In this study, we observed immune remodeling in the lung over time in a mouse model of mild peritoneal sepsis. Four days after mild peritoneal sepsis, when mice were in clinical recovery, a significantly elevated lung water content and significant monocyte and neutrophil accumulation in the lung interstitium were observed. Intratracheal LPS-induced lung inflammation appeared to be more severe in the septic lung, as indicated by increased leukocyte trafficking and protein leakage into the alveoli. However, the lung mRNA levels of inflammatory cytokines and chemokines, namely, tnf, il6, ccl2, and cxcl1, decreased over time by post-CLP day 4 to approximately one-tenth of their respective pre-CLP levels. The lungs on day 4 post-CLP also showed suppressed tnf, il6, ccl2, and cxcl1 mRNA expression upon LPS instillation. RNA sequencing results revealed that the expression levels of p21, socs1, il4, il11, and tgf were significantly decreased on day 4 post-CLP compared with the control, indicating a global suppression of mediators, except for socs3. Although the LPS-induced monocyte influx into the alveolar space was increased in the septic lung, the levels of MCP-1, KC, IL-6, and TNF-α expression in BALF were suppressed.

Monocytes isolated from septic lungs and stimulated with LPS in vitro showed increased tnf and il10 mRNA expression, while expression of il6, ccl2, and cxcl1 was highly suppressed.

The monocyte proportion in the lung interstitium was shifted to Ly6Clo dominant post-sepsis. Systemic recombinant IFN-β administration to septic mice partially restored lung mRNA expression, reverted the Ly6Chi/Ly6Clo ratio to pre-CLP values, and revived LPS reactivity in monocytes isolated from the lungs.

Lung injury as a form of CLP-induced remote organ damage has been reported in many studies [24,25,26,27]. In the present study, we characterized this phenomenon through careful evaluation of immunocompetent cell clustering in the lungs over time. Sepsis-mediated disruption of epithelial/endothelial cell integrity has been well documented [26] and was confirmed by the high degree of intra-alveolar protein leakage following LPS stimulation in our study.

We found that the LPS stimulation-induced myeloid cell influx into the alveoli was significantly higher in the CLP group than the control group, despite the relatively low KC and MCP-1 levels. This phenomenon implies a disruption in the integrity of the vascular endothelium and the alveolar epithelium during sepsis. By contrast, the cytokine/chemokine profile after LPS stimulation indicated suppression of these factors (Suppl. Figure E8).

In our previous report using the same mild peritoneal sepsis model, we suggested that decreased bacterial clearance due to impaired neutrophil migration caused by decreased KC secretion from the alveoli led to ARDS [18]. Consistent with this, this current study’s findings suggested that the lung injury state, as revealed by inflammatory cell infiltration and protein leakage into the alveoli, may not correspond to a typical hyperinflammatory state, which is usually recognized as increased host-defense capacity including inflammatory cytokine secretion.

In our previous study, i.t. administration of Pseudomonas aeruginosa in the same mild peritoneal sepsis model as used in the present study resulted in a mortality rate exceeding 90% [18], whereas in the present study, i.t. LPS instillation, an aseptic irritant, rarely caused death in mice. Therefore, these model mice appeared to have a significant impairment in the host-defense capacity against pathogens, rather than excessive inflammation.

Discrepancies between pathological findings and the immunological response have been demonstrated in recent studies. Analysis of COVID-19 pneumonia patients revealed that the immune phenotype is immunosuppressive, even in fulminant ARDS patients [28]. A randomized controlled trial of simvastatin in ARDS patients has shown that there is an immunosuppressive population and that the drug has little effect in that group [29]. Using mouse models of sepsis with relatively low mortality, previous studies have demonstrated impaired innate immunity up to 4 days after CLP [30,31,32]. By contrast, other studies have reported evidence of inflammatory activation during this period [25, 33, 34]. We have compiled a summary of studies that evaluated pulmonary immune function in mouse models, including our previous research (Table 1) [18, 25, 30,31,32,33,34,35,36,37,38]. The findings are inconsistent, with some studies suggesting immune suppression even in the chronic phase (10–21 day post-CLP) [38], while others indicate immune activation [35]. Our current study, which focuses on the later stage of the subacute phase (day 4 post-CLP), demonstrated a decline in immune responsiveness.

Series of publications have reported the properties of Ly6Chi and Ly6Clo monocytes; specifically, monocytes recruited by signals such as CCL-2 differentiate into Ly6Chi macrophages, which produce factors such as transforming growth factor-β, platelet-derived growth factor, TNF-α, and IL-1β that activate inflammation [39]. Ly6Chi monocytes have been reported to exacerbate renal injury in sepsis [40], experimental cerebral malaria [41], and lung injury caused by diffuse pulmonary hemorrhage [42]. In the LPS-induced acute lung injury mouse model, Ly6C+ monocytes reportedly exacerbated inflammation [43]. In addition, M1-dominant monocytes in the bloodstream have been associated with a worse prognosis in a baboon model of peritoneal sepsis [44]. Denstaedt et al. found Ly6Chi monocyte accumulation in the lungs 3 weeks after CLP, suggesting it as a potential cause of enhanced inflammation after the second insult with LPS stimulation [35]. By contrast, Ly6Clo monocytes exert anti-inflammatory effects, suppress T-cell function through CD52–HMGB1 binding, and express matrix metalloproteinases to prevent fibrosis [39].

