IL-8 contributes to postoperative adhesion formation through the crosstalk of neutrophils and mesothelial cells.

Before assessing neutrophil chemotaxis, we confirmed that recombinant IL-8 activated both CXCR1 and CXCR2, whereas recombinant CXCL1 selectively activated CXCR2 using β-arrestin recruitment assays in U2OS-CXCR1 and CHO-K1-CXCR2 cells (Fig.  1 A). Fig. 1 Evaluation of functional differences between IL-8 and CXCL1 in neutrophils. ( A ) Evaluation of IL-8 and CXCL1 activity in the U2OS-CXCR1 reporter cells ( left ) and CHO-K1-CXCR2 reporter cells ( right ). ( B ) Flow cytometric analysis of CXCR1 and CXCR2 surface expression on neutrophils. ( C ) Neutrophil chemotaxis assay toward IL-8 or CXCL1. RLU, relative luminescence units. Data in ( A ) and ( C ) are presented as the mean ± SD (n = 3). Evaluation of functional differences between IL-8 and CXCL1 in neutrophils. ( A ) Evaluation of IL-8 and CXCL1 activity in the U2OS-CXCR1 reporter cells ( left ) and CHO-K1-CXCR2 reporter cells ( right ). ( B ) Flow cytometric analysis of CXCR1 and CXCR2 surface expression on neutrophils. ( C ) Neutrophil chemotaxis assay toward IL-8 or CXCL1. RLU, relative luminescence units. Data in ( A ) and ( C ) are presented as the mean ± SD (n = 3). We next verified the surface expression of CXCR1 and CXCR2 on neutrophils by flow cytometry (Fig.  1 B). Finally, transwell chemotaxis assays revealed that IL-8 exhibited stronger chemotactic activity for neutrophils than CXCL1 (Fig.  1 C). First, we assessed the surface expression of CXCR1 and CXCR2 on the human mesothelial cell line MeT-5A and detected CXCR1 but not CXCR2, indicating that IL-8 can potentially act directly on these cells (Fig.  2 A and B). Fig. 2 Reciprocal amplification of IL-8 and TNF-α in MeT-5A cells. ( A ,  B ) Flow cytometric analysis of CXCR1 ( A ) and CXCR2 ( B ) surface expression on MeT-5A cells. Left: representative overlaid histograms. Right: quantification of MFI across biological replicates. P  values were determined using Student’s t test. *** indicates P   <  0.001. ( C ) Evaluation of TNF-α concentrations in culture medium. P  values were determined using Tukey’s multiple comparison test. *** indicates P   <  0.001. ( D ) Evaluation of IL-8 concentrations in culture medium. P  values were determined using Williams’ test. * indicates P   <  0.025. Quantitative data are presented as individual data points with mean ± SD. (n = 3). Reciprocal amplification of IL-8 and TNF-α in MeT-5A cells. ( A ,  B ) Flow cytometric analysis of CXCR1 ( A ) and CXCR2 ( B ) surface expression on MeT-5A cells. Left: representative overlaid histograms. Right: quantification of MFI across biological replicates. P  values were determined using Student’s t test. *** indicates P   <  0.001. ( C ) Evaluation of TNF-α concentrations in culture medium. P  values were determined using Tukey’s multiple comparison test. *** indicates P   <  0.001. ( D ) Evaluation of IL-8 concentrations in culture medium. P  values were determined using Williams’ test. * indicates P   <  0.025. Quantitative data are presented as individual data points with mean ± SD. (n = 3). Consistent with this, IL-8 treatment upregulated TNF-α production; TNF-α is a pro-inflammatory cytokine that accelerates adhesion formation 11 and its signaling pathway is upregulated in our monkey PA models 16 (Fig.  2 C). This IL-8-induced TNF-α response was blocked by the anti-IL-8 neutralizing antibody, AMY109 23 , confirming the specificity of the IL-8 effect (Fig.  2 C). Notably, CXCL1, which can only interact with CXCR2 (Fig.  1 A), failed to upregulate TNF-α (Fig.  2 C). Having observed that IL-8 stimulation increased TNF-α production, we next investigated whether TNF-α could reciprocally affect IL-8 production and found that TNF-α upregulated IL-8 production in a dose-dependent manner (Fig.  2 D). These findings demonstrate a reciprocal amplification loop between TNF-α and IL-8 in mesothelial cells, a specific interaction not observed with CXCL1. To determine whether IL-8 promotes fibrotic responses, we treated neutrophils with IL-8, but TGFB1 expression was unchanged (Fig.  3 A). In contrast, TNF-α treatment upregulated TGFB1 expression in neutrophils (Fig.  3 A). Fig. 3 Fibrotic reactions in neutrophils and MeT-5A cells. ( A ) RNA expression change of TGFB1 in neutrophils stimulated with IL-8, IL-8 + AMY109, or TNF-α. ( B ) RNA expression change of TGFB1 , COL1A1 , FN1 , CCN2 ( CTGF ), SERPINE1 (PAI-1) and VIM in MeT-5A cells stimulated with TGF-β1 (0.4, 2, 10 ng/mL), TNF-α or IL-8. Data are presented as individual data points with mean ± SD. (n = 3) P  values were determined using Dunnett’s test. * and *** indicate P   <  0.05 and P   <  0.001, respectively. Fibrotic reactions in neutrophils and MeT-5A cells. ( A ) RNA expression change of TGFB1 in neutrophils stimulated with IL-8, IL-8 + AMY109, or TNF-α. ( B ) RNA expression change of TGFB1 , COL1A1 , FN1 , CCN2 ( CTGF ), SERPINE1 (PAI-1) and VIM in MeT-5A cells stimulated with TGF-β1 (0.4, 2, 10 ng/mL), TNF-α or IL-8. Data are presented as individual