels characterize the immune response at 3 days after ORT surgery

To understand cytokine patterns driving post-ORT inflammation and recovery, we measured plasma cytokine levels of 25 patients at three time points: before surgery (T0), 24 h post-ORT (T1), and 3 days post-ORT (T2) (Fig. 1A). Distinct cytokine profiles emerged across all three time points. Interestingly, IL-18 was uniquely and significantly increased at T2 compared to both T0 and T1 (Fig. 1B), while most cytokines, including TNF-α, IL-1β and IFN-γ, the typical cytokines released during inflammation, dropped below pre-surgery levels by T1 and remained low at T2 (Fig. 1C–H). IL-6 and IL-10 peaked at T1, and returned to near pre-surgery levels by T2 (Fig. 1I, J), suggesting that the pro-inflammatory phase concludes in T1 and is not sustained into T2. Sparse Partial Least Squares Discriminant Analysis (sPLSDA) identified IL-18 as the main determinant of T2, followed by IFN-γ and IL-17A (Fig. 1K, L), which nonetheless didn’t show a T2-specific pattern. IL-18 can induce both IFN-γ and IL-17A via T-helper cell interactions [29, 30], supporting IL-18 being the key cytokine of T2. Since delayed cytokine increase might indicate recruitment of immune cells and tissue repair processes [30], T2-specific increase of IL-18 might suggest its implication in these processes. In summary, we showed for the first time a unique IL-18 peak at 3 days post-ORT surgery, characterizing T2, contrary to other cytokines that returned to baseline earlier.

Fig. 1.

Increased IL-18 levels in patient plasma characterize the immune response 3 days post-orthopedic surgery. A Workflow of patient sampling. B–J Plasma levels (pg/mL) of pro- and anti-inflammatory cytokines, including IL-18 (B), TNF-α (C), IL-1β (D), IFN-γ (E), IL17A (F), IFN-α2 (G), IL-12p70 (H), IL-6 (I) and IL10 (J) at all three time points. K, L sPLSDA loadings on components 1 and 2 of the dataset, showing the best classifiers for 3 days post-surgery (T2). Bars directional towards the left represent negative contribution and bars directional towards to right show positive contribution to T2. IL-18 shows the highest negative contribution to T2 only, thus being the best classifier to distinguish this particular time point. Data were tested for normality (Shapiro–Wilk test) and were subjected to Friedman’s ANOVA with Dunn’s post-hoc test. Data are presented as individual values, with the medians and 95% quantile ranges. *P < 0.05, **P < 0.01, ***P < 0.001.****P < 0.0001. n = 25 patients

An activated monocyte signature is evident 3 days after surgery, without pro-inflammatory gene upregulation

Innate immune cells of the monocyte–macrophage lineage both highly produce IL-18 precursor and respond to IL-18 [32]. Thus, to investigate changes in monocytes post-ORT, we performed bulk RNAseq on monocytes from 8 ORT patients at T0, T1 and T2. We compared gene expression at T2 to pre-surgery (T0 vs T2) and 24 h post-surgery (T1 vs T2). We found a distinct transcriptomic profile in T2, with a significant upregulation of most genes (Fig. 2A). Despite a minor downregulation (14%) of genes from T0 to T2 (T0 vs T2) (Fig. 2B left), even less genes (6%) were downregulated from T1 to T2 (T1 vs T2) (Fig. 2B right). The above were specific for T2, unlike T1, which showed substantial downregulation of genes compared to T0 (Fig. S1A). These findings suggest that monocytes transcription is the most active at T2. In fact, a large portion of upregulated genes were shared between T0vsT2 and T1vsT2 (Fig. 2C), but their patterns differed, as DEGs from T0vsT1 and T1vsT2 did not overlap (Fig.S1B). Overall, our findings show that monocyte gene expression changed dynamically over several days, with gene upregulation peaking at T2 after ORT.

Fig. 2.

Transcriptomic profiling of monocytes after orthopedic surgery reveals enhanced functionality without upregulated inflammation. A Scaled normalized counts of differentially expressed genes (DEGs) across time points: before surgery (T0), at 24 h (T1) and at 3 days after ORT (T2). B Number of DEGs in T0 vs T2 and T1 vs T2. C Venn diagram showing common and unique DEGs between T0 vs T2 and T1 vs T2. Upregulated genes are shown in red; downregulated genes are in blue. D–F Gene ontology analysis for DEGs. The color gradient refers to the statistical significance of the change in gene expression. D DEGs between T0 and T2. E DEGs between T1 and T2. F Combined analysis of T0 and T1 vs T2. G Spider plot showing the DEGs present in most ontologies. The three DEGs with the highest LFC are highlighted in bold letters. n = 8 patients. *P < 0.05, **P < 0.01, ***P < 0.001.****P < 0.0001. n = 8 patients

We then performed gene ontology (GO) analysis of the upregulated genes at T2 compared to T0 (Fig. 2D) and T1 (Fig. 2E), and the commonly upregulated genes between the two comparisons (T0 vs T2 and T2 vs T2, Fig. 2F). We identified pathways of monocyte activation, differentiation, adhesion and migration, but we revealed only one pathway, which was related to inflammatory regulation (Fig. 2F). These processes are activated upon clearance of inflammation [31, 32] thus creating the notion that T2 is in fact a time point of resolution, but with parallel peak of monocyte activation.

