Tissue-resident macrophages exacerbate lung injury after remote sterile damage

Abstract

Remote organ injury, which is a common secondary complication of sterile tissue damage, is a major cause of poor prognosis and is difficult to manage. Here, we report the critical role of tissue-resident macrophages in lung injury after trauma or stroke through the inflammatory response. We found that depleting tissue-resident macrophages rather than disrupting the recruitment of monocyte-derived macrophages attenuated lung injury after trauma or stroke. Our findings revealed that the release of circulating alarmins from sites of distant sterile tissue damage triggered an inflammatory response in lung-resident macrophages by binding to receptor for advanced glycation end products (RAGE) on the membrane, which activated epidermal growth factor receptor (EGFR). Mechanistically, ligand-activated RAGE triggered EGFR activation through an interaction, leading to Rab5-mediated RAGE internalization and EGFR phosphorylation, which subsequently recruited and activated P38; this, in turn, promoted RAGE translation and trafficking to the plasma membrane to increase the cellular response to RAGE ligands, consequently exacerbating inflammation. Our study also showed that the loss of RAGE or EGFR expression by adoptive transfer of macrophages, blocking the function of RAGE with a neutralizing antibody, or pharmacological inhibition of EGFR activation in macrophages could protect against trauma- or stroke-induced remote lung injury. Therefore, our study revealed that targeting the RAGE-EGFR signaling pathway in tissue-resident macrophages is a potential therapeutic approach for treating secondary complications of sterile damage.

Introduction

Sterile tissue damage caused by sterile inflammation, which occurs in the absence of invading pathogens, is involved in secondary complications of various severe illnesses, including ischemia-reperfusion injury and traumatic injury [1, 2]. Sterile tissue inflammation can lead to local tissue injury and remote organ injury [3, 4]. Tissue damage and cell death lead to the release of endogenous alarmins, such as damage-associated molecular patterns (DAMPs), which in turn activate innate immune cells and subsequently trigger an increase in cytokines and leukocyte trafficking, resulting in local sterile inflammation and tissue damage [5, 6]. Sterile tissue inflammation is strongly associated with the subsequent development of organ dysfunction distant from the site of tissue injury, especially lung damage, such as trauma-induced lung injury and stroke-induced lung injury [7–9]. Lung injury is one of the most common complications after stroke or severe trauma, has an incidence of approximately 12% or 5%, respectively, and increases the risk of mortality and disability in stroke or trauma patients and their length of hospital stay [9–11]. Although advances in the management of local tissue (such as the control of hemorrhage or reperfusion therapy) improved the survival of trauma or stroke patients, current therapies do not improve the systemic complications or prognoses of patients, and increasing attention has been given to sterile tissue inflammation and remote injury as therapeutic targets [12, 13]. However, the mechanisms linking sterile inflammation to remote injury have not been fully elucidated.

Tissue sterile injury also induces global inflammation, which depends on the immune system and can be characterized by the systemic circulation of DAMPs, inflammatory cytokines, and activated innate immune cells [1, 4, 6]. The systemic immune system can modulate tissue-resident immune cells in remote organs, resulting in an excessive immune response and subsequent organ dysfunction in locations distant from the initial site of sterile tissue injury [14]. Lung tissue-resident macrophages, including alveolar macrophages and interstitial macrophages, are the most abundant immune cells in various organ systems, where they perform critical functions in tissue homeostasis and immune surveillance [15–17]. However, the function of tissue-resident macrophages in modulating the organ response to remote damage and the mechanisms are less well understood. Our previous study and others have shown that macrophages sense circulating DAMPs, such as HMGB1, DNA, and S100 proteins, which are released from necrotic tissue, by interacting with pattern recognition receptors, such as Toll-like receptors and receptor for advanced glycation end products (RAGE) on the cell surface, thereby triggering an excessive inflammatory response [7, 18–20]. As a cell surface receptor, RAGE is a key receptor that activates innate immunity and induces the release of cytokines and depends on RAGE membrane trafficking [19, 21]. Ligand–receptor interactions can lead to receptor membrane trafficking and mediate cellular responses, including inflammatory responses. Our previous study revealed that the surface expression of RAGE on macrophages was associated with the inflammatory response [19].

In light of these findings, we hypothesized that local tissue damage caused remote organ injury via RAGE signaling in tissue-resident macrophages. In this study, we found that the RAGE- and epidermal growth factor receptor (EGFR)-driven inflammatory response in lung tissue-resident macrophages played a critical role in remote lung injury after sterile tissue inflammation. Furthermore, we found that RAGE crosstalk activated EGFR-P38 signaling through Rab5-mediated RAGE internalization to regulate RAGE intracellular trafficking and amplify inflammation. Overall, this article may guide the future development of therapeutic strategies for treating remote organ injury after sterile tissue damage.

Results

Lung-resident macrophages play an important role in remote lung injury after sterile tissue damage

To determine the role of tissue-resident macrophages in remote injury after sterile inflammation, mice were subjected to trauma (pseudofracture) in a validated model of long-bone fracture and tissue trauma [22–24] or stroke, which serves as a model for brain ischemia-reperfusion injury. These models effectively replicate local sterile inflammation and subsequent damage to remote organs, such as the lung [8, 23, 24]. We first investigated macrophage abundance after trauma and stroke in remote organs that are frequently affected by trauma and stroke, such as the lung. Interestingly, while the number of lung-resident macrophages remained unchanged after trauma or stroke, there was a significant increase in neutrophil recruitment to the lung (Fig. 1A, B). Additionally, the expression of proinflammatory cytokine genes (Tnfa, Il1b, and Cxcl1) in lung-resident macrophages significantly increased after trauma or stroke (Fig. 1C, D), and lung injury was exacerbated, which was characterized by increased neutrophil migration (Cxcr2 and Icam1) and the release of injurious neutrophil products, including proinflammatory cytokines (Tnfa and Il6) and proteases (Mmp8 and Mmp9) (Supplementary Fig. 1b). However, the expression of genes related to apoptosis (Bax, Bim, Bcl) and cell proliferation (c-Myc, Cdk4, Cdc25) in lung-resident macrophages did not significantly differ after trauma or stroke (Fig. 1C, D), indicating that the inflammatory response mediated by lung-resident macrophages was strongly correlated with lung injury following trauma or stroke. To further evaluate the role of lung tissue-resident macrophages in remote organ injury after trauma or stroke, we depleted lung tissue-resident macrophages using clodronate liposomes. Because lung tissue-resident macrophages include alveolar macrophages and interstitial macrophages [17], the mice were treated with clodronate liposomes intratracheally (i.t.) and intravenously (i.v.) to deplete lung alveolar macrophages and interstitial macrophages, respectively. Compared to PBS liposome treatment, clodronate liposome treatment depleted approximately 80% of tissue-resident macrophages in the lung (Supplementary Fig. 1c), which did not affect the abundance of neutrophils, the inflammatory state or lung homeostasis (the histological aspects of alveolar ducts, the alveolar epithelium, and blood vessels were normal in macrophage-depleted mice, and these changes were accompanied by a decrease in macrophages) (Supplementary Fig. 1c–f). However, depleting tissue-resident macrophages attenuated lung injury after trauma or stroke, as demonstrated by pathology (a significant reduction in trauma or stroke-induced inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Fig. 1E, F), lung inflammation (reduced expression of inflammatory cytokine genes and decreased neutrophil recruitment) (Fig. 1G–J) and pulmonary permeability (Fig. 1K, L). Furthermore, we investigated the contributions of alveolar macrophages and interstitial macrophages to trauma-induced lung injury. Depleting alveolar macrophages by intratracheal clodronate liposome administration did not affect the abundance of interstitial macrophages (Supplementary Fig. 2a), reduced lung inflammation (Supplementary Fig. 2b) and attenuated lung injury (pathological changes such as reductions in inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Supplementary Fig. 2c) after trauma. Similar lung protective effects were observed in trauma after interstitial macrophage depletion (Supplementary Fig. 2d–f), although these effects were relatively weaker than those observed following alveolar macrophage depletion. These results indicate that tissue-resident macrophages, including alveolar macrophages and interstitial macrophages, play a critical role in trauma- and stroke-induced lung injury by facilitating the inflammatory response.

Fig. 1.

