The prime function of classically activated macrophages (activated by Th1-type signals, such as IFN-γ) is microbial destruction. Alternatively activated macrophages (activated by Th2 cytokines, such as IL-4 and IL-13) play important roles in allergy and responses to helminth infection. We utilize a murine model of filarial infection, in which adult nematodes are surgically implanted into the peritoneal cavity of mice, as an in vivo source of alternatively activated macrophages. At 3 wk postinfection, the peritoneal exudate cell population is dominated by macrophages, termed nematode-elicited macrophages (NeMφ), that display IL-4-dependent features such as the expression of arginase 1, RELM-α (resistin-like molecule α), and Ym1. Since increasing evidence suggests that macrophages show functional adaptivity, the response of NeMφ to proinflammatory Th1-activating signals was investigated to determine whether a switch between alternative and classical activation could occur in macrophages differentiated in an in vivo infection setting. Despite the long-term exposure to Th2 cytokines and antiinflammatory signals in vivo, we found that NeMφ were not terminally differentiated but could develop a more classically activated phenotype in response to LPS and IFN-γ. This was reflected by a switch in the enzymatic pathway for arginine metabolism from arginase to inducible NO synthase and the reduced expression of RELM-α and Ym1. Furthermore, this enabled NeMφ to become antimicrobial, as LPS/IFN-γ-treated NeMφ produced NO that mediated killing of Leishmania mexicana. However, the adaptation to antimicrobial function did not extend to key regulatory pathways, such as IL-12 production, which remained unaltered.
Signals encountered by developing macrophages (Mφ)5 during migration determine their functional properties at sites of inflammation or infection. Among these signals, cytokines, which can act synergistically or have opposing effects, are responsible for the development of highly divergent Mφ phenotypes. Classical activation of Mφ is dependent on the products of activated Th1 cells, in particular IFN-γ (1). The prime function of classically activated Mφ (CAMφ) is microbial destruction, which is conducted by the production of reactive oxygen and nitrogen intermediates such as NO from inducible NO synthase (iNOS/NOSII) (2). Alternatively activated Mφ (AAMφ), activated by the Th2 cytokines IL-4 and IL-13, play important roles in allergy and responses to parasite infection (1). AAMφ and CAMφ exhibit distinct l-arginine metabolism pathways. CAMφ produce NO as a result of the up-regulation of iNOS, which is a catalyst of the l-arginine substrate (3). In AAMφ, l-arginine is converted to l-ornithine and urea by the enzyme arginase 1 (3). The induction of either arginase or iNOS is usually coupled with suppression of the opposing enzyme (4). While the NO produced by CAMφ has antimicrobial and cytotoxic properties, l-ornithine produced by AAMφ can be further metabolized to proline and polyamines, which have roles in collagen production and cell proliferation, respectively. These properties have led to the hypothesis that AAMφ are involved in wound healing (5, 6).
Although Mφ have been usefully classified in this way, the range of actual phenotypes is likely to be much broader, with AAMφ and CAMφ representing two points in a wide spectrum. Unlike Th cells where activation leads to terminal differentiation into Th1 and Th2 cells, Mφ appear to display a high degree of flexibility (7, 8, 9, 10). Plasticity of function may be an economical strategy for the immune system since, in contrast to the high turnover of T cells, Mφ are longer lived and may need to adapt their function to different pathogens or environments they will face during their lifespan.
Many studies have shown that Mφ responsiveness to a given cytokine can be altered and/or suppressed by the cytokines to which it was previously exposed (11, 12). However, the flexibility of the AAMφ phenotype is still a subject of controversy. While some reports show that IL-4 pretreatment of Mφ renders them unresponsive to Th1 activation (4, 11, 12, 13, 14), others have found that the previous IL-4 exposure enhanced the responsiveness to Th1-activating signals (15). As these results were obtained by in vitro activation with IL-4, the differences may be due to variation in the length and intensity of the activation stimuli. Indeed, Stout et al. demonstrated that two cytokines could have either antagonistic or synergistic effects on murine Mφ function dependent on the order and length of each cytokine treatment (16). In another study, human Mφ were stimulated with either pro- (e.g., TNF-α) or antiinflammatory (e.g., IL-10) cytokines and then cultured with a counterstimulatory cytokine or medium alone, and it was found that Mφ stimulated toward a specific activation state could switch their phenotype rapidly when given counterstimulatory signals or return to a quiescent state after signal arrest (9). Due to conflicting in vitro data, it is difficult to determine whether Mφ demonstrate functional adaptivity in vivo and to ascertain the physiological relevance of this phenomenon. Investigating Mφ plasticity in vivo could have important implications for therapeutic targeting of Mφ in chronic diseases but also for our general understanding of how the immune system copes with multiple infections that may require differing immune responses. In support of the hypothesis that Mφ plasticity may confer increased efficiency and flexibility for the immune response, Gratchev et al. demonstrated that in response to a second stimulation with IFN-γ, in vitro-derived human AAMφ displayed significantly higher bactericidal activity (17). Also, pretreatment of murine macrophages with the IL-13 was found to enhance LPS-induced anti-Toxoplasma gondii activity (18). However, the degree of flexibility of Mφ recruited to sites of infection is still unknown.
