Resolution of acute inflammation bridges the gap between innate and adaptive immunity

Acute inflammation is characterized by the immediate and sequential release of proinflammatory mediators resulting in the influx of polymorphonuclear leukocytes (PMNs). This early onset phase is followed by phagocytosing macrophage (Mφs) leading to leukocyte clearance and resolution.¹  Although research has traditionally focused on identifying factors that drive inflammation, emphasis has now shifted toward this latter phase of resolution to understand how immune-mediated responses switch off. Results from these studies have advanced our understanding of PMN trafficking, efferocytosis, and proinflammatory leukocyte clearance, as well as immune-suppressive eicosanoids, specialized immune-regulatory cells, and cytokine catabolism.^2-4  Such pathways terminate acute inflammatory responses and contribute to the notion that chronic inflammation/autoimmunity is avoided while homeostasis is reinstated.¹  However, we now show that these sequential and overlapping events are only part of the pathophysiological importance of resolution and that resolution creates a microenvironment conducive for the optimal development of adaptive immunity. Moreover, we present data showing that months after resolution has occurred tissues do not revert back to their preinflamed state in terms of cellular composition and phenotype. Instead, a state of “adapted homeostasis” is achieved, which impacts the severity of subsequent inflammatory responses.

Thus, after onset and resolution, we now introduce a third, postresolution phase to the acute inflammatory response dominated by macrophages and lymphocytes. These data provide a new perspective on innate immunity by highlighting the significance of proresolution processes in establishing adaptive immunity and in long-term tissue reprogramming.

Male C57Bl6/J, inducible nitric oxide synthase (iNOS)^−/−, and ccr2^−/−5 were obtained from Jackson Laboratories, whereas CX₃CR1^GFP/+6 mice were a gift from Sussan Nourshargh, The William Harvey Research Institute. Mice were maintained in accordance with United Kingdom Home Office regulations. Peritonitis was induced by the intraperitoneal injection of 0.1, 10 mg/mouse zymosan A or 50 000 Streptococcus pneumoniae^ova/mouse; S pneumoniae^ova was obtained from Dr. Krzystof Trzcinski and Dr Marc Lipsitch, Harvard School of Public Health. PKH26-PCL^red or PKH26-PCL^green (2 mL of 500 nM; Sigma) were injected intraperitoneally at time points indicated in “Results.” MC-21 (250 µL) was injected subcutaneously on days 3, 5, and 7, with mice analyzed on day 9 after zymosan. For adoptive transfer of tissue-resident macrophages Mφs (resMφs) from wild-type (WT) mice to iNOS^−/−, WT mice bearing a 0.1 mg of zymosan-induced peritonitis were injected with PKH26-PCL intraperitoneally at 72 hours. Three hours later, PKH26-PCL positive resMφs (gated as in “Results”) were isolated and 4 × 10⁶ injected into iNOS^−/− mice bearing a 0.1 mg of zymosan-triggered inflammation; spleen or lymph nodes were taken off on day 14 for immunohistochemistry. Delayed type hypersensitivity was established as previously described.⁵ 

Flow cytometry and cell sorting was done on the LSR-II/LSR-Fortessa and FACSAria (BD Biosciences), respectively. Cells were incubated with Fc-Blocker (AbD Serotec) and fluorescent-labeled antibodies. Data were analyzed with FlowJo 7.0.1 software (Tree Star) using fluorescence minus one controls as the reference for setting gates. Antibodies were obtained from BD Biosciences (F4/80, CD11b, CD11c, Ly6c, Ly6g, Gr1, CD3, CD19, CD4, CD8, CD25, FoxP3, CD62l, CD44, CD115, major histocompatibility complex (MHC)-II, Siglec-F, CD117, and CD49d). Immunohistochemistry was performed on frozen sections fixed in cold acetone for 10 minutes before staining.

Human PMNs were labeled with 5 µM carboxyfluorescein succinimidyl ester (CFSE) (Molecular Probes; Life Technologies) and aged for 48 hours (80% apoptotic, as determined by annexin V/PI labeling). The 5 × 10⁶ CFSE-labeled apoptotic PMNs were injected intraperitoneally into mice bearing a 24-hour or 72-hour peritonitis elicited by 0.1 mg of zymosan; these mice were previously injected with PKH26-PCLK^red to distinguish resident from infiltrating monocyte-derived Mφs (moMφs), as detailed in “Results.” Mice were killed after 3 hours and exudates were collected for flow cytometry.