Tissue-resident macrophages contribute to lung injury after trauma or stroke via the inflammatory response. A–D Mice were subjected to trauma (pseudofracture) or stroke. A, B Flow cytometric analysis of lung neutrophil (gated on 7AAD⁻CD45⁺) and macrophage (gated on 7AAD^-CD45⁺Ly6G⁻) numbers (n = 6). C, D The mRNA levels of genes associated with inflammatory responses (Tnfa, Il1b, and Cxcl1), genes associated with apoptosis (Bax, Bim, and Bcl), and genes associated with cell proliferation (c-Myc, Cdk4, and Cdc25) in sorted lung-resident macrophages were examined by qPCR (n = 4). E–L Mice were treated with PBS liposomes or clodronate liposomes (i.t. and i.v.), followed by trauma or stroke for 12 h (n = 4). E, F Representative H&E staining of lung sections (scale: 200 μm). G, H The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in lung tissue. I, J Flow cytometric analysis of the number of lung neutrophils. K, L The wet/dry weight ratio of the lung was determined. The data shown represent four or six independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant

Because monocyte-derived macrophage recruitment has also been reported to occur during organ injury via the inflammatory response, we investigated whether monocyte-derived macrophages contributed to lung injury associated with sterile damage. The expression of chemokine (CC-motif) ligand 2 (Ccl2), which recruits monocyte-derived macrophages to organs, did not significantly differ after trauma or stroke (Fig. 1C, D), and consistently, the abundance of macrophages in the lung remained unchanged after trauma and stroke (Fig. 1A, B). To further determine whether the increase in organ inflammation was due to tissue-resident macrophages but not monocyte-derived macrophage recruitment, we used Ccr2-knockout mice (Ccr2^−/−), which inhibited monocyte recruitment to organs [25]. Compared to those in WT mice, the number and inflammatory response of lung-resident macrophages in Ccr2^−/− mice did not change after trauma (Supplementary Fig. 3a, b), and there were no significant organ inflammatory responses or injury (pathological alterations, including increased inflammatory cell infiltration, thickened alveolar septae, alveolar and interstitial edema, and hemorrhage) (Supplementary Fig. 3a, c, d), indicating that tissue-resident macrophages but not monocyte-derived macrophages contributed to remote organ injury via the inflammatory response underlying sterile damage. Taken together, these data indicate that resident macrophages, which are distant from the site of sterile tissue damage, sense and respond to remote events, leading to remote organ injury via the inflammatory response.

The release of alarmins after sterile tissue damage induces remote organ injury via resident macrophage-specific RAGE signaling

We next investigated how tissue-resident macrophages sensed remote sterile tissue damage. Acute trauma or stroke leads to the massive release of alarmins, such as HMGB1 [7, 26, 27], AGE [20, 28], or S100 proteins [26, 29], which contribute to inflammation at the primary site via RAGE signaling and are subsequently released from the site of tissue damage into the circulation [2, 7]. However, whether the role of tissue-resident macrophages in remote organ injury is dependent on the response to circulating alarmins has not been determined. We first confirmed increased plasma concentrations of HMGB1, AGE and S100A9, which are RAGE ligands, after trauma (Fig. 2A). To determine the response of tissue-resident macrophages to circulating alarmins, we reduced systemic alarmins in vivo using soluble RAGE (sRAGE) as a decoy receptor and found that sRAGE inhibited the inflammatory response in lung-resident macrophages (Fig. 2B), which was accompanied by reduced lung inflammation (Fig. 2C) and attenuated lung injury (pathological changes included reductions in multiple indices of inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Fig. 2D) after trauma. Furthermore, we inhibited HMGB1, AGE, and S100A9 in vivo using an HMGB1 neutralizing antibody, LR-90, and paquinimod, respectively. Consistently, we found that the HMGB1 neutralizing antibody, LR-90, and paquinimod treatment had similar protective effects on the lungs to differing degrees (Supplementary Fig. 4). To confirm the effect of DAMPs on macrophages, we treated bone marrow-derived macrophages (BMDMs) from WT or Rage^−/− mice with RAGE ligands (HMGB1, AGE, or S100A9) or plasma from mice exposed to sham surgery or trauma. We found that the mRNA levels of inflammatory response genes and the secretion of inflammatory cytokines by WT BMDMs were increased by HMGB1, AGE, S100A9, as well as the plasma of mice subjected to trauma, however, these increases were not observed in Rage^−/− BMDMs or cells treated with the HMGB1 neutralizing antibody (Fig. 2E, Supplementary Fig. 5, Supplementary Fig. 6). We ensured that cytokine secretion was not due to endotoxin contamination of AGE, HMGB1, or S100A9 (Supplementary Fig. 5d). These results demonstrated that tissue-resident macrophages could sense circulating alarmins, such as HMGB1, AGE, and S100A, which were released from sites of remote primary sterile tissue damage and triggered an inflammatory response.

Fig. 2.

Alarmin release after sterile tissue damage induces an inflammatory response in resident macrophages via the RAGE signaling pathway. A Mice were subjected to trauma (pseudofracture). The plasma levels of HMGB1, AGE, and S100A9 were measured (n = 6). B–D WT mice were intraperitoneally injected with soluble RAGE (sRAGE, 4 mg/kg) or solvent 30 min before and 4 h after surgery, and 12 h after trauma, the mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in sorted lung-resident macrophages (B) and lung tissue (C) were examined (n = 3). D Representative H&E staining of lung sections (scale: 200 μm). E WT or Rage^−/− BMDMs were treated with AGE (100 μM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for 4 h, and the mRNA expression of inflammatory response genes (Tnfa, Il1b, and Cxcl1) was measured by quantitative PCR. F–H APC-labeled lung alveolar macrophages (AMs) and BMDMs from Rage^−/− mice (donor mice) were transferred to WT mice (recipient mice) by intratracheal (i.t., AMs) and intravenous (i.v., BMDMs) administration, and after 7 days, the recipient mice were subjected to trauma. G Donor macrophages were detected in lung tissue on Day 7. H The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in sorted donor macrophages (APC⁺) and nondonor macrophages (APC⁻) were examined (n = 4). I–M AMs (i.v.) and BMDMs (i.t.) were transferred from WT or Rage^−/− mice (donor mice) to WT mice (recipient mice), which were treated intratracheally (i.t.) or intravenously (i.v.) with clodronate to deplete endogenous macrophages. After 7 days, the recipient mice were subjected to trauma (n = 4). J Representative H&E staining of lung sections (scale: 200 μm). K The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in lung tissue were examined (n = 4). L The wet/dry weight ratio of the lung was determined (n = 4). M The plasma levels of HMGB1, AGE, and S100A9 were measured (n = 4). The data shown represent three, four or six independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant

To explore the role of RAGE signaling in remote organ injury after sterile tissue damage, we used WT and Rage^−/− mice. After trauma surgery, the plasma levels of HMGB1, AGE, S100A9 (Supplementary Fig. 7a) and the abundance of lung-resident macrophages (Supplementary Fig. 7b) in Rage^−/− mice and WT mice did not differ significantly, although Rage^−/− mice exhibited decreased inflammatory responses in lung resident macrophages (Supplementary Fig. 7c), which was accompanied by reduced lung inflammation (Supplementary Fig. 7b, d) and the attenuation of lung injury (Supplementary Fig. 7e), indicating that RAGE signaling in resident macrophages plays an essential role in trauma-induced remote injury.

To explore the role of RAGE signaling in tissue-resident macrophages in remote organ injury, we performed a macrophage adoptive transfer experiment before trauma was induced. Lung alveolar macrophages (AMs) and BMDMs labeled with APC were isolated from Rage^−/− mice (donors) and adoptively transferred into WT mice (recipients) (Fig. 2F). Donor-derived APC⁺ AMs and BMDMs were present in the lung, where they represented approximately 25% of the total lung-resident macrophages in recipient mice (Fig. 2G). Compared with WT lung-resident macrophages (APC⁻macrophages), the injected Rage^−/− lung-resident macrophages (APC⁺ macrophages) had reduced mRNA levels of inflammatory response genes after trauma (Fig. 2H). Next, we adoptively transferred WT or Rage^−/− macrophages (donors) into WT mice (recipients), after which the recipient mice were subjected to trauma (Supplementary Fig. 8a). The recipients of WT or Rage^−/− macrophages did not significantly differ in terms of lung homeostasis (histology of alveolar ducts, the alveolar epithelium, and blood vessels appeared normal and was not accompanied by a decrease in macrophages) (Supplementary Fig. 8b). However, compared with the recipients of WT macrophages, lung inflammation in the recipients of Rage^−/− macrophages was reduced (Supplementary Fig. 8c), and lung injury was attenuated (pathological change in the reduction in inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Supplementary Fig. 8b, d) after trauma, although there were no significant differences in HMGB1, AGE, or S100A9 concentrations (Supplementary Fig. 8e). A similar effect was observed in macrophages that were adoptively transferred after endogenous macrophage depletion (Fig. 2I–M). These results revealed that the adoptive transfer of Rage^−/− macrophages protected against remote organ injury after sterile tissue damage. Collectively, these results demonstrate that tissue-resident macrophages sense circulating alarmins, which are released from the site of remote primary tissue damage, and initiate an inflammatory response via RAGE signaling, leading to distal organ inflammation and injury after sterile tissue damage.