In this study, we have used a murine model for filarial infection as a source of Mφ, which we have previously termed nematode-elicited Mφ (NeMφ) (19, 20, 21). Mice are surgically implanted i.p. with the adult stage of the nematode Brugia malayi. By 1 wk postinfection, the peritoneal exudate cell (PEC) population is dominated by Mφ that display IL-4-dependent features such as the expression of arginase 1, RELM-α (resistin-like molecule α), and Ym1, as well as the ability to suppress the proliferation of neighboring cells (19, 22). This phenotype is sustained for many weeks in vivo. We have previously shown that NeMφ share many properties with Mφ differentiated in vitro in response to IL-4 or IL-13 (hereafter termed in vitro AAMφ) (19). However, NeMφ are recruited during a chronic Th2-dominated nematode infection and experience much longer term exposure to IL-4 and IL-13 than in vitro-AAMφ. Furthermore, NeMφ are influenced by parasite-secreted products as well as host factors other than IL-4, such as IL-10 and glucocorticoids. In light of the increasing evidence that Mφ show functional adaptivity, we decided to study the NeMφ response to Th1-activating signals as a model to investigate whether Mφ differentiated in a chronic Th2 infection setting could alter their established phenotype and whether this would translate into an ability to control an intracellular microbial infection.
Despite the long-term exposure to Th2 cytokines and antiinflammatory signals in vivo, we found that NeMφ were still responsive to LPS and IFN-γ, exhibiting dramatic changes in gene expression profile. Additionally, LPS/IFN-γ treated NeMφ were able to effectively control infection with Leishmania mexicana. However, LPS/IFN-γ treated NeMφ still maintained the IL-4-dependent capacity to suppress proliferation and failed to up-regulate IL-12. Taken together, these results demonstrate that Mφ differentiated in vivo can make radical alterations in phenotype to accommodate a new challenge. However, functional plasticity may be limited to effector rather than to regulatory pathways.
Materials and Methods
Mice and infection
Wild-type (WT) or IL-4−/− mice on the C57BL/6 background were bred in house or purchased from Harlan U.K. Mice were 6–8 wk old at the start of the experiment, and all work was conducted in accordance with the Animals (Scientific Procedures) Act of 1986. Adult B. malayi parasites were removed from the peritoneal cavity of infected gerbils purchased from TRS Laboratories or maintained in house. Mice were surgically implanted i.p. with five to six live adult female B. malayi. Three weeks later, mice were euthanized and PECs harvested by washing of the peritoneal cavity with 15 ml of ice-cold DMEM. As a control for non-Th2-polarized inflammation, mice were injected i.p. with 0.8 ml of 4% thioglycollate medium, brewer modified (BD Biosciences). Three days later, the PECs were harvested, as above.
PECs were cultured in DMEM, supplemented with 10% FCS, 2 mM l-glutamine, 0.25 U/ml penicillin, and 100 mg/ml streptomycin. Thioglycollate-elicited PECs were plated in 9-cm petri dishes and left untreated or treated for 18–24 h with IL-4 (20 ng/ml; BD Pharmingen). Nonadherent cells were subsequently washed off and the remaining adherent Mφ were left untreated or treated with LPS (100 ng/ml; Escherichia coli 0111:B4; Sigma-Aldrich) and IFN-γ (10 U/ml; BD Pharmingen) together or separately for 18–24 h. NeMφ were similarly recovered and plated in medium alone or with LPS and/or IFN-γ. Following treatment, the nonadherent cells were washed off and the adherent Mφ were recovered by a 15-min incubation at 37°C in warm 10 mM glucose and 3 mM EDTA in PBS. For cell division analysis, cells were washed in PBS and then resuspended at 1 × 107/ml in 5 μM CFSE in serum-free DMEM for 8 min at 37°C. The reaction was quenched with an equal volume of FCS and cells were washed several times in serum-free DMEM. Bone marrow-derived Mφ (BMMφ) were prepared by harvesting bone marrow from the femur and tibia of C57BL/6 mice. Erythrocytes were lysed using 3 ml of RBC lysis buffer (Sigma-Aldrich) for 5 min. Differentiation into Mφ was performed according to published protocols (23). Cells were plated onto Petri dishes at 7.5 × 106 cells/plate and cultured in DMEM, supplemented with 25% FCS (Invitrogen), 25% L929 supernatant as a source of M-CSF, 2 mM l-glutamine, 0.25 U/ml penicillin, and 100 μg/ml streptomycin. The medium was replaced after 4–6 days to generate a pure population of Mφ at day 7. BMMφ were then plated in medium alone or with LPS and/or IFN-γ (as above).
Cells were incubated at 4°C for 15 min in blocking buffer (2% mouse serum, 10 μg/ml anti-CD16/32 in FACS buffer (PBS supplemented with 2 mM EDTA and 0.5% BSA)), followed by staining for 20 min on ice with the Abs of interest at the appropriate dilution as determined by titration. Abs included tricolor-conjugated anti-F4/80 (Caltag Laboratories), PE-conjugated anti-B7.1/B7.2/MHC class II (BD Pharmingen), PE-conjugated anti-programmed death ligand (PD-L)1/PD-L2 (eBioscience), as well as appropriate isotype control Abs. Cells were washed three times in FACS buffer and fixed in 0.8% paraformaldehyde before acquisition and analysis (FACStation and FlowJo software; BD Biosciences).