Spleens were taken from C57/Blk6J, crushed, and passed through a 70-µm cell strainer followed by a 35-µm cell strainer. Red blood cells were lysed in ACK lysis buffer (Enzo Life Sciences). Spleenocytes were purified using Miltenyi Biotec CD4⁺ T-cell isolation kit II and labeled with 5-µM CFSE (Molecular Probes, Life Technologies). CD4⁺/CFSE⁺ cells were seeded at 200 000 cells/well in RPMI 1640 containing penicillin/streptomycin, 10% fetal bovine serum, 2 mM l-glutamine (Life Technologies), 30 U/mL rIL-2 (Miltenyi), and 200 000 CD3/28 beads (T-cell activation/expansion kit; Miltenyi). Then 60 000 resMφs or myeloid-derived suppressor cells (MDSCs) were added. After 3 days, lymphocytes were stained with 10 µg/mL 7AAD and proliferation measured using the LSRFortessa.

Transgenic CD4⁺ T cells expressing I-Ab restricted T-cell receptor recognizing tyrosinase related protein-1 (Trp-1) were isolated by magnetic bead sorting from Treg-depleted Foxp3-DTR-Trp-1 transgenic mice previously treated with diphtheria toxin. They were cocultured with moMφs, resMφs, or monocyte-derived dendritic cells (moDCs) for 5 days with Trp-1 peptide (SGHNCGTCRPGWRGAACNQKILTVR) + transforming growth factor (TGF)-β and analyzed for the presence of Foxp3⁺ cells by flow cytometry.

Bone marrow was flushed using a 26 g needle with Dulbecco’s modified Eagle medium + l-glutamine, fetal bovine serum, penicillin/streptomycin and N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid. Cells were filtered and seeded onto a petri dish at 1 × 10⁶ cells/mL in the above media supplemented with 20 ng granulocyte macrophage–colony stimulating factor and IL-4; media and growth factors were replenished on day 4. On day 7, dendritic cells were plated at 60 000 cells/well. At 60 000, moMφ, resMφ, and moDCs (from day 9 on 0.1 mg of zymosan) were sorted according to the gating strategy described in “Results,” and along with bone marrow-derived dendritic cells, were pulsed overnight with 20 nM ovalbumin (OVA)^323-339 and 100 ng lipopolysaccharide (Sigma). These cells were then washed and incubated with CFSE-labeled CD4 T cells. After 4 days cells were stained with anti-CD4 and 7AAD and processed on the LSRFortessa. For the determination of lymphocyte proliferation from resolving bacterial peritonitis, sorted cells were pulsed overnight with 100 ng/mL LPS only, whereas bone marrow-derived dendritic cells were pulsed with 100 ng/mL LPS and OVA.

Low-dose zymosan (0.1 mg) caused a transient peritonitis peaking at around 12 hours (onset) followed by cell clearance (Figure 1A) . In contrast, injecting 10 mg of zymosan caused a more severe response that lasted up to day 5 (Figure 1A) and was associated with systemic inflammation.⁶  Alhough inflammation after 0.1 mg of zymosan had resolved (as defined by reduced total cell numbers) from day 6, there was another wave of cell influx into the peritoneum that was not observed after 10 mg of zymosan (Figure 1A). Polychromatic flow cytometry analysis revealed Ly6c^int/F4-80^int cells from 24 hours, followed by F40/80^hi Mφs from day 6 (Figure 1B). In resolving inflammation, we also noticed a population of cells expressing Ly6g and/or CD11b (Figure 1C), comprising a mixture of PMNs and eosinophils (region A), eosinophils (region B), F4/80^hi monocytic cells (region C), and mature PMNs (region D). Cells in region C persisted throughout resolution, and possessed abundant cytoplasm and a ring-like nuclear structure⁷  (Figure 1D), which along with their ability to inhibit T-cell proliferation (Figure 1E) and cell surface markers (supplemental Figure 1, available on the Blood Web site) suggest that they are MDSCs. In contrast, in response to 10 mg of zymosan, monocyte/macrophage populations appeared more homogeneous, expressing Ly6c⁺/F4-80⁺ and F4/80⁺ from day 6 onward (Figure 1F), with a phenotype contrasting dramatically with Mφs and MDSC that infiltrated into postresolving tissues (supplemental Figure 2). The temporal profiles of monocyte/Mφ populations andPMNs are shown in Figure 1G-H.

