Tumoristatic effects of anti-CD40 mAb-activated macrophages involve nitric oxide and tumour necrosis factor-α

Authors


Dr Alexander L. Rakhmilevich, University of Wisconsin-Madison, K4/413 Clinical Science Center, Department of Human Oncology, 600 Highland Avenue, Madison, WI 53792, USA.
Email: rakhmil@humonc.wisc.edu
Senior author: Alexander L. Rakhmilevich

Summary

Effector cells of the innate immune system have diverse functions that can result in tumour inhibition or tumour progression. Activation of macrophages by CD40 ligation has been shown to induce antitumour effects in vitro and in vivo. Here we investigated the role of nitric oxide (NO) and tumour necrosis factor-α (TNF-α) as mediators in the tumoristatic effects of murine peritoneal macrophages activated with agonistic anti-CD40 monoclonal antibody (αCD40) alone and following further stimulation with bacterial lipopolysaccharide (LPS). We found that macrophages activated in vivo by αCD40 exhibited tumoristatic activity in vitro against B16 melanoma cells; the tumoristatic effect correlated with the level of NO production and was enhanced by LPS. Use of the NO inhibitor l-nitro-arginine-methyl esterase (L-NAME) and evaluation of macrophages from inducible NO synthase (iNOS)-knockout (KO) mice following αCD40 activation showed reduced tumoristatic activity. CD40 ligation enhanced expression of TNF-α. Macrophage tumoristatic activity following αCD40 treatment was reduced by TNF-α mAb or use of macrophages from TNF-α-KO mice. However, further stimulation of αCD40-activated macrophages with LPS resulted in strong tumoristatic activity that was much less dependent on NO or TNF-α. Taken together, these results suggest that NO and TNF-α are involved in, but not solely responsible for, the antitumour effects of macrophages after activation by CD40 ligation.

Abbreviations:
[3H]-TdR

[3H]-thymidine

αCD40

anti-CD40 monoclonal antibody

CD40L

CD40 ligand

Fas-L

Fas-ligand

GM-CSF

granulocyte–macrophage colony-stimulating factor

IFN

interferon

IL

interleukin

iNOS

inducible nitric oxide synthase

i.p.

intraperitoneal

KO

knockout

L-NAME

l-nitro-arginine-methyl esterase

LPS

lipopolysaccharide

mAb

monoclonal antibody

NO

nitric oxide

PEC

peritoneal exudate cells

TAM

tumour-associated macrophage

TLR4

Toll-like receptor 4

TNF-α

tumour necrosis factor-α

Introduction

Activated macrophages are multifunctional cells of the innate immune system which can play a role in neoplastic or antineoplastic processes.1 Tumour-associated macrophages (TAMs) have been characterized as M2 phenotypic macrophages, and can produce growth factors and enzymes that promote tumour cell proliferation, angiogensis, and tissue remodelling, all leading to tumour progression.2 In contrast, proinflammatory, antitumour macrophages that are an M1 phenotype secrete inflammatory mediators that are directly involved in tumour cell killing as part of the innate immune response against tumour cells.3 The tumour cytotoxic functions of these macrophages required activation by stimuli such as bacterial lipopolysaccharide (LPS) or cytokines [i.e. interferon (IFN)-γ or granulocyte–macrophage colony-stimulating factor (GM-CSF)].3

CD40 ligation has been investigated for its role in the induction of antitumour immunity. The cross-linking of CD40 on antigen-presenting cells by agonistic αCD40 has been shown to eliminate the requirement for T helper cell stimulation in the production of a cytotoxic T lymphocyte-dependent antitumour response.4,5 Furthermore, we have shown that αCD40 can induce antitumour (including antimetastatic) effects involving natural killer (NK) cells or macrophages (data not shown).6 Although ligation of CD40 on macrophages appears to play a role in their antiproliferative and apoptogenic activity against tumour cells7 as well as direct tumour cell killing,8 the specific mechanism(s) involved in this antitumour activity remains to be determined, and is the focus of this study.

The activation of macrophages and monocytes by CD40 ligation can lead to the production of nitric oxide (NO)9 and several cytokines, including IFN-γ, interleukin (IL)-12, IL-1β, IL-6, IL-8 and TNF-α.7,10,11 Production of NO following CD40 ligation of murine peritoneal macrophages requires IFN-γin vitro,9 and decreased NO production by macrophages from CD40 ligand (CD40L)-KO mice results in reduced macrophage effector function in vivo.12 Similarly, ligation of CD40 on monocytes has been shown to stimulate TNF-α release in an IFN-γ-dependent manner.13 Studies involving the activation of murine macrophages with IFN-γ and LPS have demonstrated that TNF-α and NO can act synergistically as killing mechanisms employed by activated macrophages.14,15 Thus, pathways involving NO and TNF-α represent mechanisms of contact-independent macrophage-mediated tumoritoxicity,16,17 although contact-dependent cytotoxicity via membrane-bound TNF-α has also been reported.18

IFN-γ has pluripotent and well-characterized roles in the activation of macrophages, especially in priming macrophages to further stimulation by LPS.19 We have found that IFN-γ is required for the activation of macrophages by αCD40 and their priming to LPS.7 The purpose of the current study was to determine whether NO and TNF-α, known macrophage cytotoxic effector molecules, are involved in the antitumour effects of αCD40-activated macrophages against the B16 melanoma.