Mφ purified by adherence were cocultured (1 × 105 cells/well) in 96-well flat-bottom plates with EL-4 cells (1 × 104/well) in the presence or absence of the inhibitors NG-monomethyl-l-arginine (l-NMMA; 100 μM; Sigma-Aldrich) or Nω-hydroxy-nor-l-arginine (nor-NOHA; 250 μM; Sigma- Aldrich). After 48 h, 100 μl of supernatant was removed to measure NO production. For each well, 1 μCi of [3H]TdR was added and plates were incubated overnight before harvesting and counting using a liquid scintillation counter (MicroBeta 1450; TriLux). Quadruplicate measurements per sample were performed. Results were plotted in cpm.
Quantification of NO production and arginase activity
NO production was assessed by nitrite accumulation in the culture media using the Griess reagent. In brief, 100 μl of culture supernatant was mixed with 100 μl of 5.8% phosphoric acid, 1% sulfanilamide, and 0.1% N-(1-naphthyl)ethylenediamine dihydrochloride. Absorbance was measured at 490 nm with background correction at 650 nm using a microplate reader. Concentration was determined according to a standard curve of sodium nitrite solution. Arginase activity was measured according to previously published protocols (3). Briefly, 1–2 × 105 cells were lysed with 100 μl of 0.1% Triton X-100. Following a 30-min incubation with shaking, 100 μl of 25 mM Tris-HCl (pH 7.2) and 20 μl of 10 mM MnCl2 were added and the enzyme was activated by heating to 56°C for 10 min. l-Arginine hydrolysis was conducted by incubating 100 μl of this lysate with 100 μl of 0.5 M l-arginine (pH 9.7) at 37°C for various time points between 15 and 60 min. The reaction was then stopped with 800 μl of H2SO4 (96%)/H3PO4 (85%)/H2O (1/3/7, v/v/v), and 40 μl of 9% isonitrosopropiophenone was added, followed by heating to 99°C for 30 min before reading on the microplate reader at 540 nm. A standard curve of urea solution was used to determine concentrations. One unit of arginase enzyme activity is defined as the amount of enzyme that catalyzed the formation of 1 μmol of urea per minute at 37°C. Unless otherwise stated, all reagents were obtained from Sigma-Aldrich.
RNA extraction and real-time RT-PCR
RNA was recovered from cells by resuspension in TRIzol reagent (Invitrogen). Total RNA was extracted according to the manufacturer’s instructions. Following DNAse1 treatment (Ambion), 1 μg of RNA was used for the synthesis of cDNA using Moloney murine leukemia virus reverse transcriptase (Stratagene). Relative quantification of the genes of interest was measured by real-time PCR, using the Roche LightCycler. PCR amplification conditions and primers have been described previously (20). Five serial 1/4 dilutions of a positive control sample of cDNA were used as a standard curve in each reaction, and the expression levels were estimated from the curve. Real-time PCR of the housekeeping gene β-actin allowed normalization of the expression of the genes of interest.
For measurement of cytokine production, 5 × 105 PECs were plated in 24-well plates for 2–4 h followed by removal of nonadherent cells. The remaining adherent Mφ were left untreated or were treated for 24 h with LPS/IFN-γ as described above, followed by recovery of the supernatants for cytokine quantification. TNF-α was measured using the DuoSet TNF-α ELISA kit (R&D Systems) according to the manufacturer’s instructions. IL-6 and IL-12p40 were measured according to standard sandwich ELISA protocols, using Ab pairs (unconjugated and biotinylated) from BD Pharmingen, and ExtrAvidin-alkaline phosphatase conjugate in conjunction with SigmaFast p-nitrophenyl phosphate tablet substrate (both from Sigma-Aldrich). IL-10 and MCP-1 were quantified using the cytokine bead array (CBA) kit (BD Pharmingen) according to the manufacturer’s instructions.
Mφ infection with L. mexicana
L. mexicana (strain MNYC/BZ/62/M379 or MNYC/BZ/62/M379 with DsRed integrated into sRNA locus) (24) promastigotes were cultured in vitro in semidefined medium (25) with 10% FCS and 1% penicillin-streptomycin (complete semidefined medium) at 26°C. Promastigotes were added to Mφ, purified by adherence, and plated on coverslips (BDH Laboratory Supplies) in 24-well plates, at a ratio of 10:1 for 3 h or 30:1 for 3 h, in the case of the red strain. The Mφ were then washed twice with complete DMEM to remove nonphagocytosed parasites and wells were replenished with complete medium. Where indicated, the NO inhibitor l-NMMA was added to a final concentration of 400 μM. At day 3 postinfection, cells were fixed with 2% formaldehyde for 20 min at room temperature, washed twice with PBS, and stained with 0.5–1 ml of Giemsa for 5–10 min, followed by two washes with distilled H2O. Coverslips were removed from wells, allowed to dry, and mounted onto slides with DPX mount (BDH Laboratory Supplies) before microscopic examination. Approximately 200 cells per group were counted and the percentage infection was recorded. Unless otherwise stated, all reagents were obtained from Sigma-Aldrich. Alternatively, a leishmanicidal assay was conducted at 3 h postinfection as previously described (26). The cells were lysed using 0.01% SDS in 100 μl of DMEM (FCS free) for 30 min and pipetted up and down 5–10 times. Released amastigotes were resuspended in complete semidefined medium in a total of 600 μl per well and cultured for 72 h at 26°C. Four aliquots of 150 μl for each sample were then transferred to quadruplicate wells of a 96-well plate and pulsed with [methyl-3H]thymidine (1 μCi/well) for a further 18 h at 26°C, and incorporation was assessed on a liquid scintillation counter (MictoBeta 1450; TriLux). Leismanicidal activity was measured as reduction in the incorporation of [3H]thymidine by surviving parasites. Unless otherwise stated, all reagents were obtained from Sigma-Aldrich. For immunofluorescence assays, Mφ infected on coverslips were fixed in 1–2 ml of 3% paraformaldehyde (in PBS) for 20 min and then washed in PBS. They were permeabilized in 1–2 ml of 0.25% Triton X-100 (in PBS) for 10 min and washed in PBS again. Blocking for 30 min was conducted with 1–2 ml of 1–3% (w/v) BSA in PBS. The primary Ab, anti-Ym1 (StemCell Technologies; 1/25 dilution in BSA/PBS) was incubated for 1 h. Coverslips were washed three times in PBS for 5 min with shaking. The secondary Ab (anti-rabbit Alexa Fluor 488 (Molecular Probes); 1/200 in BSA/PBS) was incubated for 1 h and then cells were washed in PBS. For nuclear staining, 1 μl of 10 mg/ml 4′,6-diamidino-2-phenylindole was added during the first wash and then cells washed three more times for 5 min with shaking. Coverslips were washed in dH2O and placed on top of gel/mount on slides and allowed to dry in the dark. Cells were visualized using an Olympus BX50 microscope, and images were taken with a MicroColor RGB-MS-C camera.