We also detected an increase in lymphocytes in the resolving peritoneum (0.1 mg of zymosan). Importantly, these cells were quantitatively fewer in inflammation driven by 10 mg of zymosan (Figure 1I-J). Using the gating strategy in supplemental Figure 3A, we detected an increase in monocyte-derived cells (moMφs and moDCs) in mesenteric lymph nodes between day 6 and day 13 (supplemental Figure 3B). These cells were classified as R1 (CD11c⁺ most likely plasmacytoid dendritic cells [pDCs]), R2 (CD11c⁺/CD11b⁺ resident and migratory DCs), R3 (CD11b Mφs), and R4 (monocytes). Lymph node expansion (supplemental Figure 3B) was associated with increased peripheral blood proliferating T and B lymphocyte numbers (supplemental Figure 3C), along with an accumulation of these cells in the peritoneum including Tregs, memory (CD44^hi/CD62^lo), and effector (CD44^lo/CD62^hi) CD4 and CD8 T cells and populations of B1a-, B1b- and B2-B cells (supplemental Figure 3D).

Hence, resolving inflammation is now characterized into 3 phases including onset (up to 12 hours), resolution (24-72 hours), and a third postresolution phase of monocytes/Mφ and MDSCs alongside lymphocyte infiltration occurring from day 6.

We used the gating strategy in supplemental Figure 4 to distinguish Mφs from dendritic cells (DCs) in the naïve peritoneum.⁸  Naive mice were injected with PKH26-PCL^red, which preferentially labeled more than 70% of peritoneal DCs and 99.1% CD11b^hi/F4-80^hi positive, so-called large peritoneal Mφs, Figure 2Ai-ii, respectively. Then 0.1 mg of zymosan was injected into mice given PKH26-PCL^red 1 hour earlier and inflammation was allowed to progress for 72 hours. Only 3.91% of PKH26-PCL^red-positive DCs were found in the cavity at this time (Figure 2Aiii), indicating that these cells disappeared upon stimulation. However, there was the appearance of PKH26-PCL^red-negative (96.1%) DCs, suggesting the infiltration of blood-derived DCs (Figure 2Aiii). In contrast, approximately 30% of CD11b^hi/F4-80^hi Mφs labeled positively for PKH26-PCL^red indicating that these cells were peritoneal-resident Mφs and the remaining 71.1% unlabeled cells were moMφs (Figure 2Aiv). To confirm this, zymosan was injected into CX₃CR1^gfp/+ mice whose peritoneal Mφs were pre-labeled with PKH26-PCL^red. In heterozygous CX₃CR^gfp/+ mice, 1 allele for the gene encoding CX₃CR1, the receptor for fractalkine (CX₃CL1), has been replaced with the gene encoding green fluorescent protein (GFP), resulting in all circulating monocytes cells labeling positively for GFP.⁹  At 72 hours post-zymosan, 30% to 40% of CD11b^hi/F4-80^hi Mφs were positive for PKH26-PCL^red, whereas 60% to 70% were only GFP^pos (Figure 2B). These data show that during resolution (72 hours) both resident and blood monocyte-derived cells occupy the peritoneum, a trend maintained up to day 17 (Figure 2C). Back-gating these populations (resident Mφs, mMφs, and DCs) revealed the altering expression of MHC-II on these cells (Figure 2D) and confirmed that although resident DCs disappeared within a few hours of stimulation (Figure 2Aiii), they are replaced by infiltrating DCs.

The majority of Ly6c^hi monocytes are retained within the bone marrow of ccr2^−/− mice,¹⁰  resulting in substantially reduced numbers of these cells in blood (Figure 2E). Indeed, numbers of MHCII⁺ CD11c⁺ cells in the naive peritoneum of ccr2^−/− mice were also substantially reduced, suggesting that the vast majority of peritoneal DCs are Ly6c^hi-derived (Figure 2E). Injecting 0.1 mg of zymosan into ccr2^−/− mice confirmed the data, obtained with dye-tracking experiments and CX₃CR1^gfp/+ mice, that virtually no Ly6c^hi moMφs were detectable in ccr2^−/− mice from days 3 to 17, whereas DCs numbers were reduced (Figure 2F).