To identify key monocyte genes in T2, we identified the most differentially expressed genes across the GOs: chemokines CXCL, CXCL2, CXCL3 and PPBP; integrins ITGB3, ITGA2B and ITGB5; the junction adhesion molecule 3 (JAM3); the complement receptor 1 (CR1) and T-Cell Acute Lymphocytic Leukemia 1 (TAL1) (Fig. 2G). Among those, ITGB3, JAM3 and TAL1 had the highest upregulation and the highest significance in T2, compared to both other timepoints (Fig. S1C). We validated the upregulation of these genes in independent patient samples (Fig.S1D). Although a role for ITGB3 and JAM3 in tissue repair through altering endothelial function or permeability has been proposed [31, 32], their role in clearance of inflammation is not yet explored. According to our data, it is possible that these genes drive T2 -related processes by influencing monocyte activation and lineage maturation (TAL1) as well as mobility (ITGB3 and JAM3).

In summary, we observed progressive and continuous monocyte activation that peaked at 3 days post-surgery, dominated by migration and chemotaxis. These results suggest monocyte activation in T2 is tied to tissue repair phenotype.

Monocyte activation and migration marker levels peak 3 days after surgery

To determine whether the observed transcriptomic changes are reflected at the functional level, we performed immunophenotyping of patient blood samples across the three time points (Fig. 3). Monocytes were identified after gating for CD45 + CD3-CD56-CD66b-CD14 + and/or CD16 + cells (Fig. S1E, F). Classical monocyte frequency significantly decreased at T2, while intermediate monocytes increased (Fig. 3A) suggesting the acquisition of distinct functional characteristics in T2. This might have been either a by-product of IL-10 increase [33] in T1, or to a T2-specific effect. To determine this, we looked at the expression levels of monocyte activation markers. CD64 and CD86 were increased at T2 compared to T0 (Fig. 3B, C) and adhesion molecule CD11b was upregulated at T2, compared to both T0 and T1 (Fig. 3D), suggesting a T2-specific effect in monocyte activation. We aimed to determine if these markers are co-expressed, tSNEs based on these markers (Fig. 3E) highlighted a functionally activated T2-specific cell population co-expressing all three markers (CD64, CD86, CD11b), indicating enhanced activation and mobility [34] of these cells in 3 days post-ORT. These findings suggest monocyte alterations peak at T2, with enhanced activation and adhesion. Together with the previously described data, these results raise the question of the potential involvement of IL-18.

Fig. 3.

Monocyte activation increases 3 days post-orthopedic surgery. A Frequency of monocyte subsets at the three time points. B–-D Cell surface expression of monocytic markers CD64, CD86 and CD11b at the three time points: before surgery (T0), at 24 h (T1) and at 3 days after ORT (T2). E tSNE plots showing monocyte subsets at the three time points, with clustering based on the CD14 and CD16 lineage markers. Overlays show CD64, CD86 and CD11b expression. Dimensionality reduction was performed by taking into account the 30 nearest neighbors and iterating 1000 times with the Barnes-Hut gradient algorithm. Data were tested for normality (Shapiro–Wilk test) and analyzed using Friedman’s ANOVA with Dunn’s post-hoc test. Data are presented as individual values, with medians and 95% quantile ranges. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 23 patients

Monocyte stimulation with IL-18 promotes the upregulation of a unique functional marker signature

Thus far, our analysis revealed a unique monocyte signature involving activation and mobility, coinciding with a peak in IL-18 levels post-ORT surgery. To investigate whether IL-18 drives these changes, we stimulated healthy donor monocytes with IL-18 in vitro and analyzed gene expression and cell surface markers that we previously associated with T2 in ORT patients (Fig. 4). TNF-α was used as a pro-inflammatory control.

Fig. 4.