LPS/IFN-γ causes a switch in l-arginine metabolism from arginase to iNOS and down-regulates the expression of alternative activation markers
Preferential expression of arginase over iNOS is a consistent feature of NeMφ (19). We decided to investigate whether the Th1-activating signals (LPS and/or IFN-γ) could alter the NeMφ iNOS/arginase balance. Thioglycollate-elicted Mφ (ThioMφ) and NeMφ were left untreated or were treated overnight with LPS and/or IFN-γ and recovered for gene expression analysis by real-time RT-PCR (Fig. 1, A and B). As expected, untreated NeMφ showed high arginase 1 expression but low iNOS RNA levels, while untreated ThioMφ showed none or low expression of both genes. Both IFN-γ and LPS could act alone or in concert to reduce the arginase 1 expression in NeMφ. This decrease was paralleled by an increase in iNOS expression. In contrast, ThioMφ required both signals for the induction of iNOS. This could reflect the less mature activation state of this Mφ type (27), which may require signaling through a pathogen recognition receptor such as TLR4 before becoming responsive to IFN-γ. For subsequent experiments we decided to use LPS and IFN-γ together to ensure classical activation of the ThioMφ group.
Although ThioMφ exhibited increases in arginase 1 mRNA in response to LPS/IFN-γ, consistent with previous studies (3), this induction in gene expression did not result in increased arginase enzyme activity (Fig. 1,C). Arginase and iNOS enzyme activities of NeMφ were altered by LPS/IFN-γ and, although not as fully, they did reflect the gene expression data (Fig. 1, C and D). This included a down-regulation in arginase activity in LPS/IFN-γ-treated NeMφ and an increase in NO production. In contrast to the mRNA levels, the NO produced by LPS/IFN-γ-treated NeMφ was lower than by LPS/IFN-γ-treated ThioMφ. These differences may reflect timing, as mRNA levels precede protein production, or other posttranslational controls. For example, the higher level of arginase in LPS/IFN-γ-treated NeMφ may compete with iNOS for l-arginine. Nevertheless, the increase in NO production coupled by a decrease in arginase activity demonstrates that NeMφ can alter the balance of arginine metabolism in response to proinflammatory signals.
Having identified RELM-α and Ym1 as the main IL-4-dependent genes expressed in NeMφ (19), we investigated whether RELM-α/Ym1 expression was also altered by LPS/IFN-γ treatment. As a control for nondifferentiated Mφ, we measured RELM-α and Ym1 expression in ThioMφ (Fig. 1,E). We observed induction of both genes by LPS/IFN-γ in a similar pattern to arginase 1 expression, although the expression levels were lower than in untreated NeMφ. Upon overnight treatment with LPS/IFN-γ, RELM-α and Ym1 expression in NeMφ was reduced by up to 90% (Fig. 1 E). Subsequent experiments always demonstrated reduced expression of these markers following LPS/IFN-γ treatment but the extent of this reduction was variable, ranging from 30% to 90% (data not shown).
LPS/IFN-γ-induced activation switch is not due to the outgrowth of a minority cell population
To rule out the possibility that the phenotypic changes in LPS/IFN-γ-treated NeMφ represented an outgrowth of a naive Mφ population rather than a change in the existing NeMφ population, we monitored the cell proliferation following treatment. Before overnight treatment with LPS/IFN-γ, cells were labeled with CFSE and then recovered and analyzed by flow cytometry. This allowed the comparison of proliferation between untreated and LPS/IFN-γ-treated Mφ, as identified according to side scatter and F4/80 expression (Fig. 2 A). Following overnight treatment with LPS/IFN-γ, ThioMφ and NeMφ groups did not undergo any cell divisions and remained CFSEhigh. Thus, the switch in gene expression and iNOS/arginase activity represents a nondividing NeMφ population and not the outgrowth of another cell population. Additionally, fluorescent staining for Ym1 and RELM-α by immunofluorescence assay (see below) or intracellular FACS (data not shown) suggested that this was a highly homogeneous starting population with all F4/80+ Mφ expressing the alternative activation markers.