Collectively, these experiments show that there is a third and more prolonged postresolution phase, subsequent to inflammatory onset and resolution, comprising tissue-resident Mφs, Ly6c^hi moMφs, and DCs, hereafter referred to as resMφs, moMφs, and moDCs, respectively. The temporal profile of these cells in WT vs ccr2^−/− is shown in Figure 2G.

Injecting OVA-labeled S pneumoniae^ova intraperitoneally resulted in a similar profile of resMφs, moMφs, and moDCs to that seen with zymosan (Figure 3A). Indeed, memory T cells were seen to accumulate in the peritoneum after PMN clearance and proliferated specifically to bone marrow-derived DCs loaded with OVA (supplemental Figure 5). We sorted postresolution-phase Mφ populations from S pneumoniae^ova and 0.1 mg of zymosan-driven inflammation, and subjected them to reverse transcription polymerase chain reaction for gene products previously determined by transcriptomic/bioinformatic analysis to be highly expressed in resolution-phase Mφs.¹¹  TIMD4 (receptor for PS, which is expressed on apoptotic cells¹² ) and ALOX15 (secretion of lipids that facilitate phagocytosis of apoptotic cells^13,14 ) were primarily expressed in PKH26-PCL^red-labeled resMφ from both models (Figure 3B). Indeed, TGF-β and IL-10, which were upregulated in Mφs during phagocytosis of apoptotic PMNs, were also expressed in these cells (Figure 3C). Injecting CFSE-labeled apoptotic PMNs into mice, bearing 0.1 mg of zymosan-induced peritonitis previously injected with PKH26-PCL^red, demonstrated that resMφs preferentially phagocytosed apoptotic PMNs during resolution (Figure 3D). It was confirmed that blood moMφs were not involved in the clearance of apoptotic PMNs in ccr2^−/− mice, which showed no accumulation of PMNs up to day 13 (Figure 3E).

The resMφs were also enriched for iNOS (Figure 4A) and arginase (Figure 4B) with iNOS expressed in vesicles in postresolution resMφs (Figure 4C). Consistent with an immune suppressive phenotype, resMφs suppressed T-cell proliferation in an iNOS-dependent manner¹⁵  (Figure 4D). Taking this further, we found PKH26-PCL^red-positive resMφs in the mesenteric lymph node and spleen on day 9 post-zymosan, located in both the T and B cell areas (Figure 4E). The presence of these immune-suppressive cells in lymphoid organs suggested a role in lymph node contraction seen from day 13 onward(supplemental Figure 3B). Indeed, although there were equivalent numbers of CD3- and CD19-positive T and B cells in the naive cavity of WT and iNOS^−/− mice (Figure 4F), there were increased numbers of these cells in the peritoneum (Figure 4G) and spleen (Figure 4H) of iNOS knockouts 14 days post-zymosan; these were effects that persisted up to 6 weeks (Figure 4I). This increase in T and B cells was reversed by adoptively transferring PKH26-PCL^red-positive resMφs taken from WT mice bearing a resolving inflammation and transferred into iNOS^−/− bearing zymosan peritonitis (Figure 4G-H) with resMφs^WT found in the spleen on day 14 (Figure 4J).

The resMφs, moMφs, and moDCs from both S pneumoniae and zymosan-triggered inflammation revealed increased expression of CD74 (HLA class II histocompatibility antigen γ chain), H2Aa (histocompatibility 2, class II antigen A, and α), and Clec2i on all 3 populations (supplemental Figure 6A). Despite these findings and data showing that monocyte-derived cells from S pneumoniae-injected mice bear a TipDC-like phenotype (supplemental Figure 6A for tumor necrosis factor (TNF)-α and Figure 4A for iNOS anddata presented in Figure 2, namely CD11c/MHC-II expression on moMφs and moDCs), incubating all 3 Mφ populations from S pneumoniae and zymosan-treated mice overnight with OVA/LPS did not stimulate CD4 T cells from OT-II mice in comparison with bone marrow-derived DCs (supplemental Figure 6B). However, the accumulation of Tregs in the peritoneum peaking 9 days after zymosan with FoxP3 and Ki67 expression was greater than that seen within peripheral blood CD4 cells (Figure 5A) and lead us to investigate whether postresolution Mφs enrich for local Tregs. To test this, we incubated resMφ, moM, and moDCs with CD4 T cells with and without TGFβ/Trp1 peptide. The resMφs and moMφs triggered FoxP3 expression within CD4 T cells, whereas moDCs caused Tregs and CD4⁺/FoxP3⁻ effector T cells to proliferate (Figure 5B).