Materials and methods

Mice and cell culture

Female C57BL/6, inducible NO synthase knockout (iNOS-KO) and TNF-α-KO mice (7–10 weeks old) were obtained from Harlan Sprague Dawley, Madison, WI (wild-type animals) or the Jackson Laboratory, Bar Harbor, ME (KO animals). All animals were housed in the university-approved facilities and were handled according to National Institutes of Health and University of Wisconsin-Madison Research Animal Resource Center guidelines. The murine B16 melanoma cell line was grown in RPMI-1640 medium (Mediatech, Herndon, VA), supplemented with penicillin (100 U/ml), streptomycin (100 µg/ml), l-glutamine (2 mm) (all from Life Technologies, Inc., Grand Island, NY), and 10% heat-inactivated fetal calf serum (FCS; Sigma Chemicals, St Louis, MO). Cells were maintained at 37° in a humidified 5% CO2 atmosphere.

Agonistic αCD40 and in vivo macrophage activation

The FGK 45·5 hybridoma cells producing an agonistic rat anti-mouse αCD4020 were a gift from Dr Fritz Melchers (Basel Institute for Immunology, Basel, Switzerland). αCD40 was purified and confirmed to be specific as previously described.6 For in vivo activation studies, mice were injected intraperitoneally (i.p.) with 0·25–0·5 mg of either αCD40 or control rat immunoglobulin G (IgG) (Sigma).

Peritoneal macrophage preparation and cytostatic assay

Antitumour cytostatic activity of macrophages was determined by the inhibition of [3H]-thymidine ([3H]-TdR) incorporation into target tumour cells, as described previously.6 Briefly, peritoneal exudate cells (PEC) were collected by peritoneal lavage with 5 ml of cold RPMI-1640 complete medium from groups of three or four mice treated the indicated number of days earlier with αCD40 or rat IgG. PEC were plated at 2·0 × 105 cells in 0·1 ml per well of a 96-well flat-bottom plate (Corning Inc, Corning, NY). After 2 hr, the monolayer was washed three times with warm RPMI to remove non-adherent cells. Flow cytometry revealed that 95% of adherent cells were macrophages, based on F4/80 expression. B16 tumour cells (1 × 104 cells/well) were added in triplicates for 48 hr to wells in the absence or presence of 10 ng/ml LPS (Sigma). As a control, B16 cells were cultured in the absence of macrophages. To estimate tumour cell proliferation, [3H]-TdR (1 µCi/well; PerkinElmer, Boston, MA) was added to cultures for the last 6 hr of the incubation period. [3H]-TdR incorporation by proliferating tumour cells was determined by beta-scintillation of total cells harvested from the wells onto glass fibre filters (Packard, Meriden, CT), using the Packard Matrix 9600 Direct beta-counter (Packard). Results are expressed as mean counts per 5 min of triplicate wells ± standard error (SE). In these assays, macrophages cultured in the absence of tumour cells showed negligible [3H]-TdR incorporation (data not shown). In vitro inhibition studies to block NO and TNF-α were performed where l-nitro-arginine-methyl esterase (L-NAME, 5 mm; Sigma) or anti-mouse TNF-α monoclonal antibody (mAb) (10 µg/ml; eBioscience, San Diego, CA) was added at the initiation of the experiment and remained present in the media throughout the 48-hr cytostatic assay.

Nitrite production

Peritoneal macrophages were prepared and cocultured with B16 cells for 48 hr in media with or without LPS, as described above for the macrophage cytostatic assay. Supernatants were collected after 42 hr and nitrite accumulation, an indicator of NO production, was determined using Griess reagent (Sigma). Equal volumes (50 µl) of supernatants and Griess reagent were mixed for 10 min, and the absorbance at 540 nm was measured by a microplate reader and compared to a standard nitrite curve ranging from 0 to 120 µm.