The mean fluorescence intensity (MFI) for F4/80 was higher in NeMφ (176) than in ThioMφ (52) (Fig. 2 B). Since F4/80 has been implicated in immunological tolerance through the generation of regulatory T cells (28), the higher surface expression on NeMφ may point to a regulatory function. Mφ are a prominent cell type in other filarial nematode infection models such as Litomosoides sigmodontis, where regulatory T cells play an important role in immunoregulation (21, 29). Interestingly, LPS/IFN-γ treatment affected F4/80 surface staining intensity, and we observed a reduction in MFI in NeMφ (from 176 to 113) but not in ThioMφ (from 52 to 58). Down-regulation in F4/80 expression upon Th1 activation could have an impact on the regulatory properties of NeMφ.
Suppressive phenotype of NeMφ is not altered by LPS/IFN-γ
Since proliferative suppression is one of the well-defined regulatory properties of NeMφ (22, 30), we decided to investigate whether LPS/IFN-γ affected the ability of NeMφ to suppress the proliferation of cocultured EL-4 thymoma cells. We included the NO inhibitor l-NMMA as a control since LPS/IFN-γ treatment resulted in NO production from all Mφ groups (see Fig. 1), and we wanted to distinguish the known suppressive effects of NO (31, 32, 33) from the suppressive mechanism mediated by NeMφ. Having previously shown that the NeMφ suppressive phenotype is dependent on IL-4 (22), we also included NeMφ generated in IL-4-deficient mice as a negative control for proliferative suppression. Finally, we included ThioMφ pretreated with IL-4, allowing a comparison between in vitro-derived AAMφ and in vivo-derived NeMφ.
As expected, the EL-4 cell proliferation was suppressed when cocultured with NeMφ in comparison to control ThioMφ (Fig. 3 A, open bars). ThioMφ pretreated with IL-4 (in vitro AAMφ) were equally suppressive while IL-4−/− NeMφ were unable to suppress EL-4 proliferation, confirming that IL-4 was essential for the development of this suppressive phenotype. When treated with LPS/IFN-γ (black bars), all Mφ cell types could suppress proliferation. However, l-NMMA treatment (gray bars) reversed the LPS/IFN-γ-induced suppressive phenotype in the ThioMφ and the IL-4−/− NeMφ, demonstrating that this suppression was NO mediated. In response to LPS/IFN-γ treatment, NeMφ and in vitro AAMφ still suppressed EL-4 thymoma proliferation but this was not NO mediated, as it was not altered by l-NMMA. LPS/IFN-γ is thus not sufficient to reverse the suppressive function of either in vitro- or in vivo-derived AAMφ.
While all Mφ groups produced NO in response to LPS/IFN-γ, NeMφ produced less NO than did ThioMφ (Fig. 3,B), consistent with our previous data (Fig. 1 D). Of note, WT NeMφ showed increased responsiveness to LPS/IFN-γ in comparison to IL-4−/− NeMφ and produced higher levels of NO. This implies that activation by IL-4 may in fact enhance responsiveness to classical activation stimuli as previously reported (15, 16).
The IL-4-dependent mechanism of proliferative suppression is still unknown but one possibility is that arginase acts by depleting arginine needed for cell growth (34). To test this possibility, WT NeMφ were treated with nor-NOHA to inhibit arginase 1 activity. This led to partial reversal of the suppressive phenotype, with nor-NOHA treatment enhancing proliferation of the EL-4 cells cocultured with NeMφ (Fig. 3 C). However, EL-4 cells cocultured with nor-NOHA-treated ThioMφ also exhibited dramatically enhanced proliferation. This suggests that inhibition of even constitutive levels of arginase has significant effects on cellular proliferation. In the presence of nor-NOHA, NeMφ permitted only one-fifth the level of EL-4 proliferation as similarly treated ThioMφ, suggesting that arginine consumption is not the dominant mechanism of suppression in NeMφ.
NeMφ express cell surface activation markers that can be further up-regulated in response to LPS/IFN-γ
We studied the surface expression of MHC class II and costimulatory molecules B7.1 and B7.2 to determine the activation status of NeMφ before and after LPS/IFN-γ treatment (Fig. 4 A). Untreated NeMφ expressed higher levels of MHC class II, B7.1, and B7.2 than did ThioMφ, as shown by MFI. The high expression of these markers is consistent with the original classification of AAMφ as “activated” (35) and the data that ThioMφ are relatively “inactive” (27). It is also consistent with previous studies showing that NeMφ are efficient APCs (36). Upon LPS/IFN-γ stimulation, the MFI of all markers was increased in both ThioMφ and NeMφ, and the MFI for MHC class II and B7.2 was highest in LPS/IFN-γ treated NeMφ.