Further analysis of resMφs, moMφs, and moDCs suggest that not only are their phenotypes similar between zymosan and bacterial infection, but that moMφ and moDCs expressing CCR2, CCR7, and CCL2 (Figure 5C) may also possess a migratory capacity. We tested this by injecting PKH26-PCL^green into mice on day 6 post-zymosan (0.1 mg), which also had PKH26-PCL^red injected into their naïve peritoneum to label resMφs. This resulted in infiltrating blood moMφs and moDCs labeling positively for only PKH26-PCL^green, whereas resMφs were labeled with both PKH26-PCL^red and PKH26-PCL^green. Analysis of mesenteric lymph nodes on day 9 post-zymosan revealed the presence of PKH26-PCL^green-labeled cells among the CD3⁺ and CD19⁺ positive lymphocyte populations (Figure 5D).

We wished to determine whether monocyte-derived cells specific to resolving inflammation bridges the gap between acute inflammation and adaptive immunity. Therefore, 0.1 mg of zymosan was injected into ccr2^−/− mice. Not only was there a significant reduction in the number of cells within R1, R2, and R3 in lymph nodes of these animals (Figure 6A), but there was also a significant reduction in lymph node (Figure 6A) and peritoneum (Figure 6B) CD3 and CD19 lymphocytes on day 9 compared with WTs. Importantly, these effects arose from eliminating Ly6c^hi monocyte-derived resident DCs (MHC-II⁺/CD11c⁺) and postresolution Ly6c^hi-derived moMφs and moDCs. To discern the relative contribution of tissue-resident DCs from postresolution moMφs and moDCs to lymph node expansion, we used MC-21, which depletes Ly6c^hi monocytes.¹⁶  MC-21 was given therapeutically every second day starting on day 3 post-zymosan (0.1 mg) and was found to deplete blood Ly6C⁺ monocytes on day 9 post-zymosan (Figure 6C). MC-21 also caused a reduction in lymph node CD3 and CD19 cells and R1, R2, and R3 (Figure 6D), and peritoneal T- and B-cell numbers (Figure 6E). These effects were not as pronounced as that observed with ccr2^−/− mice, reflecting the relative contribution of Ly6c^hi monocyte-derived cavity-resident DCs vs postresolution moMφs and DCs to lymph node expansion. To attribute a functional role to postresolution moMφs and moDCs, ccr2^−/− mice bearing a delayed type hypersensitivity were found to have a dampened adaptive immune reaction compared with WT mice (Figure 6F).

PKH26-PCL^red was injected into the naïve cavity to label resident cells. This was followed 2 hours later by 0.1 mg of zymosan intraperitoneally. Seven days after the zymosan injection, then PKH26-PCL^green was injected. This resulted in resMφs labeling with both PKH26-PCL^red and PKH26-PCL^green, but with postresolution moMφs and moDCs labeling with only PKH26-PCL^green. Examination of the cavity 60 days post-zymosan detected a population of infiltrating PKH26-PCL^green Ly6c^hi-derived moMφs whose numbers were reduced in ccr2^−/− mice as expected (Figure 7A). In contrast, moDCs were unchanged in WT mice compared with naïve controls, but were also reduced in ccr2^−/− mice (Figure 7B). We found an increase in effector CD4 T cells (Figure 7C) whose presence in the cavity was partially dependent on Ly6c^hi-derived moMφs as numbers of CD44⁺/CD62L^- CD4 T cells were partly reversed in ccr2^−/− mice (Figure 7C). Fluorescence-activated cell sorting on day 60 Ly6c^hi PKH26-PCL^green moMφs and PKH26-PCL^red/green resMφs revealed that both populations were phenotypically distinct from one another (Figure 7D). Indeed, the phenotypes of these cells on day 60 post-zymosan were also different to their phenotypes at 0 hours (resMφs) and 72 hours (resMφs and moMφs), respectively (Figure 7D). To determine the significance of these findings in terms of responses to secondary infection and/or injury, 60-day zymosan peritonitis mice were given S pneumoniae and inflammation was determined 4 hours later. Responses to S pneumoniae were dampened compared with mice that were not previously injected with zymosan 60 days earlier in terms of PMN numbers (Figure 7E).

