p50 Nuclear Factor-κB Overexpression in Tumor-Associated Macrophages Inhibits M1 Inflammatory Responses and Antitumor Resistance

Introduction

The presence of inflammatory cells in solid tumors, along with epidemiologic evidence, has long suggested that cancers arise at sites of chronic inflammation ( 1– 3). This hypothesis has recently received molecular confirmations in inflammation-associated cancer models ( 4– 6), which provide in vivo evidence supporting a causal relationship between nuclear factor-κB (NF-κB)–mediated inflammation and tumorigenesis. In this context, tumor-associated macrophages (TAM) represent the major inflammatory component of the tumor infiltrate and are considered prominent part of the inflammatory circuits that promote tumor progression ( 7). TAMs are recruited at the tumor site by tumor-derived chemotactic factor for monocytes, later identified as the chemokine CCL2/MCP-1 ( 7– 10). These cells have a dual function in their interaction with neoplastic cells, as they can express activities that can prevent the establishment and spread of tumor cells and, simultaneously, have functions to support tumor growth and dissemination ( 1, 3, 7, 11). This ambivalent relationship reflects the elevated functional plasticity of macrophages, able to express different functional programs in response to different microenvironmental signals, as exemplified in the M1-M2 paradigm of macrophage polarization ( 7, 12– 16). M1 macrophages are potent effector cells that kill microorganisms and tumor cells and produce copious amounts of proinflammatory cytokines. In contrast, M2 macrophages tune inflammatory responses and adaptive Th1 immunity, scavenge debris, and promote angiogenesis, tissue remodeling, and repair ( 15). The M1/M2 dichotomy of macrophage polarization is illustrated in TAMs, which can elicit both protumoral and antitumoral activities, as previously suggested in the “macrophage balance hypothesis” ( 11). During tumor progression, TAMs acquire a skewed M2 phenotype and express protumoral functions ( 7).

We have previously reported that TAMs from both murine and human tumors express defective interleukin-12 (IL-12) production and NF-κB activity, along with high level of the immunosuppressive cytokine IL-10 ( 17), and speculated that this molecular pathway may underlie other alterations of TAM functions dependent on NF-κB activation ( 7). The NF-κB family is a key player in controlling both innate and adaptive immunity ( 18). NF-κB proteins are present in the cytoplasm in association with inhibitory proteins that are known as inhibitors of NF-κB (IκBs). After activation by different stimuli, the IκB proteins become phosphorylated, ubiquitylated, and, subsequently, degraded by the proteasome ( 18). The degradation of IκB allows NF-κB proteins to translocate to the nucleus and bind their cognate DNA-binding sites to regulate the transcription of a large number of genes, including antimicrobial peptides, cytokines, chemokines, stress response proteins, and antiapoptotic proteins. The mammalian NF-κB family include RELA (p65), NF-κB1 (p50 and p105), NF-κB2 (p52 and p100), c-REL, and RELB ( 18). Unlike the c-REL, RELB, and RELA proteins, p50 and p52 do not contain the COOH-terminal transactivation domain and may form inhibitory homodimers able to bind NF-κB consensus sequences and to function as transcriptional repressors ( 18). Altered expression and activation of NF-κB family members in immunocompetent cells has been reported in various pathologic situations. Although its constitutive activation is associated with chronic inflammatory diseases ( 19), defective NF-κB activity has been reported in other pathologies, such as cancer and sepsis ( 19– 21).

Here, we provide evidence that TAMs are characterized by a status of tolerance in response to lipopolysaccharide (LPS) and other proinflammatory signals, assessed as defective expression of proinflammatory genes, IL-12 and tumor necrosis factor-α (TNF-α) in particular, and this state is dependent on nuclear overexpression of the p50 NF-κB inhibitory homodimer. We further show in p50 NF-κB–deficient mice, and in lethally irradiated mice transplanted with p50^−/− bone marrow, reversion of the anergic phenotype, delayed tumor progression and increased survival.