data points with mean ± SD. (n = 3) P  values were determined using Dunnett’s test. * and *** indicate P   <  0.05 and P   <  0.001, respectively. Next, we evaluated the effects of these cytokines on MeT-5A cells. IL-8 did not alter the expression of fibrosis-related genes ( TGFB1 , COL1A1 , FN1 , CCN2 ( CTGF ), SERPINE1 ) or the mesenchymal marker ( VIM ) (Fig.  3 B). By contrast, TNF-α selectively upregulated SERPINE1 and VIM , while TGF-β1 upregulated all assessed fibrosis-related genes ( TGFB1 , COL1A1 , FN1 , CCN2 , SERPINE1 ) as well as VIM (Fig.  3 B). Collectively, these results indicate that IL-8 does not directly induce fibrosis-related genes but may enhance neutrophil-derived TGF-β1 production through TNF-α induction in mesothelial cells, which could contribute to fibrotic responses. We first examined the time course of IL-8 expression in the injured abdominal wall of cynomolgus monkeys following caesarean section (C-section), which exhibits histopathological changes similar to those in the monkey PA model 16 , and confirmed that IL-8 levels increased rapidly, peaked at 6 h, and returned toward baseline by day 7 (Fig.  4 A). Similarly, the monkey PA model exhibited an almost identical temporal profile after surgery, with IL-8 concentrations peaking at 6 h and decreasing by Day 7, although they remained above baseline (Fig.  4 B). Fig. 4 Evaluation of IL-8 protein expression in injured abdominal walls from monkeys following cesarean sections (C-sections) and in postoperative adhesions (PA) model. ( A ) IL-8 concentrations in injured abdominal walls from monkeys following C-sections. Data are presented as individual data points with mean values indicated (n = 10 lesions for 0 h, 2 lesions for 6 h, Day 1, 3, 7, and 1 month). ( B ) IL-8 concentrations in injured abdominal walls from monkey PA models. Data are presented as individual data points with mean ± SD. (n = 6 lesions for 0 h, 8 lesions for 6 h, 4 lesions for Day 7) P  values were determined using Steel’s test. * and ** indicate  P   <  0.05 and  P   <  0.01, respectively. Evaluation of IL-8 protein expression in injured abdominal walls from monkeys following cesarean sections (C-sections) and in postoperative adhesions (PA) model. ( A ) IL-8 concentrations in injured abdominal walls from monkeys following C-sections. Data are presented as individual data points with mean values indicated (n = 10 lesions for 0 h, 2 lesions for 6 h, Day 1, 3, 7, and 1 month). ( B ) IL-8 concentrations in injured abdominal walls from monkey PA models. Data are presented as individual data points with mean ± SD. (n = 6 lesions for 0 h, 8 lesions for 6 h, 4 lesions for Day 7) P  values were determined using Steel’s test. * and ** indicate  P   <  0.05 and  P   <  0.01, respectively. Given this rapid upregulation of IL-8 expression following surgery, the anti-IL-8 antibody AMY109 was administered prophylactically to assess whether IL-8 is involved in PA formation. Laparoscopic scoring of adhesions on the injured abdominal wall demonstrated that AMY109 significantly ameliorated abdominal adhesion formation compared with the non-treatment group (Fig.  5 A–C and Supplementary Table S1). Adhesion formation on the uterus was also evaluated, and AMY109 showed a trend toward reducing the uterine adhesion score ( P  = 0.0907; Fig.  6 A and B). Notably, two of the five AMY109-treated monkeys (Animal ID: 209 and 215) exhibited no detectable uterine adhesions (Fig.  6 A and Supplementary Table S2). Fig. 5 Efficacy of AMY109 on postoperative adhesions (PA) at abdominal walls in monkeys. ( A ,  B ) Representative laparoscopic images of PA at abdominal walls in the non-treatment ( A ) and AMY109-treated ( B ) monkeys at 26 days after PA induction surgery. Left: wide-field laparoscopic image. Right: laparoscopic evaluation of adhesion width. Dotted lines indicate the induced adhesion area. ( C ) Adhesion scores for the non-treatment and AMY109-treated groups. Adhesion score was determined as follows: 0 = no adhesion; 1 = adhesion width < 1 mm; 2 = 1 mm ≤ adhesion width < 5 mm; 3 = adhesion width ≥ 5 mm. Data are presented as individual data points with mean ± SD. (n = 6 lesions for non-treatment group, 10 lesions for AMY109 treated group) P  value was determined using Pearson’s chi-square test. * indicates P   <  0.05. Fig. 6 Efficacy of AMY109 on postoperative adhesions (PA) on the uterus in monkeys. ( A ) Laparoscopic images of postoperative adhesions at uterus for all tested animals. ( B ) Adhesion score of the non-treat and AM109 groups (scored as in Fig.  5 C: 0 = no adhesion; 1 = adhesion width < 1 mm; 2 = 1 mm ≤ width < 5 mm; 3 = width ≥ 5 mm). Data are presented as individual data points with mean values ± SD. (n = 3 lesions for non-treatment group and 5 lesions for AMY109 treated group) P  value was determined using Pearson’s chi-square test. Efficacy of AMY109 on postoperative adhesions (PA) at abdominal walls in monkeys. ( A ,  B ) Representative laparoscopic images of PA at abdominal walls in the non-treatment ( A ) and AMY109-treated ( B ) monkeys at 26 