Enzyme linked immunosorbent assay (ELISA) kits were obtained from Multi-Science (Hangzhou, Zhejiang, China). Primary antibodies were supplied by ABclonal (Wuhan, Hubei, China), while horseradish peroxidase (HRP)/fluorescein-conjugated secondary antibodies were procured from Boster (Wuhan, Hubei, China). Cell culture reagents were supplied by Thermo Fisher Scientific (Rockford, IL, USA). Lipopolysaccharide (LPS) from Escherichia coli 055 was purchased from Sigma-Aldrich (St. Louis, MO, USA). Recombinant human proteins IGFBP5 and IGF1 were sourced from Fine Biotech (Wuhan, Hubei, China). Other reagents were all obtained from Jiancheng Bioengineering Institute (Nanjing, Jiangsu, China). Ultrapure water was produced by using a Milli-Q system (Millipore, Bedford, MA, USA).

The dataset GSE146895 was downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/), which includes the normalized expression matrices of M0, M1 and M2 macrophages, with three samples in each cell subset. To identify the regulators involved in macrophage polarization, differentially expressed genes (DEGs) between M0 and M1/M2 macrophages were screened out (fold change > 2, p < 0.05). This analysis was performed based on R software (version 4.1.1) with a “Limma” package (https://cran.r-project.org/).

The Search Tool for Retrieving Interacting Genes (STRING) is an online tool for PPI network information assessment (http://www.string-db.org) (11). We used this tool to construct a PPI network based on the above DEGs. Only interaction with a combined score over 0.4 was illustrated in the diagram. The visualization of PPI network was achieved using Cytoscape software 3.10.1. A built-in function for molecular complex detection was used to identify modules of the PPI network. We set the parameters as follows: degree cutoff, 2; node score cutoff, 0.2; K-core, 2; max depth, 100.

THP-1 cells were used for functional validation, which grew in complete RPMI 1640 medium under standard conditions (37°C, humidified atmosphere with 5% CO₂). A portion of cells grown in 6-well plates were stimulated with LPS (500 ng/ml) or in the combination with IGFBP5 (50 ng/ml) for 6 h. The medium was used for ELISA test of IGFBP5, and the harvested cells were analyzed by western blot (WB) to assess expression of immune regulators and IGFBP5. Subsequently, IGFBP5 expression was silenced in normal THP-1 cells by siRNA, which was synthesized by Genepharma Co., Ltd. (Shanghai, China) with the sequences in Supplementary Table S1 . The cells were serum-starved for 12 h, before being cultured in the medium containing a transfection mixture (200 μl Opti-MEM + 5 μl Lipofectamine 3000) and siRNA solution (5 μl siRNA in 200 μl Opti-MEM). Four hours later, the supernatant was replaced with complete medium, and the cells were incubated for an additional 48 h to complete the transfection. The medium was collected for ELISA analysis of IL-1β and IL-6. In the flow cytometry (FCM) analyses, the cells were first incubated by PE-CD86 antibody, followed by fixation, permeabilization and APC-CD206 antibody staining procedures. Then, they were subjected to a flow cytometer (FC500, Beckman). CD86 and CD206 served as the marker for M1 and M2 cell subsets, respectively.

This service was provided by Biotree Tech (Shanghai, China). Proteins were extracted by lysis buffer with the aid of sonication at 4°C, and quantified by a kit. The samples were diluted by acetone, kept overnight at -20°C, and centrifuged at 13,000 rpm for 10 min. The precipitation was washed with 80% acetone, and resolved. Dithiothreitol was added to achieve a concentration of 5 mM. The solution was incubated at 55°C for 20 min. The product reacted with iodoacetamide (15 mM) in the dark for 30 min. Subsequently, the proteins were digested with trypsin at 37°C overnight. Afterwards, the reaction was ceased, and the mixture was centrifuged at 13,000 rpm for 10 min. The supernatant was taken as polypeptide samples. Desalting was performed by using C₁₈ mini tubes. The eluents were collected, and dried at 4°C in vacuum.