Materials and Methods

Cell culture. Peritoneal exudate cells (PEC) and TAMs were incubated in RPMI 1640 containing 10% FCS, 2 mmol/L glutamine, and 100 units/mL penicillin-streptomycin. For the different treatments (stated in the text), we used the following concentrations: IFN-γ, 100 units/mL; LPS, 100 ng/mL; CD40 (to induce activation, TAMs were cocultured for 48 hours with CD40L-transfected J558L cells at 5:1 ratio; ref. 22); TNF-α, 20 ng/mL; IL-1β, 20 ng/mL; IL-10, 20 ng/mL; prostaglandin E₂ (PGE₂), 10⁻⁵ mol/L. Murine recombinant IFN-γ was kindly donated by Dr. G. Garotta (Roche, Basilea, Switzerland). IL-1β was a gift of Dr. Boraschi (Dompè, L'Aquila, Italy); TNF was from BASF/Knoll (Ludwighafen, Germany); LPS (E. Coli 055:B5) was purchased from Difco (Detroit, MI); IL-10 from Schering-Plough (Kenilworth, NJ); transforming growth factor-β (TGF-β) was from Peprotech (London, United Kingdom); and PGE₂ was from Sigma (St. Louis, MO).

Preparation of PECs and TAMs. The study was reviewed and approved by the Istituto di Ricerche Farmacologiche Mario Negri Animal Care and Use Committee in compliance with the NIH Guide for the Care and Use of Laboratory Animals and EU directives and guidelines. p50-deficient mice were kindly donated by Drs. Michael Karin and Giuseppina Bonizzi (School of Medicine, University of California San Diego, La Jolla, CA). p50-deficient mice were backcrossed for 15 generations to C57/Bl6 background before being used in the experiments. Tumor inoculation and take was carry out as previously described ( 17). PECs and TAMs were obtained 3 weeks after tumor cells injection and isolated and identified as previously described ( 23). Experimental evaluation of animal survival had to be interrupted on day 50 for ethical reasons, according to the directives and guidelines for the use and welfare of laboratory animals.

Preparation of human monocyte-derived macrophages and TAMs from human ovarian cancer. Human monocytes were separated as previously described ( 24). Monocyte-derived macrophages were derived from freshly isolated monocytes as described ( 24). For the preparation of human TAM, solid tumors were collected from untreated patients with histologically confirmed epithelial ovarian carcinoma admitted to the Department of Obstetrics and Gynecology, San Gerardo Hospital (Monza, Italy). All patients had cancer classified as stage II, III, or IV. Purification of TAM from solid tumor and omentum metastasis was done as described previously for murine tumors.

Bone marrow transplantation. C57/BL6 mice were lethally irradiated with a total dose of 900 cGy. Two hours later, mice were injected in tail vein with 15 × 10⁶ of nucleated cells obtained from C57/BL6 p50^+/+ or p50^−/− donors as follows. Bone marrow–derived cells were obtained by flushing the cavity of freshly dissected femurs with saline. Recipient mice received 0.4 mg/mL gentalyn added to the drinking water starting 1 weeks before irradiation and maintained thereafter. At 8 weeks after bone marrow transplantation (BMT), 10⁴ tumor cells were injected i.m. in the left hind limb of chimeric mice. After 3 weeks of experimental period, to verify engraftment, DNA extracted from PECs of BMT mice was analyzed by PCR for p50 locus.

ELISA. Determination of murine IL-12 and IFN-γ in the cell supernatants was carried out by using ELISA kits purchased from R&D System, Inc. (Minneapolis, MN).

Real-time PCR. Cells were cultured in medium alone or supplemented with the indicated agents, and total RNA was purified and analyzed as described ( 23). Briefly, reverse transcription reaction from 1 μg RNA template was done using Taqman reverse transcription reagents (Applied Biosystems, Piscataway, NJ) as per manufacturer's instructions. Real-time PCR was done using SyBr Green PCR Master Mix (Applied Biosystems, Piscataway, NJ) and detected by ABI-Prism 5700 Sequence Detector (Applied Biosystems, Foster City, CA). Data were processed using the GeneAmp software (Applied Biosystems, Foster City, CA) and normalized by actin gene expression levels. All real-time results were expressed as fold changes in mRNA expression with respect to the control cells. All results were normalized to the expression of the housekeeping gene β-actin in the PCR reactions. Data represented are from three independent experiments done in triplicate.

Immunostaining for p50 NF-κB and laser confocal microscopy. PECs or TAMs were grown on sterile glass coverslips at a concentration of 2 × 10⁵ per coverslip in a final volume of 1 mL complete RPMI. Confocal microscopy analysis for the p50 and p65 NF-κB was carried out as previously described ( 23).