RAGE signaling activation in tissue-resident macrophages depends on RAGE membrane trafficking

Next, we investigated the mechanism of RAGE signaling activation in tissue-resident macrophages after sterile tissue damage. The activation of RAGE signaling following ligand recognition by RAGE induces RAGE transcription, resulting in an increase in RAGE expression via positive feedback [18, 30]. After sorting lung-resident macrophages from mice subjected to sham surgery or trauma, we observed no significant differences in RAGE protein or mRNA levels (Fig. 3A, B). However, after trauma, we found a significant increase in the surface expression of RAGE on lung-resident macrophages but no change in lung neutrophils (Fig. 3C, Supplementary Fig. 9a). The amount of receptor on the cell surface determines the sensitivity of the receptor to the ligands to some extent. To confirm the role of the surface expression of RAGE on lung-resident macrophages in remote organ injury after sterile tissue damage, we used a RAGE neutralizing antibody, which inhibited RAGE membrane function. The RAGE neutralizing antibody significantly inhibited trauma-induced the surface expression of RAGE (Fig. 3D, Supplementary Fig. 9b) and decreased the inflammatory response in lung-resident macrophages (Fig. 3E), which was accompanied by a decrease in lung tissue inflammation (Fig. 3F) and the attenuation of lung injury (pathological change in the reduction in inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Fig. 3G), although the concentrations of HMGB1, AGE, and S100A9 did not significantly differ (Supplementary Fig. 10). Additionally, treatment with FPS-ZM1, a RAGE antagonist that inhibits RAGE-ligand binding, reduced the expression of inflammatory response genes in lung-resident macrophages and the lung and improved pulmonary permeability after trauma (Supplementary Fig. 11). To confirm the role of RAGE on the surface of macrophages after trauma, we treated WT BMDMs with plasma from mice subjected to sham surgery or trauma after RAGE neutralizing antibody pretreatment. We found that the RAGE neutralizing antibody decreased the expression of Tnfa, Il1b, and Cxcl1 mRNA after BMDMs were stimulated with traumatic plasma (Fig. 3H). In addition, the secretion of TNF-α and CXCL1 in murine macrophages (BMDMs) and human macrophages (THP1 cells) was significantly reduced by pretreatment with the RAGE neutralizing antibody following stimulation with AGE, HMGB1, or S100A9 (Fig. 3I, Supplementary Fig. 9c, d). These results indicate that RAGE signaling activation in resident macrophages is dependent on the surface expression of RAGE.

Fig. 3.

Posttraumatic remote organ injury depends on the surface expression of RAGE on resident macrophages. A–C Mice were subjected to trauma (n = 6). The protein (A) or mRNA (B) levels of RAGE in FACS-sorted lung-resident macrophages were measured. C RAGE surface expression on lung-resident macrophages was analyzed. D–G WT mice were injected with 10 mg/kg mouse IgG isotype control (IgG, by the vein) or 10 mg/kg RAGE neutralizing antibody (RAGE Ab, by the vein), followed by sham surgery or trauma for 12 h (n = 4). D RAGE surface expression on lung-resident macrophages was analyzed. The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in sorted lung-resident macrophages (E) and lung tissue (F) were examined. G Representative H&E staining of lung sections (scale: 200 μm). H WT mice were subjected to sham surgery or trauma. After 4 h, plasma was collected and used as conditioned media to culture WT BMDMs. The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) were measured after the cells were incubated with sham or trauma plasma or treated with an IgG isotype control (IgG, 10 µg) or RAGE neutralizing antibody (RAGE Ab, 10 µg). P/S. plasma from sham mice, P/T. plasma from trauma mice. I BMDMs were treated with AGE (100 μM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for 4 h with or without pretreatment with the RAGE neutralizing antibody (10 μg). The concentrations of TNF-α and CXCL1 were measured by ELISA (n = 4). The data shown represent four or six independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant, nd not detected

The interaction between ligands and receptors induces receptor membrane trafficking, leading to intracellular responses such as inflammation [31]. Therefore, we investigated whether RAGE signaling activation promoted RAGE membrane trafficking in macrophages. We observed a reduction in the cell surface expression of RAGE at 0.5 h (Fig. 4A; Supplementary Fig. 12a–c), but there was no effect on the protein or mRNA levels of RAGE after incubation with RAGE ligands (AGE or HMGB1) for 0.5 h (Fig. 4B, C; Supplementary Fig. 12d); these ligands can be released after trauma or stroke and can act as alarmins to active RAGE signaling in macrophages (Fig. 2). Consistent with the effect of AGE stimulation, plasma from mice subjected to trauma reduced the surface expression of RAGE (Fig. 4D) but did not change the mRNA or protein levels of RAGE at 0.5 h (Fig. 4E, F); this change was not observed in the plasma from sham mice (Fig. 4E, F), indicating that RAGE was internalized in the early phase. To further confirm the internalization of RAGE, we pretreated BMDMs or THP1 cells with dynasore, an inhibitor of active internalization processes, which prevented RAGE internalization (Fig. 4G, Supplementary Fig. 13a, b) but did not affect the total protein or mRNA expression levels after AGE stimulation (0.5 h) (Fig. 4H, I). A similar phenomenon was observed after silencing Rab5, which is localized to early endosomes and plays a role in early receptor internalization [32, 33], 0.5 h after AGE stimulation (Supplementary Fig. 13c, d). In addition, the surface expression of RAGE began to increase at 3 h (in BMDMs) or 6 h (in THP1 cells), and this change was accompanied by an increase in total RAGE protein and mRNA expression at 24 h (Fig. 4A–F; Supplementary Fig. 12), suggesting that RAGE was trafficked to the plasma membrane and that RAGE was translated during the intermediate and late phases. We then examined the effect of RAGE internalization in the early phase on the trafficking and translation of RAGE in the intermediate and late phases. We found that dynasore pretreatment or Rab5 silencing significantly decreased the surface expression of RAGE at 6 h and 24 h (Fig. 4J; Supplementary Fig. 13d, e) and inhibited RAGE mRNA and protein expression at 24 h (Fig. 4I, K) after AGE stimulation, indicating that the internalization of RAGE in the early phase is a critical step for RAGE trafficking and translation in the intermediate and late phases. These data indicate that once RAGE signaling is activated by binding to its ligands, RAGE is internalized into the cytoplasm in the early phase, leading to the translation and subsequent trafficking of RAGE to the plasma membrane.

Fig. 4.

RAGE signaling triggers inflammation by promoting RAGE membrane trafficking. A–F WT BMDMs were exposed to AGE (100 μM) (A–C) or plasma from mice subjected to sham surgery or trauma (D–F) for different times. A, D The surface expression of RAGE was analyzed by FACS at the indicated times. The protein (B, E) or mRNA (C, F) levels of RAGE were measured at the indicated times. G–L BMDMs were pretreated with dynasore (20 μM) for 1 h and then exposed to AGE (100 μM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for different times. The surface expression of RAGE was analyzed by FACS at 0.5 h (G), 6 h and 24 h (J). The protein (H, K) and mRNA (I) levels of RAGE were measured at the indicated times. L The concentrations of TNF-α and CXCL1 were measured by ELISA at 4 h. M BMDMs were treated with AGE (100 μM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for 3 h and then treated with or without a RAGE neutralizing antibody (RAGE Ab). The secretion of TNF-α and CXCL1 at 3 h was measured by ELISA. The data shown represent three (A–K) or four (L, M) independent experiments. *P < 0.01, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant, nd not detected

To investigate the role of RAGE membrane trafficking in the inflammatory response, we measured the secretion of inflammatory cytokines and chemokines after inhibiting RAGE internalization by dynasore pretreatment. We found that both dynasore pretreatment and Rab5 silencing significantly decreased the secretion of TNF-α and CXCL1 after AGE, HMGB1, and S100A9 stimulation (Fig. 4L; Supplementary Fig. 13f, g). To further confirm the impact of the increase in the surface expression of RAGE on RAGE-induced inflammation, we treated the cells with a RAGE neutralizing antibody after the surface expression of RAGE was increased by activating RAGE signaling and evaluated the secretion of inflammatory cytokines. RAGE neutralizing antibody treatment reduced the secretion of TNF-α or CXCL1 after AGE, HMGB1, and S100A9 stimulation in BMDMs and THP1 cells (Fig. 4M, Supplementary Fig. 13h), indicating the pivotal role of the increase in RAGE expression in RAGE-induced inflammation. Taken together, these results suggest that activated RAGE signaling induces RAGE internalization in the early phase, leading to subsequent RAGE translation and trafficking, which facilitate inflammation, resulting in organ injury.