IL-18 stimulation of healthy monocytes induces a unique transcriptomic and functional profile, attenuated by TNF-α and accompanied by Akt activation. A Frequency of monocyte subsets after in vitro stimulation with IL-18 and/or TNF-α. B Cell surface expression of the functional monocyte markers CD64, CD86 C Cell surface expression of the adhesion molecule CD11b after IL-18 and/or TNF-α stimulation. D Dendrogram clustering based surface markers composition, using the Minimum spanning tree algorithm embedded in the FlowJo FlowSOM plug-in. E mRNA expression of TAL1, ITGB3 and JAM3, determined by qPCR. F Western blot quantification and representative picture of Akt phosphorylation at Ser473 and Akt1, normalized to β-actin. Data were tested for normality (Shapiro–Wilk test) and analyzed by Friedman’s ANOVA with Dunn’s post-hoc test. Data are presented as individual values, with medians and 95% quantile ranges. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 8–9 healthy donors

Monocytes were identified by gating for CD45 + CD3-CD19-CD20-CD235a-CD56- and CD14 + and/or CD16 + cells (Fig. S2A). Both TNF-α and IL-18 reduced classical monocytes and increased intermediate ones, with their combination amplifying these effects (Fig. 4A). However, only IL-18 uniquely upregulated CD64 (activation) and CD11b (adhesion) compared to controls and TNF-α, while also inducing CD86 compared to TNF-α (Fig. 4B, C). The above suggested that only IL-18 alone is able to drive the expression of surface markers that are upregulated in ORT patients in T2. Interestingly, TNF-α seemed to antagonize the effect of IL-18 regarding functional marker expression (Fig. 4B, C), indicating a both synergistic and antagonistic interplay mechanism.

To analyze the distribution of functional markers across monocyte subsets, we performed FlowSOM clustering analysis based on the activation markers CD64, CD86 and CD11b and the lineage markers CD14 and CD16. We identified four functionally distinct monocyte clusters (Fig. S2B, C). Cluster 2, dominated all conditions, while the other clusters represented smaller portions of the populations (Fig. 4D). IL-18 increased the size of clusters 1 and 4, where all three activation markers were upregulated (Fig.S2B-C) compared to Control and/or TNF-α. Importantly, only IL-18 stimulation resulted in upregulation of all three markers, in these two clusters (Fig. S2C). TNF-α plus IL-18 did not significantly alter cluster size compared to the Control, but did change their arrangement (Fig. 4D). These findings suggest that TNF-α, IL-18 and their combination affect the monocytic subpopulations differently, despite having overlapping effects on the monocyte subset distribution, but only IL-18 gives rise of functionally distinct, activated monocyte populations with upregulated CD64, CD86, CD11b.

At the transcriptional level, IL-18 significantly upregulated the expression of TAL1 (monocyte differentiation), ITGB3 and JAM3 (monocyte migration) compared to the unstimulated control (Fig. 4E). Combined stimulation with TNF-α, attenuated these effects (Fig. 4E). These results indicate that IL-18 drives functional and transcriptomic changes in monocytes similar to those observed in ORT patients at T2, while combination with TNF-α dampens IL-18’s effects.

To gain the mechanistic insight on how IL-18 might the above changes, we determined the phosphorylation of NF-κB and Akt, key signaling molecules associated with IL-18. IL-18 did not alter the levels of phosphorylated NFκΒ/p50 (Fig.S2D). However, it significantly induced phosphorylation of Akt at Ser473 and upregulated total Akt1 levels (Fig. 4F) [29, 30]. The combination of the two cytokines resulted in decreased Akt but increased NFκΒ phosphorylation, suggesting again an antagonistic effect of TNF-α. Given the overlapping effector functions of Akt and NFκΒ signaling [35] we analyzed the mRNA expression of TNFA and IL1B as proinflammatory products of the signaling (Fig. S2E). We also measured IL-1β protein levels by ELISA (Fig. S2F, left). IL-18 stimulation had no effect on the mRNA levels of TNFA and IL1B, nor on IL-1β protein levels alone (Fig. S2E, F) but only in combination with TNF-α (Fig. S2E, F). The above suggested that the effect of IL-18 on monocytes involved Akt1 rather than NFκΒ signaling, which is an important finding in the context of the already established IL-18 signaling. Importantly, Akt1 has also been associated with alternatively activated macrophages [36, 37]. In combination with the induction of TAL1 by IL-18, these results suggest that IL-18 might influence monocyte-to-macrophage transition.

TNF-α and IL-18 share significant overlapping signaling, involving NF-κΒ, and seem to synergize to amplify inflammatory phenotypes [38–40]. Finally, to determine how TNF-α antagonizes IL-18 effects, we measured the mRNA expression of IL18 and the protein levels of its regulatory peptide IL-18BPα, which binds soluble IL-18 and inhibits its function. IL-18 had no effect on its own mRNA (Fig.S2E) but significantly reduced IL-18BPα protein levels, suggesting that it can enhance its own effect by downregulating its inhibitory partner (Fig.S2F, right). TNF-α restored IL-1BPα levels only in combination with IL-18 (Fig. S2F, right) partially justifying the observed attenuation of IL-18 effects, and further suggesting that the potential interplay between the two cytokines, leads to modulated IL-18 activity.