Detection of intracellular cytokines by flow cytometry

C57BL/6 mice were injected i.p. with 0·5 mg αCD40 or rat IgG. PEC were obtained 1, 2, 5, 8 or 11 days after αCD40 treatment and 1 day after rat IgG treatment. PEC were seeded into 6-well cell culture clusters (Costar; Corning Inc, Corning, NY) at a concentration of 1 × 106 cells/ml, 5 ml/well, and enriched for macrophages by adherence to plastic for 1·5 hr prior to staining. To enable accumulation of cytoplasmic cytokines in the endoplasmic reticulum, cells were incubated in medium containing monensin (1 µl/ml; eBioscience) for 4 hr. Cells were harvested by gentle scraping with a rubber policeman and assayed for intracellular TNF-α as described elsewhere21 and according to the eBioscience 2004 Catalog & Reference Manual. Briefly, macrophages were resuspended in phosphate-buffered saline (PBS) with 2% FCS (flow buffer) at a concentration of 1 × 106 cells/ml, and stained with anti-F4/80-allophycocyanin (APC) mAb (BM8; eBioscience), 2 µg/1 × 106 cells at 4° for 40 min. Rat IgG2a-APC was used as an isotype control. Cells were centrifuged, the cell pellet was resuspended in 0·1 ml of flow buffer, and 0·1 ml of IC Fixation Buffer (eBioscience) was added for 20 min. After fixation, cell membranes were permeabilized with Permeabilization Buffer (eBioscience) for 5 min, and washed in the same buffer two additional times. After the final cell wash, the pellet was resuspended in 0·1 ml Permeabilization Buffer, and cells were stained with anti-TNF-α-fluorescein isothiocyanate (FITC) mAb (MP6-XT22) or isotype control rat IgG1-FITC, 2 µg/1 × 106 cells, at 4° for 40 min. All mAbs were purchased from eBioscience. Finally, cells were resuspended in 0·3 ml of flow buffer and analysed using a FACScan cytofluorometer (Becton Dickinson, San Jose, CA). Data collected for 10 000 events/sample were analysed with the CellQuest software (Becton Dickinson, San Jose, CA).

Statistical analysis

All data are presented as mean ± SE. Statistical significance was determined by the two-tailed Student's t-test. The relationship between the tumoristatic activity and levels of nitrites produced by PEC was examined by regression analysis.

Results

Tumoristatic activity of αCD40-activated macrophages correlates with NO production in a time-dependent manner

Previous work in our laboratory has shown that αCD40 treatment in vivo activates macrophages to inhibit tumour cell growth in vitro via induction of apoptosis.7 To elucidate the antitumour effector mechanisms of αCD40-activated macrophages, we used a [3H]-TdR incorporation assay to measure the antiproliferative activity of macrophages on cocultured B16 melanoma cells, and compared it with the production of NO, a known soluble cytotoxic effector molecule. Macrophages collected 24 hr to 11 days after in vivoαCD40 treatment demonstrated increased tumoristatic activity against B16 melanoma cells compared with control macrophages from rat IgG-treated animals (Fig. 1a). The greatest tumoristatic activity was observed for macrophages collected on day 5 after αCD40 injection. NO production from αCD40-activated macrophages was detectable between days 3 and 11, with the highest concentration detected on day 5 (Fig. 1b). Thus, both tumoristatic activity and NO production, as measures of macrophage activation following αCD40 treatment, peaked on day 5 and decreased by day 11, as measured in medium without LPS. There was a correlation (P < 0·0062) between the tumoristatic activity and levels of nitrites in the supernatants after coculture of macrophages and tumour cells. NO was undetectable in cultures consisting of B16 melanoma cells in the absence of macrophages, or rat IgG-stimulated macrophages in the presence of B16 tumour cells. As additional controls, macrophages collected 3 or 5 days after rat IgG administration did not show evidence of activation, based on assessment of tumoristatic activity or NO production (data not shown).

Figure 1.

 αCD40-activated macrophages are tumoristatic and produce nitric oxide (NO) in a time-dependent manner. C57BL/6 mice (four per group) were treated with 0·5 mg αCD40 on various days (1, 2, 3, 5, 8 or 11 days) before peritoneal exudate cells (PEC) were harvested. Another group was treated with 0·5 mg rat immunoglobulin G (IgG) 1 day before PEC were harvested. (a) Macrophage tumoristatic activity was measured by inhibition of B16 tumour cell proliferation based on [3H]-TdR incorporation in tumour cells as described in ‘Materials and methods’. Lipopolysaccharide (LPS) (10 ng/ml) was added to the assays as indicated. # Indicates counts below 1000. (b) Supernatants from the same combinations shown in (a) were tested for nitrite levels with the Griess test. # Indicates the absence of detectable NO. Correlation of [3H]-TdR counts incorporated by B16 cells (a) and NO production (b) by αCD40-activated macrophages, in medium without LPS: y = −0·5894x + 41·1, r = −0·9372, P < 0·0062; or in the presence of LPS: y = −7·354x + 76·15, r = −0·8754, P < 0·027.