RAGE-EGFR signaling through RAGE membrane trafficking mediates inflammation in macrophages

We next explored RAGE membrane trafficking during RAGE-induced inflammation in macrophages. Our previous studies demonstrated that epidermal growth factor receptor (EGFR), which is activated by pattern recognition receptors such as TLR4, promotes receptor internalization and triggers multiple signal transduction processes, including inflammation, thereby contributing to tissue inflammation and infectious inflammatory diseases [34, 35]. However, it remains unclear whether RAGE, which is also a pattern recognition receptor, regulates receptor membrane trafficking by activating EGFR. Thus, we analyzed EGFR activation in RAGE-negative and RAGE-positive lung-resident macrophages after trauma. We found that RAGE-positive lung-resident macrophages had higher levels of EGFR activation than RAGE-negative lung-resident macrophages (Fig. 5A), suggesting a correlation between EGFR activation and RAGE signaling. Moreover, we found that the phosphorylation of EGFR was increased in sorted lung-resident macrophages after trauma, whereas EGFR phosphorylation was inhibited in Rage^−/− lung-resident macrophages after trauma (Fig. 5B). In parallel, we performed adoptive transfer of APC-labeled lung AMs and BMDMs from Rage^−/− mice into WT mice. We observed that the phosphorylation of EGFR was increased in WT lung-resident macrophages (APC^- macrophages) after trauma, and this increase was inhibited in injected Rage^−/− lung-resident macrophages (APC⁺ macrophages) (Fig. 5C), further indicating the strong relationship between EGFR activation and RAGE signaling in lung-resident macrophages after trauma.

Fig. 5.

EGFR is required for RAGE-induced sterile inflammation and the surface expression of RAGE. A WT mice were subjected to trauma, and the frequency of lung-resident macrophages harboring p-EGFR within the RAGE-positive or -negative population was determined by FACS. B WT or Rage^−/− mice were subjected to trauma, after which the protein levels of phosphorylated and total EGFR (p-EGFR and t-EGFR) in FACS-sorted lung-resident macrophages were measured. C APC-labeled sorted lung AMs and BMDMs were transferred to WT mice, which were then subjected to trauma after 7 days. The protein levels of phosphorylated and total EGFR (p-EGFR and t-EGFR) were measured in sorted APC⁺ macrophages and APC⁻ macrophages. D WT or Rage^−/− BMDMs were stimulated with AGE (100 μM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for 0.5 h. Phosphorylated and total EGFR (p-EGFR and t-EGFR) were examined. E WT BMDMs were stimulated with AGE (100 μM) for the indicated times. The interaction between RAGE and EGFR was examined by immunoprecipitation. WT BMDMs and Egfr^−/− BMDMs (F, H, I, K, L, N) or erlotinib (20 μM)- or PD168393 (10 μM)-pretreated (G, J, M) BMDMs were stimulated with AGE (100 μM) for the indicated times. The surface expression of RAGE was analyzed by FACS (F, G, K, L, M). The protein level of RAGE was measured at 0.5 h (H) or 24 h (N). The mRNA level of RAGE was measured (I, J). O, P WT BMDMs, Egfr^−/− BMDMs, erlotinib- (20 μM) or PD168393-pretreated (10 μM) BMDMs were stimulated with AGE (100 µM), HMGB1 (4 µg/ml), or S100A9 (500 ng/ml) for 4 h. The concentrations of TNF-α and CXCL1 were measured by ELISA (O), and the mRNA expression of inflammatory response genes (Tnfa, Il1b, and Cxcl1) was examined by quantitative PCR (P). Q WT or Egfr^−/− BMDMs were stimulated with plasma from mice subjected to trauma, after which the mRNA expression of inflammatory response genes (Tnfa, Il1b, and Cxcl1) was examined. The data shown represent three (A–N) or four (O–Q) independent experiments. **P < 0.01, ***P < 0.001, ****P < 0.0001, ns not significant, nd not detected

To determine the relationship between EGFR activation and RAGE signaling in macrophages, we analyzed EGFR activation in WT and Rage^−/− BMDMs after RAGE signaling activation. We found that EGFR was activated by AGE, HMGB1, or S100A9 stimulation in WT BMDMs but was significantly reduced in Rage^−/− BMDMs (Fig. 5D). A similar phenomenon was observed after AGE stimulation in cells treated with a RAGE neutralizing antibody (Supplementary Fig. 14a). In addition, we found that EGFR phosphorylation was increased after AGE or HMGB1 stimulation in THP1 cells and that this change could be significantly inhibited by RAGE neutralization (Supplementary Fig. 14b). Furthermore, AGE induced an interaction between RAGE and EGFR, which was initiated at 10 min, peaked at 20 min, and gradually diminished thereafter (Fig. 5E). These data indicate that ligand-activated RAGE directly interacts with EGFR, leading to EGFR activation.

We further explored whether EGFR activation influenced RAGE membrane trafficking in macrophages. We found that EGFR deletion or pharmacologic inhibition (PD168393 or erlotinib) increased the surface expression of RAGE on BMDMs or THP-1 cells 0.5 h after AGE stimulation (Fig. 5F, G; Supplementary Fig. 14c) but did not change the mRNA or protein levels of RAGE 0.5 h after AGE stimulation (Fig. 5H–J), suggesting that EGFR is required for RAGE internalization into the cytoplasm but does not affect translation in the early phase. However, inhibiting RAGE internalization by Rab5 silencing did not change EGFR activation after AGE stimulation (Supplementary Fig. 14d), indicating that EGFR serves as an upstream activator of RAGE internalization. Furthermore, we found that EGFR deletion or pharmacologic inhibition decreased the surface expression of RAGE on BMDMs or THP1 cells 6 h and 24 h after AGE stimulation (Fig. 5K–M, Supplementary Fig. 14e), and this change was accompanied by reduced RAGE mRNA and protein levels 24 h after AGE stimulation (Fig. 5I, J, N), suggesting that EGFR is required for RAGE translation and trafficking to the membrane in the intermediate and late phases.

To evaluate the role of EGFR in RAGE-induced inflammation, we measured the production of inflammatory cytokines in BMDMs with EGFR deletion or that were pretreated with EGFR pharmacologic inhibitors and subjected to RAGE signaling activation. We found that the secretion of TNF-α and CXCL1 in response to AGE, HMGB1, or S100A9 stimulation was significantly attenuated by EGFR deletion or pharmacologic inhibition in BMDMs or THP1 cells (Fig. 5O, Supplementary Fig. 14f, g), which was accompanied by reduced expression of inflammatory response genes (Tnfa, Il1b and Cxcl1) (Fig. 5P, Supplementary Fig. 14h). A similar phenomenon was observed in EGFR-deleted BMDMs in response to traumatic plasma stimulation (Fig. 5Q), indicating that EGFR activation is required for RAGE-induced inflammation. Taken together, these results show that RAGE signaling activates EGFR t direct interaction, which in turn mediates RAGE membrane trafficking and translation, including internalization to the cytoplasm in the early phase and trafficking to the plasma membrane in the intermediate and late phases, thereby facilitating inflammation.

RAGE crosstalk activates EGFR to recruit P38 and activate P38 signaling to regulate RAGE membrane trafficking and inflammation

To understand the mechanism of RAGE-EGFR signaling-induced inflammation, we explored the downstream signaling pathway through which EGFR is activated by RAGE in macrophages. Because EGFR activation can trigger P38 phosphorylation [34], we examined the activation of P38 in EGFR-deleted BMDMs or cells that were pretreated with EGFR pharmacologic inhibitors (PD168393 or erlotinib) and subjected to AGE stimulation. We found that EGFR deletion or inhibition by PD or erlotinib significantly suppressed AGE-induced P38 phosphorylation (Fig. 6A, B), suggesting that EGFR mediated RAGE-induced P38 phosphorylation. In addition, we observed an AGE-induced interaction between EGFR and P38, which was inhibited by RAGE neutralizing antibodies or RAGE deletion (Fig. 6C). Consistent with these findings, phosphorylated EGFR colocalized with phosphorylated P38 and RAGE 20 min after AGE stimulation (Fig. 6D). Furthermore, Rab5 silencing significantly inhibited AGE-induced P38 phosphorylation (Fig. 6E). These data suggest that Rab5-mediated RAGE internalization through the phosphorylation of EGFR, which colocalizes with P38 and activates P38 after RAGE signaling activation.