IL-18 drives monocyte migration and tissue remodeling, while preserving the tissue integrity of human intestinal organoids

Our findings so far suggest that IL-18 -possibly via Akt signaling- upregulates monocyte activation and migration and possibly alters maturation. To assess these effects, we used 3D human intestinal organoids (IOs) derived from human induced pluripotent stem cells (iPSCs) (Fig. 5). We chose this model since the intestine is where monocytes typically migrate and differentiate into macrophages. Furthermore, IOs form a complex and immune-competent environment that allows co-cultivation and study of immune cells [17]. Thus, despite the limitations caused by its differences with tissue in surgical sites or bone tissue in ORT surgery, IOs provide a potent environment to study monocyte migration.

Fig. 5.

IL-18 alters tissue remodeling and enhances monocyte migration and monocyte-to-macrophage transition in human organoid and monocyte co-cultures. A Single-cell RNAseq -derived clustering on UMAP, annotated for Control or IL-18 conditions, demonstrating the distribution of cells and clusters (left). Annotation of cell types for monocyte and IOs co-cultures on the UMAP (right). n = 5 organoid-monocyte co-cultures were used per condition. Processing and clustering of the single-cell RNAseq data was performed with the Seurat pipeline in the R v4.3 environment. B Gene ontology analysis demonstrating the significantly upregulated ontologies in IL-18 in comparison to Control for epithelial cells of the intestinal organoids. Genes demonstrating a |LFC|≥ 0.6 and p ≤ 0.05 were considered as significantly DEG. Significance was established according to the Seurat package statistical testing. C Immunofluorescence staining of IOs co-cultivated with monocytes for DAPI (blue color), E-cadherin (red color) and zonulin 1 (ZO-1) (yellow color). D Fluorescence intensity (MFI) of the immunofluorescent data for ZO-1. For each condition, six independent regions of interest (ROIs) were taken into account. E Frequency of monocyte subsets for monocytes that migrated into IOs after 24 h of stimulation with IL-18 and/or TNF-α. F Cell surface expression of monocyte tissue residence markers after 24-h co-cultivation with IOs and stimulation with IL-18 and/or TNF-α G tSNE plots of monocyte populations after co-cultivation with IOs and stimulation with IL-18 and/or TNF-α. Clustering analysis was based on functional markers CD68 and CD206, and the monocyte lineage markers CD14 and CD16. The lower panel shows CD68 and CD206 expression overlaid on the tSNE plots. All data in bar plots are represented as individual values along with the median value and 95% quantile range. n = 5–6 IOs for the flow data; n = 3 healthy donors for monocyte isolation; n = 5 IOs for each condition for scRNAseq. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Thus, we performed single-cell RNA sequencing (scRNAseq) on IOs co-cultivated with monocytes and stimulated with IL-18 (Fig. 5A, B). IL-18 significantly altered cell distribution across the UMAP (Fig. 5A, left), enhanced immune cell abundance in IOs (Fig. 5A, right), suggesting enhanced monocyte migration, and upregulated tissue remodeling and chemotaxis pathways in IOs, without upregulating inflammatory pathways (Fig. 5B). To further confirm the above, we determined the mRNA expression of pro-inflammatory cytokines and chemokines typically expressed in IOs in inflammation (Fig.S3A). IL-18 did not induce any of the pro-inflammatory mediators, and even reduced IL-8 expression in the presence of TNF-α (Fig.S3A), highlighting a tissue-specific interplay that contrasts with effects observed in blood cells. We further validated these findings with immunofluorescence (Fig. 5C). To confirm the above, we performed immunofluorescence on IOs co-cultivated with monocytes (Fig. 5C, D). IL-18 preserved epithelial barrier integrity, visualized with E-cadherin (epithelium) and ZO-1 (tight junctions), unlike TNF-α which caused extensive tissue disorganization and barrier damage, evidenced by disrupted ZO-1 junctions (Fig. 5C). Further quantification of the fluorescence of ZO-1, revealed that IL-18 stimulation did not result in a loss of ZO-1, unlike TNF-α stimulation (Fig. 5D). TNF-α also led to depletion of E-cadherin for a given number of cells, while the other stimulations had no effect, evidenced by consistent E-cadherin and DAPI fluorescence (Fig.5C). Interestingly, co-stimulation with IL-18 and TNF-α did not disturb intestinal organoid structure (Fig. 5C, D). Together, these results suggest that IL-18 drives increased monocyte migration into IOs along with the expression of genes linked to chemotaxis and tissue remodeling rather than tissue damage.

Focusing on the immune cell cluster from scRNAseq, we identified only 16 DEGs between control and IL-18. GO analysis indicated that IL-18-activated monocytes primarily engaged in chemotaxis and phosphoatidyl-inositol phosphate binding, processes mediated by Akt [41, 42], but not inflammation (Fig.S3B). Key genes included vascular endothelial growth factor A (VEGFA) pleckstrin homology domain containing A5 (PLEKHA5), and notably CXCR4, all directly associated to Akt signaling [43, 44].

