Platelets orchestrate the resolution of pulmonary inflammation in mice by T reg cell repositioning and macrophage education

Timely resolution of inflammation is important to impede uncontrolled host tissue destruction, and neutrophils need to be efficiently removed after clearance of the invading microorganisms to avoid excessive tissue damage (Arienti et al., 2019; Levy and Serhan, 2014). Neutrophil apoptosis with subsequent engulfment by macrophages is the major route by which the host clears neutrophils. Efficient phagocytosis of apoptotic neutrophils by macrophages (efferocytosis) not only prevents their secondary necrosis but also turns pro-inflammatory macrophages into cells with an anti-inflammatory, reparative signature (Doran et al., 2020). Dysfunction in the neutrophil apoptosis machinery is considered critical for the pathogenesis of many chronic human inflammatory diseases, e.g., pulmonary fibrosis after the acute respiratory distress syndrome (ARDS; Potey et al., 2019).

ARDS with accompanying pulmonary inflammation is a common life-threatening disease (Matthay et al., 2012). Despite improved supportive care, the mortality rate of ARDS remains high and may reach up to 45% in severe ARDS (Ranieri et al., 2012). This syndrome may develop in response to several causes, including direct injury of the lung by bacterial or viral infection or aspiration of gastric content, or due to indirect causes, such as sepsis (Matthay et al., 2012). Pulmonary inflammation is characterized by an increased number of neutrophils in the lung and increased permeability, leading to lung edema and, consequently, decreased pulmonary gas exchange (Matthay et al., 2012; Semple et al., 2019). Indeed, neutrophil activation and recruitment participating in host defense are key events in the development of pneumonia and ARDS (Grommes and Soehnlein, 2011; Rebetz et al., 2018). Recruitment of neutrophils into inflamed tissue is required for eliminating invading pathogens, but can also cause tissue destruction by releasing a variety of enzymes and toxic granule content (Phillipson and Kubes, 2011).

The pathogenesis of pulmonary inflammation has been shown to be additionally platelet dependent in several lung injury models (Kornerup et al., 2010; Looney et al., 2009; Rossaint et al., 2014; Sreeramkumar et al., 2014; Zarbock et al., 2006). Platelet–neutrophil aggregates can be found in the lung microvasculature during pulmonary inflammation (Grommes et al., 2012; Rossaint et al., 2014; Zarbock et al., 2006), and the exchange of mediators between these two cell types amplifies inflammation (Rossaint et al., 2016). While it is now commonly accepted that the interaction of platelets and neutrophils is a prerequisite for neutrophil recruitment in many entities of ARDS, it remains unclear whether this interaction is restricted to the intravascular compartment or rather extends its influence in the tissue parenchyma. Beyond the importance for intravascular platelets for neutrophil recruitment, recent studies showed that platelet–neutrophil complexes are dynamically formed in the blood during pulmonary inflammation and that these complexes can also be found in the intra-alveolar space, thus indicating a role for intra-alveolar platelets in the pathogenesis of ARDS (Amison et al., 2018; Ortiz-Muñoz et al., 2014). It has been shown that macrophages are involved in the initial inflammatory phase of ARDS, but recent reports demonstrated that these cells are also involved in the resolution of pulmonary inflammation (Kapur et al., 2019). Regulatory T (T reg) cells represent a T cell subpopulation with predominantly immune regulatory functions and are immunosuppressive. T reg cells are a source of anti-inflammatory cytokines IL-10 and TGFβ. T reg cells have been shown to exert anti-inflammatory functions during the onset of lung injury with IL-10 playing an important role (Kapur et al., 2017a), and low IL-10 levels may represent a risk factor for the development of ARDS (Kapur et al., 2017b). Here, IL-2 appears to be a protective mediator triggering IL-10 release by T reg cells as a counterbalance in acute inflammation (e.g., TRALI; He et al., 2019); however, whether and how platelets, macrophages, and T reg cells may cooperate during the resolution of pulmonary inflammation is unknown.

The aim of the present study was to investigate the role of platelets during inflammation resolution and to identify cellular partners involved in this process. We find that intravascular platelets are crucially required for the recruitment of T reg cells into the lung and that intra-alveolar platelet relocation is indispensable for macrophage polarization toward an anti-inflammatory phenotype, aiding in the clearance of alveolar neutrophils and termination of pulmonary inflammation.

Bacterial pneumonia represents clinically the most important etiology of ARDS. Therefore, we chose the infectious model of bacterial-induced pneumonia for this study (Cilloniz et al., 2018). Platelets are required for neutrophil recruitment into the lung during pneumonia (Rossaint et al., 2016). After induction of bacterial pneumonia, neutrophil recruitment into the alveoli, accompanied by increased MIP2⍺ serum levels, plateaued at day 2 and declined subsequently (Fig. 1 A and Fig. S1 A). To assess time dependency, we depleted platelets by intravascular injection of a platelet-targeting α-GPIbα antibody at day 2 (i.e., after peak neutrophil recruitment), which lead to a reduction of >98% of circulating platelets over a period of 3 d (Fig. S1 B). Alveolar neutrophil numbers remained high on days 4 and 5 in platelet-depleted mice (Fig. 1 A). Additionally, we used Pf4^iDTR mice (Pf4Cre;Rosa^iDTR mice) to selectively deplete platelets (Fig. 1 A; Wuescher et al., 2015). We chose to start diphtheria toxin (DT) treatment 3 d before pneumonia induction to match depletion and the peak of neutrophils in the lung (Fig. S1 C). DT administration did not affect the blood counts of erythrocytes or leukocyte subsets (Fig. S1 D). In Pf4^iDTR mice, the number of alveolar neutrophils also remained high on days 4 and 5 (Fig. 1 A). If platelets were depleted before the induction of bacterial pneumonia, all mice died within a 48 h of observation period due to abolished, initial neutrophil recruitment and overwhelming bacterial dissemination (Fig. S1 E). Neutrophil apoptosis in the bronchoalveolar lavage (BAL) remained low in Pf4^iDTR and platelet-depleted mice until day 5 compared with elevated apoptosis in control mice (Fig. 1 B). Consistently, we observed that platelet depletion caused persistent high neutrophil counts and decreased neutrophil apoptosis in the lung interstitial space at day 5 (Fig. 1, C and D; for exemplary FACS plots, see Fig. S2 A) and increased lung wet/dry ratios (Fig. S1 F), which is altogether indicative of delayed resolution of inflammation in the absence of platelets. To assess whether continuing polymorphonuclear granulocyte (PMN) recruitment after day 2 following induction of pneumonia affects the balance between PMN influx and clearance in the lungs, we depleted PMNs by DT administration in Mrp8-iDTR mice on day 2 after pneumonia induction and observed similar counts of viable and apoptotic PMN in the BAL and lung interstitial compartment on days 3–5 as in the respective non–PMN-depleted groups (Fig. 1, A–D). Notably, the effects of platelet depletion on the counts of viable and apoptotic PMNs in the BAL and lung interstitial compartment on days 3–5 were comparable in control mice and PMN-depleted mice, thus showing that PMN clearance is the major determinant of net PMN accumulation during the resolution of pulmonary inflammation in our model. To exclude that PMN-bound, activated platelets binding annexin V by surface-expressed phosphatidylserine falsely mimics PMN apoptosis, we analyzed CD41⁻ PMNs from control mice after pneumonia induction and could still observe similar apoptosis rates (Fig. S2 D).