Fig. 6.

EGFR-P38 signaling is critical for RAGE-induced sterile inflammation and RAGE membrane trafficking. A WT or Egfr^−/− BMDMs were treated with AGE (100 μM) for 0.5 h. The protein levels of phosphorylated and total P38 (t-P38 and p-P38, respectively) were measured by western blotting. B WT BMDMs were pretreated with or without erlotinib (20 μM) or PD168393 (10 μM) for 1 h, followed by AGE (100 μM) stimulation for 0.5 h. The protein levels of phosphorylated and total P38 (t-P38, p-P38) were measured by western blotting. C WT BMDMs, Rage^−/− BMDMs, or RAGE neutralization antibody-pretreated BMDMs were exposed to AGE (100 μM) for 20 min. The interaction between P38 and EGFR was examined by immunoprecipitation. D WT BMDMs were stimulated with AGE for 20 min. The interactions between RAGE, phosphorylated P38 (p-P38) and phosphorylated EGFR (p-EGFR) were analyzed by fluorescence microscopy (white scale: 50 μm; red scale: 20 μm). E WT BMDMs were stimulated with AGE (100 μM) for 0.5 h after being transfected with si-NC or si-Rab5. The protein levels of phosphorylated and total P38 (t-P38, p-P38) were measured by western blotting. WT BMDMs were pretreated with or without SB203580 (10 μM) for 1 h, followed by AGE (100 μM) stimulation for 4 h (K), 6 h (F, G, H), or 24 h (H, I, J). RAGE surface expression was analyzed by FACS (F, J). The protein (G, I) and mRNA (H) levels of RAGE were measured by western blotting and qPCR, respectively. The concentrations of TNF-α and CXCL1 were measured by ELISA (K). The data shown represent three (A–J) or four (K) independent experiments. **P < 0.01, ***P < 0.001, ns not significant, nd not detected

To assess the role of P38 activation in enhancing the surface expression of RAGE during AGE-induced inflammation, the P38 phosphorylation SB203580 was used to inhibit P38 activation. We found that SB203580 significantly reduced the surface expression of RAGE at 6 h (Fig. 6F) without affecting RAGE total protein or mRNA expression at 6 h (Fig. 6G, H). Notably, SB203580 reduced RAGE mRNA and protein levels at 24 h (Fig. 6H, I), and this change was accompanied by a decrease in the surface expression of RAGE (Fig. 6J). Next, we evaluated whether P38 inhibition reduced the secretion of inflammatory cytokines during RAGE-induced inflammation. SB203580 significantly reduced the production of TNF-α and CXCL1 in BMDMs in response to AGE, HMGB1 or S100A9 stimulation (Fig. 6K, Supplementary Fig. 15). Collectively, these findings suggest that RAGE promotes inflammation by activating EGFR, which colocalizes with P38, thereby driving RAGE membrane trafficking through Rab5-mediated RAGE internalization.

Inhibiting EGFR signaling reduces the inflammatory response of tissue-resident macrophages and prevents remote organ injury underlying sterile tissue damage

Our findings suggested that EGFR signaling in tissue-resident macrophages might be a valuable therapeutic target for trauma-induced remote organ injury. We examined the effect of adoptive transfer of EGFR-deleted macrophages on trauma-induced lung injury. We adoptively transferred WT or Egfr^−/− macrophages (AMs and BMDMs) into WT mice (recipients), which were depleted of endogenous macrophages by clodronate, and after 7 days, the recipient mice were subjected to trauma (Fig. 7A). Compared with WT macrophages, Egfr^−/− macrophages decreased lung inflammation and lung injury (as shown by the reductions in trauma-induced inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Fig. 7B, C) after trauma, although there was no significant difference in the concentrations of HMGB1, AGE, or S100A9 (Supplementary Fig. 16a), suggesting that EGFR signaling in tissue-resident macrophages contributed to tissue injury.

Fig. 7.

InhibitingEGFR protects against trauma- or stroke-induced lung-resident inflammatory responses and lung injury. A–C Adoptive transfer of AMs (i.v.) and BMDMs (i.t.) from WT or Egfr^−/− mice (donor mice) into WT mice (recipient mice), which were treated intratracheally (i.t.) or intravenously (i.v.) with clodronate to deplete endogenous macrophages. After 7 days, the recipient mice were subjected to trauma (A). B The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in lung tissue were examined (n = 4). C Representative H&E staining of lung sections (scale: 200 μm). D–K WT mice were pretreated with erlotinib (Er, 100 mg/kg body weight) or solvent by oral gavage for 3 days and then subjected to trauma or stroke for 12 h. D RAGE surface expression in lung-resident macrophages was examined by FACS (n = 3). E The protein levels of phosphorylated and total P38 (t-P38, p-P38) were measured in sorted lung-resident macrophages (n = 3). The mRNA levels of inflammatory response genes (Tnfa, Il1b, and Cxcl1) in sorted lung-resident macrophages (F, H) or lung tissue (G, I) were measured (n = 4). J, K Representative H&E staining of lung sections (scale: 200 μm). The data shown represent three or four independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001

Additionally, in light of the essential role of EGFR signaling in tissue-resident macrophages in remote organ injury, we examined the therapeutic potential of targeting EGFR signaling with drugs in animal models of sterile tissue damage induced by trauma or stroke. To evaluate the therapeutic effect of inhibiting EGFR in vivo, mice subjected to trauma or stroke were pretreated with erlotinib. Erlotinib treatment reduced the surface expression of RAGE (Fig. 7D) and P38 phosphorylation (Fig. 7E) in lung-resident macrophages after trauma or stroke, which was accompanied by a reduced inflammatory response in lung-resident macrophages and lung tissue (Fig. 7F–I). Additionally, erlotinib improved lung injury (pathological changes were demonstrated by reductions in inflammatory cell infiltration, alveolar and interstitial edema, hemorrhage, and alveolar septal thickening) (Fig. 7J, K), although no significant difference was observed in the concentrations of HMGB1, AGE, or S100A9 (Supplementary Fig. 16b), which was consistent with our in vitro findings (Fig. 5). Taken together, these data indicate that EGFR signaling, which is activated by RAGE signaling in lung-resident macrophages, causes an inflammatory response, thereby contributing to lung injury. Inhibition of EGFR with an inhibitor such as erlotinib reduces the inflammatory response and lung pathology in vitro and in vivo, highlighting EGFR as a potential therapeutic target for the treatment of remote organ injury underlying sterile tissue damage.

Discussion

Remote organ injury, which is a major secondary complication of sterile tissue damage, remains the main cause of poor prognoses [8, 12]. Therefore, we should search for treatment targets by examining critical nodes underlying remote organ injury secondary to sterile tissue damage. However, the mechanisms of remote organ injury secondary to sterile tissue inflammation are incompletely understood, and effective treatment options are rare. In this study, we demonstrated that tissue-resident macrophages played a crucial role in remote organ injury through the RAGE membrane trafficking-EGFR-Rab5-P38 pathway, which underlies sterile tissue damage (Fig. 8).

Fig. 8.