Total peptides (2 μg) in samples were separated by an EASY-nLC1200 UHPLC system, and detected by a Q Exactive HFX Orbitrap detector with an electrospray ion source (Thermo Fisher Scientific). Separation was achieved on a column (100 μm × 15 cm, Reprosil-Pur 120 C18AQ, 1.9 μm, Dr. Maisch). Aqueous solution 0.1% formic acid + 2% acetonitrile and 80% acetonitrile + 0.1% formic acid served as the phase A and B, respectively. A 120 min gradient elution program running at 300 nl/min was adopted: 25% (phase B) for 2 min, 5-22% for 88 min, 22-45% for 26 min, 45-95% for 2 min, 95% for 2 min. Data dependent acquisition was performed at the profile and positive mode. The resolution for MS1 and MS2 detection was set at 120,000 and 15,000, respectively. Top 20 abundant ions were fragmented with the normalized collision energy of 27%. The isolation window was set at 1.2 m/z. The peaks with charge(s) of only one or over 6 were excluded in the following analyses.

Data mining was achieved by Proteome Discoverer (Version 2.4.0.305) and the built-in Sequest HT search engine. Protein signals in the raw data were identified by searching a UniProt FASTA database (uniprot-Homo sapiens-9606-2022-11). In addition, proteins were quantified in samples based on the signal intensity of unique peptides and razor peptides. Differentially expressed proteins (DEPs) were visualized in a heatmap and used for clustering analysis. The DEPs were then mapped onto Gene Ontology (GO) terms as well as Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. Key nodes in the PPI network were highlighted in a quantification diagram.

For this immunofluorescence (IF) experiment, untreated and LPS-stimulated THP-1 cells were fixed in 4% formaldehyde for 30 min, permeabilized by 0.3% Triton X-100, and then incubated with antigen repair solution for 10 min. Subsequently, these cells were blocked with 5% bovine serum albumin solution in an incubator at 37°C for 1 h. After that, they were incubated overnight at 4°C with diluted anti-IGFBP5 (1:100) or IGF1 (1:200) antibody solutions, followed by treatments with fluorescein-conjugated secondary antibodies (1:500) at room temperature for 1 h. Finally, the cellular nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI). These cells were observed using a SP8 LIGHTNING confocal microscope (Leica).

Cells were lysed in NP-40 lysis buffer on ice. After centrifugation at 12,000 rpm for 5 min, the supernatant was collected. An antibody specific for the target protein was added into the cell lysate, and incubated overnight at 4°C under occasionally shaking. On the second day, pre-washed protein A/G magnetic beads were incubated with the sample for 1.5 h. These beads were retrieved in a magnetic field, and washed 5 times with NP-40 buffer. Proteins binding to the beads were then eluted, and boiled at 95°C in diluted loading buffer. In normal co-IP assays, samples obtained before and after the above steps were subjected to normal WB assays.

To detected IGFBP5-bound proteins, the eluate from the beads was separated by electrophoresis. The separation gels were stained with Coomassie Brilliant Blue. The bands exhibiting a distinct density difference between control and IGFBP5-IP sample were excised and cut into pieces (1 mm). The gel fragments were decolorized and kept in acetonitrile for 10 min. Thereafter, they were treated with dithiothreitol and iodoacetamide, followed by digestion, desalting and Liquid Chromatography Mass Spectrometry (LC-MS) detection, as those described in the proteomics study. Proteins exclusively detected in IGFBP5-IP sample were subjected to PPI analysis. The most abundant protein in center of the network was selected for further experiments.