Unresolved acute inflammation may lead to sustained, chronic inflammation in the lung, with increased deposition of collagen fibers in the lung (Levy and Serhan, 2014). The lack of platelets during the resolution phase (days 3–5) led to a significantly increased collagen content and reduced lung compliance of the lung after 2 wk of pneumonia induction (Fig. 1, E and F). This was reversed by intratracheal (i.t.) instillation of 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB), a specific cyclin-dependent kinase 7/9 inhibitor that induces neutrophil apoptosis in the lung after i.t. instillation (Leitch et al., 2012; Rossi et al., 2006). Histological analysis showed persistent pulmonary inflammation, fibrotic tissue remodeling, and elevated Ashcroft scores (Ashcroft et al., 1988) in platelet-depleted mice after 2 wk (Fig. 1, G and H). In comparison, neutrophil depletion prevented fibrotic tissue remodeling (Fig. 1, G and H). These data demonstrated an unexpected and absolute requirement for platelets during the resolution phase of pulmonary inflammation.

T reg cells are recruited into the lung during the resolution of inflammation (Ehrentraut et al., 2013). Platelet–neutrophil aggregates in peripheral blood were rapidly detectable and their frequency steadily decreased from day 2, whereas platelet–T reg cell aggregates were first detected after 2 d and steadily increased until day 5, corresponding to the resolution phase (Fig. 2, A–C; for exemplary FACS plots, see Fig. S2, B and C). Platelet interactions with monocytes and CD4 T cells did not show significant alterations, whereas the percentage of CD41⁺ CD8 cells slightly increased on day 5 following the induction of pulmonary inflammation (Fig. 2 A). T reg cell counts in blood increased after 2 d (Fig. 2 D), and T reg cell recruitment in lungs started 2–3 d after pneumonia induction and was nearly completely abolished following platelet depletion at day 2 (Fig. 2 E). To address this dichotomous behavior and the switch from platelets preferably binding to neutrophils during the onset of inflammation toward mainly binding T reg cells during the resolution phase, we analyzed the surface expression of adhesion molecules on neutrophils, platelets, and T reg cells. Interestingly, PSGL-1 and Mac-1 expression on circulating neutrophils significantly decreased after the onset of pulmonary inflammation following day 2 (Fig. 2, F and G), while the surface expression of the counter receptors P-selectin and GPIb, respectively, on platelets remained steady after initial inflammatory stimulus (P-selectin; Fig. 2 H) or unaltered, as in the case of GPIb (Fig. 2 I). In contrast, expression of PSGL-1 (Fig. 2 J) on circulating T reg cells was increased 2 d after pneumonia, suggesting that this was the rate-limiting step in platelet binding. The sheddase ADAM8 is externalized by neutrophils following activation and aids in the migration through the lung tissue (Dreymueller et al., 2017), but cleaves PSGL-1 on neutrophils (Domínguez-Luis et al., 2011). Circulating neutrophils showed an increase of ADAM8 expression 3 d after pneumonia induction compared with noninfected mice (Fig. 2 K). Incubation of isolated neutrophils with rADAM8 (300 pg/ml) inhibited platelet–neutrophil aggregate formation to a level similar to PSGL-1 blockade (Fig. 2 L), and blocking ADAM8 activity (inhibitor BK-1361, 25 µg/g body weight) preserved the prevalence of platelet–neutrophil aggregates in vivo (Fig. 2 M), whereas platelet–T reg cell interactions remained unaltered (Fig. 2 N). Thus, there appears to be a molecular transition from platelets preferentially binding to neutrophils early during inflammation, whereby ADAM8 cleaves PSGL-1 and allows the preferential binding of platelets to T reg cells during the resolution phase. ADP (adenosine diphosphate) is a weak platelet agonist and it has been published that ADP induces a dose-dependent release of platelet ⍺-granules and increases platelet P-selectin surface expression in a dose-dependent manner (Dong et al., 2015; Elaïb et al., 2016; Lu et al., 2015), which we also observed in dose–response experiments in vitro (Fig. S2, F and G).

The effect of intravascular platelet depletion on neutrophil clearance from the lung at day 2 after onset of initial neutrophil accumulation suggested the imminent importance of platelets entering the alveoli at a later time point. Thus, we injected WT mice with red fluorescent platelets isolated from DsRed⁺ mice at day 3 after pneumonia induction. We detected red fluorescent platelets in the BAL fluid on day 5 (Fig. 3 A), indicating that, although neutrophil recruitment into the lung did not continue after day 2, platelets still translocated into the alveoli with similar ratios of WT to DsRed⁺ platelets in the blood and BAL (Fig. 3, B and C). Analysis of the platelet activation status in the BAL revealed platelets in the BAL to, at least in part, express P-selectin on their surface (Fig. S2 H). In a reciprocal experiment, we injected isolated DsRed⁺ platelets i.v. into WT mice before induction of pneumonia and found progressive decline in the BAL fluid up to day 5 (Fig. 3 C). We next transferred platelets (10⁸ to 5 × 10⁸ platelets/mouse) into platelet-depleted mice to determine whether this could rescue normal leukocyte dynamics. Platelet-reconstituted Pf4^iDTR mice displayed a concentration-dependent rescue of pulmonary T reg cell recruitment (Fig. 3 D). The rationale for choosing day 3 for the platelet reconstitution was not to interfere with the role of platelets during the initiation phase of pulmonary inflammation or with the administration of platelet-depleting antibodies at day 2 after induction of pneumonia.

T reg cell recruitment coincided with the appearance of anti-inflammatory cytokines IL-10 and TGFβ in the BAL starting from day 3 (Fig. 3, E and F; compare with Fig. 2 E). DT-mediated depletion of intravascular platelets in Pf4^iDTR mice, depletion of T reg cells in FoxP3^DTR mice (DEREG mice; Fig. S3 A), as well as blocking of VLA-4 (an integrin required for T reg migration; Klann et al., 2018) all prevented the increase of IL-10 in the BAL (Fig. 3 E). Importantly, depletion of T reg cells by using DEREG mice did not cause neutrophil accumulation by itself (Fig. S3 B); however, upon induction of inflammation, T reg cell depletion caused persistently high neutrophil numbers in the BAL and decreased neutrophil apoptosis and platelet numbers in the BAL, both of which were reversed by T reg cell reconstitution or administration of recombinant IL-10 (Fig. 3, G and H). T reg cell depletion also led to increased lung collagen content and reduced lung compliance (Fig. 3, I and J). These data demonstrate that platelets drive the recruitment of T reg cells in the alveoli, which, in turn, contributes to resolve pulmonary inflammation.