Immunohistochemical analysis of tumor infiltrating macrophages. Frozen sections from five wild-type (wt) and five knockout (KO) neoplastic mouse tissues were used for immunohistochemical evaluations. Immunohistochemistry was done using the following antibodies: F4/80 monoclonal antibody (clone C1:A3-1; dilution of 1:50), anti-CD4 (clone 6K1.5; undiluted), and anti-CD8 (clone TIB; dilution 1:5). The sections were rehydrated with PBS and then incubated with primary antibodies for 1 hour. The reactions were revealed by means of an indirect immunoperoxidase technique, with 3,3′-diaminobenzidine free base as chromogen. A quantitative evaluation of immunostaining was applied by using WinRec Program (Image Pro Plus, Media Cybernetics, Silver Spring, MD) and by counting immunopositive cells (identified by the presence of a brown staining) in 10 randomly selected high power fields for each case.

Western blot analysis. The nuclear and cytosolic extracts were analyzed by SDS/PAGE (10% acrylamide) as described ( 23). Immunoblotting was done with rabbit anti-p50 (#1141), anti-p65 (#1226) antisera ( 25), or by phospho-IκBα antibody (Santa Cruz Biotechnology, Santa Cruz, CA); MyD88 antibody (IMGENEX, San Diego, CA) or actin antibody (Santa Cruz Biotechnology).

Flow cytometric analysis. Macrophages were stained using monoclonal antibodies: R. phycoerythrin–conjugated anti-F4/80 (AbD Serotec, Oxford, United Kingdom), allophycocyanin-conjugated anti-TLR2 (eBioscience, San Diego, CA), Alexa Fluor 488 conjugated anti-TLR4 (eBioscience), and their relative isotype controls according to the manufacturer's instructions. Detectors were set to logarithmic amplification, and 10,000 events were analyzed using BD DIVA software. All samples were processed using BD FACS CANTO flow cytometer. Natural killer (NK) cells were stained by using monoclonal anti-NK-1.1 antibody (BD Biosciences, PharMingen, San Diego, CA).

Electrophoretic mobility shift assay. Nuclear proteins were prepared and analyzed as described ( 17). Oligonucleotides were purchased from Life Technologies (Milan, Italy). An IL-6κB double-strand oligonucleotide was used in electrophoretic mobility shift assay (EMSA) as previously reported ( 26). The following antisera were used in supershift analysis: #1157, raised against the human and murine p50 NF-κB ( 25); #1226, raised against the human and murine p65 ( 25); #1266, raised against the murine c-Rel ( 25).

Transient transfection. Transfections were done using Effectene Transfection Reagents (Qiagen, Milan, Italy). Plasmid containing the IL-12p40 promoter linked to the luciferase reporter gene was obtained from Dr. Stephen Male (Howard Hughes Medical Institute and Department of Microbiology, Immunology and Genetics, University of california, Los Angeles, CA). The cytomegalovirus (CMV) construct expressing the murine p50 NF-κB subunit was a kind gift of Dr. Howard Young (National Cancer Institute, Frederick, MD). The total amount of DNA transfected in the different groups was normalized with the empty CMV construct. Luciferase reporter assays were done in 96-well optiplates (Packard Instrument, Inc., Meriden, CT) using the luciferase assay system (Promega, Inc., Madison, WI) according to the manufacturer's instructions.

Results

Defective expression of inflammatory cytokines in TAMs. In agreement with the transcriptome profile of TAMs from a murine fibrosarcoma, which we recently reported ( 23), Fig. 1A shows the defective expression of a few representative inflammatory cytokine genes (IL-12p40, IL-12p35, TNF-α, and IL-6) and high levels of the immunosuppressive cytokine IL-10, in LPS-activated TAMs. We rule out possible defects in the expression level of the LPS receptors TLR2 and TLR4 ( 27, 28) by showing no significant differences in their surface expression level, in TAMs versus PECs ( Fig. 1B). MyD88, was first characterized as an essential component for the activation of innate immunity by all the TLRs ( 27). Based on this, we evaluated the level of MyD88 protein in TAMs versus PECs and observed no differences in their respective level of MyD88 protein expression ( Fig. 1C). Moreover, to evaluate the functional activity of these receptors, the LPS-dependent phosphorylation of the IκBα inhibitor of NF-κB, an event which precedes NF-κB nuclear translocation ( 18), was analyzed at different times ( Fig. 1C). IκBα phosphorylation in TAMs was delayed, thus correlating with the defective expression of NF-κB–dependent cytokine genes.