We show that there is a previously overlooked third phase of leukocyte influx into tissues after onset and resolution of acute inflammation. These cells comprise Ly6c^hi-derived moMφs, moDCs, and MDSCs. In addition, tissue-resident (prenatal-derived^17-19 ) resMφs that disappear during the early phase of the inflammatory response,²⁰  reappear postresolution. These diverse populations of mononuclear phagocytes were observed alongside lymph node expansion and increased numbers of peripheral blood and peritoneal memory and regulatory lymphocytes. Based on our data and supported by others,^21-24  we conclude that in response to resolving, but not hyperinflammation, DCs residing in naïve tissues take up antigens and migrate to local lymph nodes (including peritoneal milky spots²⁵ ) to initiate lymphocyte activation. The latter is amplified by postresolution CCR2-expressing monocytes; whether these effector monocytes²⁶  exert their effects after migrating into the cavity and/or directly from the blood²¹  remains to be clarified. Concomitantly, as inflammation resolves resMφs and moMφs trigger FoxP3 expression within CD4 T cells while moDCs trigger their proliferation. The resMφs repopulate the cavity with a proportion migrating to the draining lymph nodes to bring about lymph node contraction in an iNOS-dependent manner from day 9 to day 13 onward. Finally, populations of moMφs remain in tissues months after the inflammation has resolved, dictating the magnitude of subsequent acute innate inflammatory stimulation (see hypothesis shown in supplemental Figure 7).

The importance of IL-10 and TGF-β in zymosan-generated Mφs, in terms of counterbalancing adaptive immune responses, was reported by others.²⁷  The resMφs expressing TGF-β and IL-10 as a consequence of phagocytosing apoptotic PMNs²⁸  migrate to the lymph node and spleen expressing immune-suppressive iNOS. These cells are likely to remain in lymphoid organs with a role in terminating adaptive immune responses and long–term, tempering of the severity of future antigen-specific immunity. A role for iNOS in this setting was illustrated in the context of “adjuvant immunogenicity,” first described more than 40 years ago, which was found to be dependent on the presence of Mycobacterium tuberculosis within the adjuvant.²⁹  Exposure to complete Freund’s adjuvant can impair the subsequent expression of autoimmune disease in rodents. This immunoprotective effect was demonstrated in multiple autoimmune disease models, both spontaneous and induced and in multiple species, including rats,^30-33  mice,³⁴  and guinea pigs.^29,35  In each case, preimmunization with Freund’s complete adjuvant alone up to a month before the disease induction by immunization resulted in decreased incidence and severity of disease^29-31,35,36  in an iNOS-dependent manner.³⁷  Similarly, Mφ-mediated immunosuppression has been reported after bacteria, fungi, and parasite³⁸  infection. For instance, mice immunized with attenuated Salmonella typhimurium (SL3235), although protected against virulent challenge, are unable to mount in vivo and in vitro antibody responses to non-Salmonella antigens, such as tetanus toxoid and sheep red blood cells, and exhibit profoundly suppressed responses to B- and T-cell mitogens. It transpires that suppression of antibody responses is mediated by iNOS within Mφs.^39-42  Collectively, we argue that, in addition to bridging the gap between innate and adaptive immunity, resolution may also establish a phase of immunologic tolerance. It is unclear why a PMN-driven, acute onset phase of inflammation should be accompanied, paradoxically, by a prolonged phase of immune suppression. One possibility is to suppress the development of maladaptive immune response leading to autoimmunity.