T reg cells express CD40 ligand (CD40L), which is shed and released as soluble CD40L (sCD40L). Platelets express the counter receptor CD40 binding sCD40L, causing platelet activation. The integrins α_IIbβ₃ and α₅β₁ expressed on the platelet surface then mediate platelet aggregation. We detected increased sCD40L plasma levels after pneumonia induction (Fig. 4 A). Similar sCD40L levels were detected in platelet-depleted mice, indicating that T reg cells rather than platelets themselves are the source of sCD40L in this condition (Fig. 4 A). Incubation of isolated WT platelets with sCD40L significantly increased platelet P-selectin surface expression levels, platelet fibrinogen binding, and shape change, as indicated by analysis of platelet spreading stages on serum-coated coverslips, indicating platelet activation (Fig. 4, B–D; and Fig. S3 C). Furthermore, CD40L stimulation caused mild platelet aggregation as detected by aggregometry (Fig. 4 E). Stimulation with sCD40L induced platelet–T reg cell aggregate formation in vitro, which was significantly decreased by pretreatment with a CD40-blocking antibody (Fig. 4 F). T reg cells express PSGL-1, the counter receptor for platelet P-selectin and a major determinant of platelet interaction with leukocytes (Rossaint and Zarbock, 2015; Sreeramkumar et al., 2014). Pretreatment with P-selectin or PSGL-1–blocking antibodies significantly decreased platelet–T reg cell aggregate formation (Fig. 4 F). Coculture of T reg cells and platelets increased the release of IL-10 and TGFβ, and the levels of both cytokines were reduced after blocking either P-selectin, PSGL-1, or CD40 (Fig. 4, G and H). Although supernatant from sCD40L-treated platelets contains TGFβ, the incubation of isolated T reg cells with supernatant from sCD40L-treated platelets alone did not increase IL10 production by T reg cells (Fig. 4, G and H). These data suggest that direct platelet–T reg cell interactions elicit the production of anti-inflammatory cytokines. To investigate possible indirect effects of platelets on T reg cells, we coincubated isolated T reg cells with supernatant from activated platelets. We could neither detect alterations in IL-10 and TGFβ release by T reg cells, nor alteration in PSGL-1 expression, indicating that the effect of platelets on T reg cells is through direct cell–cell interactions (Fig. 4, G and H). These data further illustrate the switch in cellular partnership of platelets from neutrophils during the early inflammatory response to T reg cells in the late resolution phase.

Consistent with the requirement for cell–cell contacts mediated by P-selectin and CD40, the number of T reg cells recruited into the alveoli was significantly reduced by blocking either receptor by using F(ab) fragments after the initial recruitment phase, and accordingly the numbers of neutrophils in the BAL remained high until day 5 (Fig. 4, I and J). These treatments also significantly reduced apoptotic neutrophils and IL-10 and TGFβ concentrations in the BAL and led to increased collagen content and decreased lung compliance (Fig. S4, A–D). Blocking CD40–CD40L interactions before infection did not significantly affect neutrophil or T reg cell accumulation at day 3, nor modulated neutrophil apoptosis or IL-10 concentrations in the BAL at day 5, indicating that CD40–CD40L interactions are not relevant for the initial immune response to pneumonia infection (Fig. S4, E–H).

We next investigated the role of platelets on the differential regulation of the T reg cell transcriptome. We induced pulmonary inflammation and isolated T reg cells alone and platelet–T reg cell aggregates from the BAL at day 5. For this purpose, the single platelet transcriptome was deducted from the platelet–T reg cell transcriptome. Direct platelet–T reg cell interactions caused the differential expression of 1,897 genes in T reg cells compared with T reg cells without bound platelets (Fig. 4, K and L). In particular, genes coding for anti-inflammatory mediators (Il10, Il11, Gdf15), GTPase-activating proteins (Arhgap25, Arhgap24), and genes regulating terminal cell maturation (Irf8, Ifi207, Pias1) and activation (Tspan31, Nfam1) in T reg cells were significantly downregulated in the absence of platelets. These findings led to the hypothesis that platelets modulate and boost the anti-inflammatory and reparative properties of T reg cells and contribute to resolution of inflammation in the lungs.

Previous reports have shown the intra-alveolar presence of platelets during pulmonary inflammation (Ortiz-Muñoz et al., 2014). We performed electron microscopy and could indeed detect platelets within the lung alveoli (Fig. 5, A–F), which was in agreement with our flow cytometry analyses (Fig. S2 D). Platelet counts in the BAL increased 5 d following pneumonia induction (Fig. 5 G). To investigate which cell type afforded platelet relocation into the alveoli, we induced bacterial pneumonia in control mice and Mrp8^iDTR mice in which DT administration led to neutrophil depletion before pneumonia induction (Fig. S3 D). Neutrophil depletion significantly decreased the number of detectable platelets in the BAL after pneumonia induction (Fig. 5 G). In contrast, T reg cell depletion in DEREG mice did not alter platelet relocation into the alveoli in the early stages of disease. Rather, reductions were seen at later times, which together indicated temporally restricted shuttling of platelets by each cell type, with T reg cells dominating at late times (Fig. 5 G). Consistent with the requirement of platelets for the immunomodulatory effect of T reg cells, we found that platelet depletion decreased TGFβ and IL-10 levels in the BAL (Fig. 5, H and I).

Finally, we visualized platelet relocation into the alveoli during pneumonia. We analyzed lung sections from mice 3 d (Fig. 5, J–L) and 5 d (Fig. 5, M–P) after pneumonia induction by confocal microscopy. By using confocal microscopy, we also detected platelet–T reg cell aggregates in the lung at day 5 after induction of pulmonary inflammation (Fig. 5 Q). Platelet accumulation in the alveolar space was significantly attenuated by prior neutrophil depletion in Mrp8^iDTR mice (Fig. 5 R). Platelet–neutrophil interactions in the lung tissue rose up by day 3, followed by a decrease on day 5 (Fig. 5 S). Together, these data demonstrate that platelets require neutrophils and T reg cells to relocate into the alveolar space at the onset and during the resolution of bacterial-induced inflammation.

Alveolar macrophages are among the first cells that detect bacterial infections and initiate an immune response (Kaur et al., 2015). Increased macrophage numbers in the BAL were detectable until at least 4–5 d after pneumonia induction, but neither platelet nor T reg cell depletion modulated macrophage numbers (Fig. 6 A). To assess whether alveolar macrophages were recruited from the blood, we injected isolated eGFP⁺ monocytes in WT mice and observed the appearance of alveolar macrophages in the lung after pneumonia induction, confirming at least in part a blood origin (Fig. S5, A–C). To assess the functional role of macrophages in the resolution phase of bacterial-induced pulmonary inflammation, macrophages were depleted by injecting chlodronate liposomes. The efficiency of macrophage depletion in the lung alveolar and interstitial space was verified to be >95%. Macrophage depletion on day 2 after the induction of pulmonary inflammation significantly increased neutrophil counts in the BAL and interstitial compartment and decreased neutrophil apoptosis on the consecutive days to a level similar to platelet depletion on day 2 after induction of pneumonia. In fact, combined macrophage and platelet depletion did not show an additional effect (Fig. S5, D–G).