LPS stimulation of TLR4 promotes activation of an MyD88-dependent and an MyD88-independent pathway ( 29). The MyD88-dependent pathway controls the events leading to NF-κB activation and to inflammatory cytokine production. The MyD88-independent pathway activates the IFN-regulatory factor (IRF-3) and the late phase of NF-κB activation, leading to the production of IFN-inducible genes. As shown in Fig. 1D, IRF-3 nuclear translocation was observed in both PECs and TAMs, suggesting that TAMs are characterized by selective inhibition of the MyD88-dependent pathway, as also substantiated by the observation that LPS-activated TAMs express high levels of the IFN-inducible chemokine genes CCL2, CCL5, and CXCL10 ( Fig. 1E). These data are in agreement with our previous work showing that TAMs are characterized by enhanced IRF-3/signal transducers and activators of transcription 1 activation ( 23).

Nuclear overexpression of the p50 NF-κB inhibitory homodimer in TAMs. NF-κB is classically described as an heterodimer composed by the p50 and p65 subunits, which drives transcription of a number of inflammatory genes ( 18). Hence, we investigated the intracellular distribution and DNA-binding activity of NF-κB components in TAMs versus PECs. Strikingly, resting TAMs displayed a massive nuclear localization of the p50 NF-κB protein ( Fig. 2A ). As expected, resting PECs showed low nuclear expression of the p50 NF-κB protein, which was increased by LPS. In a series of four experiments, the ratio of nuclear p50 NF-κB in TAMs versus PECs ranged between 5- and 10-fold. Noteworthy, the level of expression of p50 mRNA were equivalent between PECs and TAMs (data not shown).

Next, we evaluated the kinetics of nuclear translocation of the p50 and p65 NF-κB proteins in LPS-activated TAMs and PECs, as indicated ( Fig. 2B). Western blot analysis showed a time-dependent increase of nuclear p50 NF-κB in PECs, up to 2 hours, whereas nuclear p50 was constitutively higher in TAMs. In contrast, the LPS-induced nuclear translocation of the p65 NF-κB protein was reduced in TAMs ( Fig. 2A and B) compared with PECs. This result mirrors the delayed LPS-dependent phosphorylation of IκBα observed in TAMs ( Fig. 1C).

To investigate whether the DNA-binding activity of the NF-κB complexes in TAMs and PECs reflected the intracellular distribution of the p50 and p65 NF-κB subunits, EMSA analysis was done by using nuclear extracts in the presence of a canonical ³²P-dCTP–labeled IL-6 NF-κB double-strand oligonucleotide ( Fig. 2C). We observed the presence of three DNA/protein complexes (a, b, and c). Although complex a was induced in the nucleus of TAMs and PECs, following LPS treatment for 2 hours, complex b was constitutively expressed in TAMs and was induced at lower level in PECs. Supershift analysis was done to identify the composition of complexes a and b ( Fig. 2C). An anti-p65 antiserum supershifted only complex a, whereas the antiserum against the p50 NF-κB subunit supershifted both complexes a and b, thus proving that complex a is formed by the p50/p65 heterodimer and complex b by the p50 homodimer. These results confirmed a higher DNA-binding activity of the inhibitory NF-κB p50 homodimer in TAMs compared with PECs. An antiserum against the cRel transcription factor caused only a slight supershift. Complex c, which was equally expressed in PECs and TAMs, was not affected by the presence of the antisera and was not further analyzed.

Activation of NF-κB was also studied in PECs and TAMs by laser confocal microscopy (LCM). Consistent with the biochemical studies, the nuclei of untreated TAMs seemed to contain high constitutive level of p50, whereas in response to LPS, TAMs showed negligible p65 translocation ( Fig. 2D). Cotransfection of increasing amounts of a p50 NF-κB expression vector with a plasmid containing the IL-12p40 promoter region −350 to +55, encompassing the identified IL-12p40 NF-κB-binding site ( 30), resulted in a dose-dependent inhibition of the IL-12p40 promoter activity ( Fig. 2E), in the monocytic cell line RAW 267.4.