A schematic model of RAGE-EGFR signaling in lung-resident macrophages after sterile tissue damage. Circulating alarmins, such as HMGB1, S100, and AGE, are released from sterile tissue damage sites such as fracture trauma or stroke sites and bind to RAGE on remote lung-resident macrophages. This binding triggers RAGE interactions with EGFR, resulting in EGFR phosphorylation and leading to Rab5-mediated RAGE internalization mediated by phosphorylated EGFR. Subsequently, P38 is recruited and activated, promoting RAGE translation and trafficking to the plasma membrane, thereby intensifying the cell response to alarmins, which exacerbates inflammation and lung injury

Tissue-resident macrophages are present in almost all organs and perform diverse functions, such as maintaining tissue homeostasis and facilitating immune surveillance [15, 16]. Resident macrophages participate in the pathogenesis of various diseases, such as acute respiratory distress syndrome and COVID-19 [36], cardiac conduction and myocardial infarction [37], acute liver injury [38], inflammatory bowel disease [39], and neuroinflammation and synaptic information transfer [40], via the inflammatory response or phagocytosis. Tissue-resident macrophage populations originate from embryonic progenitors derived from the yolk sac through self-proliferation or are supported by the recruitment of circulating monocyte-derived macrophages [16]. We observed that remote organ injury after trauma or stroke was associated with an increased inflammatory response in tissue-resident macrophages rather than an increase in local death, the proliferation of resident macrophages, or the recruitment of circulating monocytes. Furthermore, major populations of lung-resident macrophages, including alveolar macrophages (AMs) and interstitial macrophages (IMs), contribute to trauma-induced lung injury. Studies have shown that AMs or IMs are activated by various stimuli and can release cytokines, such as TNF-α and IL-1β [17, 41, 42]. However, some studies have shown that IMs are associated with the production of the immunosuppressive cytokine IL-10 and regulate the immune response in the lung [43, 44]. A possible explanation for these phenomena is that our study focused only on acute injury in the early phase, and a strong proinflammatory response in resident macrophages, including AMs and IMs, plays a predominant role in the initiation of organ inflammation but not monocyte chemoattraction via the CCL2-CCR2 axis or the anti-inflammatory response in IMs [25]. It is necessary to investigate the function of tissue-resident macrophages in the late phase or in chronic inflammatory processes because of their long lifespan, self-proliferation and delayed increase in CCL2 levels, and the underlying mechanisms may include sensing and responding to the release of late mediators or exosomes [45] or crosstalk with parenchymal and stromal cells in the organ [46].

The mechanism by which resident macrophages in organs sense danger signals from distant sites of sterile tissue damage remains unclear. Recent research has suggested that pattern recognition receptors in resident macrophages detect specific ligands, which are released from damaged distant tissues, leading to an inflammatory response and organ injury. Toll-like receptors and RAGE are expressed on macrophages and are potential candidate receptors. RAGE has been shown to be involved in various conditions, such as fracture, crush injury, and brain injury [7, 19], via its immune response [18]. Considering the increase in the levels of circulating alarmins such as HMGB1, S100A9, AGE, which are ligands of RAGE, after trauma and stroke, it is not surprising that RAGE signaling in tissue-resident macrophages is associated with remote organ injury. In this study, we demonstrated the key role of RAGE signaling in resident macrophages, which sense and respond to ligands such as HMGB1, S100A9, and AGE, which are endogenous alarmins, thereby initiating inflammation and remote organ injury in sterile tissue damage such as trauma or stroke. Our study and others have shown that binding of ligands such as HMGB1, S100 or AGE to RAGE on the macrophage surface activates RAGE signaling [18, 20, 31, 47], thereby activating downstream MAPKs and nuclear factor kappa B (NF-κB) [18, 48] and leading to high levels of inflammatory cytokines and chemokines (IL-6, TNF-α, Il-1β, CXCL1, and CXCL8), which results in aberrant neutrophil infiltration into the lung [47, 49] and subsequent lung injury. In response to ligand recognition by RAGE, activated RAGE signaling induces the transcription of itself, leading to increased RAGE expression in a positive feedback manner [18, 30]. However, RAGE transcription and translation in macrophages were not affected after the acute stage of trauma in vivo or RAGE-induced inflammation in vitro, while the surface expression of RAGE was significantly increased. The amount of receptor present on the cell surface mediates various responses, including inflammation. In human microvascular endothelial cells [50] and umbilical vein endothelial cells [51], the surface expression of RAGE is increased after infection. Similarly, our previous study revealed increased the surface expression of RAGE after HMGB1 stimulation [19]. However, RAGE blockade could not completely eliminate inflammation, suggesting the activation of additional mechanisms by macrophages. Each alarmin may interact with one or more receptors. For instance, HMGB1 can also signal through other pattern recognition receptors, including Toll-like receptors 2, 4, and 9 [20, 52]. AGE can also initiate signaling via CD36 [20], and S100A9 can interact with TLR4 [26, 53]. TLRs and CD36, which are expressed on macrophages, serve as critical receptors for innate immune activation and cytokine release. Additionally, increased levels of IL33, which is an alarmin, have been detected in polytraumatized patients, leading to pulmonary complications [54]. This alarmin is known to activate RAGE/EGFR signaling in COPD epithelial cells [55]; however, it is unclear which other receptor–ligand pairs participate in alerting tissue-resident macrophages, which requires further investigation. In the present study, we found that the inflammatory response was suppressed by inhibiting RAGE function on the cell membrane during RAGE-induced inflammation in BMDMs (in vitro) or posttraumatic inflammation in tissue-resident macrophages (in vivo), indicating that the amount of RAGE on the cell surface partly determined the intensity of the inflammatory response induced by its ligands. However, most related research has focused on total protein expression rather than membrane expression and trafficking. The role of RAGE membrane trafficking after trauma or stroke and the mechanisms are not completely understood.

Ligand–receptor interactions can lead to the trafficking of surface receptors, including receptor internalization to the endoplasmic chamber and receptor transport to the cell membrane, subsequently activating specific intracellular signaling cascades [56]. Previous studies have shown that HMGB1 or AGE binds to various molecules, such as CdtB or LPS, and can be endocytosed via RAGE, leading to intracellular responses [21, 57, 58]. Here, we found that RAGE internalization occurred in the early phase, while RAGE trafficking to the cell membrane occurred in the middle and late phases after RAGE signaling activation. The internalized ligand‒receptor complex was transported to early endosomes via vesicles following endosomal sorting processes. After sorting, the receptor can be recycled to the cell membrane from the endocytic recycling compartment (ERC) [59]. Research has reported the rapid transport of RAGE to the cell membrane from intracellular storage pools in endothelial cells [60]. The binding of ligands to RAGE does not accelerate clearance or degradation but rather induces the expression of the receptor itself and amplifies an inflammatory response, leading to cellular activation and tissue dysfunction through a receptor-dependent mechanism, suggesting that the increase in the surface expression of RAGE was due to the recycling of RAGE and intracellular storage of RAGE rather than its translation in the middle phase; however, in the late phase, ligands influence the surface expression of RAGE by promoting additional RAGE translation, which traffics to the cell surface. Thus, an increase in the frequency of interactions between a cell surface receptor (RAGE) and ligands (such as HMGB1, AGE, and S100A9) results in increased RAGE translocation from the cytosol to the membrane and increases the effects of RAGE.

However, the detailed mechanism of RAGE trafficking remains unclear. Rab (Ras-related proteins in brain) GTPases, which are members of the Ras superfamily, serve as master regulators of vesicle-mediated membrane endocytosis and exocytosis, including the internalization and intracellular trafficking of receptors [61, 62]. Our study indicated that Rab5 was involved in RAGE intracellular trafficking. Rab5, which is localized to early endosomes, is implicated in early receptor internalization [32] via APPL1 [63] or PI3K signaling [64], which in turn triggers excessive NLRP3 inflammasome activation and excessive IL-1β production. In the present study, we showed that Rab5 regulated RAGE internalization; however, the mechanism by which RAGE is transported to the cell membrane in the middle and late phases is unclear. Rab4, which is localized to sorting endosomes, is involved in the rapid recycling from sorting endosomes to the membrane via PI3K signaling [32, 65], which promotes HIV infection in T cells. Rab11, which is localized to the trans-Golgi network, is implicated in slow recycling or biosynthetic trafficking from the Golgi to the membrane [32] via TLR4 signaling [66], which activates interferon regulatory factor-3 signaling and IFN-β production. Therefore, we speculate that Rab5-mediated RAGE internalization may lead to Rab4-mediated intracellular RAGE trafficking (recycling and storage), which causes RAGE translation and cytokine production and subsequent trafficking to the plasma membrane by Rab11.

Our results showed the process of RAGE internalization; however, the mechanism by which RAGE triggers inflammation through RAGE internalization and trafficking has not been determined. EGFR is the key factor in the production of infectious inflammatory cytokines during sepsis (such as TNF-α, IL-1β, and IL6) via TACE, TGF-α, Mig6 and TLR4 signaling [34, 35]. In myocardial infarction, inflammatory lung damage decreased in response to deletion of EGFR in myeloid cells [14]. In addition, previous studies demonstrated that RAGE reduced the inflammatory response by suppressing the phosphorylation of the p66^Shc protein [67], which is induced by AGE-mediated ROS via EGFR [20, 68]. The inhibition of EGFR/P38 MAPK signaling has been shown to reduce inflammatory responses [69]. In patients with COPD, oxidized IL-33, which is an alarmin found in polytraumatized patients [54], binds RAGE to activate EGFR signaling in epithelial cells [55]. Similarly, in the present study, we found that ligands could bind to RAGE, leading to RAGE interactions with EGFR and the subsequent activation of EGFR, which in turn promoted P38 activation, resulting in excessive sterile inflammation and lung injury after sterile tissue damage. Activated EGFR recruits phosphotyrosine-binding proteins via the autophosphorylation of tyrosine residues in EGFR dimers, leading to PI3K pathway activation [70], which is important for Rab5-mediated enhancement of endocytosis [32, 60]. One of the major roles of Rab5 is to establish specific membrane domains within the endosome to recruit various protein components [33, 60], such as MAPK signaling proteins. Although signaling from activated EGFR primarily occurs at the plasma membrane, the internalization of activated EGFR at endosomes mediates intracellular signaling [71] via colocalization with MERK1/2 [72]. EGFR autophosphorylation creates docking sites for the recruitment of the substrate EPS8 [73], which is a binding partner of the Rab5 GTPase-activating protein (GAP) RN-TRE [74], thereby leading to Rab5 activation through a direct interaction between RN-TRE and Rab5 [74], resulting in EGFR internalization [73, 75].