Macrophages have been shown to mediate the clearance of apoptotic neutrophils in alveoli (Fadok et al., 1998; Levy and Serhan, 2014). Consistently, we found that macrophages took up eGFP⁺ neutrophils in LysM-Cre/eGFP mice, confirming that dying neutrophils are phagocytosed by macrophages during the resolution phase (data not shown). The persisting intra-alveolar accumulation of neutrophils 5 d after pneumonia induction caused by platelet depletion on day 2 could be reversed by i.t. instillation of platelets in a concentration-dependent manner, but not by cell-free platelet supernatant or 50 µl ADP alone (10 µM; Fig. 6 B). Likewise, the amount of apoptotic neutrophils after platelet depletion was increased by instilled platelets (Fig. 6 C). As i.t. instillation would be challenging in the clinical context, we also transfused platelets i.v. in platelet-depleted mice and observed decreased intra-alveolar accumulation of neutrophils and an increased percentage of apoptotic neutrophils 5 d after inducing pneumonia (Fig. 6, B and C).

To investigate the role of platelets on differential transcriptome regulation, we induced pneumonia, depleted platelets after 3 d, and isolated pro- (CD68⁺CD80⁺) and anti-inflammatory (CD68⁺CD163⁺) macrophages from the BAL at day 5. RNA sequencing (RNaseq) analysis of pro-inflammatory macrophages revealed very few genes that were differentially expressed (DE; Fig. 6, D and G). Platelet depletion predominantly led to the upregulation of genes encoding for adenosin signaling (Adora1), apoptosis inhibition (Faim2), metabolism of GTP- and GDP-binding proteins (Rab40c), and inflammatory cell surface receptors (Ptgfr, Il9r). In contrast, platelet depletion led to the differential regulation of 811 genes in anti-inflammatory macrophages (Fig. 6, E and H), with differential downregulation of genes governing inflammatory pathways (Anxa11, Rapbp1, CD86, Pkn1, RhoC) and upregulation of genes modulating cell differentiation (Fgfr1). We could identify 61 genes in pro-inflammatory macrophages and 804 genes in anti-inflammatory macrophages that were differentially regulated following platelet depletion, and seven genes (Faim2, Gm15758, Gm40649, Lgi1, Pfn4, Ss18l1, Synpr) that were differentially regulated in the same direction in both macrophage populations after platelet depletion (Fig. 6 F). Gene ontology analysis revealed the significant regulation of nine distinct pathways, including the regulation and organization of nucleus, nucleolus, cytoplasm, protein-containing complex, trans-Golgi network, and genes regulating leukocyte apoptosis. In particular, the gene Pfn4 has been linked to playing a role in cytoskeletal rearrangement and efferocytosis (Benoit et al., 2012; Wang et al., 2018), and the gene Tspan3, a member of the tetraspanin family of genes involved in the organization of the engulfment synapse (Barth et al., 2017), was downregulated on anti-inflammatory macrophages in the absence of platelets. This finding led us to the hypothesis that the presence of intra-alveolar platelets triggered neutrophil clearance by reprogramming macrophages. We isolated macrophages from BAL fluid obtained 5 d after pneumonia induction and analyzed gene expression. The pro-inflammatory phenotype was characterized by upregulated expression of Gpr18 and Fpr2, whereas EGR2 and c-Myc were characteristic of the anti-inflammatory phenotype (Jablonski et al., 2015). Macrophages isolated from platelet-depleted mice infected with Klebsiella pneumonia showed a shift toward a pro-inflammatory polarization of macrophages, which could be reversed by injecting platelets i.t. into platelet-depleted animals (Fig. 6 I). A similar pro-inflammatory shift could be detected when isolated alveolar macrophages from untreated control mice were incubated with BAL fluid from platelet-depleted mice (Fig. 6 J). Furthermore, significantly higher levels of TNFα and IFNγ could be detected in the BAL supernatant in the absence of platelets (Fig. S5, H and I). These data indicate that platelets are crucially needed for the polarization and education of anti-inflammatory macrophages during resolution.

To test the phagocytic capability of macrophages to ingest isolated, ex vivo apoptotic neutrophils, we isolated alveolar macrophages from control mice and platelet-depleted mice 5 d after pneumonia induction (Andonegui et al., 2003). The efferocytosis index of macrophages isolated from platelet-depleted mice was significantly lower than in macrophages from platelet-competent mice (Fig. 6 K). A similar effect on the efferocytosis index could be detected in isolated alveolar macrophages from untreated mice incubated with BAL fluid obtained from control mice and platelet-depleted mice 5 d after pneumonia induction (Fig. 6 L). Efferocytosis could be restored by i.t. instillation of isolated platelets into platelet-depleted mice (Fig. 6 L).

To further characterize macrophage polarization, we isolated macrophages at day 5 after pneumonia induction and detected an upregulation of anti-inflammatory genes (Jablonski et al., 2015), whereas intra-alveolar platelet depletion led to increased inflammation-associated gene expression (Fig. 7 A). Interestingly, Gpr18 and Fpr2, both well-known receptors binding pro-resolving lipid mediators, were increasingly expressed on macrophages in platelet-depleted mice. Thus, we analyzed the expression profile of pro-resolving lipid mediators at day 5 after pneumonia induction in the lung tissue of mice that received a platelet-depleting or control antibody at day 2 after pneumonia induction. Platelet depletion significantly increased the levels of resolvin D4 (RvD4), resolvin E4 (RvE4), lipoxin A4 (LXA4), and 15-epi-LXA4, whereas the levels of resolvin D6 (RvD6), leukotriene B4 (LTB4), leukotriene C4 (LTC4), leukotriene D4 (LTD4), and leukotriene E4 (LTE4) were significantly decreased compared with isotype-treated mice (Fig. 7, B–D).

To investigate whether pro-resolving macrophage polarization is driven by direct cellular interactions with T reg cells and/or platelets, or by soluble mediators, we incubated isolated macrophages with T reg cells alone or with T reg cells and platelets, or with the supernatant taken from these cells. We observed resolving-type macrophage polarization after incubation with T reg cells and ADP- or thrombin-activated platelets, or with the supernatant from a coculture of both cells but not with T reg cells alone (Fig. 8, A and B). Consistent with these findings, efferocytosis significantly increased when isolated macrophages were incubated with T reg cells and ADP- or thrombin-activated platelets, or with their supernatant (Fig. 8, C and D).

To gain mechanistic insights into the cellular crosstalk, we next cocultured T reg cells with or without activated platelets and/or macrophages and blocked protein de novo synthesis by actynomcyin. We observed that IL-10 and TGFβ were only produced by T reg cells in the presence of ADP- or thrombin-activated platelets (Fig. 9, A and B). These data indicate that the simultaneous presence of T reg cells and activated platelets is necessary to produce IL-10 and TGFβ de novo, which, in turn, mediate the polarization of macrophages toward a pro-resolving phenotype. These data indicate that T reg cells respond transcriptionally to platelet binding.