days after PA induction surgery. Left: wide-field laparoscopic image. Right: laparoscopic evaluation of adhesion width. Dotted lines indicate the induced adhesion area. ( C ) Adhesion scores for the non-treatment and AMY109-treated groups. Adhesion score was determined as follows: 0 = no adhesion; 1 = adhesion width < 1 mm; 2 = 1 mm ≤ adhesion width < 5 mm; 3 = adhesion width ≥ 5 mm. Data are presented as individual data points with mean ± SD. (n = 6 lesions for non-treatment group, 10 lesions for AMY109 treated group) P  value was determined using Pearson’s chi-square test. * indicates P   <  0.05. Efficacy of AMY109 on postoperative adhesions (PA) on the uterus in monkeys. ( A ) Laparoscopic images of postoperative adhesions at uterus for all tested animals. ( B ) Adhesion score of the non-treat and AM109 groups (scored as in Fig.  5 C: 0 = no adhesion; 1 = adhesion width < 1 mm; 2 = 1 mm ≤ width < 5 mm; 3 = width ≥ 5 mm). Data are presented as individual data points with mean values ± SD. (n = 3 lesions for non-treatment group and 5 lesions for AMY109 treated group) P  value was determined using Pearson’s chi-square test.

PathHunter® U2OS CXCR1 β-Arrestin (U2OS-CXCR1, #93-0226C3) and PathHunter® CHO-K1 CXCR2 β-Arrestin cells (CHO-K1-CXCR2, #93-0202C2) were purchased from Eurofins DiscoverX and were used to evaluate CXCR1- and CXCR2-mediated signaling, respectively. U2OS-CXCR1 cells were maintained in AssayComplete™ Cell Culture Kit-103 (Eurofins DiscoverX). CHO-K1-CXCR2 were maintained in AssayComplete™ Cell Culture Kit-107 (Eurofins DiscoverX). For the in vitro assessment of mesothelial cell responses, MeT-5A cells (CRL-9444), derived from human pleural mesothelium, were purchased from American Type Culture Collection and maintained in Medium 199 (Thermo Fisher Scientific, Inc.) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich) and 20 mM HEPES solution (Sigma-Aldrich). Human neutrophils were isolated from healthy donor blood by EasySep Direct Human Neutrophil Isolation Kit (STEMCELL Technologies, Inc.) according to the manufacturer’s instructions. The blood sampling protocol from healthy donors was approved by the ethics committee of Chugai Pharmaceutical Co., Ltd. (Approval No. E23049). All procedures were performed in accordance with the Declaration of Helsinki. Informed consent was obtained from all participants. Recombinant human IL-8 (UniProt P10145 , residues 1—99) was produced in CHO cells by Chiome Bioscience Inc. Recombinant Human CXCL1/GRO alpha Protein (CXCL1, #275-GR-050) was purchased from R&D Systems, Inc. Recombinant Human TNF-alpha Protein (TNF-α, #210-TA-005 and Recombinant Human TGF-beta 1 Protein (TGF-β1, #240-B-010) were purchased from R&D Systems, Inc. The anti-IL-8 antibody AMY109 was developed as described in a previous report 23 . Human IgG1, Kappa Purified Myeloma Protein (hIgG1, #I5154) was purchased from Sigma-Aldrich. To assess the agonistic activity of IL-8 and CXCL1 on CXCR1 and CXCR2, we evaluated CXCR1 and CXCR2 GPCR activation via the recruitment of β-arrestin using U2OS-CXCR1 and CHO-K1-CXCR2 reporter cells, respectively. These cells were incubated with serially diluted IL-8 and CXCL1 at 37 °C with 5% CO2 for 75 min, followed by incubation at room temperature for 15 min (total 90 min). β-galactosidase activity was then measured using the PathHunter® Detection Kit (Eurofins DiscoverX) according to the manufacturer’s instructions. Surface expression of CXCR1 and CXCR2 on freshly isolated human neutrophils and MeT-5A cells was evaluated by flow cytometry. Isolated human neutrophils or MeT-5A cells were incubated with FcR Blocking Reagent, human (Miltenyi Biotech) at 4 °C for 10 min, followed by incubation with eBioscience Fixable Viability Dye eFluor 780 (Thermo Fisher Scientific, Inc.) together with one of the following primary antibodies (10 μg/mL): Purified Mouse Anti-Human CD181 antibody (BD Biosciences, clone 5A12), Purified Mouse Anti-Human CD182 antibody (BD Biosciences, clone 6C6), Purified Mouse IgG2b κ Isotype Control (BD Biosciences, clone 27–35) or Purified Mouse IgG1, κ Isotype Control (BD Biosciences, clone MOPC-21) at 4 °C for 30 min. Then, cells were washed and incubated with Goat anti-Mouse IgG (H + L) Secondary Antibody, Biotin (Thermo Fisher Scientific, Inc., # 31800, 1 μg/mL) at 4 °C for 30 min. Neutrophil samples were further stained for 30 min at 4 °C with PE-CF594 streptavidin (BD Biosciences, #562284, 2 μg/mL) together with BV421-conjugated anti-human CD66b (BD Biosciences, clone G10F5, 4 μg/mL) and PerCP/Cy5.5-conjugated anti-human CD16 (BioLegend, clone 3G8, 4 μg/mL) to allow neutrophil gating (CD66b⁺ CD16⁺). MeT-5A samples received only PE-CF594 streptavidin (2 μg/mL) under the same conditions. Samples were analyzed with Fortessa-X20 instrument (BD Biosciences), and data analysis was performed via FlowJo v10.4.1 (FlowJo, LLC). Gating strategies for the flow cytometry analysis of neutrophils and MeT-5A are shown in Supplementary Fig. S1A and B, respectively. The neutrophil chemotaxis assay was performed with transwell plates with 