Taken together, these results indicate that high expression of the p50 NF-κB subunit in TAMs acts as inhibitory mechanism of gene transcription controlled by NF-κB. To establish whether high p50 NF-κB expression in TAMs could affect other NF-κB–inducing signals ( 18), we determined the expression of TNF-α mRNA in response to either IFN-γ/LPS, IFN-γ/CD40, TNF-α, or IL-1β. As shown, inducible TNF-α mRNA levels were consistently lower in TAMs compared with PECs ( Fig. 2F).

To confirm the relevance of this observation in human tumors, TAMs from ovarian carcinoma patients were assayed by LCM for the nuclear translocation of the p50 and p65 NF-κB subunits ( Fig. 2G). Monocyte-derived macrophages were used as control cell population. Although in monocyte-derived macrophages high levels of nuclear p50 were observed only after activation with LPS, TAMs from both ovarian carcinoma and omentum metastasis displayed constitutive high levels. Higher nuclear levels of p65 NF-κB were induced by LPS in monocyte-derived macrophages compared with TAMs from both the primary tumor and the omentum metastasis. These results are in line with our previous observations on defective production of IL-12p70 by TAMs from human ovarian cancer ( 17) and were confirmed in three of three ovarian carcinoma patients.

When expressed a high levels, both the RelB and the p52 NF-κB members may act as repressors of the NF-κB transcriptional activity ( 18). We observed higher nuclear levels of both the RelB and p52 proteins in TAMs compared with PECs (data not shown), suggesting that additional inhibitory mechanisms may take place in TAMs.

Restoration of TAM responsiveness in p50^−/− mice. To establish whether impaired expression of inflammatory cytokine genes was reversible, TAMs were cultured for different periods under standard conditions and analyzed for cytokine gene expression. After 24 to 72 hours, TAMs recovered their capability to express high levels of IL-12p40 mRNA ( Fig. 3A ) and to secrete the IL-12p70 heterodimer, in response to IFN-γ/LPS ( Fig. 3B). In addition, IL-10 mRNA expression partially decreased during this period ( Fig. 3A). TAMs also recovered the capability to express higher TNF-α mRNA levels ( Fig. 3C). Strikingly, resting TAMs cultured for 24 or 72 hours displayed reduced levels of nuclear p50 NF-κB, comparable with those observed in PECs, and recovered the capability to enhance nuclear p50 NF-κB in response to LPS stimulation ( Fig. 3D).

p50 mRNA levels were next analyzed in the attempt to investigate the molecular mechanisms guiding nuclear overexpression of p50. Noteworthy, the level of expression of p50 mRNA were equivalent between PECs and TAMs (data not shown). To investigate whether signals present at the tumor microenvironment ( 15), such as IL-10, TGF-β, PGE₂, and TNF-α, could promote DNA-binding activity of the p50 homodimer, nuclear extracts from PEC activated or not, for 36 hours, were analyzed by EMSA ( Fig. 3E). Although TNF-α induced strong formation of the p50/p65 complex and only a minor activity of the p50 NF-κB homodimer, IL-10, TGF-β, and PGE₂ elicited a preferential induction of the p50 NF-κB homodimer.

To definitively assess the role of p50 NF-κB in the LPS-dependent cytokine expression by macrophages, PECs and TAMs obtained from both wt and p50^−/− mice were activated for 4 hours and analyzed for the level of IL-12p40 and TNF-α, IL-10, CCL2, and CXCL10 mRNAs. As shown in Fig. 3F, p50^−/− TAMs expressed higher levels of IL-12p40 and TNF-α mRNAs in response to LPS, comparable with wt PECs. In contrast, lack of p50 NF-κB in TAMs resulted in an appreciable decrease of the LPS-inducible levels of IL-10 mRNA. These observations were confirmed by ELISA (data not shown). Furthermore, LPS-activated p50^−/− TAMs expressed higher levels of the IFN-inducible gene CXCL10 ( 29).