Finally, to further analyze the source of IL-10 and TGFβ, we analyzed Il10 and Tgfb1 mRNA expression in isolated T reg cells and macrophages and observed a higher expression in T reg cells (Fig. 9 C). Blocking both cytokines led to persistently higher neutrophil counts and decreased neutrophil apoptosis in the BAL (Fig. 9, D and E). In contrast, the numbers of T reg cells and platelets in the BAL were not affected by IL-10 and TGFβ blockade (Fig. 9, F and G), altogether demonstrating that T reg cell–platelet coupling elicits the anti-inflammatory, pro-resolving fate of alveolar macrophages to terminate inflammation and protect the host.

Neutrophil apoptosis and macrophage education is the major route to clear neutrophils and maintain homeostasis (Bratton and Henson, 2011). Macrophage phagocytosis not only prevents secondary necrosis of apoptotic neutrophils and neutrophil debris and pro-inflammatory signal release, but also polarizes macrophages toward an anti-inflammatory phenotype (Kennedy and DeLeo, 2009). Here, we demonstrate that platelets, which have been involved in the onset of inflammation, are equally important in the resolution phase by mediating T reg cell recruitment into the alveoli, a process that is dependent on CD40–CD40L and P-selectin/PSGL-1 interactions. Furthermore, platelets and T reg cells cooperate in the resolution process by orchestrating macrophage polarization toward a reparative phenotype and increase the efferocytosis capacity of macrophages in the alveoli needed for the clearance of residual neutrophils.

Several studies have provided evidence that platelets appear in the lung alveoli (Pitchford et al., 2008; Zarbock et al., 2006). Furthermore, platelets eventually coupled to leukocytes can be detected in the BAL after induction of pulmonary inflammation (Amison et al., 2018; Lax et al., 2017; Ortiz-Muñoz et al., 2014). This is well in accordance with our findings that platelets translocate into the alveoli during the inflammatory phase and during the resolution process following pulmonary inflammation. Our study demonstrates for the first time that platelets get recruited to the lung during the onset of resolution, together with T reg cells. Platelets were long thought to be passive blood cells reaching their site of action by chance, enforced by their sheer numbers (Herter et al., 2014). Contradicting this dogma, platelets are now known to be capable of active migration (Gaertner et al., 2017). Interestingly, platelet recruitment into the lung vasculature and tissue does not entirely rely on the distinct mechanisms that mediate pulmonary neutrophil accumulation together with platelets but also appear to exist independently (Cleary et al., 2019). This fits well with our observation that the distribution of platelets is not only restricted to the intravascular compartment, but that platelets also translocate toward the alveoli.

Previous studies have suggested that platelets may not only be important for the propagation of vascular inflammation—for example, platelet-activating factor plays a role in mediating urate crystal uptake during the resolution of gouty inflammation (Yagnik, 2014). Platelets are also a major source for anti-inflammatory mediators of the lipoxin family (e.g., specialized pro-resolving mediators, such as resolvins, protectins, and maresins; Yadav and Kor, 2015). These lipoxins are produced and released already during the inflammatory onset phase of acute inflammation, and their concentrations sharply rise during the convergence toward the resolution phase (Haworth et al., 2008). Interestingly, these lipoxins also promote phagocytic clearance of apoptotic immune cells during resolution (Mitchell et al., 2002). This evidence aligns with our observations that platelets and T reg cells in the alveoli are needed to induce a shift toward the anti-inflammatory macrophage phenotype to support the clearance of inflammatory immune cells and resolution. Interestingly, expression of two major receptors (Gpr18 and Fpr2) for pro-resolving lipid mediators was found to be increased on macrophages in the presence of platelets. Likewise, major resolvin-class lipid mediators were found to be increased on day 5 in platelet-depleted animals in which resolution appeared to be impaired. One possible explanation for this finding may be an as-yet-unknown feedback mechanism that leads to a prolonged anti-inflammatory lipid mediator profile in the lung to compensate for the loss of the pro-resolving action of platelets during this stage. Furthermore, it is also possible that these particular mediators are not involved in the resolution process. However, it has to be acknowledged that the analysis of lipid pro- and anti-inflammatory mediators on day 5 in this study only represents a snapshot and warrants further experiments to better characterize the underlying molecular mechanisms in more detail.

Platelets have been previously described to interact with T reg cells under inflammatory conditions. It was shown that platelets are needed to control the anti-inflammatory actions of CD4⁺ T reg cells following burn injury (Bergmann et al., 2016). These findings are in accordance with ours, including the time periods (2–7 d) after which platelets regulate T reg cell–mediated responses. Beyond the lungs, platelets have also been shown to interact with CD4⁺ T cells in the liver following ischemic injury and during atherosclerosis (Khandoga et al., 2006; Li et al., 2013). Moreover, the CD40/CD40L axis is crucial for the interplay between platelets and T reg cells in atherosclerosis (Lievens et al., 2010). These findings again agree with the observations made here showing that platelet P-selectin, PSGL-1 on T reg cells, and the CD40/CD40L axis are needed for platelet–T reg cell complex formation as well as recruitment of these cells into the alveolar space during the resolution phase. We observed a time-dependent transition in platelet binding from neutrophils to T reg cells. We show that initial platelet binding to neutrophils appears to be diminished over time by decreasing surface expression of PSGL-1, a phenomenon mediated by the sheddase ADAM8, leading to decreased neutrophil–platelet complex formation, as reported previously (Davenpeck et al., 2000; Domínguez-Luis et al., 2011). Mirroring the response in neutrophils, T reg cells upregulate PSGL-1 on the cell surface during the resolution phase that enables physical complexing with platelets. In this regard, it is interesting that platelets are thought to be capable of inducing CD4⁺ T cell differentiation by both the release of distinct cytokines and by direct cell–cell contact with T cells increasing IL-10 production and release by T cells (Gerdes et al., 2011). Indeed, we observed that the interaction of platelets and T reg cells is necessary for cytokine production (IL-10 and TGFβ) by T reg cells, and that these are the mediators that induce macrophage polarization to enforce alveolar immune cell clearance.

In has recently become apparent that innate immune cells are constantly being reprogrammed in order to adjust their phenotypic behavior to the environmental needs (Adrover et al., 2020). While this concept is relatively new to neutrophils and T reg cells, macrophage reprogramming in response to various inflammatory or tissue conditions has long been known (Michaeloudes et al., 2020). This is the first study to demonstrate that direct platelet–T reg cell interactions modulate the T reg cell transcriptome, induce the secretion of cytokines that, in turn, reprogram macrophages for resolution. Further, we observed differential regulation of genes modulating efferocytosis only in anti-inflammatory macrophages, a finding in line with previously reports (Benoit et al., 2012; Wang et al., 2018).

T reg cell differentiation and turnover is critically dependent on dendritic cells (DCs). It has been shown that the suppression of the T reg cell/IL-10 axis of lung injury is DC dependent (Kapur et al., 2017a), yet the role of DCs on the regulation of the platelet–T reg cell interplay during the resolution of pulmonary inflammation remains unknown and warrants further research.