3-μm pores (CORNING). A solution containing serially diluted IL-8 and CXCL1 was placed in each lower compartment, and isolated human neutrophil suspensions were placed in the upper compartment. After incubating at 37 °C with 5% CO2 for 2 h, the number of migrated neutrophils was measured using CellTiter-Glo ® 2.0 (Promega) according to the manufacturer’s instructions. MeT-5A cells were incubated with IL-8 (100 ng/mL) with or without AMY109 (10 μg/mL) or hIgG1 (10 μg/mL), CXCL1 (100 ng/mL), TNF-α (16–10,000 pg/mL) and TGF-β1 (0.4–10 ng/mL) at 37 °C with 5% CO2 for 24 h. After incubation, the culture media were collected, and total RNA was extracted using the RNeasy Mini Kit. The isolated human neutrophils were incubated in RPMI-1640 (Sigma-Aldrich) supplemented with 5% FBS containing IL-8 (100 ng/mL) with or without AMY109 (10 μg/mL) and TNF-α (5 ng/mL). After incubation at 37 °C with 5% CO2 for 18 h, total RNA was extracted using the RNeasy Mini Kit (QIAGEN) according to the manufacturer’s instructions. The culture supernatants from cytokine-treated neutrophils or MeT-5A were subjected to an Inflammation 20-Plex Human ProcartaPlex Panel (Thermo Fisher Scientific Inc.) for evaluating cytokine production and was performed according to the manufacturer’s instructions. The total RNA from neutrophils or MeT-5A cells treated by cytokines was reverse-transcribed to cDNA using SuperScript™ IV VILO™ Master Mix (Thermo Fisher Scientific, Inc.) according to the manufacturer’s instructions. qRT-PCR was performed using TaqMan™ Fast Advanced Master Mix for qPCR (Thermo Fisher Scientific, Inc.) and QuantStudio 12 K Flex (Thermo Fisher Scientific, Inc.) according to the manufacturer’s instructions. PCR primers for TGFB1 (Hs00998133_m1), COL1A1 (Hs00164004_m1), FN1 (Hs01549976_m1), CCN2 ( CTGF ) (Hs00170014_m1) and GAPDH (Hs02786624_g1) were purchased from Thermo Fisher Scientific, Inc. PCR primers for SERPINE1 (Hs.PT.58.3938488.g) and VIM (Hs.PT.58.38906895) were purchased from Integrated DNA Technologies, Inc. RNA expression change in TGFB1 , COL1A1 , FN1 , CCN2 , SERPINE1 or VIM was calculated via the ΔΔCt method and GAPDH was used as an endogenous control gene. The animal studies were conducted using cynomolgus monkeys in the breeding colony of Tsukuba Primate Research Center (TPRC) at the National Institutes of Biomedical Innovation, Health and Nutrition (NIBN) in Ibaraki, Japan. Cynomolgus monkeys were originally brought from the Philippines, Indonesia, and Malaysia. Although colonies were sometimes crossbred, the monkeys were basically bred within a colony of animals from the same geographic origin. The monkeys which were used for this study were second to fourth generation descendants of the imported monkeys. This study was conducted using sixteen female and six male cynomolgus monkeys of 6–21 years old in the breeding colony of TPRC from 2017 to 2025. The environment of the animal room was set with 25 ± 3 °C room temperature, 60 ± 5% relative humidity, and a 12 h light-and-dark cycle. The animals were given water ad libitum and fed daily with 70–100 g of commercially available solid food (CMK-2; CLEA Japan, Inc.) and 100 g apples. Animal studies were approved by the Institutional Animal Care and Use Committee of the NIBN (approval number: DS28-53R7, DS30-37R8). All monkeys used in this study were cared for according to procedures approved by the Animal Care and Use Committee of the NIBN, which reviews study plans according to the guidelines in Japan specified in the ‘Ministry of Health, Labor and Welfare: Basic Policies for the Conduct of Animal Experimentation’. In addition, protocols for all experiments involving animals complied with the guidelines set by the same institute for the care, use, and biological hazard countermeasures of laboratory animals. The study is reported in accordance with ARRIVE guidelines. Ten pregnant female cynomolgus monkeys in the breeding colony were used to determine the postoperative time course of IL-8 protein in the injured abdominal wall. Cesarean sections and subsequent sample acquisition were performed as follows, in accordance with the previous report 16 : General anesthesia was induced with intramuscular ketamine hydrochloride (10 mg/kg; Ketalar, Daiichi Sankyo Propharma Co., Ltd.) and atropine sulfate hydrate (0.05 mL/kg; ATROPINE SULFATE Injection, Nipro ES Pharma Co., Ltd.). Additionally, the local anesthetics lidocaine (Xylocaine, Sandoz Pharma K.K.) and the hemostatic agent (Bosmin, Daiichi Sankyo Co., Ltd.) were subcutaneously injected. After a midline laparotomy, a sample of the abdominal wall was collected as a 0 h sample. Following fetal delivery, inhalational general anesthesia was administered using sevoflurane (Sevoflurane Inhalation Solution, Viatris Healthcare G.K.). At 6 h (n = 2), 1 day (n = 2), 3 days (n = 2), 7 days (n = 2), 30 days (n = 1) and 31 days (n = 1) after cesarean section, abdominal walls at a midline incision site were collected under anesthesia via intramuscular administration of a mixture containing ketamine hydrochloride and atropine