LPS activates macrophages via Toll-like receptor 4 (TLR4) and may serve as an additional stimulus to further activate IFN-γ-primed macrophages.15,22 To determine the responsiveness of αCD40-activated macrophages to this second stimulus, αCD40-activated macrophages were cocultured with B16 melanoma cells in the presence of LPS and evaluated for in vitro tumoristatic activity and NO production. Figures 1(a) and (b) show that the presence of LPS resulted in augmented tumoristatic activity and augmented NO production. Moreover, these LPS-augmented cultures revealed a correlation (P < 0·027) between enhanced tumour growth inhibition and NO production, where increased nitrite levels were detectable as early as day 1, peaked on day 3, and remained elevated until day 11. Thus, in vivoαCD40-activated macrophages demonstrated significant in vitro tumoristatic activity in a time-dependent fashion that correlated with NO release.

L-NAME inhibits in vitro tumorstatic activity of αCD40-activated macrophages

To examine the role of NO in the tumoristatic activity of αCD40-activated macrophages in vitro, a [3H]-TdR incorporation assay was carried out in the presence of L-NAME, an inhibitor of NOS. The presence of this inhibitor substantially inhibited the tumoristatic effect of αCD40-activated macrophages against B16 melanoma cells compared with non-inhibited αCD40-activated macrophages (Fig. 2a; P < 0·05). NO production was not detectable by Griess assay in the presence of L-NAME (Fig. 2b). The possibility of direct L-NAME toxicity to macrophages was ruled out by the observance of intact tumoristatic activity and NO production by αCD40-activated macrophages treated in vitro with D-NAME, the inactive isomer of L-NAME (data not shown). The presence of L-NAME also did not affect the growth of B16 cells alone (data not shown). NO appears to play a major role in the antitumour effector mechanism of these αCD40-activated macrophages, as the presence of L-NAME resulted in undetectable levels of NO and prevented much of the tumoristatic activity induced by αCD40 activation of macrophages. However, L-NAME did not completely block the tumoristatic activity induced by activation with αCD40 + LPS, despite complete inhibition of detectable NO production (Figs 2a and b). Under the conditions tested, L-NAME inhibited approximately 46% (mean of three experiments with a range of 28·7–63·5% inhibition) of the LPS-enhanced αCD40-activated macrophage-mediated tumoristatic effect despite abrogation of NO production. Therefore, LPS stimulation of CD40-ligated macrophages involved an additional, non-NO-mediated effector mechanism(s).

Figure 2.

 Nitric oxide (NO) is involved in the tumoristatic activity of αCD40-activated macrophages. Peritoneal exudate cells (PEC) were collected from C57BL/6 mice (four per group) 5 days after injection of 0·25 mg αCD40 or rat immunoglobulin G (IgG). (a) Macrophage tumoristatic activity was assessed in a [3H]-TdR assay as described in ‘Materials and methods’. l-nitro-arginine-methyl esterase (L-NAME) (5 mm) was added to the adherent macrophages and B16 cell cocultures in the presence or absence of 10 ng/ml lipopolysaccharide (LPS). As a control, B16 cells cultured in the absence of effector cells incorporated an average of 125·3 × 103 (media alone) or 130·3 × 103 (with LPS) [3H]-TdR counts. *P < 0·05. (b) NO production was measured using Griess reagent for the same cultures shown in (a). # Indicates the absence of detectable NO. Results shown in (a) and (b) are representative of three similar experiments. (c) C57BL/6 (control) or inducible nitric oxide synthase knockout (iNOS-KO) mice (three per group) were injected with 0·25 mg αCD40 or rat IgG and PEC were harvested 5 days later. Macrophage tumoristatic activity was measured in a [3H]-TdR assay as described in ‘Materials and methods’. As a control, B16 cells cultured in the absence of effector cells incorporated a mean of 115·3 × 103 (media alone) or 120·3 × 103 (with LPS) [3H]-TdR counts. Results are representative of four similar experiments. **P = 0·109.