In summary, we have shown that platelets are recruited to the pulmonary alveoli and are actively involved in resolving inflammation, together with T reg cells. This interaction triggers a phenotypic macrophage shift toward a reparative phenotype and prevents excessive pulmonary injury after an infection. These findings are of clinical importance, as antiplatelet drugs are common in the clinic. Thus, a deeper understanding of the role of platelets in these situations may provide new opportunities to develop and adapt new treatment strategies to improve patient outcomes after pulmonary infectious and inflammatory diseases.

We used 8–12-wk-old male C57BL/6 mice. Mice were kept in a barrier facility under specific pathogen–free conditions. Unless otherwise stated, all reagents were obtained from Sigma-Aldrich. We used Pf4^iDTR, Mrp8^iDTR, and DEREG mice on a C57BL/6 background to deplete platelets, neutrophils, or T reg cells, respectively, by administration of 400 ng DT, as previously described (Buch et al., 2005; Lahl and Sparwasser, 2011; Wuescher et al., 2015). If not indicated otherwise, littermates were used as controls. All animal experiments were approved by the animal research committee of the Landesamt für Natur, Umwelt und Verbraucherschutz.

We used a murine pneumonia model as published by Ittner et al. (2012), with some modifications. Overnight cultures (37°C) of of K. pneumoniae (strain 13883; American Type Culture Collection) were grown in tryptic soy medium, washed, and resuspended in sterile saline (0.9%). Mice were anaesthetized by intraperitoneal injection of ketamine (125 µg⋅g^-1 body weight; Pfizer) and xylazine (12.5 µg⋅g⁻¹ body weight; Bayer). Animals were challenged with 1.5 × 10⁷ viable K. pneumoniae per mouse. At this inoculation dose, all mice survived the observation period. After the indicated time points, mice were sacrificed and the lungs were lavaged five times with 0.7 ml physiological saline solution. The number of leukocytes in the BAL was counted using kimura staining with an improved Neubauer counting chamber and an inverted cell culture microscope (Primovert; Carl Zeiss) equipped with a 10×/0.75 NA objective. A total of four fields with 16 standardized subfields in each individual field were counted for each sample. CFUs in the lung, blood, and spleen were counted by serial plating on tryptic soy agar plates (Ittner et al., 2012).

Analysis of viable lung sections was performed as described before (Hasenberg et al., 2011). Mice were injected i.t. with K. pneumoniae. After 4 h, animals were injected with Alexa Fluor 488–coupled anti-Gr1 antibody (clone RB5-8C6; 5 µg/mouse, purified from hybridoma supernatant), Alexa Fluor 568–coupled anti-PECAM antibody (clone 390; 50 µg/mouse; BD Biosciences), and BV421-coupled anti-CD41 antibody (clone MWReg30; 5 µg/mouse; BioLegend) to stain neutrophils, endothelial cells, and platelets. Mice were sacrificed and lungs were filled with 1 ml of low-melting agarose. After removal, lungs were cut using a vibratome. Lungs were fixed in a cell culture dish and submersed in PBS and z-stacks were recorded using a spinning disc confocal microscope (CellObserver SD; Carl Zeiss) equipped with a 20×/1.0 NA objective.

Approximately 10⁵ isolated macrophages were seeded in a 24-well plate. Macrophages were incubated with 5-chloromethylfluorescein diacetate (CMFDA)–labeled neutrophils at a ratio of 1:1 and incubated at 37°C for 1 h. Excessive extracellular fluorescence was quenched with Trypan blue (0.04% in PBS). The proportion of macrophages (CD11b⁺F4/80⁺) with increased CMFDA fluorescence (indicating uptake of fluorescent-labeled apoptotic cells) was analyzed by flow cytometry on a FACSCantoII (BD Biosciences). The efferocytosis index was calculated as the percentage of macrophages showing increased CMFDA fluorescence as an indicator of neutrophil ingestion.

Apoptotic neutrophils in the BAL fluid were analyzed by flow cytometry on a FacsCantoII (BD Biosystems) using an Annexin V staining. Macrophages in the BAL were analyzed by flow cytometry by an F4/80 antibody, and T reg cells were identified based on their expression of CD4/CD25/FoxP3. Specific neutrophil accumulation in the intravascular, interstitial, and alveolar compartment was analyzed by a flow cytometry–based technique as described before (Reutershan et al., 2005). Neutrophils were depleted in some mice by injection of anti-Ly6G antibody (clone 1A8, 200 µg/mouse i.p.; BioLegend). Platelets were depleted in some mice by i.v. injection of a platelet-depleting antibody (clone R300; 2 µg/g bodyweight; Emfret). T reg cells were depleted in some mice by i.v. injection of an α-CD25 antibody (clone PC61; BioLegend; Setiady et al., 2010). Some mice were injected with blocking antibodies against P-selectin (clone RB40.34; 50 µg/mouse, purified from hybridoma supernatant) or CD40 (clone HM40-3; 50 µg/mouse; BioLegend).

Recruitment of neutrophils into the intravascular and interstitial compartment of the lung was determined as described previously (Zarbock et al., 2006). Briefly, Alexa Fluor 633–labeled Gr1 antibody (clone RB6-8C5, staining kit; Invitrogen) to murine neutrophils was injected i.v. 5 min before euthanasia. This procedure labels only intravascular neutrophils, as the time period is not sufficient for the antibody to react with neutrophils located in the interstitial or intra-alveolar compartment. Thoracotomy was performed, the inferior vena cava was opened, and nonadherent neutrophils were removed from the pulmonary vasculature by perfusing the right ventricle with 3 ml of saline with a constant pressure of 25 cmH₂O. After collection of the BAL, lungs were removed, minced, and digested with an enzyme cocktail (collagenase type XI, hyaluronidase type I-s, DNase1; Sigma-Aldrich) containing 65 µg of unlabeled Gr1 antibody at 37°C for 60 min. A cell suspension was made by passing the digested lung tissue through a 70-µm cell strainer (BD Falcon). RBCs were lysed and the remaining leukocytes were resuspended in PBS and counted. The fraction of neutrophils in the suspension was determined by flow cytometry (FacsCantoII; BD Biosciences). Neutrophils were identified by their typical appearance in the forward/side scatter and their expression of CD45 (clone 30-F11), 7/4 (clone 7/4; both BD PharMingen), and Gr1 (clone RB6-8C5, purified from hybridoma supernatant). The labeled Gr1 antibody was used to differentiate between intravascular (CD45⁺7/4⁺Gr1⁺) and interstitial (CD45⁺7/4⁺Gr1⁻) neutrophils.

To quantify the collagen content in the lung (as a measure for pulmonary fibrosis), the right lung lobes were excised, weighted, and homogenized. The tissue was hydrolyzed by addition of 12 N HCL and incubation at 120°C for 3 h. Samples were subsequently incubated with 4-dimethylaminobenzaldehyde for 90 min at 60°C. The absorbance of oxidized hydroxyproline was analyzed at 560 nm using a photometer.