Li et al. proposed the idea that Ly6Chi macrophages are of the M1 phenotype and that Ly6Clo macrophages are of the M2 phenotype [39]. Based on previous reports [45], we can reasonably infer that Ly6Chi monocytes are inflammatory, while Ly6Clo monocytes are anti-inflammatory.

Using i.t. LPS, we found that monocytes entering the alveoli were CCR2+Ly6Chi, indicating the M1 phenotype. Despite this, the TNF-α, IL-6, KC, and MCP-1 levels in BALF were reduced.

It has been reported that monocytes, like macrophages, can exhibit phenotypes other than M1 and M2 [46, 47] and that Ly6Chi and Ly6Clo switching occurs in response to the surrounding environment [48, 49]. Furthermore, the lung environment may change in response to the systemically secreted cytokine environment, with plasticity in monocyte reactivity. These may be the reason for the reduced cytokine secretion by intraalveolar monocytes in the sepsis lung despite being Ly6Chi.

In the present study, systemic IFN-β administration restored reduced cytokine mRNA (including tnf, il6, ccl2, cxcl1, and il10) expression on day 4 post-CLP. The impaired response to LPS in monocytes isolated from septic lungs was restored by coincubation with IFN-β. The Ly6Chi/lo ratio also returned to control levels with systemic IFN-β. We have previously shown that IFN-β promotes re-expression of inflammatory mediators via IFN-α/β receptor in macrophages during the LPS-tolerant phase, when TLR4 signaling is suppressed by SOCS signaling [18, 37]. The observation that, at day 4 post-CLP, the expression of mediators in the lung tissue—both pro-inflammatory and anti-inflammatory—was globally suppressed, with the exception of socs3, which was upregulated, supports this interpretation. The present results also suggest that IFN-β partially restores the immunological remodeling of monocytes in septic mouse lungs.

In contrast to the highly suppressed synthesis of il6, ccl2, and cxcl1 following CLP, the synthesis of tnf, another inflammatory cytokine, was relatively preserved in the present study. We previously reported that alveolar macrophages stimulated with LPS synthesize intracellular socs3 mRNA, which renders cells tolerant to LPS [37]. SOCS3 suppresses the MyD88-related signaling pathway but not the TRIF pathway among the TLR4 signaling pathways. Since tnf mRNA can also be synthesized via the TRIF pathway, preserved TRIF pathway signaling may be the mechanism explaining the observed phenomenon [50, 51].

In the present study, systemic administration of IFN-β restored the suppressed cytokine response without causing excessive activation. Combined with its previously reported ability to inhibit increased pulmonary permeability via CD73 [52], the administration of IFN-β in the current model may enhance host defense mechanisms without inducing further lung injury due to excessive inflammation.

Although some patients may certainly benefit from immunosuppressive therapy in severe sepsis-related conditions, the host-defense capacity may be compromised in others; thus, it is crucial to evaluate the immune status of individual patients when implementing immunomodulatory therapy.

Immunological reprogramming after sepsis potentially may not provide an effective biological defense against pathogens in the lung and may represent one mechanism for developing ventilator-associated pneumonia, which frequently occurs in patients recovering from sepsis. To prevent nosocomial pneumonia, restoration of adequate host-defense capacity within the lungs is necessary. Furthermore, the partial restoration of the LPS response capacity in immunocompetent cells by systemic IFN-β administration is a promising result for future clinical applications.

A randomized controlled trial involving ARDS patients classified as moderate to severe according to the Berlin definition failed to demonstrate the efficacy of IFN-β administration [53], leading to recommendations against its routine use until new evidence is reported [54]. In addition, our group's research in a different sepsis mouse model has shown that the timing of administration significantly influences outcomes [37], highlighting the clear need for indicators to identify responsive patient populations.

The sequential sepsis → pneumonia model included only male mice; however, sex is a known risk factor for ARDS [55], and the influence of female hormones has not been examined in this model. A variety of other factors should be considered in a heterogeneous clinical setting.

In the present study, the characterization of Ly6Chi/lo monocytes, including their host-defense capacity and intracellular protein expression, was not performed. This remains a crucial area for investigation in future studies. Furthermore, the tissue concentrations of systemically administered IFN-β, as well as soluble CD73—a marker of IFN-β efficacy [52]—were not measured in this study. The effects of IFN-β on the ARDS mouse model include the CD73-mediated prevention of vascular permeability, as reported by Kiss et al. [52]. Given that our previous study demonstrated attenuated lung injury in a P. aeruginosa pneumonia-induced ARDS model, it is anticipated that IFN-β may also exert beneficial effects on lung injury in the current model. This mechanism was not explored in this study but warrants further investigation in future studies.

Minimally invasive surrogate markers to detect the immune status of the host should be investigated for clinical application. The indication for IFN-β administration should be further studied because the reverse effect of immunosuppression by IFN-β, as we have reported in the past [18, 37], may produce opposite results depending on the timing of administration.