As the production of inflammatory cytokines by mononuclear phagocytes is essential to promote development of the adaptive immunity and expression of Th1 cytokines by CD4⁺ T helper lymphocytes ( 31), splenocytes from wt and p50^−/− tumor-bearing mice were tested for the secretion of IFN-γ. As shown in Fig. 3G, phorbol 12-myristate 13-acetate/Concanavalin–activated splenocytes from wt tumor bearers (tum) secreted lower levels of IFN-γ (4 ng/mL) compared with splenocytes from healthy mice (N; 17 ng/mL). Noteworthy, splenocytes from tumor-bearing p50^−/− mice significantly recovered their capability to secrete IFN-γ (10.5 ng/mL). To evaluate whether possible changes in the number of IFN-γ-producing NK cells could account for the observed recovery of IFN-γ secretion by splenocytes from p50 KO mice, we determined the number of NK cells in the spleen from wt and p50 KO tumor-bearing mice. As shown ( Fig. 3H), cytofluorimetric analysis showed no differences in the number of NK cells.

Tumor growth retardation in p50^−/− mice. Higher expression of inflammatory mediators by p50^−/− TAMs suggested possible differences in the level of recruitment of infiltrating leukocytes. Immunohistochemical analysis of serial areas of the MN/MCA1 fibrosarcoma showed similar levels of infiltrating macrophages (F4/80⁺ cells) in wt versus p50^−/− mice ( Fig. 4A ). In contrast, a significant up-regulation of the numbers of CD4⁺ and CD8⁺ T lymphocytes was observed in tumors grown in p50^−/− mice. This observation is in line with the higher expression of the T cell–specific chemokine CXCL10 observed in p50^−/− TAMs ( Fig. 3F).

To establish whether p50 NF-κB–mediated tolerance in TAMs could play a role in the control of tumor growth, C57B1/6 wt and p50^−/− mice were implanted with the MN/MCA-1 fibrosarcoma. Throughout the entire experimental period (3 weeks) tumor growth in p50^−/− mice was significantly reduced ( Fig. 4B). In addition, we monitored the growth rate of the B16 melanoma as we previously showed that TAMs from this tumor share defective IL-12 and enhanced IL-10 phenotype ( 17). In agreement, a significant reduction of B16 tumor growth was observed in p50^−/− mice ( Fig. 4C). Accordingly, the survival of p50^−/− mice was prolonged with 6 of 10 p50^−/− mice (60%) alive on day 50, when all wt controls had died with a median survival of 42 days (P < 0.01). Finally, to test whether tumor growth retardation observed in p50-deficient mice could be ascribed to the absence of p50 in hematopoietic cells, bone marrow from either wt or p50^−/− mice was transplanted in lethally irradiated C57/Bl6 wt mice. As shown in Fig. 4D, mice transplanted with p50^−/− bone marrow displayed reduced tumor growth, similar to what observed in p50^−/− mice.

Discussion

The presence of inflammatory cells in solid tumors has long suggested that cancers arise at sites of chronic inflammation ( 1), and recent molecular confirmations in inflammation-associated cancer models ( 4– 6) support a causal relationship between NF-κB–mediated inflammation and tumorigenesis. Here, we confirm and extend previous observations ( 17) that TAMs have defective activation of NF-κB in response to different agonists (LPS, CD40L, TNF-α, and IL-1β). The defect of TAMs was selective, in that the activation by LPS of MyD88-independent IRF-3 was normal, and TAMs produced copious amounts of IL-10 and chemokines downstream of this pathway. Defective M1 activation of TAMs was associated with elevated nuclear localization of p50 homodimers. Finally, TAMs isolated from tumors transplanted in p50^−/− deficient mice showed essentially normal M1 activation, associated with significant tumor growth retardation. Bone marrow chimeras reconstituted with p50^−/− hematopoietic cells showed reduced tumor growth, suggesting that indeed mononuclear phagocyte activity is a major player in growth retardation. These results indicate that p50 plays a pivotal role in tuning NF-κB–dependent M1 activation of TAMs and in the subversion of macrophage responsiveness in established neoplasia.