To further examine monocyte functional changes, we performed flow cytometry analysis of the co-cultures of monocytes and IOs (Fig. 5E–G). Monocytes were identified by gating for CD45 + cells, CD3-CD19-CD20-CD235a-CD56- and CD14 + and/or CD16 + cells (Fig.S3C). IL-18 increased intermediate monocyte frequency and decreased classical monocytes (Fig. 5E). We hypothesized that the migrated monocytes, would gain tissue-resident macrophage features [31]. Consistent with this, IL-18 significantly increased the surface expression of CD68 (Fig. 5F, left) as well as CD206 expression (Fig. 6F, right), both of which are markers associated with tissue resident macrophages [45].

Fig. 6.

TNF-α/IL-18 ratio demonstrates diagnostic and predictive potential in ORT and in septic patients. A Spearman pairwise correlations between the ORT patient biochemical parameters and TNF-α/IL-18 ratio. B Spearman correlations between the SOFA score of septic patients, the biochemical parameters of these patients and their TNF-α/IL-18 ratios. C ROC curve analysis evaluating the potential to predict high or low risk of mortality, determined as low (1–6) or high (> 6) SOFA score respectively. D sPLSDA contributions on compounds 1 and 2 of septic patient cytokine data, demonstrating the most characteristic cytokine for each patient subpopulation. Bars pointing to the left indicate negative contribution and bars pointing to the right indicate positive contribution of each cytokine in the cluster phenotype. E Graphical abstract of the findings of the study. For the correlograms, positive correlations are displayed in red and negative correlations in blue. The color intensity and size of the squares is proportional to the Spearman correlation coefficients. Visualization was done in R studio by corrplot package. *P < 0.05, **P < 0.01, ***P < 0.001. AUC area under curve, SOFA sequential organ failure assessment,. TROMB thrombocytes, MPV mean platelet volume, PCT platelet crit

As tissue-resident macrophages co-express CD68 and CD206, we visualized the distribution of functional markers with tSNE plots (Fig. 5G). We found an enriched population of monocytes positive for both CD68 and CD206, only following IL-18 stimulation (Fig. 5G). These results were exclusive to IO co-cultures and were absent in monocytes cultured alone (Fig.S3D-F), emphasizing the pleiotropic nature of IL-18 when acting on monocytes alone, versus within IO co-cultures. Interestingly, the expression of CD206 is inhibited upon Akt1 blockade [46] thus highlighting once more the role of Akt in IL-18 -driven changes, and binding the scRNAseq and flow cytometry data together.

TNF-α/IL-18 ratio has diagnostic and predictive potential in ORT and in septic patients

In the final part of our study, we assessed the prognostic and diagnostic potential of our findings by analyzing the connection between plasma markers in ORT patients, and TNF-α/IL-18 ratio, due to their dual regulation. TNF-α/IL-18, MPV, PCT, and THROMB showed significant correlations with leukocyte counts in T2 (Fig. 6A). Given the usage of leukocyte counts in predicting immune response and potential infection, the TNF-α/IL-18 ratio may hold diagnostic potential in the monitoring patient recovery post-surgery. We also analyzed septic patients from our previous study, where IL-18 elevation in early-stage sepsis correlates with survival in T1 [6]. We opted for this cohort, since sepsis in patients represents the ultimate and most severe immune dysregulation instead of recovery. In this cohort, TNF-α/IL-18 and urea negatively correlated with SOFA scores at T2 (Fig. 6B). Regression analysis showed that higher TNF-α/IL-18 ratios correlated with lower SOFA scores, although this plateaued beyond certain values (Fig. S4A), suggesting that IL-18 beyond a certain threshold may have a different effect on SOFA score. We then evaluated the ability of TNF-α/IL-18 to predict the clinical state severity, at 3 days after septic patient’s hospitalization to ICU, represented by low (1–6) or high (> 6) SOFA score, with ROC analysis. Thrombocytes, Urea, which are part of SOFA score calculation, as well as TNF-α/IL-18 showed significant predictive potential of mortality risk (AUC = 0.7353, 0.7941, and 0.7412, respectively) (Fig. 6C). As a result, the above findings suggest a strong predictive potential for the ratio TNF-α/IL-18 in septic patient recovery.

Finally, to explore IL-18 in the determination of patient subpopulations, we performed PCA and sPLSDA analysis with septic patient cytokine data. We identified 2 clusters of patients (Fig.S4B) with common signatures. The remaining patients did not have any common cytokine determinants and stayed un-clustered. Cluster 1 was characterized specifically and uniquely by IL-18 while Cluster 2 was characterized by the combination of IL-1β, IFN-y, TNF-α ad IL-17A (Fig. 6D). This finding is interesting given that in ORT patients, IFN-γ and IL-17A are part of T2 -related signature, led by IL-18. Other cytokines did not show any specific clustering potential. Interestingly, patients of Cluster 1, characterized by IL-18, demonstrated lower SOFA scores compared to the unclustered patients (Fig. S4C) with the exception of one patient with a different source of sepsis (data not shown). In total, the above results suggest that evaluation of TNF-α/IL-18 ratio may hold clinical diagnostic or predictive potential in the monitoring of patients. Despite these promising findings however, the information regarding the full clinical potential of TNF-α/IL-18 remains mostly preliminary. Thus, further research is required in order to fully determine the prognostic/diagnostic potential of the ratio in the clinic. Future studies benefitting from a bigger cohort size, longer study time points and more in-depth and large-scale validation of the TNF-α/IL-18 ratio in combination with other cytokines and patient parameters, can shed more light into the IL-18 clinical significance.