PD-L1 and PD-L2 are members of the B7 family of costimulatory molecules and have been reported as useful markers to distinguish between classical and alternative activation of Mφ in vitro (37). Additionally, they have been implicated in the proliferative suppression observed by Mφ during infection with the platyhelminths Taenia crassiceps and Schistosoma mansoni (38, 39). We thus measured surface expression of both PD-L1 and PD-L2 in NeMφ (Fig. 4,B). PD-L1 surface expression was similar in both ThioMφ and NeMφ and was up-regulated in both Mφ groups in response to LPS/IFN-γ, consistent with reports that classically activated Mφ show increased PD-L1 expression (37). Surprisingly, despite previous studies reporting PD-L2 as a marker for in vitro AAMφ (37), neither untreated or LPS/IFN-γ-treated NeMφ expressed PD-L2. To confirm that the PD-L2 staining was optimal, we demonstrated that in vitro IL-4 treatment could up-regulate PD-L2 on ThioMφ (Fig. 4 B) as previously shown (37). Our finding that in vivo AAMφ display different surface expression profiles to in vitro-derived AAMφ is consistent with previous studies that reported differences in gene expression profiles and cell morphology (40), and it reiterates the importance of in vivo models for the study of these activated Mφ subsets. Furthermore, in contrast to suppressive Mφ found in platyhelminth infection (38, 39), these data suggest that the PD-L1/PD-L2 costimulatory molecules may not be responsible for the suppression we observe in nematode infection.
TNF-α, IL-6, and IL-10, but not IL-12p40, production is enhanced in response to LPS/IFN-γ
ThioMφ, in vitro AAMφ, and NeMφ were again left untreated or were treated overnight with LPS/IFN-γ and then supernatants were taken for cytokine measurement by sandwich ELISA or CBA. Fig. 5 shows that TNF-α was absent in all groups but was induced in response to LPS/IFN-γ in both ThioMφ and ThioMφ plus IL-4. Although TNF-α was induced to a much lesser extent in NeMφ than in ThioMφ in the experiment shown, TNF is consistently up-regulated in NeMφ by LPS/IFN-γ treatment and can sometimes exceed levels seen in similarly treated ThioMφ (data not shown). Therefore, pretreatment with IL-4, or induction in an IL-4 environment, did not prevent TNF-α from being induced by type 1 stimuli, consistent with previous reports from Stout et al. (16). Untreated NeMφ produced high levels of IL-6 relative to ThioMφ, and all groups displayed an increase in IL-6 expression with LPS/IFN-γ treatment (Fig. 5).
Enhancement of TNF-α and IL-6 together provided more evidence that NeMφ can be reprogrammed to increase their proinflammatory capacity. However, the switch to a more classical activation state was not complete. Whereas ThioMφ and in vitro-derived AAMφ produced IL-12p40 in response to LPS/IFN-γ, NeMφ completely failed to do so (Fig. 5). Consistent with previous reports that IL-10 is produced by AAMφ (41), IL-10 was produced to the greatest extent by NeMφ and increased following LPS/IFN-γ treatment.
LPS/IFN-γ enables NeMφ to control Leishmania infection
As LPS/IFN-γ treatment could induce alternatively activated NeMφ to a produce NO and TNF, we asked whether this flexibility was reflected in the functional ability to control infection by an intracellular pathogen. To address this question, L. mexicana promastigotes were used to infect untreated and LPS/IFN-γ-stimulated NeMφ and BMMφ. Parasite survival after 3 h was determined by lysis of the Mφ and measurement of the parasite proliferation by [3H]thymidine incorporation (Fig. 6). Untreated NeMφ had the most surviving promastigotes after 3 h. This likely correlates with enhanced parasite uptake. The NeMφ group recorded more than 10,000 cpm compared with BMMφ with or without LPS/IFN-γ, which had less than one-tenth the cpm (Fig. 6; 1000 cpm). NeMφ may have encountered signals in vivo that increase expression of the receptors involved in the phagocytosis of Leishmania (e.g., complement receptors) (42). The thymidine count for NeMφ treated with LPS/IFN-γ was 2500 cpm, which suggests that this group only allowed uptake of approximately one-fourth the number of promastigotes, perhaps due to the absence or down-regulation of these receptors (Fig. 6).
To assess the capacity of different Mφ populations to control L. mexicana growth, we determined the percentage of Mφ infected after 3 days by microscopy. Fig. 7 A shows that LPS/IFN-γ treatment, and to a lesser extent IFN-γ alone, conferred resistance to L. mexicana in both BMMφ and NeMφ. Indeed, NeMφ appeared better able to clear infection when treated with IFN-γ alone than did BMMφ.
In a separate experiment we examined the role of NO in the leishmanicidal activity observed by NeMφ treated with LPS/IFN-γ. l-NMMA was used to inhibit iNOS activity, as verified by nitrite levels of the day 3 supernatant (Fig. 7,B), and the Mφ were again examined after 3 days of infection (Fig. 7,C). In agreement with the previous experiment, LPS/IFN-γ treatment of NeMφ conferred an increased ability to kill Leishmania, but in the presence of iNOS inhibitors, NeMφ treated with LPS/IFN-γ were rendered as susceptible to infection as those not treated (Fig. 7 C). These data demonstrate that NO is playing an important role in microbicidal activity of LPS/IFN-γ-activated NeMφ.
To visualize these results, experiments were also conducted with promastigotes of a L. mexicana strain that displays red fluorescence (24). In these experiments ThioMφ were used as controls. Fig. 7,D shows that in the untreated groups there is an abundance of red intracellular parasites at day 3, but in all LPS/IFN-γ-treated groups we saw an almost complete absence of L. mexicana. Thus, these type 1 stimuli again conferred full resistance to this intracellular parasite, regardless of Mφ history. These results were confirmed by microscopic examination (Fig. 7,E). Mφ were also stained with anti-Ym1 to confirm their AAMφ status before infection. To our surprise, Ym1 staining was still detectable following LPS/IFN-γ treatment (Fig. 7,D, green). As this contrasted with our previous data in which Ym1 mRNA decreases upon LPS/IFN-γ treatment (Fig. 1,E), we chose to repeat these experiments to measure both protein and RNA simultaneously. The finding confirmed those in Figs. 1,E and 7 D by demonstrating that while mRNA levels decrease in response to LPS/IFN-γ, protein is still detectable (data not shown).