In conclusion, our data indicate that tissue-resident macrophages trigger inflammation and organ injury by sensing circulating alarmins, such as HMGB1, AGE, and S100A9, released from the site of distant sterile injury through the RAGE-EGFR signaling pathway. In addition, RAGE activates EGFR by binding to ligands on the membrane, thereby driving Rab5-mediated RAGE internalization via phosphorylated EGFR, which colocalizes with P38 and activates P38; this in turn promotes RAGE translation and trafficking to the plasma membrane to increase the cellular response to RAGE ligands, subsequently resulting in exacerbated inflammation. Thus, our work provides insight into the pathogenesis of remote organ injury underlying sterile tissue damage and suggests that targeting the RAGE-EGFR signaling pathway in tissue-resident macrophages is a potential strategy for treating secondary complications of sterile damage.

Materials and Methods

Animals

C57BL/6 mice (8–10 weeks old, 24–28 g) were purchased from the Central Animal Facility of Southern Medical University (Guangzhou, China). Rage-knockout male mice (8 to 10 weeks old) were obtained from the Animal Core Facility of Nanjing Medical University (Nanjing, China). Ccr2-knockout mice and Egfr-knockout mice were obtained from Cyagen Biosciences (Guangzhou, China). All animal experimental protocols conformed to the National Institutes of Health guidelines and were approved by the committees of Guangdong Medical University (Guangzhou, China).

Reagents and antibodies

Recombinant mouse HMGB1 (764006), recombinant human HMGB1 (557804), neutralizing anti-HMGB1 (clone 3E8, 651402), anti-MerTK-APC-CY7 (151519), MerTK-PE (151505), CD64-FITC (161007), CD64-PE (161004), CD11c-APC (117323), and CD11b (101211) were obtained from BioLegend. AGE (51995), anti-RAGE (ab216329, for western blot analysis and immunoprecipitation), mouse Alexa Fluor 488-conjugated RAGE antibody (FAB11795G, for flow cytometry), and human Alexa Fluor® 488 anti-RAGE (ab237361, for flow cytometry) were obtained from Abcam. PD168393 (PZ0285), dynasore (D7693), and paquinimod (SML2883) were obtained from Sigma. Erlotinib was obtained from Selleck.cn. LR-90 (HY-76383) was obtained from MedChem Express. Soluble RAGE (RD272590100) was obtained from Biovendor. Antibodies specific for total EGFR (4267), phospho-EGFR (3777), phospho-P38 (4511, 9216), total P38 (8690) and β-actin (3700), horse anti-mouse IgG (H&L) (7076) and goat anti-rabbit IgG (H&L) (7074) were obtained from Cell Signaling Technology. A RAGE antibody (sc-365154 AF647, for flow cytometry and immunofluorescence) was purchased from Santa Cruz Biotechnology. Anti-Rab5 (PA5-88260), mouse IgG isotype control (02-6502), goat anti-rabbit IgG (H + L) Alexa Fluor 488 (A-11008), donkey anti-rabbit IgG (H + L) Alexa Fluor 594 (A-21203), goat anti-mouse IgG (H + L) Alexa Fluor 488 (A-11001), donkey anti-mouse IgG (H + L) Alexa Fluor 594 (A-21203), a live/dead violet vitality kit (L34958), siRab5 (152489), and CellTracker Deep Red dye (C34565) were purchased from ThermoFisher Scientific. Recombinant mouse S100A9 protein (2065-S9-050), FPS-ZM1 (6237), mouse Alexa Fluor 488-conjugated RAGE antibody (FAB11795G for flow cytometry), rat Alexa Fluor 488-conjugated IgG2A isotype control (IC006G for RAGE-AF488), mouse Alexa Fluor 647-conjugated RAGE antibody (FAB11795R for flow cytometry), rat IgG2A Alexa Fluor 647-conjugated isotype control (IC006R for RAGE-AF647), mouse RAGE antibody (MAB11795 for flow cytometry), and neutralizing anti-RAGE (AF1145 for neutralization) were obtained from R&D Systems. Anti-mouse CD45-PE/Cy7 (552848), CD45-BUV395 (567451), CD45-FITC (553079), Ly6G-APC/Cy7 (560600), Ly6G-PE (551461), F4/80-AF488 (567201), and F4/80-PE (565410) were obtained from BD Biosciences.

Trauma mouse model

A trauma mouse model was generated by pseudofracture, which is a validated murine model of sterile musculoskeletal trauma to evaluate posttraumatic early and late immune responses [22–24]. Male mice (8–10 weeks old) were euthanized, and 2 femurs and 2 tibias were harvested and crushed using a sterile mortar in 2 ml of sterile PBS under sterile conditions to create the ‘bone solution’. The mice were anesthetized with ketamine (50 mg/kg BW) and xylazine (5 mg/kg BW). An 18 cm hemostat was clamped onto each posterior thigh muscle at the midpoint along the femur for 2 min to induce tissue injury, followed by the injection of 0.3 ml of bone solution into the injured posterior muscles of each thigh. In the sham group, 0.3 ml of sterile saline was injected into the posterior thigh muscle. After 6 h, serum and lung tissue were collected for further analysis after pseudofracture or sham surgery. None of the mice died during the surgery, and the number of mice used is shown in the figures.

Stroke mouse model

A mouse model of stroke was generated by middle cerebral artery occlusion (MCAO) [76]. Focal cerebral ischemia was induced by MCAO in this study. The mice were anesthetized with ketamine (50 mg/kg BW) and xylazine (5 mg/kg BW), and their temperature was maintained at 36 ± 0.5 °C using a homoeothermic blanket. To establish MCAO, a 6-0 silicon-coated monofilament suture (Doccol, Redland, CA, USA) was inserted into the left external carotid artery, advanced into the internal carotid artery, and occluded the origin of the MCA. After 45 min of MCAO, the suture was removed to allow reperfusion. Sham-operated animals underwent internal carotid artery exposure without MCAO. Thirty-eight mice underwent MCAO surgery, and four mice that died due to illness were excluded from the analyses. The remaining 34 mice were included in the analyses.

Bone marrow-derived macrophage (BMDM) isolation and culture

BMDM isolation was performed as previously described [19]. Briefly, bone marrow was flushed from male mice (8–10 weeks old) and cultured in BMDM culture medium (DMEM supplemented with 50 µg/ml penicillin/streptomycin and 10 ng/ml recombinant macrophage colony stimulating factor). BMDMs were completely differentiated and ready for use by Day 7.

Cell line and cell culture

The human monocyte cell line THP1 was obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The cells were cultured in RPMI-1640 medium (Gibco, USA) supplemented with 10% FBS and penicillin/streptomycin (50 µg/ml). To obtain THP1-derived macrophages, THP1 cells were pretreated with 100 ng/ml phorbol myristate acetate (PMA; Sigma‒Aldrich) for 48 h.