Overall, our study demonstrates that IL-18, characterizes the immune response 3 days after ORT and drives the observed monocyte transcriptome and functional profiles. Instead of inducing inflammation, IL-18 modulates monocyte migration, maturation and chemotaxis, predisposing monocytes towards pro-resolution phenotype (Fig. 6E).

Discussion

Unraveling the molecular mediators that govern recovery or post-surgery adverse event development is crucial for advancing therapeutic and diagnostic strategies for management of patients recovering from surgery [1]. Here we analyzed patient immunophenotypes, complemented by bulk RNAseq from monocytes of patients, before and after surgery. Key findings were validated using organoid models and scRNAseq, defining IL-18 as a mediator/determinant of inflammation resolution and subsequent successful recovery.

Sterile inflammation induced after surgery triggers the release of both pro- and anti-inflammatory cytokines [1–4]. Our data show for first time that IL-18 is a unique determinant of the resolution phases of inflammation, observed 3 days after surgery, while other analyzed cytokines including TNF-ɑ, IL-6 and IL-10 have already returned to pre-surgery baseline levels. IL-18 might in fact drive the cytokine patterns that characterize the timepoint 3 days after surgery, since on one hand it can induce IFN-y production alone, and on the other hand, it has been shown to correlate directly with IL-17 levels in Th17 cells in rheumatoid arthritis [39, 47]. Additionally, while it is possible that other cytokine patterns co-exist in this time point, this finding is in direct contrast with the already established role of IL-18, which was primarily associated with immune system dysfunction, simultaneous TNF-α or IL-1β production, and inflammatory damage, as well as the upregulation of markers linked to many inflammatory pathologies [10, 39, 48, 49] or even to sepsis [38, 40, 50]. Increased cellular activation and cytokine patterns have been primarily reported for the immediate post-operative period of 24 h after surgery [51]. On the contrary, our data show that in fact the peak of monocyte activation is at 3 days post-surgery and doesn’t involve inflammatory actions, but migration and cellular activation, potentially suggesting the initiation of a pro-resolution program. This observation is also supported by RNAseq data showing increased expression of genes involved in monocyte migration and the transition to macrophages, potentially indicating the initiation of a pro-resolution program. Our findings indicate a divergent role of IL-18 in the immune response over time and in conjunction with different signaling partners, which activate downstream signaling pathways.

IL-18 alone can induce the monocyte activated status observed at 3 days post-surgery, as suggested by in vitro restimulation. However, this response may be further influenced by other pro-resolution signaling factors or the prior actions of IL-10, which could, for example, contribute to the additional upregulation of CD64 expression [52]. We observed increased surface expression of CD64, CD11b and CD86 at 3 days after surgery, indicating enhanced activation and mobility [34]. The boosted monocyte activation status is likely driven by IL-18 downstream signaling, as signaling pathways downstream of IL-18R lead mainly to phosphorylation of NFκB [35], furthermore IL-18R trigger also leads to activation of MAPK/STAT pathway or in our context, to a more important PI3K/Akt/mTOR activation in hematopoietic and endothelial cell types [35, 53]. Along with these findings, we further examine the role of IL-18, by showing that the changes in monocyte migration and differentiation upon IL-18 administration were accompanied by Akt, not NFκB phosphorylation. This finding is of particular importance, as Akt can activate NFκB via phosphorylation of IκB inhibitor [54] but also promote anti-inflammatory or even pro-resolution events [39, 55, 56]. Thus, the observed IL-18 -driven effects can either be mediated by Akt directly, indirectly by modulation of MAPK/STAT [54, 57] or other, IL-18R -independent mechanisms. Our data further suggest that transcriptional changes caused by IL-18 alone are attenuated in co-stimulation with TNF-α via the regulation of IL-18BPα levels, resulting in different signatures [38]. These findings support that IL-18 might not just activate monocytes and induce migration, but promote the acquisition of a pro-resolution profile. Our findings correspond to the described role of PI3k/Akt/mTOR in the attenuation of LPS/TLR signaling in macrophages [36] and in the rise of non-inflammatory monocytes that resolve inflammation [39, 55, 56].