Increased p50 NF-κB expression is functionally associated with the LPS-induced tolerance of macrophages ( 32), and peritoneal macrophages from p50^−/− mice have been shown to be incapable of undergoing endotoxin tolerance-mediated suppression of TNF-α production ( 33). We observed that TAMs from p50^−/− tumor-bearing mice displayed normal expression of IL-12 and TNF-α, an event paralleled by decreased tumor growth and increased animal survival. Although during bacterial infection, or sepsis, up-regulation of the p50 NF-κB homodimer is an adaptive response providing protection from pathologic hyperactivation of the innate immune system ( 33), our results suggest that cancer may exploit this mechanism to impair type 1 inflammation and effective activation of adaptive Th1 immunity. In support of this and in agreement with previous reports ( 31, 33, 34), we observed that overexpression of the p50 NF-κB subunit in the monocytic cell line RAW 267.4 results in the inhibition of the IL-12p40 promoter transcriptional activity. Moreover, TAMs cultured in standard conditions for a period of 24 to 72 hours reduced their level of p50 NF-κB protein to physiologic levels and recovered their capability to express IL-12p70 and TNF-α. Thus, the unique phenotype of TAMs is induced by the tumor microenvironment and overexpression of p50 NF-κB in these cells, and possibly in other components of the immune system, drives inhibition of immune functions in cancer bearers. Constitutive nuclear overexpression of the p50 NF-κB was also observed in TAMs from human ovarian cancer, that are characterized by defective IL-12p70 production ( 17). NF-κB seems a main target of tumor-mediated immune dysfunction observed in cancer bearers. It was reported that anergic T cells from cancer bearers display defective NF-κB activation ( 35) and Th1 response ( 36), and that in vivo anergized T cells express a predominant formation of the p50 NF-κB inhibitory homodimer, likely responsible for repression of cytokine expression ( 37).

The mechanisms controlling p50 biogenesis have been elucidated in part. Lin et al. have shown that cotranslational folding of the NH₂-terminal portion of p105 protects this region that forms p50 from degradation by proteasome ( 38). In addition, the p50 NF-κB activity is also controlled by Bcl3, a close homologue of IκB proteins ( 39), which displays versatile functions, including cytoplasmic activation of p50 homodimers and their nuclear translocation and modulation of the NF-κB activity in the nucleus ( 39– 42). Interestingly, it was observed that IL-10 selectively induces nuclear translocation and DNA binding of p50/p50 homodimers in human monocytic cells ( 43). This well correlates with our previous observation of high production of IL-10 by TAMs of the MN/MCA1 fibrosarcoma, which we reported in vitro ( 17) and in vivo ( 23). Moreover, our data also show that in addition to IL-10, PGE₂ and TGF-β, which are expressed by TAMs ( 7) and promote M2-type polarized inflammation ( 15), are microenvironmental signals promoting p50 NF-κB homodimer activity. Of interest, PGE₂ was shown to mimic the phenotype of TAMs, being able to enhance p50 and inhibit p65 translocation and DNA-binding activity ( 44). Of interest, peritoneal macrophages from tumor-bearing mice were impaired in the p65 NF-κB nuclear translocation ( 45).

We observed that TAMs from p50^−/− mice express higher level of the T-cell recruiting and IFN-inducible chemokine CXCL10 ( 22, 29, 46, 47), which correlates with increased infiltration of CD4⁺ and CD8⁺ T lymphocytes in tumors from p50^−/− mice. This observation is in line with the established antitumor role of these cells as well as with inhibition of tumor growth and increased survival of p50^−/− mice.

Overall, our data indicate that the tumor microenvironment tolerates macrophages by inducing high nuclear localization of the p50 NF-κB inhibitory homodimer. This observation is consistent with previous results in mouse and human tumors ( 7, 17), whereas it seems in contrast with recent results showing a key role of the NF-κB pathway in liver and colon carcinogenesis ( 4, 5). In particular, lineage-restricted inactivation of IKKβ revealed a nonreduntant role of NF-κB activation in myeloid cells in colon carcinogenesis ( 5), an observation generally consistent with a protumor function of inflammatory reactions and TAMs in particular ( 1– 3). We speculate that the apparent discrepancy between results reported here of detective p50 homodimer-driven NF-κB activation in TAMs and these recent observations ( 4, 5) reflects differential involvement of NF-κB–driven full-blown inflammation in early steps of carcinogenesis compared with established neoplasia. Although NF-κB activation in myeloid cells would facilitate the early steps of carcinogenesis, tumor-mediated increased expression of p50 NF-κB in TAMs would guide smoldering inflammation and progression of established neoplasia. Inhibition of p50 NF-κB expression in TAMs may represent an innovative anticancer strategy, complementary to immunostimulatory strategies.