Drug administration

sRAGE: The mice received two intraperitoneal injections of sRAGE (4 mg/kg), one 30 minutes before surgery and another 4 hours after surgery. LR-90: The mice were treated with 50 mg/l LR-90 in their drinking water for 4 weeks prior to trauma induction. Paquinimod: The mice were administered paquinimod (10 mg/kg) or vehicle (DMSO) for 3 days by oral gavage prior to trauma induction. FPS-ZM1: The mice were treated with FPS-ZM1 (5 mg/kg) or vehicle (DMSO) for 3 days by oral gavage prior to trauma induction. HMGB1 or RAGE neutralizing antibody: The mice were administered 10 mg/kg mouse IgG isotype control or different doses of HMGB1 or RAGE neutralizing antibodies for 2 h by tail vein injection prior to trauma induction. BMDMs or THP-1 cells were pretreated with 10 µg of mouse IgG isotype control, HMGB1 or RAGE neutralizing antibodies for 0.5 h and then exposed to HMGB1, AGE or S100A9 for different times. Erlotinib: The mice were administered erlotinib (100 mg/kg suspended in polyethylene glycol (PEG) 400) or the solvent (PEG400) for 3 days by oral gavage prior to trauma or stroke induction. BMDMs or THP-1 cells were pretreated with or without 20 µM erlotinib for 1 h and then exposed to HMGB1, AGE or S100A9 for different times. Dynasore, PD168393, or SB203580: BMDMs or THP1 cells were pretreated with dynasore (20 μM), PD168393 (10 μM), or SB203580 (10 μM) for 1 h and then exposed to AGE, HMGB1, or S100A9 for different times.

Mouse alveolar macrophage (AM) isolation

Primary mouse AMs were isolated from euthanized mice by bronchoalveolar lavage (BAL) using 3 ml of PBS supplemented with 1 mM EDTA. The lavage fluid was centrifuged at 300 × g for 10 min. The pellet was suspended and cultured in RPMI 1640 medium supplemented with 10% FBS. After 45 min, the cells suspended in the medium were discarded, and the adherent macrophages were used for further studies. The percentage of alveolar macrophages was confirmed to be greater than 95% by flow cytometry.

Lung wet/dry weight measurement

The severity of pulmonary edema was assessed by determining the lung wet/dry ratio. Briefly, after the mice were sacrificed, lung tissues were collected and immediately weighed to determine the wet weight (W), after which the lung tissues were placed in an oven at 80 °C for 48 h and subsequently weighed again to obtain the dry weight (D). The ratio of wet weight to dry weight was calculated.

Macrophage depletion

For macrophage depletion, clodronate liposomes (LIPOSOMA) or PBS liposomes were delivered via the tail vein (100 µl for systemic depletion) and by intratracheal injection (60 µl for lung alveolar macrophage depletion) 24 h prior to trauma or stroke induction.

Adoptive macrophage transfer

For the adoptive transfer experiments shown in Figs. 2F and 3C, WT mice were used as the recipients. The recipients received two injections of 1*10⁶ BMDMs (intravenously) or 3*10⁵ AMs (intratracheally) one day apart, and the cells were labeled with 10 µM CellTrace Deep Red before being transferred. For the adoptive transfer experiments shown in Figs. 2I and 7A, macrophage-depleted WT mice were used as recipients and received two injections of 1 × 10⁶ BMDMs (intravenously) or 3 × 10⁵ AM (intratracheally) one day apart. In both experiments, the recipient mice were used 7 days after macrophage transfer.

Flow cytometry (FACS) and sorting

For surface staining, single-cell suspensions were stained with relevant fluorescent antibodies. For intracellular p-EGFR staining, after Fc receptor blocking and cell surface staining, the cells were fixed and permeabilized using a kit (eBioscience, San Diego, CA) according to the manufacturer’s instructions. Rabbit p-EGFR antibodies and the rabbit IgG isotype control were used as primary antibodies, and AF488-conjugated goat anti-rabbit was used as the secondary antibody. Dead cells were excluded from the analysis using the LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Scientific, Waltham, MA). The cells were examined by a BD LSRFortessa™ flow cytometer (BD Bioscience, Franklin Lakes, NJ, USA).

Lung single-cell suspensions were gated on the FCS/SSC plot. Subsequently, singlet live cells were selected, and CD45+ cells were gated for further cell selection. Neutrophils were identified as Ly6G-positive among the sorted cells, while resident macrophages were identified as Ly6G-negative and positive for MerTK and CD64. All cell sorting was conducted using a SORP-aria (BD Biosciences, San Jose, CA). The samples were stained with the following antibodies: live/dead (violet or 7AAD), CD45 (BUV395 or PE-CY7), Ly6G (APC-CY7 or PE), MerTK (APC-CY7 or PE), CD64 (FITC or PE), and RAGE (FITC or AF647).

Cell transfection

To silence Rab5 expression, BMDMs (1 × 10⁶ cells) were seeded in 6-well plates and transfected with Rab5 siRNA using Lipofectamine 3000 (Invitrogen) according to the manufacturer’s instructions. After 24 h or 48 h, the silencing efficiency was measured by quantitative PCR and immunoblotting.

Immunoblotting

BMDMs or lung macrophages were lysed in RIPA buffer. The lysates were centrifuged, and the protein concentrations of the supernatants were determined using a BCA kit (Thermo Fisher Scientific). The proteins were separated by SDS‒PAGE and transferred to PVDF membranes (Millipore). After being blocked with 5% skim milk for 1 h at room temperature, the membranes were incubated with primary antibodies overnight at 4 °C. After the membranes were washed three times with TBST, they were incubated with the corresponding secondary antibodies at room temperature for 1 h, after which the proteins were visualized using increased chemiluminescence (ECL) reagent (Millipore). Densitometric analysis of the proteins was performed with Image-Pro Plus.

Real-time quantitative PCR (q-PCR)

Total RNA was extracted from BMDMs using TRIzol RNA Isolation Reagent (Life Technologies, Pittsburgh, PA, USA) according to the manufacturer’s instructions. Reverse transcription from RNA to cDNA was performed using an RT reagent kit (Takara). The cDNA was subsequently amplified by q-PCR using specific primers. The primers used for qPCR are listed in Supplementary Table 1.

Enzyme-linked immunosorbent assay (ELISA)

BMDMs (1 × 10⁶ cells) were seeded in 6-well plates, and the cell culture supernatants were collected after the indicated treatments. ELISA kits for TNF-α (88-7324-88 (mouse), 88-7346-88 (human), Thermo Fisher Scientific) and CXCL1 (DY453, R&D Systems) were used to measure cytokine levels in the supernatants or serum according to the manufacturer’s instructions.

Biochemical analysis

The serum was collected, and HMGB1, AGE, and S100A9 levels were measured using an HMGB1 expression kit (30164033, IBL International), an AGE assay kit (ab238539, Abcam), and a mouse S100A9 ELISA kit (2065-S9-050, R&D Systems) according to the manufacturer’s instructions. The endotoxin concentrations of HMGB1 (764006 or 557804), AGE (51995) and S100A9 (2065-S9-050) were assessed using the Pierce™ Chromogenic Endotoxin Quant Kit (A39552S, Thermo Fisher Scientific) according to the manufacturer’s instructions.

Immunoprecipitation

Cells (8 × 10⁶ cells) were lysed in nondenaturing lysis buffer (Millipore), and the cell lysates were cleared by centrifugation. The protein concentrations were measured by a BCA kit. Equal amounts of protein were precleared with protein A/G beads (Thermo Fisher Scientific) at 4 °C and incubated overnight with primary antibodies or irrelevant IgG at 4 °C. Immune complexes were captured by incubation with protein A/G beads for 3 h at 4 °C, after which the agarose/sepharose beads were extensively washed with lysis buffer containing NP40. Finally, the proteins were eluted in 2X SDS sample buffer and analyzed by immunoblotting.

Immunofluorescence analysis

Cells were seeded onto glass slides. After the indicated treatments, the cells were fixed with 4% paraformaldehyde (10 min at room temperature), permeabilized with 0.5% Triton X-100 in PBS (10 min), and blocked with 5% bovine serum albumin (BSA) for 2 h. Then, the cells were incubated with the relevant primary antibodies overnight at 4 °C, followed by incubation with the appropriate secondary fluorescent antibodies (1:400) for 1 h at room temperature. Nuclei were counterstained with DAPI (10 min) before images were captured using a fluorescence microscope (Nikon Corporation).

Statistical analysis

The data were analyzed using GraphPad Prism 9.0 (GraphPad Software). For comparisons between two experimental groups, the unpaired two-tailed Student’s t test was used. One-way ANOVA with Tukey’s post hoc test was performed for multiple groups. The results are presented as the mean ± SD. P values < 0.05 were considered to indicate statistical significance.

Study approval

All animal procedures conformed to the guidelines of Animal Research: Reporting of In Vivo Experiments and were approved by the Animal Research Center of Guangdong Medical University.

Author contributions

JT and XYH conceived and supervised the study. HHZ and JJJ designed and performed the experiments and analyzed the data with JLZ, who also prepared the figures. ZYX, XLL, PYX, WLT, and JDZ performed the experiments. JT revised the manuscript. All the authors performed critical reviews of the manuscript. HHZ, XYH, and JT wrote the manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41423-024-01125-1.