Tumoristatic activity of iNOS-deficient macrophages activated by αCD40

Having observed that the tumoristatic activity induced by αCD40 activation of macrophages was nearly abrogated by inhibiting NO production in vitro with L-NAME, we next addressed the effect of CD40 ligation on macrophages from iNOS-KO mice. iNOS-deficient mice and control C57BL/6 mice were injected with αCD40 or rat IgG. A macrophage tumoristatic assay against B16 melanoma cells showed that αCD40 activation of macrophages from both control and iNOS-KO mice induced tumoristatic activity (Fig. 2c). However, the tumoristatic effect of macrophages from iNOS-KO mice was somewhat less than that exhibited by macrophages from control C57BL/6 mice in this experiment (P = 0·109). Assessment of all four separate similar experiments showed that the degree of tumoristatic activity (αCD40 macrophages versus rat IgG macrophages) was significantly less in iNOS-deficient mice (mean percentage inhibition = 42·8%) compared with wild-type mice (mean percentage inhibition = 62·7%) (P = 0·030). As expected, macrophages from wild-type mice, but not iNOS-KO mice, produced NO following activation with αCD40 (data not shown). Addition of LPS to αCD40-activated macrophages further stimulated the effector cells from both iNOS-KO and C57BL/6 mice, although the degree of enhanced activation was minimal in iNOS-deficient macrophages (Fig. 2c). Together these results show that activated macrophages from iNOS-KO mice had reduced tumoristatic activity, supporting the hypothesis that NO is involved in, but not the sole mediator of, the antitumour effect induced by αCD40-activated macrophages.

In vivo CD40 ligation on macrophages induces production of endogenous TNF-α

As some tumoristatic activity persisted in iNOS-deficient macrophages, the role of TNF-α as an additional effector mechanism of αCD40-activated macrophages was investigated. In our previous experiments, cytokines (such as IFN-γ) released by αCD40-activated macrophages were not reproducibly detected by enzyme-linked immunosorbent assay (ELISA), whereas intracellular staining of macrophages yielded much clearer results.7 Therefore, to determine the effect of CD40 ligation on TNF-α production by activated macrophages, intracellular staining for TNF-α and flow cytometric analyses were carried out. Mice were treated with αCD40 and PEC were harvested between 24 hr and 11 days later. PEC collected from animals 24 hr after treatment with rat IgG served as a control. Evaluation of F4/80+ macrophages showed that the expression of TNF-α peaked on day 1 and slowly declined over the next 10 days (Fig. 3). Furthermore, 3 days after αCD40 treatment, 68% of F4/80+ macrophages showed detectable TNF-α expression, compared with 5% in rat IgG-treated macrophages (data not shown); similar data were obtained in two separate experiments. Thus, our data corroborate studies13 that have shown CD40-induced production of TNF-α by macrophages.

Figure 3.

 Time-dependent tumour necrosis factor (TNF)-α expression in αCD40-activated macrophages. Peritoneal exudate cells (PEC) were collected at the indicated number of days after intraperitoneal (i.p.) injection of C57BL/6 mice (three per group) with 0·5 mg αCD40 or rat immunoglobulin G (IgG) (day 1). Staining of F4/80+ macrophages for intracellular TNF-α was performed as described in ‘Materials and methods’. The data are presented as histograms where filled peaks represent isotype control and open peaks represent specific staining for endogenous TNF-α in F4/80+ macrophages. Mean fluorescent intensities of TNF-α+ staining for cells that express F4/80+ are indicated for each time-point.

Inhibition or lack of TNF-αin vitro decreases the tumoristatic effect of αCD40-activated macrophages

Although TNF-α production was greatest 24 hr after αCD40 administration and tumoristatic activity was most prominently observed around day 5 following treatment, we hypothesized that TNF-α may still be involved in the antitumour activity through direct actions or by mediating autocrine or paracrine effects. To further clarify the cytotoxic effect of αCD40-activated macrophages, we investigated the role of TNF-α. αCD40-activated macrophages from C57BL/6 mice were evaluated for their antitumour effects in vitro against B16 melanoma tumour targets in the presence of neutralizing TNF-α mAb. Figure 4(a) shows that the tumoristatic activity of αCD40-activated macrophages was virtually eliminated by neutralizing TNF-α mAb, compared with the tumoristatic activity of CD40-ligated macrophages in the absence of blocking TNF-α mAb (P < 0·001). The efficacy of the neutralizing TNF-α mAb was verified by the ability of the blocking mAb to prevent the cytotoxic effect of soluble murine TNF-α on B16 melanoma cells over a 48-hr period in a [3H]-TdR assay (data not shown). Additionally, anti-TNF-α mAb did not affect the growth of B16 cells alone, in the absence of macrophages (data not shown). As other studies have shown that TNF-α can induce the expression of iNOS RNA and NO production in macrophages,23,24 a Griess assay was performed on the supernatants from macrophage–tumour cocultures containing anti-TNF-α mAb to determine whether NO production was affected. In three separate experiments, samples containing neutralizing anti-TNF-α mAb had significantly reduced nitrite levels (data not shown). Thus, in summary, these experiments clearly showed that the tumoristatic effect of αCD40-activated macrophages involved TNF-α, which may also influence the production of NO.

Figure 4.