The high plasticity of monocytes and macrophages enables their differentiation into a diverse spectrum of effector cells in response to the inducing stimuli highlighting their functional versatility [58, 59]. However, the systemic or tissue -specific mediators responsible for the pro-resolution monocyte and macrophage phenotypes in post-surgery patients is scarce. The peak of IL-18 levels observed 3 days after ORT, coincided with monocyte activation—documented by their high transcriptional activity and upregulation of activation and functional markers, associated with increased migration into tissues and differentiation. IL-18 has been found to induce the expression of monocyte chemotactic molecules such as MCP-1, CXCL8 and CXCL9 in vitro, either alone or in conjunction with IL-12 [60, 61], supporting the notion that it is the driver of observed changes in 3 days after surgery. These findings were supported by the upregulation of CXCL chemokines and integrins, which are key mediators of monocyte chemotaxis and migration [62, 63].

To prove the enhanced migration followed by transition to macrophages, we chose the intestine microenvironment. The intestine is where monocytes frequently migrate and differentiate into short-lived macrophage populations [64, 65], co-existing with long-living tissue-resident macrophages, each having distinct roles in intestine homeostasis [66, 67]. The homing of monocytes is mediated by various integrins and chemokines [66, 67]. Therefore, we used our state-of-the-art 3D human IOs cultivated with monocytes. We investigated the impact of IL-18 stimulation using scRNAseq, where we observed various ontologies associated with migration and tissue remodeling. The monocytes co-cultured with IOs and treated with IL-18 showed increased expression of tissue resident macrophage markers (CD206, CD68) [68, 69]. This was in line with our previously reported IL-18-induced upregulation of TAL1, a regulator of the monocyte-to-macrophage transition [70], along with integrins and other functional markers involved in monocyte activation and migration identified in our 3D IO model. Apart from its effects on immune cells, IL-18 is also known to regulate epithelial barrier integrity in the intestine [71]. We observed that IL-18 preserves epithelial barrier integrity, whereas TNF-α stimulation disrupted the barrier integrity. Consistent with this, previous studies have shown that IL-18 restores intestinal homeostasis by inducing specific immunometabolic configurations [72]. These results, supported by the systemic data from post-surgery patients and IL-18's direct in vitro effects on monocytes, indicate that IL-18's pro-resolution actions are not limited to IOs but can be probably extended to other tissues and systems. Furthermore, our findings raise the question if this mechanism is translatable more to a systemic level or if its effect is just paracrine and tissue-specific. This evidence contradicts the previously suggested role of IL-18 in different biological setups, as a strictly pro-inflammatory cytokine.

Identifying the association of IL-18 with inflammation resolution after ORT led us to uncover specific molecular mechanisms of how IL-18 contributes to monocyte and macrophages'role in tissue regeneration and inflammation recovery. Previous studies have reported an important association with IL-18 levels and sepsis. Accumulating evidence suggests that IL-18 may hold prognostic and diagnostic potential in sepsis and post-surgical outcomes, while higher circulating IL-18 levels correlate with increased disease severity, organ dysfunction, and higher mortality rates [29, 38, 40, 50, 73]. However, this has not been explored in combination with other cytokines that might influence IL-18 levels and ultimately patient outcomes. We suggest that in the absence of TNF-α, IL-18 signaling attenuates inflammation and contributes to immune homeostasis re-establishment. Hence, we propose the TNF-α/IL-18 ratio as a novel marker of inflammation resolution, which assesses the ability of IL-18 to signal alone or with different signaling partners. We further validated our findings using an independent septic patient cohort, as a representation of the most severe immune dysregulation. In this cohort the cluster of patients uniquely determined just by IL-18 was associated with lowest SOFA score and clinical severity state, while the patients did not group together based on any other clinical characteristic. Sepsis is a multi-faceted condition that involves complex cytokine interactions and several molecular partners. Variability factors such as source of sepsis, comorbidities and level or manifestation of immune dysregulation, influence the cytokine levels, combinations and ultimately the outcome for patients. Our findings highlight the emerging need to analyze IL-18 in the broader context of other present cytokines and subsequent signaling pathways [38]. To further validate our findings, we propose in vitro trans-well assays where monocytes migrate upon stimulation with IL-18 alone or with its major signaling partners, thus clarifying how additional cytokines influence its effects. Ultimately, our findings can be translated to a novel diagnostic marker showing resolution of inflammation, perhaps with multiplex assays, determining TNF-α/IL-18 ratio, major cytokine partners of IL-18 and key biochemical characteristics like SOFA parameters and leukocyte counts, to assess patient recovery status.