For the measurement of lung compliance, an endotracheal tube was placed by tracheotomy and connected to a precision syringe connected to a pressure gauge. Static lung compliance was assessed by measuring the required inspiratory pressure to inflate the lung by an inspiratory volume of 1 ml room air and lung compliance was calculated (in mbar/ml).

Whole blood samples were withdrawn from mice and stained with Alexa Fluor 633–coupled anti-Gr1 antibody (clone RB6-8C5, purified from hybridoma supernatant), PE-coupled anti-CD41 antibody (clone MWReg30; BD Biosciences), FITC-coupled anti-Ly6B2 antibody (clone 7/4; AbD Serotec), and PerCP-coupled anti-CD45 antibody (clone 30-F11; BD Biosciences). Platelet–neutrophil aggregates were quantified by measuring the percentage of CD41⁺ neutrophils (CD45⁺Gr1⁺7/4⁺) using a flow cytometer (BD FACSCantoII; BD Biosciences). To analyze aggregates of platelets and T reg cells, whole blood samples were stained with Alexa Fluor 633–coupled anti-CD4 antibody (clone GK1.5; BioLegend), PE-coupled anti-CD41 antibody (clone MWReg30; BD Biosciences), FITC-coupled anti-CD25 antibody (clone PC61; BioLegend), and PerCP-coupled anti-CD45 antibody (clone 30-F11; BD Biosciences). Platelet–T reg cell aggregates were quantified using a flow cytometer (BD FACSCantoII; BD Biosciences).

Serum and coculture supernatant levels of IL-10, TGFβ, and CD40 ligand were analyzed by commercially available ELISA sets following the manufacturer’s instructions (R&D Systems).

Macrophages were isolated from BAL as described (Andonegui et al., 2003), adherence purified, and cultured in RPMI supplemented with 10% FCS, 1% penicillin, streptomycin, and fungizone.

Femurs and tibias of mice were dissected, and the marrow was flushed with PBS and layered on a three-step Percoll gradient (72%, 64%, and 52%) and centrifuged at 1,060 ×g for 30 min.

Citrate-anticoagulated whole blood samples were withdrawn from donor mice. Platelet-rich plasma (PRP) was separated by from whole blood by repetitive slow-speed centrifugation at 68 ×g for 8 min. Platelets were isolated from PRP by centrifugation at 1088 ×g for 5 min. Washed platelets were resuspended in Tyrode’s buffer and kept at 37°C before use. Platelet purity after isolation was >97%. Before i.t. injection, some platelets were stimulated with 10 µM ADP and injected i.t. in 50 µl Tyrode’s buffer.

T reg cells were isolated from single-cell suspensions of the mouse spleen by using the CD4⁺CD25⁺ Regulatory T Cell Isolation Kit (Miltenyi Biotec).

For electron microscopy, the lung was perfused with PBS via the right ventricle, followed by 2% glutaraldehyde, 2% paraformaldeyde in 0.2 M cacodylate buffer (pH 7.4). The lung was removed and small samples were further fixed under low vacuum until they settled down on the bottom. The specimen was postfixed with 1% osmiumtetroxide and 1.5% potassium ferrocyanide, dehydrated, and embedded in epon. 60-nm ultrathin sections were cut (Leica UC6 ultramicrotome) and counterstained with uranyl acetate and lead. Samples were inspected on a transmission electron microscope at 80 kV (Fei-Tecnai12; FEI) and pictured with a CCD camera (Megaview; SIS).

BAL fluid was obtained from mice 5 d after induction of pneumonia by flushing the lungs with 5× ice-cold PBS/0.6 mM EDTA (Mg and Ca free). After centrifugation (450 ×g, 10 min), cells were resuspended in RPMI and allowed to adhere to the bottom of plastic 6-well plates during cultivation at 37°C/5% CO₂ for 45 min. Adherent macrophages were harvested after removal of supernatant. After RNA extraction using Trizol reagent and cDNA transcription (cDNA synthesis kit; Sigma-Aldrich), gene expression was analyzed by quantitative real-time PCR (qRT-PCR) using SYBR Green Mastermix (Life Technologies) on an Applied Biosystems 7600 Real-Time PCR device. The primer sequences were as follows: Gpr18: forward, 5′-GAC​AGA​CAG​GAG​GTT​CGA​CAT​ACA-3′; reverse, 3′-TGT​ATT​CCT​CTG​GGT​GTG​GAG​CCA-5′; Fpr2: forward, 5′-TCT​ACC​ATC​TCC​AGA​GTT​CTG​TGG-3′; reverse, 5′-TTA​CAT​CTA​CCA​CAA​TGT​GAA​CTA-3′; Egr2: forward, 5′-CCT​TTG​ACC​AGA​TGA​ACG​GAG​TG-3′; reverse, 5′-CTG​GTT​TCT​AGG​TGC​AGA​GAT​GG-3′; and cMyc: forward, 5′-CTT​CTC​TCC​GTC​CTC​GGA​TTC​T-3′; reverse, 5′-GAA​GGT​GAT​CCA​GAC​TCT​GAC​CTT-3′. Data were normalized to GAPDH (forward, 5′-GGC​AAA​TTC​AAC​GGC​ACA​GT-3′; reverse, 5′-GAT​GGT​GAT​GGG​CTT​CCC-3′).

RNA was extracted from sorted cells using an RNeasy Mini Kit (Qiagen). Total RNA samples were mixed with oligo-dT and dNTPs, incubated at 72°C, and immediately put back on ice. Reverse transcription into cDNA was performed based on polyA tail, followed by template switch at the 5′ end of the RNA and amplification of the full-length cDNA by PCR. The average molecule length was analyzed using the Agilent Technologies 2100 bioanalyzer instrument. For library construction, PCR products were purified and selected with the Agencourt AMPure XP-Medium kit. DNA was quantified by Agilent Technologies 2100 bioanalyzer. Double-stranded PCR products were the heat denatured and circularized by the splint oligo sequence. Single-stranded circle DNA was formatted as the final library. The library was amplified to make DNA nanoball, which had more than 300 copies of one molecule. DNA nanoballs were loaded into the patterned nanoarray and single-end 50-base reads were generated in the way of sequenced by combinatorial Probe-Anchor Synthesis. The RNaseq data underwent thorough quality control with FastQC and MultiQC, (Ewels et al., 2016) before sequencing adapter trimming was performed with trim-galore. Read mapping and quantification was performed with Salmon (Patro et al., 2017) using the Gencode (Frankish et al., 2019) mouse reference transcriptome vM23. The Salmon quasi-index was created with a default k-mer length of 31 and the Gencode flag. Subsequent quantification was performed while accounting for sequence-specific biases, like random hexamer priming, which often results in lower base-call quality of the first few bases of a read. Additional parameters included --validateMappings and --rangeFactorizationBins 4 to potentially improve the quantification accuracy. Additionally, the --gcBias flag was used to correct for the slightly skewed GC content of the reads observed in the quality control step. The transcript counts were summarized to gene level with tximport (Soneson et al., 2015) and supplementary annotation data from Ensembl (Zerbino et al., 2018) through the biomaRt package (Durinck et al., 2005; Durinck et al., 2009). During this summarization step, allosomal and mitochondrial genes were excluded to decrease (sex-specific) biases. Differential expression analysis was performed by DESeq2 (Love et al., 2014) on gene-level counts. The reported log fold changes were shrunk with apeglm (Zhu et al., 2019) Gene-set enrichment analysis was performed for the DE genes with goseq (Young et al., 2010) both for Gene Ontology (Ashburner et al., 2000; The Gene Ontology Consortium, 2019) terms and Kyoto Encyclopedia of Genes and Genomes (Kanehisa and Goto, 2000; Kanehisa et al., 2019) pathways. The P values of the enrichment analyses were multiple testing corrected and a significance threshold of false discovery rate < 0.05 was used to determine the terms and pathways with a significantly altered number of DE genes. The transcriptomic data in this publication have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus and are accessible through accession no. GSE171989. All original data are accessible through the corresponding author upon reasonable request.