 Tumour necrosis factor (TNF)-α is involved in the tumoristatic activity of αCD40-activated macrophages. (a) Peritoneal exudate cells (PEC) were collected from C57BL/6 mice (three per group) 5 days after injection of 0·25 mg αCD40 or rat immunoglobulin G (IgG). The macrophage tumoristatic assay was performed as described in ‘Materials and methods’. Anti-TNF-α neutralizing monoclonal antibody (mAb) (10 µg/ml) was added to the cocultures as indicated at the initiation of the experiment. As a control, B16 cells cultured in the absence of effector cells incorporated an average of 116·7 × 103[3H]-TdR counts. *P < 0·001. Results are representative of five similar experiments. (b) Control C57BL/6 mice and TNF-α knockout (KO) mice (three per group) were treated with either 0·5 mg αCD40 or rat IgG. Three days later, macrophages were cocultured with B16 tumour cells in the [3H]-TdR incorporation assay in medium alone or in the presence of 10 ng/ml lipopolysaccharide (LPS). # Indicates counts below 1000. (c) Cell culture supernatants from (b) were tested for nitric oxide (NO) by Griess reagent. # indicates the absence of detectable NO. Results presented in (b) and (c) are representative of two similar experiments.

To evaluate the requirement for TNF-α involvement in the tumoristatic activity of αCD40-activated macrophages, TNF-α-KO and control mice were treated with αCD40, and macrophages were evaluated for their tumoristatic activity against B16 melanoma cells. TNF-α-KO mice activated with αCD40 showed significantly decreased tumoristatic activity against B16 cells compared with αCD40-activated wild-type macrophages (Fig. 4b; P = 0·025). Nevertheless, these αCD40-activated macrophages from TNF-α-KO mice still mediated tumoristasis compared with rat IgG-treated macrophages also from TNF-α-KO mice (P = 0·0053). The degree of tumour inhibition by αCD40-activated macrophages compared with rat IgG-treated macrophages was stronger in wild-type mice (75·7%) than in TNF-α-deficient mice (45·3%) (P = 0·031). This suggests that TNF-α is involved in tumoristasis by wild-type macrophages.

However, as some tumoristatic activity persisted despite the lack of TNF-α in the TNF-α-KO mice, additional mechanisms must account for the antitumour activity of αCD40-activated, TNF-α-KO macrophages, suggesting that other mechanisms may also be used by αCD40-activated macrophages from C57BL/6 mice. Additional stimulation with LPS of αCD40-activated macrophages from TNF-α-KO mice resulted in strong tumoristatic activity. The enhanced tumoristatic activity following LPS stimulation seen for rat IgG-treated macrophages may reflect an undetected, subclinical infection in these animals which primed macrophages to LPS. NO production by αCD40-activated macrophages from TNF-α-KO mice was also detected (Fig. 4c), showing that αCD40 treatment can result in iNOS expression and activity that is independent of TNF-α. As CD40 ligation can stimulate macrophage production of several proinflammatory cytokines (i.e. IL-1β, IL-6, IL-8, IL-12 and IFN-γ), the presence of these factors in combination with LPS triggering is a likely explanation for the production of NO by αCD40-activated macrophages from TNF-α-KO mice.

To determine whether CD40-ligated macrophage tumoristatic activity can occur in the absence of both NO and TNF-α, we tested macrophages from wild-type mice in the presence of L-NAME and anti-TNF-α mAb (Fig. 5a) and iNOS-deficient mice and wild-type mice stimulated with αCD40 or rat IgG and cultured in the presence or absence of neutralizing TNF-α mAb (Fig. 5b). In Fig. 5(a), the tumoristatic activity of αCD40-activated macrophages was reduced in the presence of both L-NAME and anti-TNF-α mAb (P = 0·009). However, even in the presence of these agents, αCD40-activated macrophages exerted an antiproliferative effect on B16 cells, compared with B16 growth in the absence of any macrophages (P = 0·001). Similarly, the inability of L-NAME and anti-TNF-α mAb to abrogate the tumoristatic effects of αCD40-activated macrophages to the level of control rat IgG-treated macrophages suggested that other factors may be involved in the tumoristatic activity of these macrophages (P = 0·0019). In another model system, neutralization of TNF-α secreted by αCD40-activated macrophages from iNOS-KO mice significantly reduced the tumoristatic activity exhibited by these macrophages in the absence of LPS (Fig. 5b; P = 0·036). The relatively low proliferation of B16 cells in this assay possibly reflects a suboptimal growth pattern of B16 cells at the time of cell collection. Still, the activated macrophages caused significant tumour growth inhibition when compared with [3H]-TdR incorporation (67·6 × 103 counts) by B16 cells cultured without any macrophages (P = 0·0003). Futhermore, LPS was still able to activate greater tumoristasis. Proliferation of B16 cells cultured with these iNOS-KO, αCD40-activated macrophages and neutralizing TNF-α mAb in the presence of LPS was less than the proliferation of B16 cells in the presence of the same macrophage population, but without LPS (P = 0·028), and was dramatically less than the proliferation of B16 cells cultured without any macrophages (P = 0·0001). Together, these findings document the involvement of NO and TNF-α, as well as some additional factors, in the tumoristatic activity of αCD40-activated macrophages.