However, this study has some limitations, which can be addressed by follow up studies. Further experiments should further elaborate the interplay between IL-18 and its signaling partners. IL-18 shows signaling divergence including activation of different transcription factors with the matter of time of IL-18 activation. Examining these interactions in more depth allows better understanding of the specific role of IL-18 in inflammation resolution. In terms of modeling IL-18 effects on monocyte migration and maturation, while IOs are a valuable tool to model the human intestine in a controlled setting and dissecting aspects of monocyte function, they inherently simplify the context in which monocytes interact, migrate, and differentiate into macrophages. This is because IOs is lacking the full cellular heterogeneity, tissue architecture, mechanical queues and overall systemic influences of the human intestine. Thus, further research is necessary to help translating these findings into clinical contexts. In terms of patients, we acknowledge that the findings regarding TNF-α/IL-18 require further validation and, despite being promising, are preliminary. Another limitation of this study is relatively high age of our cohort (median age 73 years), which could influence IL-18's role in inflammation resolution. Furthermore, we acknowledge that the exclusion of patients undergoing ongoing chronic immuno-suppression therapy or oncological disorders, might limit the generalization of our findings for those groups. Our results are also limited to short-term recovery periods, it may not capture the full impact of IL-18 in long-term recovery or chronic inflammation resolution.

Conclusions

In summary, we demonstrate a novel role for IL-18 in post-surgical recovery, highlighting its involvement in monocyte remodeling and inflammation resolution. Unlike its established pro-inflammatory role, IL-18 in the context of post-surgical recovery and in the absence of TNF-α co-stimulation, promotes a pro-resolution profile, potentially through Akt signaling, facilitating monocyte migration and differentiation. These findings emphasize the importance of interpreting IL-18 as part of a broader cytokine network and highlight its potential as a marker of recovery prognosis.

While our study provides significant insights, the relatively small and age-restricted cohort and the short-term focus on recovery limit the generalizability of our findings. Future studies should validate these observations in larger, more diverse populations, investigate the long-term effects of IL-18 on chronic inflammation resolution, and explore its potential as a therapeutic target. Personalized cytokine profiling, incorporating the TNF-α/IL-18 ratio, may enable patient stratification and inform tailored therapeutic strategies, advancing the management of post-surgical recovery.

Abbreviations

BCA

Bicinchoninic acid assay

BSA

Bovine serum albumin

CCL

C–C motif ligand

CD

Cluster of differentiation

cDNA

Complementary DNA

DE

Differential expression

DEG(s)

Differentially expressed gene(s)

DNA

Deoxyribonucleic acid

EGF

Epidermal growth factor

FACS

Fluorescence activating cell sorting

FBS

Fetal bovine serum

FcR

Fc receptor

GO

Gene ontology

HBSS

Hank's balanced salt solution

IF

Immunofluorescence

IL-18BPα

Interleukin 18 binding protein subunit α

IO(s)

Intestinal organoid(s)

iPSCs

Induced pluripotent stem cells

ITGB

Integrin Subunit Beta

JAM

Junction adhesion molecule

LFC

Log-fold change

MACS

Magnetic-activated cell sorting

MCP-1

Monocyte chemoattractant protein 1

MFI

Mean fluorescence intensity

mRNA

Messenger RNA

mTOR

Mechanistic Target of Rapamycin

ORT

Orthopedic surgery

PBS

Phosphate-buffered saline

Pen/Strep

Penicillin and streptomycin

PFA

Paraformaldehyde

phospho-Akt

Phosphorylated form of Akt (alpha serine/threonine-protein kinase)

phospho-NFκΒ1/p50

Phosphorylated form of nuclear factor κΒ subunit 1 / p50

PI3k

Phosphoinositide 3-Kinase

PPBP

Pro-platelet basic protein

PVDF

Polyvinylidene fluoride

qPCR

Quantitative polymerase chain reaction

RNAseq

RNA sequencing

RT

Room temperature

SOFA

Sequential organ failure assessment

sPLSDA

Sparse Partial Least Squares Discriminant Analysis

TBST

Tris buffered saline with tween

TAL-1: T-cell

Acute leukemia 1

TLR

Toll-like receptors

TNF-α

Tumor necrosis factor α

Author contributions

IP, GB, VB, ZT conducted the experiments and acquired the data. IP, GB, VB, MHK, KB, PK, PO, RK participated in the analysis of the data. MHe, JS, MD, JE, MV contributed to patient recruitment and clinical sample processing. AM, MD contributed to clinical data collection. MHe, TT, VS supervised the clinical part of study. PK supervised bulk RNAseq analysis. IP, VB, ES, JF and MHK wrote the manuscript. MHK and JF designed the study, MV, MHo, MHK and JF acquired the funding and supervised the study. All authors participated in editing and reviewing the manuscript.

Funding

This research was supported by the Ministry of Health of the Czech Republic (NU21-06-00408 awarded to MV and MHK), all rights reserved and DRO (Institute of Hematology and Blood Transfusion—UHKT, 00023736 awarded to JF).

Availability of data and materials

The data that support the findings of this study are available on request from the corresponding author. Part of the data is not publicly available due to privacy or ethical restrictions. The full RNAseq and scRNAseq datasets have been deposited to GEO database under the accession numbers GSE282927 and GSE282929, respectively.

Declarations

Ethics approval and consent to participate

Written informed consents were obtained from all recruited patients, and all procedures and protocols were approved by the institutional ethics committee of St Anne's University Hospital Brno (11G/2021).

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.