We cultured normal THP-1 cells beforehand. The protein samples collected from their lysates before and after ANXA2/IGFBP5 co-IP were analyzed by WB. In subsequent experiments, IGFBP5 was silenced in some of the cells, or THP-1 cells were treated with LPS (500 ng/ml) combined with IGFBP5 (50 ng/ml) for 6 h. co-IP assays were performed once again to assess the binding status of ANXA2 and TLR4 under these differed conditions. Afterwards, the experiment involving LPS + IGFBP5 treatments was repeated. The medium and cells were collected for ELISA and WB experiments, respectively. Based on the results, the experiment was optimized. The concentrations of LPS and IGFBP5 were increased by 2-fold. The treatment duration was extended to 12 h. The replicate experiments used both normal and ANXA2-silenced THP-1 cells, and the medium was collected to measure IL-1β levels.

For WB assays, total cellular proteins were extracted using RIPA buffer. Protein samples were separated by electrophoresis, and transferred onto PVDF membranes. The membranes were then sequentially incubated with skim milk, primary antibodies and HRP-conjugated secondary antibodies. Protein bands were visualized by using an enhanced chemiluminescence substrate solution.

To clarify the role of IGFBP5 in inflammatory diseases, we measured its blood levels in adjuvant-induced arthritis (AIA) rats and RA patients. The samples were from our previous studies (12–14). Anticoagulated blood was centrifuged. The obtained plasma was used for ELISA assays. Red blood cells in the sediment were lysed. Protein and RNA in remaining white blood cells (WBCs) were extracted for WB and polymerase chain reaction (PCR) analyses, respectively. Additional blood from RA patients and healthy participants was processed to isolate monocytes by a commercial kit. IGFBP5 expression in these monocytes was analyzed by PCR method. Briefly, the samples were lysed in TRIzol reagent. Total RNA was extracted with chloroform, purified with isopropanol and ethanol, and used to synthesize cDNA. Quantitative PCR analysis was conducted on a Life Tech 7500 Real Time PCR detection system (Carlsbad). Those experiments involving human blood samples were approved by the Institutional Ethics Research Committee of Yijishan Hospital (No. 2022-20), and the written consent was obtained from all participants. Next, 3T3-L1 cells were induced to mature according to a reported method (9), and cultured with sera from the aforementioned rats and human volunteers for 12 h. The medium was then collected for ELISA analysis of IGFBP5. Some mature 3T3-L1 adipocytes were treated with IGFBP5 (100 ng/ml) for 12 h, and levels of TNF-α and IL-6 in the medium were detected using ELISA kits.

In vivo experiments were performed on male C57 mice (8 weeks old, obtained from Tianqin Biotechnology, Changsha, Hunan, China). Mice were housed in groups of four per cage in a specific pathogen-free laboratory, and provided with commercial rodent chow and boiled tap water. After one week of acclimation, 18 mice were randomly divided into 3 groups: healthy, ALI, and ALI + IGFBP5. LPS-induced ALI progresses rapidly, and will eventually cause mortality. Hence, experiment time window is short. IGFBP5-brought consequences were uncertain. It would amplify experiment risks. By taking these risks into consideration, we pre-treated mice with IGFBP5 (10 μg/kg) for 3 days. Then, ALI and ALI + IGFBP5 groups received an intraperitoneal injection of LPS (5 mg/kg). This experiment design highlights importance of IGFBP5-related environment difference during ALI progress. Twelve hours after LPS administration, all mice were euthanized via an overdose injection of pentobarbital sodium. Blood and organs were collected. Lung weight was recorded to assess edema severity. Levels of IGFBP5 in plasma and tissue homogenates along with IL-6, IL-1β, and MCP-1 were determined by ELISA kits. Anticoagulated blood was used for FCM and complete blood count (CBC) analyses. Monocytes were identified by FITC-CD11b antibodies, and those expressing high levels of Ly6c were classified as classical monocytes. CBC was performed on a PE-6800 VIT blood cell counter (Pukang Biotech). Lungs were sectioned, and stained by hematoxylin and eosin (H&E) for histological examination with standard procedures (13). The animal experiments were approved by the Ethical Committee of Wannan Medical College (No. LLSC-2022-223).