Figure 5.

 The combined role of tumour necrosis factor (TNF)-α and nitric oxide (NO) in the tumoristatic activity of macrophages activated by αCD40. (a) Peritoneal exudate cells (PEC) were collected from C57BL/6 mice (two per group) 3 days after injection of 0·25 mg αCD40 or rat immunoglobulin G (IgG). Macrophage tumoristatic activity was assessed in a [3H]-TdR assay as described in ‘Materials and methods’. l-nitro-arginine-methyl esterase (L-NAME) (5 mm) and anti-TNF-α monoclonal antibody (mAb) (5 µg/ml) were added to the adherent macrophages and B16 cell cocultures in the presence or absence of 10 ng/ml lipopolysaccharide (LPS). (b) PEC were collected from inducible nitric oxide synthase knockout (iNOS-KO) mice (three per group) 5 days after injection of 0·25 mg rat IgG or αCD40. The macrophage tumoristatic assay was performed as described in ‘Materials and methods’. Anti-TNF-α neutralizing mAb (10 µg/ml) was added to the cocultures in the presence or absence of 10 ng/ml LPS. As a control, B16 cells cultured in the absence of effector cells incorporated an average of 67·6 × 103 (media alone) or 65·4 × 103 (with LPS) [3H]-TdR counts. Results are representative of two similar experiments.

Discussion

We previously showed that the activation of macrophages by αCD40 resulted in tumour cell growth inhibition in vitro7 and antitumour effects in vivo (data not shown). The tumoristatic activity by αCD40-activated macrophages observed in this study, as measured by [3H]-TdR incorporation into B16 melanoma cells, was previously shown to reflect apoptotic death of tumour cells.7 In this study, we have shown that in vivo activation of C57BL/6 peritoneal macrophages by αCD40 resulted in production of the proinflammatory molecules NO and TNF-α, and that these factors are involved as cytotoxic mediators of the tumoristatic effects of macrophages. Together with our previous results, which showed that macrophages activated by αCD40 secreted proinflammatory cytokines (IFN-γ and IL-12) (data not shown), this suggests that CD40 ligation on macrophages promotes the development of antitumour M1 macrophages.

Activated antitumour macrophages have been shown to exert their antitumour effects through a number of cytotoxic mediators, including NO and TNF-α.15 Here, we have demonstrated that NO and TNF-α play a major role in the tumoristatic activity of αCD40-activated macrophages, as in vitro blockade of NO and TNF-α functions dramatically decreased the level of macrophage-mediated tumoritoxicity. However, these factors are only partially involved in the tumoristatic mechanisms of αCD40 + LPS-activated macrophages. We believe that αCD40 + LPS-stimulated macrophages resemble the characteristics of αCD40-activated antitumour macrophages in vivo, which probably utilize several mechanisms of tumour cytotoxicity. In fact, our in vivo experiments testing the antitumour efficacy of αCD40 in the treatment of subcutaneous B16 melanoma in iNOS-KO mice showed that the antitumour effects of αCD40 occurred even in the absence of NO (data not shown). As exogenous LPS was not included as part of this in vivo experimental protocol, it is possible that factors from the tumour microenvironment, including tumour expressed ligands or factors, endogenous endotoxin, or αCD40-activated macrophage-derived factors (IFN-γ, TNF-α, etc.) could substitute for LPS in further stimulating αCD40-activated macrophages to result in the observed NO-independent antitumour effects. In a separate study, we found that coculture of αCD40-activated macrophages with B16 melanoma cells resulted in increased NO production as a measure of macrophage activation (data not shown).

Classical activation of macrophages by priming with IFN-γ and further stimulation with LPS has been well characterized.25 A direct comparison of genes induced by priming of macrophages by αCD40 versus IFN-γ has not been reported, though it is presumed that each macrophage stimulus can elicit a distinct pattern of new gene expression.26 An infection model showed that CD40 stimulation induced macrophage functions that were distinct from those activated by IFN-γ and LPS.27 Our findings similarly indicate a difference in macrophage cytotoxicity mechanisms, where further activation with LPS of αCD40-ligated macrophages did not seem to substantially involve NO and TNF-α in the augmented tumoristasis of B16 melanoma cells. Additionally, in some experiments, we have observed baseline activation of non-CD40-ligated macrophages, as in Fig. 2, where rat IgG-treated macrophages inhibited B16 cell growth by 65%. Notably, the rat IgG-treated macrophages did not produce NO, and their tumoristatic activity was not affected by L-NAME, which alludes to other mechanisms involved in this tumour growth inhibition.