In contrast, injecting high-dose zymosan (10 mg) resulted in a substantial inflammatory cell infiltrate comprising classically activated (M1)-like Mφs secreting high levels of TNF-α, IFN-γ, and IL-6 (>2000 pg/mL). Compared with resolving inflammation, such an inflammatory insult triggered substantially reduced numbers of Tregs and effector/memory lymphocytes. Proinflammatory cytokines, such as interferon (IFN) and TNF-α, interfere with antigen-specific T-cell responses^43-45  and clearance of viral and mycobacterial infections in mice.^46,47  Furthermore, inhibition of these cytokines enhances pathogen clearance and resolution of disease, an effect that is dependent on the presence of T cells. This suggests that excessive inflammation inhibits antigen-specific T-cell function and therefore immunity to pathogens.

Although the M1/M2 classification of macrophages was largely borne out of isolated monocytes incubated with defined growth factors in vitro, the phenotype of Mφ populations is likely to be more complex and overlapping, contingent on tissue, phase of inflammation, and the nature of the inciting inflammatory stimuli. We identified at least 4 distinct populations of monocyte/Mφs, each with a role in the resolution and in the development and control of adaptive immunity. In terms of individual cell phenotypes, we found that resMφs expressed TGF-β1 and IL-10, presumably as a consequence of phagocytising apoptotic PMNs; these cells also expressed ALOX-15 and TIMD4 to facilitate the recognition and uptake of apoptotic cells.^12-14  In contrast, moMφs expressed IL12p35, whereas moDCs were enriched for IL1β, CCR7, CCR2, and CCL7. Taking this further, populations of moMφs that persisted in the cavity for months postinflammation were phenotypically different from resMφs. These data underline the fact that despite experiencing the same inflammatory cues, different monocyte/macrophage populations possess diverse phenotypes that are neither M1 nor M2, but are commensurate with the phase of inflammation. Indeed, this probably extends to macrophage populations occupying different tissue niches under physiological^48,49  and disease conditions.

As reported by others using thioglycollate-induced peritonitis,¹⁷  we also found a population of moMφs that persisted in the peritoneum for at least 2 months postresolution. The phenotype of these cells was different to that of resMφs at this time; indeed the phenotype of day 60 moMφs was also different than the phenotype early in the inflammatory response (at 72 hours). Moreover, the phenotype of resMφs 60 days post-zymosan was different to the inflammatory characteristics in the naïve cavity. This emphasizes functional plasticity in macrophage phenotype congruent with the environment. Specifically, despite coexisting in the same inflammatory milieu, macrophages of different origins acquire distinct phenotypes that change throughout inflammation. This emphasizes that far from revering back to the state, the tissue experience before resolution, postresolution tissues experience a state of “adaptive homeostasis,” which dictates the magnitude of subsequent inflammatory stimuli. This is an area that requires further exploration in the future.

In summary, we show that resolution is not the end of the immune response to infection/injury but that it acts as a bridge between innate and adaptive immunity, thereby adding a third phase to acute inflammation after acute and resolution, namely postresolution. Disruption of proresolution pathways by hyperinflammatory stimuli, for instance, impairs the development of specific immunity. These data redefine resolution as the creation of a tissue microenvironment that facilitates interaction between the innate and adaptive arms of the immune system and that postresolution tissue that acquires a state of “adapted homeostasis.”

D.W.G. is a Wellcome Trust Senior Research Fellow. M.S. salary was supported by a studentship from the Medical Research Council, United Kingdom. J.N. was supported by D.W.G.’s fellowship.

This work was supported by a grant from the Wellcome Trust and Medical Research Council, United Kingdom (D.W.G.).

Contribution: M.S. and J.N. performed the majority of experiments; E.K. designed primer sequences and carried out polymerase chain reaction; F.A.-V. and S.Q. performed tolerance experiments; S.Y. designed monocyte ablation studies and performed studies in which mice were exposed to a second inflammatory stimulation; M. Mack provided MC-21 and expertise on monocyte biology depletion; M. Motwani performed lymphocyte proliferation and phenotyping studies; T.A. performed; and D.W.G. designed experiments and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Derek W. Gilroy, Division of Medicine, Centre for Clinical Pharmacology and Therapeutics, University College London, 5 University St, London, United Kingdom WC1E 6JJ; e-mail: d.gilroy@ucl.ac.uk.