Samples were extracted and lipid mediators were identified and quantified as described (Gomez et al., 2020). In brief, tissues were placed in 1 ml of ice-cold methanol containing deuterated internal standards (d₈-5S-HETE, d₄-LTB₄, d₅-LXA₄, d₄-PGE₂, d₅-RvD2, d₅-MaR1, d₅-MaR2, d₅-RvD3, d₄-RvE1, d₅-17R-RvD1, d₅-LTC₄, d₅-LTD₄, and d₅-LTE₄) representing each chromatographic region of identified lipid mediator. Following protein precipitation (−20°C for a minimum of 45 min), supernatants were extracted on an ExtraHera instrument (Biotage) using solid-phase extraction with Isolute C18 500-mg columns (Biotage). Methyl formate and methanol fractions were collected, brought to dryness, and resuspended in phase (methanol:water, 1:1 vol/vol) for injection on a Shimadzu LC-20AD HPLC and a Shimadzu SIL-20AC autoinjector, paired with a QTrap 5500 or QTrap 6500+ (Sciex). In the analysis of mediators eluted in the methyl formate fraction, an Agilent Poroshell 120 EC-C18 column (100 mm × 4.6 mm × 2.7 µm) was kept at 50°C and mediators eluted using a mobile phase consisting of methanol:water:acetic acid of 20:80:0.01 (vol/vol/vol) that was ramped to 50:50:0.01 (vol/vol/vol) over 0.5 min and then to 80:20:0.01 (vol/vol/vol) from 2 min to 11 min, maintained until 14.5 min and then rapidly ramped to 98:2:0.01 (vol/vol/vol) for the next 0.1 min. This was subsequently maintained at 98:2:0.01 (vol/vol/vol) for 5.4 min, and the flow rate was maintained at 0.5 ml/min. QTrap 5500 was operated in negative ionization mode using a multiple reaction monitoring method. In the analysis of mediators eluted in the methanol fraction, an Agilent Poroshell 120 EC-C18 column (100 mm × 4.6 mm × 2.7 µm) was kept at 50°C and mediators were eluted using a mobile phase consisting of methanol:water:acetic acid 55:45:0.5 (vol/vol/vol) over 5 min, ramped to 80:20:0.5 (vol/vol/vol) for 2 min, maintained at 80:20:0.5 (vol/vol/vol) for the successive 3 min, and ramped to 98:2:0.5 (vol/vol/vol) over 3 min. This condition was kept for 3 min. QTrap 6500+ was operated in positive ionization mode using a multiple reaction monitoring method. Each lipid mediator was identified using established criteria, including: (1) the presence of a peak with a minimum area of 2,000 counts, (2) matching retention time to synthetic or authentic standards, (3) four or more data points, and (4) matching of at least six diagnostic ions to that of the reference standard, with a minimum of one backbone fragment being identified. Calibration curves were obtained for each mediator using synthetic compound mixtures at 0.78, 1.56, 3.12, 6.25, 12.5, 25, 50, 100, and 200 pg that gave linear calibration curves with r² values of 0.98–0.99. The results of the lipid mediator quantification in mouse lung tissue are presented in Table S1. The original raw data are accessible through the corresponding author upon reasonable request.

Statistical analysis was performed with SPSS (version 22.0) using Mann-Whitney test where appropriate. More than two groups were compared using one-way ANOVA followed by Bonferroni testing. Data distribution was assessed using Kolmogorov-Smirnov test or Shapiro-Wilks test. All data are represented as whisker box pots with 25th to 75th percentile (interquartile ranges). A P value <0.05 was considered significant. For in vivo experiments, the provided n is the number of animals used per experiment.

Fig. S1, related to Fig. 1, shows MIP2⍺ serum levels and systemic platelet and leukocyte count after injection of platelet-depleting antibody, platelet depletion scheme, erythrocytes, and white blood cell counts in peripheral blood of PF4^+/+iDTR mice after control and DT treatment, survival analysis, and lung wet/dry ratios. Fig. S2, related to Figs. 1, 2, 3, and 5, shows exemplary flow cytometry plots and gating strategy of lung homogenates, platelet-neutrophil aggregates, apoptosis rates in CD41^neg (without bound platelet), and CD41^pos (with bound platelet) neutrophils and P-selectin surface expression on platelets. Fig. S3, related to Figs. 3, 4, and 5, shows a scheme of T reg cell depletion and T reg cell and PMN counts in DEREG mice, exemplary images from platelet-spreading experiments, and a scheme of PMN depletion and PMN counts in Mrp8-iDTR mice. Fig. S4, related to Fig. 4, shows data from bacterial pneumonia in CD40L^−/− mice and mice treated with blocking antibodies against CD40 and P-selectin. Fig. S5, related to Fig. 6, shows monocyte recruitment from the circulation into the lung during resolution, data on the effect of macrophage depletion on the resolution of bacterial pneumonia, and concentrations of TNF⍺ and IFNγ in the BAL supernatant. Table S1 lists the results of the lipid mediator quantification in mouse lung tissue.

This work was supported by Deutsche Forschungsgemeinschaft grants ZA428/14-1, INST 211/604-2, Za2/001/18, RO4537/3-1, RO4537/4-1, RO4537/5-1, and KFO342 and the German-Israeli Foundation for Scientific Research and Development (grant I-1470-412.13/2018).

Author contributions: J. Rossaint: Conceived of the study, performed the experiments, analyzed the data, and wrote the manuscript; K. Thomas, S. Mersmann, J. Skupski, and A. Margraf: Performed experiments; T. Tekath: Performed biostatistical analysis; C. C. Jouvene and J. Dalli: Performed lipid mediator analysis. A. Hidalgo, S.G. Meuth, O. Soehnlein, and A. Zarbock: Conceived of the study, analyzed the data, and contributed to writing the manuscript.

The transcriptomic data in this publication have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus and are accessible through accession no. GSE171989. All original data are accessible through the corresponding author upon reasonable request.