THP-1 cells were treated with IGFBP5 protein at various concentrations for 12 h, and cell viability was assessed using Cell Counting Kit-8. In the subsequent experiment, THP-1 cells were stimulated with LPS (1 μg/ml). A subset was co-treated by IGFBP5 within the nontoxic concentration range. IL-1β levels in the medium were measured by ELISA kits. This experiment was repeated, and IGF1 was used instead of IGFBP5. Next, we separately silenced IGFBP5 and IGF1R in THP-1 cells. The cells and their culture medium were collected for WB and ELISA analyses, respectively. Afterwards, some LPS-primed THP-1 cells were treated with IGFBP5 in the presence of IGF1 or not. The resulting medium was analyzed for IL-1β levels, while IGF1 localization was examined by IF method. In addition, some LPS-stimulated THP-1 cells were treated with IGFBP5, IGF1, picropodophyllin (PPP, a selective IGF1R inhibitor) or their combinations. Levels of IL-1β and ARG-1 in the medium were measured by using the appropriate ELISA kits.

Sample size in experiments was calculated by G*power (www.psychologie.hhu.de). Results of WB and IF experiments were quantified by Image J (1.52a, NIH, Bethesda, MD). Data were presented in the form of mean and standard deviation. Shapiro-Wilk test and Bartlett’s test were adopted to check normality and homogeneity of variance, respectively. Statistical difference among/between groups was evaluated by one-way analysis of variance followed by Tukey’s post hoc test or t-test by the aid of GraphPad Prism 8.0 (Cary, NC).

Macrophages at different polarization states display distinct gene expression profiles. The DEGs between M0 and M1/M2 subsets are shown in Figure 1A , and we focused on those showing the opposite change trends between M1 and M2 subsets by taking M0 cells as the reference. A PPI network was constructed using these DEGs ( Figure 1B ). It reveals IGFBP5 as a central node, a gene that is regulated by metabolic hormones. Its expression decreased in M1 subset and increased in M2 subset ( Figure 1C ). IGFBP5 showed strong correlations with key immune genes such as IL18R1, IL12A, CX3CL1, and CCL22. Different macrophage subsets were clearly distinguishable in scatter plots illustrated by IGFBP5 and these immune genes ( Figure 1D ).

LPS skewed THP-1 monocytes toward an inflammatory phenotype, indicated by up-regulation of IRF7 and concurrent down-regulation of IGFBP5 ( Figures 2A, B ). This change of IGFBP5 was confirmed by an ELISA test ( Figure 2C ). When monocytes were sufficiently activated by LPS, IGFBP5 treatment suppressed inflammation with the high efficiency. It suppressed expression of MMP3, iNOS and p-p65, while promoted expression of ARG-1 ( Figures 2D, E ). Conversely, IGFBP5 silencing led to increased secretion of IL-1β and IL-6 ( Figure 2F ). Meanwhile, the ratio of M1/M2 cell subset was increased ( Figures 2G, H ). In this experiment, IGFBP5 was silenced in normal untreated THP-1monocytes. The collective evidence demonstrates that IGFBP5 basically plays an anti-inflammatory effect in vivo regardless of cellular immune status difference.

Principal component analysis reveals distinct protein expression profiles between normal and IGFBP5-silenced THP-1 cells ( Figure 3A ). All the samples were correctly assigned into different groups in the clustering analysis ( Figure 3B ). Most DEPs were localized in the cytoplasm and nucleus, with a smaller portion belonging to secreted proteins ( Figure 3C ). It suggests that IGFBP5 potently influences intracellular signaling and cytokine secretion. GO enrichment exhibits that IGFBP5 is basically an immune regulator. IGFBP5 silencing affected many immune pathways in monocytes ( Figure 3D ). KEGG enrichment provides more insights into immune-regulatory effects of IGFBP5 ( Figure 3E ). Except ferroptosis, all those identified pathways are involved in immune regulation. NF-κB ranked as the 12^th most enriched pathway ( Supplementary Table S2 ). It is a common downstream target of many immune pathways. This finding underscores broader significance of IGFBP5 in the immune system. Expression of node proteins from the PPI network is shown in Figure 3F . They are all with inflammatory properties and up-regulated, when IGFBP5 was silenced. It confirms the anti-inflammatory role of IGFBP5 in monocytes.