sulfate hydrate. The collected samples were placed in liquid nitrogen. For postoperative management of animals, Cefazolin Sodium (Nichi-Iko Pharmaceutical Co., Ltd.) as an antibiotic and Buprenorphine Hydrochloride (NISSIN, Nissin Pharmaceutical Co., Ltd.) as an analgesic were intramuscularly administered for 3 days after surgery. Eight female and six male monkeys were subjected to PA induction surgery. Three of eight female monkeys were the same individuals that participated in our previous study 16 . Adhesion induction surgeries and subsequent sample acquisition were performed as follows, in accordance with the previous report 16 . Monkeys were given general anesthesia by intramuscular administration of a mixture of ketamine hydrochloride and xylazine hydrochloride (Seractal, Bayer Yakuhin, Ltd., Osaka, Japan), followed by a midline incision. Two 10 mm incisions were made on the abdominal wall, approximately 20 mm to the left and right of the midline incision site and these incisions were abraded five times with gauze and sutured with four stitches. 10 mm incision, five-time abrasion and suture were also conducted on the uterus. Following surgery, these monkeys underwent the same postoperative management as those that had undergone C-sections. The abdominal wall was sampled immediately after the midline laparotomy as a control sample (0 h sample). At 6 h (four male monkeys) and 7 days (two male monkeys) after surgery, monkeys were anesthetized with the same protocol used for PA induction surgery. For tissue sampling, some animals were subsequently euthanized via intravenous administration of pentobarbital, an AVMA-approved method for non-human primates. Following either anesthesia or euthanasia, a midline abdominal incision was made, the PA-induced areas of the abdominal wall were excised, and the samples were immediately snap-frozen in liquid nitrogen. Tissue lysates from surgery sites were prepared as follows. The frozen tissue samples were segmented and ground with a Multi-beads shocker MB100-4A (Yasui Kikai Corp.). T-PER™ tissue protein extraction reagent (ThermoFisher Scientific, Inc.) was mixed with cOmplete™, Mini, EDTA-free protease inhibitor cocktail (F. Hoffmann-La Roche Ltd.) and PhosSTOP™ EASYpack inhibitors (F. Hoffmann-La Roche Ltd.). With this solution, the ground samples were mixed and homogenized twice, and the resulting mixture was pooled. After centrifugation, the supernatants (tissue lysate sample) were stored frozen until IL-8 measurement described below. The concentrations of IL-8 in the tissue lysates were measured by sandwich immunoassay using two noncompetitive anti-human IL-8 antibodies (Chugai Pharmaceutical Co., Ltd.). Multi-array® 96-well plate (MESO SCALE DIAGNOSTICS, LLC (MSD)) was coated with AMY109 to prepare AMY109-coated plate. The tissue lysate samples and IL-8 standard solutions were applied to the AMY109-coated plate, and captured IL-8 was incubated with anti-IL-8 antibody followed by Anti Rabbit Antibody Goat SULFO-TAG Labeled (MSD). Electrochemiluminessence signals were measured by MESO® SECTOR S600 (MSD) after incubation with MSD Read Buffer T (MSD). The detection limit was 2.05 pg/mL of the tissue lysates. Three female monkeys that underwent PA induction surgery in the previous study 16 were utilized as a control (non-treatment) group. Five additional female monkeys were utilized as an anti-IL-8 antibody, AMY109 treatment group. In this group, 10 mg/kg of AMY109 was administered intravenously two days prior to PA induction surgery. Adhesion formation was assessed by laparoscopy 26 days after PA induction surgery, a timepoint at which we had previously confirmed PA establishment on injured abdominal walls and uterus 16 . The laparoscopic procedure was performed as previously described 16 . Monkeys were anesthetized via a mixture of ketamine hydrochloride and xylazine hydrochloride, after which a small midline abdominal incision was made, through which a laparoscope was inserted. Next, a high-flow insufflation unit was used to insufflate carbon dioxide gas into the peritoneal cavity and inflate the abdomen. To quantify postoperative adhesions, a small ventral incision was then made to insert a scaled bar. The width of adhesions on the injured abdominal wall and uterus in each monkey was evaluated according to the following scoring system: score 0, no adhesion; score 1, adhesion width < 1 mm; score 2, 1 mm ≤ adhesion width < 5 mm; score 3, adhesion width ≥ 5 mm. Following the laparoscopic observation, both incisions were sutured closed with Vicryl Rapide suture 4–0. The monkeys then underwent the same postoperative treatment as those that had undergone C-sections. All statistical analysis was performed using JMP version 17. The data are presented as the means ± SDs unless otherwise noted. Detailed information on the statistical tests used, sample sizes(n), and P values, which are indicated by asterisks, are provided in each figure legend.