Mφ can be involved in both pro- and antiinflammatory responses, in tissue destructive activities as well as in restorative activities (1, 43). Increasing evidence suggests that Mφ activated in a Th2 setting have tissue repair as a primary function. This hypothesis has been driven by the knowledge that proline, which is an important precursor of collagen and polyamines that are involved in cell proliferation, is produced by AAMφ under the control of arginase (5) and is supported by subsequent data that AAMφ are involved in fibrosis (6). Recent data from our laboratory strongly support this hypothesis, as the key features of alternative activation are induced solely in response to tissue injury (20, 44). It is understandable that a strong wound healing response would occur in the context of helminth infection, as tissue migratory or tissue invasive parasites often lead to physical trauma to host tissue. Importantly, the entire process of tissue repair requires the production of antiinflammatory factors (45). Consistent with this, NeMφ produce many factors associated with tissue repair and reduced inflammation (e.g., arginase 1, IL-10 (Figs. 1,A and 5, respectively) and TGF-β (data not shown)) (46, 47). We wanted to determine if the immune system, faced with a potentially more life-threatening situation (i.e., a bacterial pathogen), had the capacity to rapidly switch from a wound healing to a proinflammatory response. Such functional plasticity would allow AAMφ involved in a tissue repair to respond appropriately to a microbial challenge, and this would be beneficial before the recruitment of new cells to the site of dual infection.
We show herein that a Mφ population generated in the Th2 environment of helminth infection in vivo can respond to proinflammatory stimuli ex vivo. Following overnight incubation of NeMφ with IFN-γ and LPS, there was reduced expression of transcripts for the alternative activation markers arginase, RELM-α, and Ym1. This reduction was accompanied by a dramatic up-regulation of iNOS mRNA. Enzyme activity changes were also seen but to a lesser degree than in control macrophages. Nonetheless, the amount of NO produced was sufficient to fully control infection with an intracellular pathogen, suggesting that NeMφ can divert their resources toward antimicrobial activity if needed. We also observed an increase in surface expression of class II and costimulatory molecules (Fig. 4,A) under the influence of LPS/IFN-γ. This suggests that these AAMφ have enhanced APC function. However, analysis of the NeMφ cytokine profile (Fig. 5) presented an intriguing picture with implications for APC function. Although LPS/IFN-γ treatment appeared to promote a more classically activated phenotype in the AAMφ through elevated TNF-α, IL-6, and NO production, this switch was not complete, and they failed to produce any detectable IL-12p40 (Fig. 5). Thus, although these cells had enhanced antimicrobial killing and increased markers of classical activation, they would not induce Th1 cell development (48). Consistent with this, LPS/IFN-γ treatment caused the up-regulation of IL-10 in NeMφ (Fig. 5), which inhibits Th1 cell responses (49). Lamina propria Mφ from the intestine of mice, which like NeMφ express relatively high IL-10 and TGF-β, also produce little or no IL-12 after stimulation with TLR ligands (50). Although the authors concluded that the IL-10-producing Mφ are hyporesponsive to TLR ligands, only IL-12 was assessed, and thus it is possible that the defect, as in this study, was specific for IL-12.
Consistent with a specific defect in LPS-mediated IL-12 production but not NO production, recent work from Foster et al. has shown that TLR-induced genes with different functions may have diverse regulatory requirements (51). On an epigenetic level, gene-specific modifications of chromatin can lead to the priming of antimicrobial effectors while proinflammatory mediators, such as IL-6, are silenced. This would potentially allow the immune system to effectively deal with microbial pathogens while controlling the pathology associated with inflammation (51). The fact that we have seen the products of some genes associated with an antimicrobial response up-regulated in response to LPS/IFN-γ in NeMφ (i.e., iNOS) but not to IL-12 suggests that these AAMφ may have undergone similar epigenetic modifications in vivo. Indeed, a recent study by Wen et al. has shown that dendritic cells recovered from mice that have survived severe peritonitis-induced sepsis display a chronic suppression of IL-12 expression. The IL-12 deficiency of dendritic cells from this setting is due to histone modification (52). Similarly, the propensity of NeMφ to produce large amounts of IL-10 following LPS/IFN-γ stimulation may reflect chromatin modifications to the IL-10 promoter locus, as shown for superinduction of IL-10 by immune complexes (53).
We have shown that LPS/IFN-γ treatment could confer resistance of NeMφ to infection with L. mexicana promastigotes. Models of murine Leishmania spp. involve a dominant Th1 response leading to the classical activation of Mφ and elimination of the parasites through the production of NO (54). NeMφ were able to control infection almost as well as BMMφ given the same treatment of LPS/IFN-γ (Fig. 7,A). The ability of NeMφ given IFN-γ alone to control parasite numbers more effectively than for BMMφ may be explained by signals that NeMφ encounter in vivo causing the up-regulation of the receptor for IFN-γ. Consistent with this hypothesis, NeMφ produced more iNOS mRNA in response to IFN-γ treatment than ThioMφ (Fig. 1 B).