IGFBP5 was localized on the surface of THP-1 monocytes ( Figure 4A ). Its distribution was decreased after LPS stimulus ( Figure 4B ). Hence, IGFBP5 may interact with certain receptors on cells. Then, a comparative analysis of IGFBP5-IP and control samples was performed based on LC-MS analysis, identifying 163 proteins exclusively bound to IGFBP5 ( Figure 4C ). The detailed protein information is provided in Supplementary Table S3 . Among the top 30, ANXA2 was the only membrane protein ( Figure 4D ). In addition, it was a central node in the PPI network constructed by these 163 proteins ( Figure 4E ). The results show the key role of ANXA2 in IGFBP5-cuased immune consequences.

Co-IP assays confirmed a strong interaction between IGFBP5 and ANXA2 ( Figures 5A, B ). ANXA2 potently activates TLR4 signaling (15). We assumed that intervention of IGFBP5 into their interaction would a cause for the observed immune changes. To test this hypothesis, we silenced IGFBP5, but observed that the interaction between ANXA2 and TLR4 was unaffected ( Figure 5C ). Being a typical agonist of TLR4, LPS reinforced its affinity to ANXA2. IGFBP5 exerted no impact on this outcome ( Figure 5D ). Hence, IGFBP5 does not influence immune status of monocyte/macrophage via modulation of ANXA2-mediated TLR4 activation. Despite this, IGFBP5 treatment effectively suppressed LPS-induced inflammatory polarization of THP-1 monocytes, evidenced by decreased IL-1β level and increased ARG-1 expression ( Figure 5E ). It was reported that ANXA2 can activate NF-κB directly (16). We investigated whether this mechanism is involved in immune-regulatory functions of IGFBP5. IGFBP5 reduced expression of p-ANXA2 and p-p65, leading to IRF7 down-regulation ( Figures 5F, G ). Of note, anti-inflammatory effects of IGFBP5 disappeared when ANXA2 was silenced, but this condition didn’t affect the outcome of LPS stimulus ( Figure 5H ). These findings confirm that ANXA2 is essential for IGFBP5-brought immune changes in monocytes, which are irrelevant to LPS/TLR4 signaling.

Based on these findings above, we hypothesized that IGFBP5 expression negatively correlates with inflammation, and its decrease might serve as a diagnostic marker for inflammatory diseases. Indeed, IGFBP5 expression in AIA rats’ WBCs was lower than healthy controls. This result was confirmed by both WB ( Figures 6A, B ) and PCR ( Figure 6C ) analyses. But unexpectedly, IGFBP5 levels in their blood were increased by 5-fold approximately ( Figure 6D ). IGFBP5 expression declined in RA patients’ WBCs too, at both protein ( Figures 6E, F ) and mRNA ( Figure 6G ) levels. Meanwhile, blood levels of IGFBP5 were notably increased ( Figure 6H ). PCR analyses using the in vitro cultured human monocytes confirm that IGFBP5 expression was impaired in RA conditions ( Figure 6I ). It suggests that increased IGFBP5 in the blood might originate from other tissues. White adipose tissue (WAT) is the largest secretion organ, and participates in both metabolism and immune regulation (17). It would be a major source of IGFBP5. Indeed, 3T3-L1 cells secreted IGFBP5 at large amounts. This secretion capacity was promoted by the sera of AIA rats and RA patients significantly ( Figure 6J ). But IGFBP5 did not alter immune status of adipocytes. Secretion of TNF-α and IL-6 in these cells was unchanged after IGFBP5 stimulus ( Figure 6K ). The above findings show that WAT contributes more to blood IGFBP5 pool than monocytes.