PA can cause serious complications for patients after surgical interventions, creating an urgent need for the development of effective therapeutic strategies. In this study, we revealed a previously unappreciated role for IL-8 in the progression of PA and identified it as a potential therapeutic target. We found that (1) IL-8 showed superior capacity for neutrophil migration compared to its functional homologue, CXCL1; (2) IL-8 reciprocally amplified TNF-α in MeT-5A cells; (3) TNF-α, in turn, upregulated TGF-β1 in neutrophils, which can subsequently contribute to the fibrotic responses; and (4) blockade of IL-8 ameliorated PA formation in the monkey PA model. Collectively, these data indicate that IL-8 orchestrates both early inflammatory responses, including neutrophil migration and TNF-α upregulation, and the subsequent fibrotic cascade through bidirectional crosstalk between neutrophils and mesothelial cells (Fig.  7 ). Targeting IL-8 could be a novel therapeutic strategy for PA by simultaneously regulating both the inflammatory and fibrotic phases of adhesion development. Fig. 7 Possible role of IL-8 and AMY109 in postoperative adhesion formation. IL-8 promotes neutrophil infiltration and forms a reciprocal amplification with TNF-α in mesothelial cells. Elevated TNF-α upregulates TGF-β1 in neutrophils, driving adhesion formation. Neutralization of IL-8 by AMY109 blocks neutrophil recruitment and subsequent inflammatory responses and fibrotic reactions, thereby preventing adhesion formation. Possible role of IL-8 and AMY109 in postoperative adhesion formation. IL-8 promotes neutrophil infiltration and forms a reciprocal amplification with TNF-α in mesothelial cells. Elevated TNF-α upregulates TGF-β1 in neutrophils, driving adhesion formation. Neutralization of IL-8 by AMY109 blocks neutrophil recruitment and subsequent inflammatory responses and fibrotic reactions, thereby preventing adhesion formation. Neutrophil depletion has been reported to reduce PA formation in mouse models, suggesting that blocking neutrophil migration to surgical areas could be a new strategy for PA prevention 9 , 10 . CXCL1 and CXCL2, acting via CXCR2, have been reported to be involved in neutrophil migration in mouse PA models 9 – 11 . Because both IL-8 and CXCL1 were highly upregulated in our monkey PA model 16 , we compared their chemotactic activities using human neutrophils in vitro and found that IL-8 showed much stronger migratory activity than CXCL1. Neutrophil migration induced by IL-8 has been reported to be predominantly controlled via CXCR1 rather than CXCR2 22 , suggesting that CXCR1 might be the more important chemokine receptor in this process. Altogether, these findings indicate that IL-8 is likely to be an effective therapeutic target for inhibiting neutrophil recruitment in humans. Importantly, mice lack the IL-8 gene, whereas humans and monkeys possess it, highlighting fundamental species-specific differences in chemokine biology and the value of our monkey PA model for translational research. Subsequently, building upon work by Uyama et al. which reported that the crosstalk of neutrophils and mesothelial cells mediated by IL-6, CXCL2, TNF-α and TGF-β1 contributed to the progression of PA 10 , we investigated whether IL-8 played a role in this crosstalk. Notably, we found that IL-8 treatment upregulated TNF-α expression in MeT-5A cells, whereas CXCL1 did not. This difference may in part reflect the receptor expression pattern: MeT-5A cells express CXCR1 but not CXCR2. While IL-8 can activate both CXCR1 and CXCR2, CXCL1 can only activate CXCR2, which may account for its lack of effect in this context. Furthermore, TNF-α upregulated IL-8 expression in MeT-5A cells, supporting a reciprocal amplification between TNF-α and IL-8. Given that a previous study reported that TGF-β1, the master regulator of fibrotic responses, was expressed in neutrophils and myofibroblasts that may include cells of mesothelial origin in PA in mice 10 , we evaluated whether IL-8 and TNF-α upregulate TGF-β1 in neutrophils and mesothelial cells. TNF-α upregulated TGFB1 in neutrophils, consistent with a previous report 10 , whereas IL-8 did not. In MeT-5A cells, IL-8 did not induce fibrosis-related genes, whereas TNF-α upregulated SERPINE1 and the mesenchymal marker VIM . Furthermore, TGF-β1 upregulated these markers together with broader fibrosis-associated genes ( TGFB1 , COL1A1 , FN1 , and CCN2 ). Upregulation of VIM in MeT-5A cells suggests that mesothelial cells may undergo mesothelial-to-mesenchymal transition (MMT), which has been implicated in adhesion development 24 . These cytokine-driven mesenchymal shifts, together with the upregulation of fibrosis-associated genes, may allow mesothelial cells to contribute to PA formation. Taken together, our findings delineate a dual role for IL-8 in PA formation: it not only acts as a primary chemoattractant for neutrophils but also may amplify TNF-α production in mesothelial cells, which in turn promotes neutrophil-derived TGFB1 and activates MMT or fibrosis-related pathways in mesothelial cells. This dual mechanism offers a potential explanation for how IL-8 blockade may attenuate PA formation by suppressing