Previous work in the field has shown that alternative Mφ activation can promote Leishmania growth (55, 56). This is because l-ornithine synthesized from arginase 1 activity, which is largely controlled by IL-4, can be used by the parasites to generate polyamines and to proliferate (56, 57). The role of IL-4 for promoting susceptibility in nonhealing murine L. mexicana infections has been demonstrated, since susceptible mice lacking IL-4 or STAT6 acquire a Th1 response and fail to develop lesions (58, 59, 60). Although these above findings may seem at odds with the results presented herein, IL-4 does not always promote susceptibility. IL-4 has been shown to provide a strong stimulus for the killing of intracellular amastigotes of L. major, as long as low concentrations of IFN-γ are present (61). Similarly, pretreatment with IL-13 has been shown to increase the ability of Mφ to effectively deal with the intracellular protozoan parasite T. gondii after LPS activation (18), and this was found to be due to enhanced NO production (62). We found that susceptibility of NeMφ to L. mexicana decreased dramatically when treated with even IFN-γ alone, as well as with LPS/IFN-γ (Fig. 7,A). Although NeMφ from our B. malayi infection model produced lower levels of NO after LPS/IFN-γ treatment than did ThioMφ (Fig. 1,D) or IL-4 pretreated ThioMφ (data not shown), they were still enough to confer resistance. Even when NeMφ were infected at a 30:1 ratio of parasites/Mφ, they still managed to clear the parasite upon LPS/IFN-γ treatment (Fig. 7, D and E). Thus, despite extensive exposure to type 2 cytokines in vivo, the NeMφ maintained an IFN-γ-responsive phenotype that allowed the manifestation of a fully antimicrobial phenotype. As these experiments were performed ex vivo, they do not allow us to conclude the full infection outcome in the context of a Th2-mediated response that would be maintained in vivo. As such, the data are not directly comparable to studies in which helminth infection exacerbated leishmania infection (55, 63). Nonetheless, they do allow us to dissect the potential for functional adaptation of in vivo-generated AAMφ.
A consistent feature of NeMφ and other helminth-induced Mφ is that they act as potent suppressors of cellular proliferation (22, 64), and several different mechanisms for the suppression have been proposed. We recently provided evidence that TGF-β may be involved to some degree (21) and have now seen that arginase contributes to this suppression since inhibition with nor-NOHA caused a partial reversal of this phenotype. However, both arginase (Fig. 1, A and C) and TGF-β (data not shown) production were partially reversed upon treatment with LPS/IFN-γ, and yet these cells retained full suppressive abilities (Fig. 3,A), so additional factors must be involved. PD-L1 has been implicated in the proliferative suppression observed by Mφ during infection with the platyhelminths T. crassiceps and S. mansoni (38, 39) and would be consistent with a contact-dependent mechanism (22). However, we did not find evidence for up-regulation of PD-L1 on NeMφ compared with the ThioMφ controls (Fig. 4,B). Also, NeMφ did not up-regulate another marker of AAMφ generated in vitro, PD-L2 (Fig. 4 B). Data now emerging from several laboratoties would suggest that the profound suppression observed may be due to multiple antiproliferative mechanisms that in addition to PD-L interactions (38, 39) include TGF-β production (21) lipid mediator release (65) and IL-10 production (66, 67).
We have shown that despite long-term exposure to Th2 cytokines and antiinflammatory signals in vivo, Mφ could respond to the Th1-activating signals LPS/IFN-γ and become effectively antimicrobial. These ex vivo findings may not translate directly to in vivo coinfection studies where cross-regulation will be occurring at the level of the systemic T cell response as well as local effector responses. However, they do provide evidence that Mφ highly polarized in vivo maintain remarkable phenotypic plasticity and are very consistent with recent studies in which the abilities of tumor-promoting AAMφ to become classically activated and promote tumor killing have been demonstrated (68, 69). Watkins et al. demonstrated that systemic treatment of tumor-bearing mice with IL-12 altered the functional response of Mφ from a predominantly antiinflammatory to a more proinflammatory phenotype, including up-regulation of TNF-α and IL-6. Importantly, these changes were sustained over many days (70). These data support the viability of therapeutic approaches to alter Mφ polarization in vivo (71, 72). Our data do not show that AAMφ make a complete conversion to CAMφ but rather that they adopt key aspects of the CAMφ phenotype necessary for a particular function, consistent with the partially overlapping phenotypes observed during in vitro studies (9, 16, 17). Before the therapeutic potential can be fully realized a much greater understanding of the mechanisms and limitations of Mφ plasticity are needed.
We thank Toni Aebischer for helpful discussion, use of reagents, and critical review of the manuscript. We also thank members of the Allen and Maizels laboratories for helpful discussions.
The authors have no financial conflicts of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Wellcome Trust and the Medical Research Council, United Kingdom.
Abbreviations used in this paper: Mφ, macrophage; CAMφ, classically activated macrophage; iNOS, inducible NO synthase; AAMφ, alternatively activated macrophage; NeMφ, nematode-elicited macrophage; PEC, peritoneal exudate cell; RELM-α, resistin-like molecule α; WT, wild type; BMMφ, bone marrow-derived macrophage; PD-L, programmed death ligand; l-NMMA, NG-monomethyl-l-arginine; nor-NOHA, Nω-hydroxy-nor-l-arginine; CBA, cytokine bead array; ThioMφ, thioglycollate-elicited macrophage; MFI, mean fluorescence intensity.