RA patients experience mild and chronic inflammation. The role of IGFBP5 in acute inflammation might be different. To thoroughly clarify the relevance of IGFBP5 to inflammation, we detected IGFBP5 levels in ALI mice, and studied their response to IGFBP5 treatment. Unlike the situation in RA, ALI mice exhibited a significant decrease of IGFBP5 in blood ( Figure 7A ). ELISA analyses of various tissues reveal that circulating IGFBP5 primarily originated from WAT in mice, which was accountable for ALI-caused blood IGFBP5 decline ( Figure 7B ). Although monocytes were not the main source of IGFBP5, they responded actively to IGFBP5. This stimulus further increased the proportion of classic subset in monocytes ( Figures 7C, D ). IGFBP5 did not affect CBC results ( Figure 7E ), but did enhance secretion of IL-6, IL-1β, and MCP-1 in ALI mice ( Figure 7F ). As a result, lung edema was aggravated ( Figure 7G ). Histological examination confirms the detrimental effects of IGFBP5 on ALI. In IGFBP5-treated ALI mice, pathological changes such as inflammatory cell infiltration, alveolar walls and cellular swelling were all worsened ( Figure 7H ).

The best known role of IGFBP5 is the regulator of IGF1. By binding to this hormone, IGFBP5 controls its accessibility to IGF1R, which has anti-inflammatory properties (10). This mechanism is possibly related to immune-regulatory functions of IGFBP5, and this theory was investigated afterwards. IGFBP5 did not affect viability of THP-1 cells significantly until reaching 1000 ng/ml ( Figure 8A ). Hence, we investigated effects of IGFBP5 on IL-1β secretion in LPS-treated THP-1 cells in a concentration range of 0.1 to 500 ng/ml. IGFBP5 exerted an anti-inflammatory effect ( Figure 8B ). The main difference of ALI mice from the conditions in vitro was the presence of IGF1. This difference would be a decisive variable for these varied outcomes. We next validated anti-inflammatory functions of IGF1/IGF1R. As anticipated, IGF1 stimulus inhibited LPS-induced IL-1β secretion in THP-1 cells in a concentration-dependent manner ( Figure 8C ). Silencing of IGFBP5 and IGF1R both increased IL-1β secretion, while decreased ARG-1 production in the monocytes ( Figure 8D ). In these conditions, p65 was highly phosphorylated ( Figures 8E, F ). Albeit both IGFBP5 and IGF1 negatively regulated inflammation, their co-existence was not harmonious. IGFBP5 binds to IGF1 with the higher affinity than IGF1R (10). Because of that, it prevented IGF1 activating IGF1R on cells ( Figures 8G, H ). Apparently, its interaction with IGF1 reduced free IGFBP5, which can interact with ANXA2 and inhibit NF-κB activation. From this sense, IGFBP5 supplement in medium led to a lose-lose situation for both of them. As such, IGFBP5 suppressed IL-1β secretion in LPS-stimulated THP-1 monocytes under normal conditions, but it exaggerated inflammation in the presence of excessive IGF1 ( Figure 8I ). In the next experiment, we found that IGF1 reduced IL-1β secretion in LPS-stimulated THP-1 cells by more than half. However, this effect was diminished by IGFBP5 and PPP (a selective IGF1R inhibitor). IGFBP5 did not reinforce the effect of PPP, indicating that they used the similar mechanism to antagonize the anti-inflammatory effect of IGF1. The change of ARG-1 mirrored the opposite pattern to that of IL-1β ( Figure 8J ). These findings demonstrate that IGFBP5 disrupts immune homeostasis in vivo by repressing anti-inflammatory functions of IGF1/IGF1R.