both the initial inflammatory cell infiltration and the subsequent fibrotic cascade. To determine the optimal treatment schedule, we evaluated the postoperative time course change of IL-8 expression in monkeys following C-sections. IL-8 rose rapidly, peaking at 6 h post-surgery and decreasing from its peak by day 7. A similar pattern was observed in our PA model. Although the limited sampling intervals may not have captured the exact peak, the data clearly indicate that the steepest increase occurred within the first few hours after the incision. This early surge provided the rationale for administering AMY109 before surgery, and the efficacy of this pre-operative treatment was subsequently assessed. In our PA model, standardized incision, abrasion, and suturing were performed on both the abdominal wall and the uterus, and the extent of adhesions that developed at these injury sites was evaluated to determine the pharmacological effect of IL-8 blockade. To obtain a reliable yet minimally invasive read-out of PA severity, we adopted laparoscopic scoring. Although adhesion scoring based on the combination of adhesion area and adhesion tenacity is often utilized to evaluate PA severity 13 , filmy and dense vascularized adhesions frequently co-existed within the same lesion (Fig.  5 A), making tenacity grading inconsistent. Adhesion severity was therefore quantified by measuring adhesion width, which was appropriate given the standardized length of the injury. Notably, while adhesion width served as a practical indicator of severity, there were technical limitations inherent to laparoscopic measurement. Because adhesion width was assessed using a scaled bar marked in 1-mm increments, lesions with widths below 1 mm could not be measured precisely, even though their mild severity was evident. Conversely, in some animals, extensive adhesions from the injury site merged with those arising from the midline laparotomy incision, making it difficult to delineate the boundary of the induced adhesion. Although the exact adhesion width could not be measured in these animals, it was still evident whether the lesion represented a very mild or very severe adhesion. To address these limitations, we adopted a categorical scoring system based on adhesion width to assess adhesion severity consistently and interpretably across animals. AMY109 significantly reduced the adhesion score of the abdominal wall and showed a tendency to reduce that of the uterus ( P  < 0.05 and P  = 0.0907, respectively). Each monkey had two abdominal wall lesions but only one uterine lesion, the smaller number of evaluable uterine sites likely contributing to the reduced statistical power in the uterine analysis. Notably, two of the five AMY109-treated monkeys exhibited no uterine adhesions, suggesting that IL-8 blockade may be effective in the uterus as well as in the abdominal wall. Taken together, these findings provide the first primate-based evidence that the pharmacological blockade of IL-8 can attenuate PA formation. Despite these encouraging findings, an important limitation must be acknowledged. Although AMY109 reduced adhesion severity in vivo, we were unable to determine whether this effect was accompanied by suppression of neutrophil infiltration or by modulation of mesothelial-neutrophil crosstalk identified in vitro. Surgical injury elevates alarmins or damage-associated molecular patterns (DAMPs), which are known to induce IL-8 and TNF-α, and these cytokines can be produced by multiple immune and stromal cell types at the surgical sites, indicating that a range of cellular sources likely contributes in vivo 20 , 25 . Thus, clarifying whether these proposed processes operate in vivo would be important and would provide deeper biological insight into how IL-8 contributes to PA formation. As PA is an excessive wound healing process, any anti-adhesion strategy must be evaluated for its potential impact on normal tissue repair. In this study, no wound abnormalities, including wound dehiscence, were observed at the surgery sites in AMY109-treated monkeys. Although CXCR2 knockout mice have been reported to show delayed wound healing 26 , this discrepancy may be explained by compensatory activity from other chemokine family members that can partially substitute for IL-8's function in wound healing. Although we confirmed that IL-8 is more potent than its functional homologue, CXCL1, the redundancy in the chemokine system may allow AMY109 to suppress excessive fibrosis leading to PA formation without compromising normal wound healing. In summary, we demonstrate a novel strategy for PA prevention using an anti-IL-8 antibody. In a monkey PA model, the anti-IL-8 antibody AMY109 significantly reduced adhesion formation, suggesting its high translational potential for human use. IL-8 inhibition, when combined with the inhibition of other reported pathways, including IL-6 10 and TNF-α 11 , may potentially yield more potent effects. Further research in this area is expected to lead to the development of more efficient therapeutic strategies.