In this study, we have examined the possible in vitro roles of NO and TNF-α to determine the extent of their involvement in the tumoristatic mechanism of αCD40-activated macrophages. We previously reported that macrophages activated in vivo by αCD40 can secrete IFN-γ, and that this endogenous IFN-γ was required for the activation of macrophages by αCD40.7 IFN-γ may also play a role in tumour cell killing, or may play a role in activating other cytotoxic effector molecules such as NO.9 Recently, we extended our previous work and observed that addition of neutralizing anti-IFN-γ mAb in vitro nearly abrogated the tumoristatic effects against B16 melanoma of macrophages from C57BL/6 mice activated with αCD40 in the absence of LPS (data not shown). Similarly, macrophages from IFN-γ-KO mice activated with αCD40 in vivo revealed their inability to mediate tumoristasis of B16 cells as compared with that of αCD40-activated wild-type macrophages.7 Notably, in both of these models, αCD40-activated macrophages did not produce any NO (data not shown). Clarification of the specific role of IFN-γ in the tumoristatic activity of αCD40-activated macrophages remains to be obtained. While these data support the importance of NO in the in vitro antitumour effects of αCD40-activated macrophages, the impaired secretion of other cytotoxic factors may also have contributed to the abolished antitumour activity.

Multiple mechanisms may be involved in macrophage-mediated tumour cell killing.28 The presence of cytostatic activity in the absence of NO or TNF-α, especially when αCD40-activated macrophages are additionally stimulated by LPS, suggests that these effector cells employ antitumour mechanisms in addition to NO and TNF-α. Other potential mediators of macrophage cytotoxic activity include the TNF-related apoptosis inducing ligand29 and Fas-ligand (Fas-L).30 Additionally, αCD40-activated peritoneal macrophages demonstrated increased Fas-L and lymphocyte function-associated antigen-1 (LFA)-1 (CD11/CD18) expression (data not shown). Both Fas-L and LFA-1 are macrophage cell surface-expressed molecules that may be involved in contact-dependent macrophage-mediated cytotoxicity.31 Tsung et al. showed that paraformadehyde-fixed, IL-12-activated macrophages were able to exert cytotoxic effects independent of perforin, Fas, and NO.32 This suggests that some components of cytotoxicity involve contact-dependent membrane mechanisms, not requiring viable macrophages.

In conclusion, we have found that αCD40-activated macrophages demonstrated tumoristatic activity against B16 melanoma cells that involved the cytotoxic effector molecules NO and TNF-α. The blockade of both effector molecules in vitro resulted in significant decreases in macrophage-mediated tumour cell death. However, the roles of these factors were seen to depend on the nature of macrophage activation; the addition of LPS caused additional stimulation of αCD40-activated macrophages such that their mechanisms of tumoritoxicity could not be solely attributed to NO or TNF-α. Therefore, while these effector molecules may be involved in the antitumour effects of αCD40-activated macrophages in vivo, further studies are needed to determine the effect of the tumour microenvironment and other secreted molecules, as well as possible contact-dependent mechanisms of macrophage killing. In addition, these findings must be appropriately considered in the context of human disease. Currently, the role of NO in the tumour microenvironment of cancer patients is poorly understood. Depending on concentration, NO may play a role in pro-tumour events, including angiogenesis and metastasis at low concentrations, or cytotoxicity and apoptosis at high concentrations.33 While TAMs have been shown to express iNOS, some studies showed that the NO levels found in tumour extracts were at least 1–2 orders of magnitude lower than the level often associated with macrophage-mediated cytotoxicity.34 As an immunotherapeutic modality, it is possible that αCD40 will activate TAMs to produce greater amounts of NO in vivo, thus contributing to the antitumour effects we have observed.35

As key members of the innate immune system, macrophages may play a central role in the complex relationship between inflammation and the development of cancer.36 M2 TAMs secrete factors that can promote tumour cell growth and suppress parts of the adaptive immune response.2,37 Some immunotherapeutic approaches have been able to convert pro-tumour M2 macrophages into antitumour M1 macrophages.38,39 In this study, we have shown that αCD40 can induce antitumour activity of macrophages that involves the cytotoxic effector molecules NO and TNF-α. Macrophage activation via CD40 ligation represents a novel immunotherapeutic strategy, particularly when combined with other macrophage-activating agents, such as CpG,35 that may act by influencing the effector functions of TAMs to convert them from a tumour-promoting M2 phenotype to a tumour-retarding M1 phenotype.

Acknowledgements

This work was supported by NIH Grants CA87025 and CA32685, a grant from the US-Israel Binational Science Foundation, and a grant from the Midwest Athletes Against Childhood Cancer Fund. We would like to thank Dr Jacquelyn Hank for helpful scientific discussions.

Ancillary