Postoperative adhesions (PA) frequently develop after abdominal surgery, and severe cases can give rise to small-bowel obstruction, chronic pain, and female infertility 1 , 2 . A substantial proportion of patients require readmission or adhesiolysis for adhesion-related symptoms, and the healthcare costs for PA are estimated to exceed US$1.3 billion annually in the United States 3 . Although several barrier materials have been approved to prevent PA formation 4 , adhesions still occur, and none of these products has achieved widespread clinical adoption 5 – 7 . The continued incidence of PA despite the availability of barrier methods underscores the limitations of current preventive strategies and the urgent need for alternative approaches. To develop such strategies, a deeper understanding of the molecular and cellular mechanisms underlying PA formation and progression is essential. The process of PA formation involves a complex series of events. Surgical trauma triggers inflammatory responses such as upregulation of inflammatory cytokines and influx of inflammatory cells, which subsequently activate coagulation cascades, resulting in fibrin deposition and, ultimately, PA formation 6 , 8 . Recent studies with mouse PA models have delineated several steps in PA progression. Mesothelial cells, comprising the epithelial lining of the peritoneum and visceral organs, are activated after surgery and secrete chemokines that attract immune cells including neutrophils and monocytes 9 . The migrated neutrophils promote the adhesion process by crosstalking with mesothelial cells through inflammatory modulators such as IL-6, CXCL2, TNF-α, neutrophil extracellular traps (NETs), and fibrotic modulator TGF-β1 9 – 11 . Furthermore, a lineage tracing study using a mouse PA model revealed that the adhesion tissues were formed from injured mesothelial cells 12 , indicating that they modulate both inflammatory responses and fibrotic responses. Despite these insights from mouse PA models, species differences limit the direct translation of rodent data to humans 13 – 15 . To overcome this difficulty, we have developed a novel PA model in cynomolgus monkeys ( Macaca fascicularis ) 16 which share close genomic and anatomical similarities with humans 17 . Our model not only reliably established PA but recapitulated their progression cascades. The adhesion sites also contained cell types similar to those in humans, including mesothelial cells, demonstrating our model’s high translatability to human biology 12 , 18 . Furthermore, we confirmed the postoperative upregulation of CXCL8 (IL-8) expression, which is not genetically conserved in mice, as well as CXCL1 at surgical sites. IL-8 is a CXC chemokine rapidly induced by danger signals or inflammatory signals 19 , 20 . Its receptors, CXCR1 and CXCR2, are expressed on neutrophils, monocytes, and endothelial cells, and their engagement promotes chemotaxis, neutrophil degranulation, and angiogenesis 20 . In rodents, CXCL1 and CXCL2, functional homologues of IL-8 14 , 20 that mainly signal through CXCR2, compensate for its absence 20 and have been implicated in mouse PA progression through the induction of neutrophil chemotaxis to adhesion sites 9 – 11 , 21 . Although CXCL1 and CXCL2 genes are also present in primates, including monkeys and humans, we hypothesized that IL-8 might be the most influential chemokine in adhesion formation. This is due to several factors: IL-8 can react to both CXCR1 and CXCR2 receptors with the highest affinity, whereas CXCL1 and CXCL2 can only react to CXCR2 20 . Second, it has been reported that neutrophil chemotaxis of IL-8 is predominantly mediated via CXCR1 22 . Therefore, IL-8, which strongly binds to both CXCR1 and CXCR2, may have more potent effects on neutrophil recruitment and subsequent adhesion formation compared to other family chemokines that interact only with CXCR2. Indeed, we previously reported that AMY109, a long-acting recycling antibody which inhibits both human and cynomolgus IL-8, reduces PA at incision sites in monkeys with surgically induced endometriosis, indicating IL-8 as a novel therapeutic target for PA 23 . However, as the pathophysiology of endometriosis may contribute to PA progression, it is necessary to elucidate the IL-8 role in PA progression in the absence of endometriosis. In this study, we aimed to elucidate how IL-8 contributes to PA progression. We first compared the chemotactic potency of IL-8 and CXCL1 toward human neutrophils in vitro to quantify their functional differences. Second, we then investigated whether IL-8 exerts additional effects beyond neutrophil chemotaxis by assessing its ability to modulate inflammatory and fibrotic responses in human mesothelial cells and neutrophils in vitro. Finally, we assessed the efficacy of prophylactic AMY109 treatment on PA formation in our monkey PA model to evaluate the role of IL-8 in vivo.

Supplementary Information. Supplementary Information.